maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/lldb/source/Symbol/ClangASTContext.cpp

6326 lines
246 KiB
C++
Raw Normal View History

Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
//===-- ClangASTContext.cpp -------------------------------------*- C++ -*-===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
Add missing stdio.h include; fix include path for ClangASTContext.h. llvm-svn: 105921
2010-06-14 03:06:42 +08:00
#include "lldb/Symbol/ClangASTContext.h"
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
// C Includes
// C++ Includes
#include <string>
// Other libraries and framework includes
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
// Clang headers like to use NDEBUG inside of them to enable/disable debug
// releated features using "#ifndef NDEBUG" preprocessor blocks to do one thing
// or another. This is bad because it means that if clang was built in release
// mode, it assumes that you are building in release mode which is not always
// the case. You can end up with functions that are defined as empty in header
// files when NDEBUG is not defined, and this can cause link errors with the
// clang .a files that you have since you might be missing functions in the .a
// file. So we have to define NDEBUG when including clang headers to avoid any
// mismatches. This is covered by rdar://problem/8691220
Added a new #define, LLVM_NDEBUG_OFF, that should be set if linking against an LLVM compiled with NDEBUG off. If it is set, we do not enable NDEBUG in any place where we include LLVM headers. llvm-svn: 143036
2011-10-27 01:46:51 +08:00
#if !defined(NDEBUG) && !defined(LLVM_NDEBUG_OFF)
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
#define LLDB_DEFINED_NDEBUG_FOR_CLANG
Moved NDEBUG from a global setting to a specific hack. llvm-svn: 107894
2010-07-09 02:16:16 +08:00
#define NDEBUG
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
// Need to include assert.h so it is as clang would expect it to be (disabled)
#include <assert.h>
#endif
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
#include "clang/AST/ASTContext.h"
#include "clang/AST/ASTImporter.h"
#include "clang/AST/CXXInheritance.h"
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
#include "clang/AST/DeclObjC.h"
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
#include "clang/AST/DeclTemplate.h"
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
#include "clang/AST/RecordLayout.h"
#include "clang/AST/Type.h"
#include "clang/Basic/Builtins.h"
I left some stray debugging messages in the source code. Removing these. llvm-svn: 149903
2012-02-07 05:28:03 +08:00
#include "clang/Basic/Diagnostic.h"
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
#include "clang/Basic/FileManager.h"
Updated to the LLVM/Clang of 2010-11-17 at 3:30pm. llvm-svn: 119677
2010-11-18 10:56:27 +08:00
#include "clang/Basic/FileSystemOptions.h"
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
#include "clang/Basic/SourceManager.h"
#include "clang/Basic/TargetInfo.h"
#include "clang/Basic/TargetOptions.h"
#include "clang/Frontend/FrontendOptions.h"
#include "clang/Frontend/LangStandard.h"
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
#ifdef LLDB_DEFINED_NDEBUG_FOR_CLANG
Moved NDEBUG from a global setting to a specific hack. llvm-svn: 107894
2010-07-09 02:16:16 +08:00
#undef NDEBUG
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
#undef LLDB_DEFINED_NDEBUG_FOR_CLANG
// Need to re-include assert.h so it is as _we_ would expect it to be (enabled)
#include <assert.h>
#endif
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
Made lldb_private::ArchSpec contain much more than just an architecture. It now, in addition to cpu type/subtype and architecture flavor, contains: - byte order (big endian, little endian) - address size in bytes - llvm::Triple for true target triple support and for more powerful plug-in selection. llvm-svn: 125602
2011-02-16 05:59:32 +08:00
#include "lldb/Core/ArchSpec.h"
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
#include "lldb/Core/dwarf.h"
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
#include "lldb/Core/Flags.h"
Logging fixes. Added a simple class to handle logging for type importing, making errors much easier to diagnose. Also removed some Objective-C related logging. llvm-svn: 117568
2010-10-29 02:19:36 +08:00
#include "lldb/Core/Log.h"
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
#include "lldb/Core/RegularExpression.h"
#include "lldb/Expression/ASTDumper.h"
Added ClangExternalASTSourceCommon, a local superclass for all our external AST sources that lets us associate arbitrary flags with the types we put into the AST contexts. Also added an API on ClangASTContext that allows access to these flags given only an ASTContext and a type. Because we don't have access to RTTI, and because at some point in the future we might encounter external AST sources that we didn't make (so they don't subclass ClangExternalASTSourceCommon) I added a magic number that we check before doing anything else, so that we can catch that problem as soon as it appears. llvm-svn: 145748
2011-12-03 11:15:28 +08:00
#include "lldb/Symbol/ClangExternalASTSourceCommon.h"
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
#include "lldb/Symbol/VerifyDecl.h"
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
#include "lldb/Target/ExecutionContext.h"
#include "lldb/Target/Process.h"
#include "lldb/Target/ObjCLanguageRuntime.h"
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
Add missing stdio.h include; fix include path for ClangASTContext.h. llvm-svn: 105921
2010-06-14 03:06:42 +08:00
#include <stdio.h>
More missing files from my previous checkin. llvm-svn: 110299
2010-08-05 09:57:25 +08:00
using namespace lldb;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
using namespace lldb_private;
using namespace llvm;
using namespace clang;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
static bool
Adding a new API call IsTypeComplete() to SBType. This call is meant to check if the type has been previously completed or not (which is mostly interesting from a performance point of view) Adding a test case that checks that we do not complete types before due time. This should help us track cases similar to the cascading data formatters. llvm-svn: 153363
2012-03-24 09:11:14 +08:00
GetCompleteQualType (clang::ASTContext *ast, clang::QualType qual_type, bool allow_completion = true)
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
{
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
case clang::Type::ConstantArray:
{
const clang::ArrayType *array_type = dyn_cast<clang::ArrayType>(qual_type.getTypePtr());
if (array_type)
Adding a new API call IsTypeComplete() to SBType. This call is meant to check if the type has been previously completed or not (which is mostly interesting from a performance point of view) Adding a test case that checks that we do not complete types before due time. This should help us track cases similar to the cascading data formatters. llvm-svn: 153363
2012-03-24 09:11:14 +08:00
return GetCompleteQualType (ast, array_type->getElementType(), allow_completion);
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
}
break;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
case clang::Type::Record:
case clang::Type::Enum:
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::TagType *tag_type = dyn_cast<clang::TagType>(qual_type.getTypePtr());
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (tag_type)
{
clang::TagDecl *tag_decl = tag_type->getDecl();
if (tag_decl)
{
<rdar://problem/11082392> Fixed an issue that could cause circular type parsing that will assert and kill LLDB. Prior to this fix the DWARF parser would always create class types and not start their definitions (for both C++ and ObjC classes) until we were asked to complete the class later. When we had cases like: class A { class B { }; }; We would alway try to complete A before specifying "A" as the decl context for B. Turns out we can just start the definition and still not complete the class since we can check the TagDecl::isCompleteDefinition() function. This only works for C++ types. This means we will not be pulling in the full definition of parent classes all the time and should help with our memory consumption and also reduce the amount of debug info we have to parse. I also reduced redundant code that was checking in a lldb::clang_type_t was a possible C++ dynamic type since it was still completing the type, just to see if it was dynamic. This was fixed in another function that was checking for a type being dynamic as an ObjC or a C++ type, but there was dedicated fucntion for C++ that we missed. llvm-svn: 153713
2012-03-30 08:51:13 +08:00
if (tag_decl->isCompleteDefinition())
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return true;
Adding a new API call IsTypeComplete() to SBType. This call is meant to check if the type has been previously completed or not (which is mostly interesting from a performance point of view) Adding a test case that checks that we do not complete types before due time. This should help us track cases similar to the cascading data formatters. llvm-svn: 153363
2012-03-24 09:11:14 +08:00
if (!allow_completion)
return false;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (tag_decl->hasExternalLexicalStorage())
{
lldb-59. llvm-svn: 132304
2011-05-30 08:49:24 +08:00
if (ast)
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
{
lldb-59. llvm-svn: 132304
2011-05-30 08:49:24 +08:00
ExternalASTSource *external_ast_source = ast->getExternalSource();
if (external_ast_source)
{
external_ast_source->CompleteType(tag_decl);
return !tag_type->isIncompleteType();
}
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
}
}
return false;
}
}
}
break;
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::ObjCObjectType *objc_class_type = dyn_cast<clang::ObjCObjectType>(qual_type);
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (objc_class_type)
{
clang::ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
// We currently can't complete objective C types through the newly added ASTContext
// because it only supports TagDecl objects right now...
System-wide summaries: - Summaries for char*, const char* and char[] are loaded at startup as system-wide summaries. This means you cannot delete them unless you use the -a option to type summary delete/clear - You can add your own system-wide summaries by using the -w option to type summary add Several code improvements for the Python summaries feature llvm-svn: 135326
2011-07-16 07:30:15 +08:00
if (class_interface_decl)
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
{
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
if (class_interface_decl->getDefinition())
return true;
Adding a new API call IsTypeComplete() to SBType. This call is meant to check if the type has been previously completed or not (which is mostly interesting from a performance point of view) Adding a test case that checks that we do not complete types before due time. This should help us track cases similar to the cascading data formatters. llvm-svn: 153363
2012-03-24 09:11:14 +08:00
if (!allow_completion)
return false;
Remove unused code. llvm-svn: 153786
2012-03-31 07:48:28 +08:00
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
if (class_interface_decl->hasExternalLexicalStorage())
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
{
several improvements to "type summary": - type names can now be regular expressions (exact matching is done first, and is faster) - integral (and floating) types can be printed as bitfields, i.e. ${var[low-high]} will extract bits low thru high of the value and print them - array subscripts are supported, both for arrays and for pointers. the syntax is ${*var[low-high]}, or ${*var[]} to print the whole array (the latter only works for statically sized arrays) - summary is now printed by default when a summary string references a variable. if that variable's type has no summary, value is printed instead. to force value, you can use %V as a format specifier - basic support for ObjectiveC: - ObjectiveC inheritance chains are now walked through - %@ can be specified as a summary format, to print the ObjectiveC runtime description for an object - some bug fixes llvm-svn: 134293
2011-07-02 08:25:22 +08:00
if (ast)
lldb-59. llvm-svn: 132304
2011-05-30 08:49:24 +08:00
{
several improvements to "type summary": - type names can now be regular expressions (exact matching is done first, and is faster) - integral (and floating) types can be printed as bitfields, i.e. ${var[low-high]} will extract bits low thru high of the value and print them - array subscripts are supported, both for arrays and for pointers. the syntax is ${*var[low-high]}, or ${*var[]} to print the whole array (the latter only works for statically sized arrays) - summary is now printed by default when a summary string references a variable. if that variable's type has no summary, value is printed instead. to force value, you can use %V as a format specifier - basic support for ObjectiveC: - ObjectiveC inheritance chains are now walked through - %@ can be specified as a summary format, to print the ObjectiveC runtime description for an object - some bug fixes llvm-svn: 134293
2011-07-02 08:25:22 +08:00
ExternalASTSource *external_ast_source = ast->getExternalSource();
if (external_ast_source)
{
external_ast_source->CompleteType (class_interface_decl);
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
return !objc_class_type->isIncompleteType();
several improvements to "type summary": - type names can now be regular expressions (exact matching is done first, and is faster) - integral (and floating) types can be printed as bitfields, i.e. ${var[low-high]} will extract bits low thru high of the value and print them - array subscripts are supported, both for arrays and for pointers. the syntax is ${*var[low-high]}, or ${*var[]} to print the whole array (the latter only works for statically sized arrays) - summary is now printed by default when a summary string references a variable. if that variable's type has no summary, value is printed instead. to force value, you can use %V as a format specifier - basic support for ObjectiveC: - ObjectiveC inheritance chains are now walked through - %@ can be specified as a summary format, to print the ObjectiveC runtime description for an object - some bug fixes llvm-svn: 134293
2011-07-02 08:25:22 +08:00
}
lldb-59. llvm-svn: 132304
2011-05-30 08:49:24 +08:00
}
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
}
several improvements to "type summary": - type names can now be regular expressions (exact matching is done first, and is faster) - integral (and floating) types can be printed as bitfields, i.e. ${var[low-high]} will extract bits low thru high of the value and print them - array subscripts are supported, both for arrays and for pointers. the syntax is ${*var[low-high]}, or ${*var[]} to print the whole array (the latter only works for statically sized arrays) - summary is now printed by default when a summary string references a variable. if that variable's type has no summary, value is printed instead. to force value, you can use %V as a format specifier - basic support for ObjectiveC: - ObjectiveC inheritance chains are now walked through - %@ can be specified as a summary format, to print the ObjectiveC runtime description for an object - some bug fixes llvm-svn: 134293
2011-07-02 08:25:22 +08:00
return false;
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
}
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
}
}
break;
case clang::Type::Typedef:
Adding a new API call IsTypeComplete() to SBType. This call is meant to check if the type has been previously completed or not (which is mostly interesting from a performance point of view) Adding a test case that checks that we do not complete types before due time. This should help us track cases similar to the cascading data formatters. llvm-svn: 153363
2012-03-24 09:11:14 +08:00
return GetCompleteQualType (ast, cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType(), allow_completion);
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated:
Adding a new API call IsTypeComplete() to SBType. This call is meant to check if the type has been previously completed or not (which is mostly interesting from a performance point of view) Adding a test case that checks that we do not complete types before due time. This should help us track cases similar to the cascading data formatters. llvm-svn: 153363
2012-03-24 09:11:14 +08:00
return GetCompleteQualType (ast, cast<ElaboratedType>(qual_type)->getNamedType(), allow_completion);
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
default:
break;
}
return true;
}
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
static AccessSpecifier
More missing files from my previous checkin. llvm-svn: 110299
2010-08-05 09:57:25 +08:00
ConvertAccessTypeToAccessSpecifier (AccessType access)
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
{
switch (access)
{
More missing files from my previous checkin. llvm-svn: 110299
2010-08-05 09:57:25 +08:00
default: break;
case eAccessNone: return AS_none;
case eAccessPublic: return AS_public;
case eAccessPrivate: return AS_private;
case eAccessProtected: return AS_protected;
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
}
return AS_none;
}
static ObjCIvarDecl::AccessControl
More missing files from my previous checkin. llvm-svn: 110299
2010-08-05 09:57:25 +08:00
ConvertAccessTypeToObjCIvarAccessControl (AccessType access)
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
{
switch (access)
{
More missing files from my previous checkin. llvm-svn: 110299
2010-08-05 09:57:25 +08:00
default: break;
case eAccessNone: return ObjCIvarDecl::None;
case eAccessPublic: return ObjCIvarDecl::Public;
case eAccessPrivate: return ObjCIvarDecl::Private;
case eAccessProtected: return ObjCIvarDecl::Protected;
case eAccessPackage: return ObjCIvarDecl::Package;
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
}
return ObjCIvarDecl::None;
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
static void
ParseLangArgs
(
LangOptions &Opts,
Updated LLVM/Clang to revision from 2010-06-13T06:00. llvm-svn: 105918
2010-06-14 01:34:29 +08:00
InputKind IK
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
)
{
// FIXME: Cleanup per-file based stuff.
// Set some properties which depend soley on the input kind; it would be nice
// to move these to the language standard, and have the driver resolve the
// input kind + language standard.
Updated LLVM/Clang to revision from 2010-06-13T06:00. llvm-svn: 105918
2010-06-14 01:34:29 +08:00
if (IK == IK_Asm) {
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
Opts.AsmPreprocessor = 1;
Updated LLVM/Clang to revision from 2010-06-13T06:00. llvm-svn: 105918
2010-06-14 01:34:29 +08:00
} else if (IK == IK_ObjC ||
IK == IK_ObjCXX ||
IK == IK_PreprocessedObjC ||
IK == IK_PreprocessedObjCXX) {
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
Opts.ObjC1 = Opts.ObjC2 = 1;
}
LangStandard::Kind LangStd = LangStandard::lang_unspecified;
if (LangStd == LangStandard::lang_unspecified) {
// Based on the base language, pick one.
switch (IK) {
Updated LLVM/Clang to revision from 2010-06-13T06:00. llvm-svn: 105918
2010-06-14 01:34:29 +08:00
case IK_None:
case IK_AST:
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
case IK_LLVM_IR:
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
assert (!"Invalid input kind!");
Updated LLVM/Clang to revision from 2010-06-13T06:00. llvm-svn: 105918
2010-06-14 01:34:29 +08:00
case IK_OpenCL:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
LangStd = LangStandard::lang_opencl;
break;
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
case IK_CUDA:
LangStd = LangStandard::lang_cuda;
break;
Updated LLVM/Clang to revision from 2010-06-13T06:00. llvm-svn: 105918
2010-06-14 01:34:29 +08:00
case IK_Asm:
case IK_C:
case IK_PreprocessedC:
case IK_ObjC:
case IK_PreprocessedObjC:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
LangStd = LangStandard::lang_gnu99;
break;
Updated LLVM/Clang to revision from 2010-06-13T06:00. llvm-svn: 105918
2010-06-14 01:34:29 +08:00
case IK_CXX:
case IK_PreprocessedCXX:
case IK_ObjCXX:
case IK_PreprocessedObjCXX:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
LangStd = LangStandard::lang_gnucxx98;
break;
}
}
const LangStandard &Std = LangStandard::getLangStandardForKind(LangStd);
Opts.BCPLComment = Std.hasBCPLComments();
Opts.C99 = Std.isC99();
Opts.CPlusPlus = Std.isCPlusPlus();
Opts.CPlusPlus0x = Std.isCPlusPlus0x();
Opts.Digraphs = Std.hasDigraphs();
Opts.GNUMode = Std.isGNUMode();
Opts.GNUInline = !Std.isC99();
Opts.HexFloats = Std.hasHexFloats();
Opts.ImplicitInt = Std.hasImplicitInt();
// OpenCL has some additional defaults.
if (LangStd == LangStandard::lang_opencl) {
Opts.OpenCL = 1;
Opts.AltiVec = 1;
Opts.CXXOperatorNames = 1;
Opts.LaxVectorConversions = 1;
}
// OpenCL and C++ both have bool, true, false keywords.
Opts.Bool = Opts.OpenCL || Opts.CPlusPlus;
// if (Opts.CPlusPlus)
// Opts.CXXOperatorNames = !Args.hasArg(OPT_fno_operator_names);
//
// if (Args.hasArg(OPT_fobjc_gc_only))
// Opts.setGCMode(LangOptions::GCOnly);
// else if (Args.hasArg(OPT_fobjc_gc))
// Opts.setGCMode(LangOptions::HybridGC);
//
// if (Args.hasArg(OPT_print_ivar_layout))
// Opts.ObjCGCBitmapPrint = 1;
//
// if (Args.hasArg(OPT_faltivec))
// Opts.AltiVec = 1;
//
// if (Args.hasArg(OPT_pthread))
// Opts.POSIXThreads = 1;
//
// llvm::StringRef Vis = getLastArgValue(Args, OPT_fvisibility,
// "default");
// if (Vis == "default")
Updated LLVM to latest version as of 10/28 at 7pm, and made minor integration fixes. llvm-svn: 117680
2010-10-30 02:38:40 +08:00
Opts.setVisibilityMode(DefaultVisibility);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
// else if (Vis == "hidden")
// Opts.setVisibilityMode(LangOptions::Hidden);
// else if (Vis == "protected")
// Opts.setVisibilityMode(LangOptions::Protected);
// else
// Diags.Report(diag::err_drv_invalid_value)
// << Args.getLastArg(OPT_fvisibility)->getAsString(Args) << Vis;
// Opts.OverflowChecking = Args.hasArg(OPT_ftrapv);
// Mimicing gcc's behavior, trigraphs are only enabled if -trigraphs
// is specified, or -std is set to a conforming mode.
Opts.Trigraphs = !Opts.GNUMode;
// if (Args.hasArg(OPT_trigraphs))
// Opts.Trigraphs = 1;
//
// Opts.DollarIdents = Args.hasFlag(OPT_fdollars_in_identifiers,
// OPT_fno_dollars_in_identifiers,
// !Opts.AsmPreprocessor);
// Opts.PascalStrings = Args.hasArg(OPT_fpascal_strings);
// Opts.Microsoft = Args.hasArg(OPT_fms_extensions);
// Opts.WritableStrings = Args.hasArg(OPT_fwritable_strings);
// if (Args.hasArg(OPT_fno_lax_vector_conversions))
// Opts.LaxVectorConversions = 0;
// Opts.Exceptions = Args.hasArg(OPT_fexceptions);
// Opts.RTTI = !Args.hasArg(OPT_fno_rtti);
// Opts.Blocks = Args.hasArg(OPT_fblocks);
// Opts.CharIsSigned = !Args.hasArg(OPT_fno_signed_char);
// Opts.ShortWChar = Args.hasArg(OPT_fshort_wchar);
// Opts.Freestanding = Args.hasArg(OPT_ffreestanding);
// Opts.NoBuiltin = Args.hasArg(OPT_fno_builtin) || Opts.Freestanding;
// Opts.AssumeSaneOperatorNew = !Args.hasArg(OPT_fno_assume_sane_operator_new);
// Opts.HeinousExtensions = Args.hasArg(OPT_fheinous_gnu_extensions);
// Opts.AccessControl = Args.hasArg(OPT_faccess_control);
// Opts.ElideConstructors = !Args.hasArg(OPT_fno_elide_constructors);
// Opts.MathErrno = !Args.hasArg(OPT_fno_math_errno);
// Opts.InstantiationDepth = getLastArgIntValue(Args, OPT_ftemplate_depth, 99,
// Diags);
// Opts.NeXTRuntime = !Args.hasArg(OPT_fgnu_runtime);
// Opts.ObjCConstantStringClass = getLastArgValue(Args,
// OPT_fconstant_string_class);
// Opts.ObjCNonFragileABI = Args.hasArg(OPT_fobjc_nonfragile_abi);
// Opts.CatchUndefined = Args.hasArg(OPT_fcatch_undefined_behavior);
// Opts.EmitAllDecls = Args.hasArg(OPT_femit_all_decls);
// Opts.PICLevel = getLastArgIntValue(Args, OPT_pic_level, 0, Diags);
// Opts.Static = Args.hasArg(OPT_static_define);
Opts.OptimizeSize = 0;
// FIXME: Eliminate this dependency.
// unsigned Opt =
// Args.hasArg(OPT_Os) ? 2 : getLastArgIntValue(Args, OPT_O, 0, Diags);
// Opts.Optimize = Opt != 0;
unsigned Opt = 0;
// This is the __NO_INLINE__ define, which just depends on things like the
// optimization level and -fno-inline, not actually whether the backend has
// inlining enabled.
//
// FIXME: This is affected by other options (-fno-inline).
Possibly too soon for this change. llvm-svn: 153944
2012-04-03 16:41:55 +08:00
Opts.NoInline = !Opt;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
// unsigned SSP = getLastArgIntValue(Args, OPT_stack_protector, 0, Diags);
// switch (SSP) {
// default:
// Diags.Report(diag::err_drv_invalid_value)
// << Args.getLastArg(OPT_stack_protector)->getAsString(Args) << SSP;
// break;
// case 0: Opts.setStackProtectorMode(LangOptions::SSPOff); break;
// case 1: Opts.setStackProtectorMode(LangOptions::SSPOn); break;
// case 2: Opts.setStackProtectorMode(LangOptions::SSPReq); break;
// }
}
More leaks detection: - fixed 3 posix spawn attributes leaks - fixed us always leaking CXXBaseSpecifier objects when we create class base classes. Clang apparently copies the base classes we pass in. Fixed some code formatting in ClangASTContext.cpp. llvm-svn: 107459
2010-07-02 09:29:13 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ClangASTContext::ClangASTContext (const char *target_triple) :
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
m_target_triple(),
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
m_ast_ap(),
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
m_language_options_ap(),
m_source_manager_ap(),
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
m_diagnostics_engine_ap(),
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
m_target_options_ap(),
m_target_info_ap(),
m_identifier_table_ap(),
m_selector_table_ap(),
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
m_builtins_ap(),
m_callback_tag_decl (NULL),
m_callback_objc_decl (NULL),
m_callback_baton (NULL)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (target_triple && target_triple[0])
Fixed a compile error. llvm-svn: 136551
2011-07-30 09:26:02 +08:00
SetTargetTriple (target_triple);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
//----------------------------------------------------------------------
// Destructor
//----------------------------------------------------------------------
ClangASTContext::~ClangASTContext()
{
m_builtins_ap.reset();
m_selector_table_ap.reset();
m_identifier_table_ap.reset();
m_target_info_ap.reset();
m_target_options_ap.reset();
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
m_diagnostics_engine_ap.reset();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
m_source_manager_ap.reset();
m_language_options_ap.reset();
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
m_ast_ap.reset();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
void
ClangASTContext::Clear()
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
m_ast_ap.reset();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
m_language_options_ap.reset();
m_source_manager_ap.reset();
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
m_diagnostics_engine_ap.reset();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
m_target_options_ap.reset();
m_target_info_ap.reset();
m_identifier_table_ap.reset();
m_selector_table_ap.reset();
m_builtins_ap.reset();
}
const char *
ClangASTContext::GetTargetTriple ()
{
return m_target_triple.c_str();
}
void
ClangASTContext::SetTargetTriple (const char *target_triple)
{
Clear();
m_target_triple.assign(target_triple);
}
Made lldb_private::ArchSpec contain much more than just an architecture. It now, in addition to cpu type/subtype and architecture flavor, contains: - byte order (big endian, little endian) - address size in bytes - llvm::Triple for true target triple support and for more powerful plug-in selection. llvm-svn: 125602
2011-02-16 05:59:32 +08:00
void
ClangASTContext::SetArchitecture (const ArchSpec &arch)
{
Fixed a compile error. llvm-svn: 136551
2011-07-30 09:26:02 +08:00
SetTargetTriple(arch.GetTriple().str().c_str());
Made lldb_private::ArchSpec contain much more than just an architecture. It now, in addition to cpu type/subtype and architecture flavor, contains: - byte order (big endian, little endian) - address size in bytes - llvm::Triple for true target triple support and for more powerful plug-in selection. llvm-svn: 125602
2011-02-16 05:59:32 +08:00
}
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
bool
ClangASTContext::HasExternalSource ()
{
ASTContext *ast = getASTContext();
if (ast)
return ast->getExternalSource () != NULL;
return false;
}
void
ClangASTContext::SetExternalSource (llvm::OwningPtr<ExternalASTSource> &ast_source_ap)
{
ASTContext *ast = getASTContext();
if (ast)
{
ast->setExternalSource (ast_source_ap);
ast->getTranslationUnitDecl()->setHasExternalLexicalStorage(true);
//ast->getTranslationUnitDecl()->setHasExternalVisibleStorage(true);
}
}
void
ClangASTContext::RemoveExternalSource ()
{
ASTContext *ast = getASTContext();
if (ast)
{
llvm::OwningPtr<ExternalASTSource> empty_ast_source_ap;
ast->setExternalSource (empty_ast_source_ap);
ast->getTranslationUnitDecl()->setHasExternalLexicalStorage(false);
//ast->getTranslationUnitDecl()->setHasExternalVisibleStorage(false);
}
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
ASTContext *
ClangASTContext::getASTContext()
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (m_ast_ap.get() == NULL)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
m_ast_ap.reset(new ASTContext (*getLanguageOptions(),
*getSourceManager(),
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
getTargetInfo(),
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
*getIdentifierTable(),
*getSelectorTable(),
*getBuiltinContext(),
0));
Made all LLDB-generated ASTContexts have valid DiagnosticClients, and removed code that was patching over the original problem. llvm-svn: 121601
2010-12-11 08:08:56 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if ((m_callback_tag_decl || m_callback_objc_decl) && m_callback_baton)
{
m_ast_ap->getTranslationUnitDecl()->setHasExternalLexicalStorage();
//m_ast_ap->getTranslationUnitDecl()->setHasExternalVisibleStorage();
}
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
m_ast_ap->getDiagnostics().setClient(getDiagnosticConsumer(), false);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return m_ast_ap.get();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Builtin::Context *
ClangASTContext::getBuiltinContext()
{
if (m_builtins_ap.get() == NULL)
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
m_builtins_ap.reset (new Builtin::Context());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return m_builtins_ap.get();
}
IdentifierTable *
ClangASTContext::getIdentifierTable()
{
if (m_identifier_table_ap.get() == NULL)
m_identifier_table_ap.reset(new IdentifierTable (*ClangASTContext::getLanguageOptions(), NULL));
return m_identifier_table_ap.get();
}
LangOptions *
ClangASTContext::getLanguageOptions()
{
if (m_language_options_ap.get() == NULL)
{
m_language_options_ap.reset(new LangOptions());
Updated LLVM/Clang to revision from 2010-06-13T06:00. llvm-svn: 105918
2010-06-14 01:34:29 +08:00
ParseLangArgs(*m_language_options_ap, IK_ObjCXX);
// InitializeLangOptions(*m_language_options_ap, IK_ObjCXX);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
return m_language_options_ap.get();
}
SelectorTable *
ClangASTContext::getSelectorTable()
{
if (m_selector_table_ap.get() == NULL)
m_selector_table_ap.reset (new SelectorTable());
return m_selector_table_ap.get();
}
Updated to the LLVM/Clang of 2010-11-17 at 3:30pm. llvm-svn: 119677
2010-11-18 10:56:27 +08:00
clang::FileManager *
ClangASTContext::getFileManager()
{
if (m_file_manager_ap.get() == NULL)
Updated to latest LLVM/Clang for external AST source changes that allow TagDecl subclasses and Objective C interfaces to complete themselves through the ExternalASTSource class. llvm-svn: 120749
2010-12-03 07:20:03 +08:00
{
clang::FileSystemOptions file_system_options;
m_file_manager_ap.reset(new clang::FileManager(file_system_options));
}
Updated to the LLVM/Clang of 2010-11-17 at 3:30pm. llvm-svn: 119677
2010-11-18 10:56:27 +08:00
return m_file_manager_ap.get();
}
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
clang::SourceManager *
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
ClangASTContext::getSourceManager()
{
if (m_source_manager_ap.get() == NULL)
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
m_source_manager_ap.reset(new clang::SourceManager(*getDiagnosticsEngine(), *getFileManager()));
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return m_source_manager_ap.get();
}
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
clang::DiagnosticsEngine *
ClangASTContext::getDiagnosticsEngine()
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
if (m_diagnostics_engine_ap.get() == NULL)
Updated to latest and greatest clang for a "print cvr-qualifiers on function declarations" fix. llvm-svn: 119847
2010-11-20 05:46:54 +08:00
{
llvm::IntrusiveRefCntPtr<DiagnosticIDs> diag_id_sp(new DiagnosticIDs());
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
m_diagnostics_engine_ap.reset(new DiagnosticsEngine(diag_id_sp));
Updated to latest and greatest clang for a "print cvr-qualifiers on function declarations" fix. llvm-svn: 119847
2010-11-20 05:46:54 +08:00
}
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
return m_diagnostics_engine_ap.get();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
class NullDiagnosticConsumer : public DiagnosticConsumer
Made all LLDB-generated ASTContexts have valid DiagnosticClients, and removed code that was patching over the original problem. llvm-svn: 121601
2010-12-11 08:08:56 +08:00
{
public:
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
NullDiagnosticConsumer ()
Made all LLDB-generated ASTContexts have valid DiagnosticClients, and removed code that was patching over the original problem. llvm-svn: 121601
2010-12-11 08:08:56 +08:00
{
m_log = lldb_private::GetLogIfAllCategoriesSet (LIBLLDB_LOG_EXPRESSIONS);
}
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
void HandleDiagnostic (DiagnosticsEngine::Level DiagLevel, const Diagnostic &info)
Made all LLDB-generated ASTContexts have valid DiagnosticClients, and removed code that was patching over the original problem. llvm-svn: 121601
2010-12-11 08:08:56 +08:00
{
if (m_log)
{
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
llvm::SmallVector<char, 32> diag_str(10);
Made all LLDB-generated ASTContexts have valid DiagnosticClients, and removed code that was patching over the original problem. llvm-svn: 121601
2010-12-11 08:08:56 +08:00
info.FormatDiagnostic(diag_str);
diag_str.push_back('\0');
m_log->Printf("Compiler diagnostic: %s\n", diag_str.data());
}
}
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
DiagnosticConsumer *clone (DiagnosticsEngine &Diags) const
{
return new NullDiagnosticConsumer ();
}
Made all LLDB-generated ASTContexts have valid DiagnosticClients, and removed code that was patching over the original problem. llvm-svn: 121601
2010-12-11 08:08:56 +08:00
private:
LogSP m_log;
};
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
DiagnosticConsumer *
ClangASTContext::getDiagnosticConsumer()
Made all LLDB-generated ASTContexts have valid DiagnosticClients, and removed code that was patching over the original problem. llvm-svn: 121601
2010-12-11 08:08:56 +08:00
{
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
if (m_diagnostic_consumer_ap.get() == NULL)
m_diagnostic_consumer_ap.reset(new NullDiagnosticConsumer);
Made all LLDB-generated ASTContexts have valid DiagnosticClients, and removed code that was patching over the original problem. llvm-svn: 121601
2010-12-11 08:08:56 +08:00
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
return m_diagnostic_consumer_ap.get();
Made all LLDB-generated ASTContexts have valid DiagnosticClients, and removed code that was patching over the original problem. llvm-svn: 121601
2010-12-11 08:08:56 +08:00
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
TargetOptions *
ClangASTContext::getTargetOptions()
{
if (m_target_options_ap.get() == NULL && !m_target_triple.empty())
{
m_target_options_ap.reset (new TargetOptions());
if (m_target_options_ap.get())
m_target_options_ap->Triple = m_target_triple;
}
return m_target_options_ap.get();
}
TargetInfo *
ClangASTContext::getTargetInfo()
{
// target_triple should be something like "x86_64-apple-darwin10"
if (m_target_info_ap.get() == NULL && !m_target_triple.empty())
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
m_target_info_ap.reset (TargetInfo::CreateTargetInfo(*getDiagnosticsEngine(), *getTargetOptions()));
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return m_target_info_ap.get();
}
#pragma mark Basic Types
static inline bool
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
QualTypeMatchesBitSize(const uint64_t bit_size, ASTContext *ast, QualType qual_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
uint64_t qual_type_bit_size = ast->getTypeSize(qual_type);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (qual_type_bit_size == bit_size)
return true;
return false;
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
More missing files from my previous checkin. llvm-svn: 110299
2010-08-05 09:57:25 +08:00
ClangASTContext::GetBuiltinTypeForEncodingAndBitSize (Encoding encoding, uint32_t bit_size)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast = getASTContext();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
assert (ast != NULL);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return GetBuiltinTypeForEncodingAndBitSize (ast, encoding, bit_size);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ClangASTContext::GetBuiltinTypeForEncodingAndBitSize (ASTContext *ast, Encoding encoding, uint32_t bit_size)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (!ast)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return NULL;
switch (encoding)
{
More missing files from my previous checkin. llvm-svn: 110299
2010-08-05 09:57:25 +08:00
case eEncodingInvalid:
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (QualTypeMatchesBitSize (bit_size, ast, ast->VoidPtrTy))
return ast->VoidPtrTy.getAsOpaquePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
break;
More missing files from my previous checkin. llvm-svn: 110299
2010-08-05 09:57:25 +08:00
case eEncodingUint:
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedCharTy))
return ast->UnsignedCharTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedShortTy))
return ast->UnsignedShortTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedIntTy))
return ast->UnsignedIntTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedLongTy))
return ast->UnsignedLongTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedLongLongTy))
return ast->UnsignedLongLongTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedInt128Ty))
return ast->UnsignedInt128Ty.getAsOpaquePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
break;
More missing files from my previous checkin. llvm-svn: 110299
2010-08-05 09:57:25 +08:00
case eEncodingSint:
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (QualTypeMatchesBitSize (bit_size, ast, ast->CharTy))
return ast->CharTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->ShortTy))
return ast->ShortTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->IntTy))
return ast->IntTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->LongTy))
return ast->LongTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->LongLongTy))
return ast->LongLongTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->Int128Ty))
return ast->Int128Ty.getAsOpaquePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
break;
More missing files from my previous checkin. llvm-svn: 110299
2010-08-05 09:57:25 +08:00
case eEncodingIEEE754:
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (QualTypeMatchesBitSize (bit_size, ast, ast->FloatTy))
return ast->FloatTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->DoubleTy))
return ast->DoubleTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->LongDoubleTy))
return ast->LongDoubleTy.getAsOpaquePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
break;
More missing files from my previous checkin. llvm-svn: 110299
2010-08-05 09:57:25 +08:00
case eEncodingVector:
rdar://problem/10611315 expression command doesn't handle xmm or stmm registers... o Update ClangASTContext::GetBuiltinTypeForEncodingAndBitSize() to now handle eEncodingVector. o Modify RegisterValue::SetFromMemoryData() to fix the subtle error due to unitialized variables. o Add a test file for "expr $xmm0". llvm-svn: 152190
2012-03-07 09:12:24 +08:00
// Sanity check that bit_size is a multiple of 8's.
if (bit_size && !(bit_size & 0x7u))
return ast->getExtVectorType (ast->UnsignedCharTy, bit_size/8).getAsOpaquePtr();
break;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
default:
break;
}
return NULL;
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
ClangASTContext::GetBuiltinTypeForDWARFEncodingAndBitSize (const char *type_name, uint32_t dw_ate, uint32_t bit_size)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast = getASTContext();
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
#define streq(a,b) strcmp(a,b) == 0
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
assert (ast != NULL);
if (ast)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
switch (dw_ate)
{
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
default:
break;
case DW_ATE_address:
if (QualTypeMatchesBitSize (bit_size, ast, ast->VoidPtrTy))
return ast->VoidPtrTy.getAsOpaquePtr();
break;
case DW_ATE_boolean:
if (QualTypeMatchesBitSize (bit_size, ast, ast->BoolTy))
return ast->BoolTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedCharTy))
return ast->UnsignedCharTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedShortTy))
return ast->UnsignedShortTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedIntTy))
return ast->UnsignedIntTy.getAsOpaquePtr();
break;
case DW_ATE_lo_user:
// This has been seen to mean DW_AT_complex_integer
if (type_name)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
if (::strstr(type_name, "complex"))
{
clang_type_t complex_int_clang_type = GetBuiltinTypeForDWARFEncodingAndBitSize ("int", DW_ATE_signed, bit_size/2);
return ast->getComplexType (QualType::getFromOpaquePtr(complex_int_clang_type)).getAsOpaquePtr();
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
break;
case DW_ATE_complex_float:
if (QualTypeMatchesBitSize (bit_size, ast, ast->FloatComplexTy))
return ast->FloatComplexTy.getAsOpaquePtr();
else if (QualTypeMatchesBitSize (bit_size, ast, ast->DoubleComplexTy))
return ast->DoubleComplexTy.getAsOpaquePtr();
else if (QualTypeMatchesBitSize (bit_size, ast, ast->LongDoubleComplexTy))
return ast->LongDoubleComplexTy.getAsOpaquePtr();
else
Fixed cases where we were translating "long long" types to use the "long" types when they are the same size. The new code will use the correct type now when converting DWARF built-in types into clang types. llvm-svn: 118003
2010-11-02 11:48:39 +08:00
{
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
clang_type_t complex_float_clang_type = GetBuiltinTypeForDWARFEncodingAndBitSize ("float", DW_ATE_float, bit_size/2);
return ast->getComplexType (QualType::getFromOpaquePtr(complex_float_clang_type)).getAsOpaquePtr();
Fixed cases where we were translating "long long" types to use the "long" types when they are the same size. The new code will use the correct type now when converting DWARF built-in types into clang types. llvm-svn: 118003
2010-11-02 11:48:39 +08:00
}
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
break;
case DW_ATE_float:
if (QualTypeMatchesBitSize (bit_size, ast, ast->FloatTy))
return ast->FloatTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->DoubleTy))
return ast->DoubleTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->LongDoubleTy))
return ast->LongDoubleTy.getAsOpaquePtr();
break;
case DW_ATE_signed:
if (type_name)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
if (streq(type_name, "wchar_t") &&
QualTypeMatchesBitSize (bit_size, ast, ast->WCharTy))
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return ast->WCharTy.getAsOpaquePtr();
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
if (streq(type_name, "void") &&
QualTypeMatchesBitSize (bit_size, ast, ast->VoidTy))
Modified version of a patch from Warren Paul that takes care of issues with indirect forms, deals with empty DW_AT_comp_dir attributes, and fixups for handling other signed integer types. llvm-svn: 125240
2011-02-10 07:39:34 +08:00
return ast->VoidTy.getAsOpaquePtr();
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
if (strstr(type_name, "long long") &&
QualTypeMatchesBitSize (bit_size, ast, ast->LongLongTy))
return ast->LongLongTy.getAsOpaquePtr();
if (strstr(type_name, "long") &&
QualTypeMatchesBitSize (bit_size, ast, ast->LongTy))
return ast->LongTy.getAsOpaquePtr();
if (strstr(type_name, "short") &&
QualTypeMatchesBitSize (bit_size, ast, ast->ShortTy))
return ast->ShortTy.getAsOpaquePtr();
if (strstr(type_name, "char"))
{
if (QualTypeMatchesBitSize (bit_size, ast, ast->CharTy))
return ast->CharTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->SignedCharTy))
return ast->SignedCharTy.getAsOpaquePtr();
}
if (strstr(type_name, "int"))
{
if (QualTypeMatchesBitSize (bit_size, ast, ast->IntTy))
return ast->IntTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->Int128Ty))
return ast->Int128Ty.getAsOpaquePtr();
}
Modified version of a patch from Warren Paul that takes care of issues with indirect forms, deals with empty DW_AT_comp_dir attributes, and fixups for handling other signed integer types. llvm-svn: 125240
2011-02-10 07:39:34 +08:00
}
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
// We weren't able to match up a type name, just search by size
if (QualTypeMatchesBitSize (bit_size, ast, ast->CharTy))
return ast->CharTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->ShortTy))
return ast->ShortTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->IntTy))
return ast->IntTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->LongTy))
return ast->LongTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->LongLongTy))
return ast->LongLongTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->Int128Ty))
return ast->Int128Ty.getAsOpaquePtr();
break;
case DW_ATE_signed_char:
if (type_name)
Fixed cases where we were translating "long long" types to use the "long" types when they are the same size. The new code will use the correct type now when converting DWARF built-in types into clang types. llvm-svn: 118003
2010-11-02 11:48:39 +08:00
{
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
if (streq(type_name, "signed char"))
{
if (QualTypeMatchesBitSize (bit_size, ast, ast->SignedCharTy))
return ast->SignedCharTy.getAsOpaquePtr();
}
Fixed cases where we were translating "long long" types to use the "long" types when they are the same size. The new code will use the correct type now when converting DWARF built-in types into clang types. llvm-svn: 118003
2010-11-02 11:48:39 +08:00
}
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
if (QualTypeMatchesBitSize (bit_size, ast, ast->CharTy))
return ast->CharTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->SignedCharTy))
return ast->SignedCharTy.getAsOpaquePtr();
break;
Added support for C++0x char16_t and char32_t types. llvm-svn: 143246
2011-10-29 07:06:08 +08:00
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
case DW_ATE_unsigned:
if (type_name)
Added support for C++0x char16_t and char32_t types. llvm-svn: 143246
2011-10-29 07:06:08 +08:00
{
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
if (strstr(type_name, "long long"))
{
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedLongLongTy))
return ast->UnsignedLongLongTy.getAsOpaquePtr();
}
else if (strstr(type_name, "long"))
{
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedLongTy))
return ast->UnsignedLongTy.getAsOpaquePtr();
}
else if (strstr(type_name, "short"))
{
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedShortTy))
return ast->UnsignedShortTy.getAsOpaquePtr();
}
else if (strstr(type_name, "char"))
{
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedCharTy))
return ast->UnsignedCharTy.getAsOpaquePtr();
}
else if (strstr(type_name, "int"))
{
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedIntTy))
return ast->UnsignedIntTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedInt128Ty))
return ast->UnsignedInt128Ty.getAsOpaquePtr();
}
Added support for C++0x char16_t and char32_t types. llvm-svn: 143246
2011-10-29 07:06:08 +08:00
}
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
// We weren't able to match up a type name, just search by size
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedCharTy))
return ast->UnsignedCharTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedShortTy))
return ast->UnsignedShortTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedIntTy))
return ast->UnsignedIntTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedLongTy))
return ast->UnsignedLongTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedLongLongTy))
return ast->UnsignedLongLongTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedInt128Ty))
return ast->UnsignedInt128Ty.getAsOpaquePtr();
break;
case DW_ATE_unsigned_char:
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedCharTy))
return ast->UnsignedCharTy.getAsOpaquePtr();
if (QualTypeMatchesBitSize (bit_size, ast, ast->UnsignedShortTy))
return ast->UnsignedShortTy.getAsOpaquePtr();
break;
case DW_ATE_imaginary_float:
break;
case DW_ATE_UTF:
if (type_name)
Added support for C++0x char16_t and char32_t types. llvm-svn: 143246
2011-10-29 07:06:08 +08:00
{
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
if (streq(type_name, "char16_t"))
{
return ast->Char16Ty.getAsOpaquePtr();
}
else if (streq(type_name, "char32_t"))
{
return ast->Char32Ty.getAsOpaquePtr();
}
Added support for C++0x char16_t and char32_t types. llvm-svn: 143246
2011-10-29 07:06:08 +08:00
}
Fixed ClangASTContext to correctly recognize wchar_t as distinct from int. llvm-svn: 153920
2012-04-03 09:10:10 +08:00
break;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
}
// This assert should fire for anything that we don't catch above so we know
// to fix any issues we run into.
Fixed an assert that could cause a crash when there was an unrecognized DW_TAG_base_type. Now it is a error printed to stderr. llvm-svn: 131473
2011-05-18 02:15:05 +08:00
if (type_name)
{
Added code in the Host layer that can report system log messages so that we don't have "fprintf (stderr, ...)" calls sprinkled everywhere. Changed all needed locations over to using this. For non-darwin, we log to stderr only. On darwin, we log to stderr _and_ to ASL (Apple System Log facility). This will allow GUI apps to have a place for these error and warning messages to go, and also allows the command line apps to log directly to the terminal. llvm-svn: 147596
2012-01-05 11:57:59 +08:00
Host::SystemLog (Host::eSystemLogError, "error: need to add support for DW_TAG_base_type '%s' encoded with DW_ATE = 0x%x, bit_size = %u\n", type_name, dw_ate, bit_size);
Fixed an assert that could cause a crash when there was an unrecognized DW_TAG_base_type. Now it is a error printed to stderr. llvm-svn: 131473
2011-05-18 02:15:05 +08:00
}
else
{
Added code in the Host layer that can report system log messages so that we don't have "fprintf (stderr, ...)" calls sprinkled everywhere. Changed all needed locations over to using this. For non-darwin, we log to stderr only. On darwin, we log to stderr _and_ to ASL (Apple System Log facility). This will allow GUI apps to have a place for these error and warning messages to go, and also allows the command line apps to log directly to the terminal. llvm-svn: 147596
2012-01-05 11:57:59 +08:00
Host::SystemLog (Host::eSystemLogError, "error: need to add support for DW_TAG_base_type encoded with DW_ATE = 0x%x, bit_size = %u\n", dw_ate, bit_size);
Fixed an assert that could cause a crash when there was an unrecognized DW_TAG_base_type. Now it is a error printed to stderr. llvm-svn: 131473
2011-05-18 02:15:05 +08:00
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return NULL;
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ClangASTContext::GetBuiltInType_void(ASTContext *ast)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return ast->VoidTy.getAsOpaquePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Added support for indicating to the expression parser that the result of an expression should be coerced to a specific type. Also made breakpoint conditions pass in the bool type for this type. The expression parser ignores this indication for now. llvm-svn: 119779
2010-11-19 10:52:21 +08:00
clang_type_t
ClangASTContext::GetBuiltInType_bool()
{
return getASTContext()->BoolTy.getAsOpaquePtr();
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
ClangASTContext::GetBuiltInType_objc_id()
{
Pulled in a new version of LLVM/Clang to solve a variety of problems with Objective-C object completion. To go along with the LLVM/Clang-side fixes, we have a variety of Objective-C improvements. Fixes include: - It is now possible to run expressions when stopped in an Objective-C class method and have "self" act just like "self" would act in the class method itself (i.e., [self classMethod] works without casting the return type if debug info is present). To accomplish this, the expression masquerades as a class method added by a category. - Objective-C objects can now provide methods and properties and methods to Clang on demand (i.e., the ASTImporter sets hasExternalVisibleDecls on Objective-C interface objects). - Objective-C built-in types, which had long been a bone of contention (should we be using "id"? "id*"?), are now fetched correctly using accessor functions on ClangASTContext. We inhibit searches for them in the debug information. There are also a variety of logging fixes, and I made two changes to the test suite: - Enabled a test case for Objective-C properties in the current translation unit. - Added a test case for calling Objective-C class methods when stopped in a class method. llvm-svn: 144607
2011-11-15 10:11:17 +08:00
return getASTContext()->getObjCIdType().getAsOpaquePtr();
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
ClangASTContext::GetBuiltInType_objc_Class()
{
Pulled in a new version of LLVM/Clang to solve a variety of problems with Objective-C object completion. To go along with the LLVM/Clang-side fixes, we have a variety of Objective-C improvements. Fixes include: - It is now possible to run expressions when stopped in an Objective-C class method and have "self" act just like "self" would act in the class method itself (i.e., [self classMethod] works without casting the return type if debug info is present). To accomplish this, the expression masquerades as a class method added by a category. - Objective-C objects can now provide methods and properties and methods to Clang on demand (i.e., the ASTImporter sets hasExternalVisibleDecls on Objective-C interface objects). - Objective-C built-in types, which had long been a bone of contention (should we be using "id"? "id*"?), are now fetched correctly using accessor functions on ClangASTContext. We inhibit searches for them in the debug information. There are also a variety of logging fixes, and I made two changes to the test suite: - Enabled a test case for Objective-C properties in the current translation unit. - Added a test case for calling Objective-C class methods when stopped in a class method. llvm-svn: 144607
2011-11-15 10:11:17 +08:00
return getASTContext()->getObjCClassType().getAsOpaquePtr();
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
ClangASTContext::GetBuiltInType_objc_selector()
{
Pulled in a new version of LLVM/Clang to solve a variety of problems with Objective-C object completion. To go along with the LLVM/Clang-side fixes, we have a variety of Objective-C improvements. Fixes include: - It is now possible to run expressions when stopped in an Objective-C class method and have "self" act just like "self" would act in the class method itself (i.e., [self classMethod] works without casting the return type if debug info is present). To accomplish this, the expression masquerades as a class method added by a category. - Objective-C objects can now provide methods and properties and methods to Clang on demand (i.e., the ASTImporter sets hasExternalVisibleDecls on Objective-C interface objects). - Objective-C built-in types, which had long been a bone of contention (should we be using "id"? "id*"?), are now fetched correctly using accessor functions on ClangASTContext. We inhibit searches for them in the debug information. There are also a variety of logging fixes, and I made two changes to the test suite: - Enabled a test case for Objective-C properties in the current translation unit. - Added a test case for calling Objective-C class methods when stopped in a class method. llvm-svn: 144607
2011-11-15 10:11:17 +08:00
return getASTContext()->getObjCSelType().getAsOpaquePtr();
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
}
Introduced support for UnknownAnyTy, the Clang type representing variables whose type must be inferred from the way they are used. Functions without debug information now return UnknownAnyTy and must be cast. Variables with no debug information are not yet using UnknownAnyTy; instead they are assumed to be void*. Support for variables of unknown type is coming (and, in fact, some relevant support functions are included in this commit) but will take a bit of extra effort. The testsuite has also been updated to reflect the new requirement that the result of printf be cast, i.e. expr (int) printf("Hello world!") llvm-svn: 131263
2011-05-13 07:54:16 +08:00
clang_type_t
ClangASTContext::GetUnknownAnyType(clang::ASTContext *ast)
{
return ast->UnknownAnyTy.getAsOpaquePtr();
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
ClangASTContext::GetCStringType (bool is_const)
{
QualType char_type(getASTContext()->CharTy);
if (is_const)
char_type.addConst();
return getASTContext()->getPointerType(char_type).getAsOpaquePtr();
}
Fixed a problem where we did not read properties correctly if the setter/getter were not present in the debug information. The fixes are as follows: - We not only look for the method by its full name, but also look for automatically-generated methods when searching for a selector in an Objective-C interface. This is necessary to find accessors. - Extract the getter and setter name from the DW_TAG_APPLE_Property declaration in the DWARF if they are present; generate them if not. llvm-svn: 154067
2012-04-05 08:12:52 +08:00
clang_type_t
ClangASTContext::GetVoidType()
{
return GetVoidType(getASTContext());
}
clang_type_t
ClangASTContext::GetVoidType(ASTContext *ast)
{
return ast->VoidTy.getAsOpaquePtr();
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
ClangASTContext::GetVoidPtrType (bool is_const)
{
return GetVoidPtrType(getASTContext(), is_const);
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ClangASTContext::GetVoidPtrType (ASTContext *ast, bool is_const)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
QualType void_ptr_type(ast->VoidPtrTy);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (is_const)
void_ptr_type.addConst();
return void_ptr_type.getAsOpaquePtr();
}
Added support to the Objective-C language runtime to find Objective-C class types by looking in the symbol tables for the individual object files. I did this as follows: - I added code to SymbolFileSymtab that vends Clang types for symbols matching the pattern "_OBJC_CLASS_$_NSMyClassName," making them appear as Objective-C classes. This only occurs in modules that do not have debug information, since otherwise SymbolFileDWARF would be in charge of looking up types. - I made a new SymbolVendor subclass for the Apple Objective-C runtime that is in charge of making global lookups of Objective-C types. It currently just sends out type lookup requests to the appropriate SymbolFiles, but in the future we will probably extend it to query the runtime more completely. I also modified a testcase whose behavior is changed by the fact that we now actually return an Objective-C type for __NSCFString. llvm-svn: 145526
2011-12-01 06:11:59 +08:00
clang::DeclContext *
ClangASTContext::GetTranslationUnitDecl (clang::ASTContext *ast)
{
return ast->getTranslationUnitDecl();
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
Updated to latest LLVM/Clang for external AST source changes that allow TagDecl subclasses and Objective C interfaces to complete themselves through the ExternalASTSource class. llvm-svn: 120749
2010-12-03 07:20:03 +08:00
ClangASTContext::CopyType (ASTContext *dst_ast,
ASTContext *src_ast,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t clang_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated to the LLVM/Clang of 2010-11-17 at 3:30pm. llvm-svn: 119677
2010-11-18 10:56:27 +08:00
FileSystemOptions file_system_options;
Updated to latest LLVM/Clang for external AST source changes that allow TagDecl subclasses and Objective C interfaces to complete themselves through the ExternalASTSource class. llvm-svn: 120749
2010-12-03 07:20:03 +08:00
FileManager file_manager (file_system_options);
ASTImporter importer(*dst_ast, file_manager,
Updated to revision 123723 of LLVM, to bring in support for minimal type import functionality. llvm-svn: 123787
2011-01-19 07:32:05 +08:00
*src_ast, file_manager,
false);
Added a named container for the source QualType in the type copy routine to make type problems easier to debug. llvm-svn: 118638
2010-11-10 06:37:10 +08:00
Updated to latest LLVM/Clang for external AST source changes that allow TagDecl subclasses and Objective C interfaces to complete themselves through the ExternalASTSource class. llvm-svn: 120749
2010-12-03 07:20:03 +08:00
QualType src (QualType::getFromOpaquePtr(clang_type));
QualType dst (importer.Import(src));
Added a named container for the source QualType in the type copy routine to make type problems easier to debug. llvm-svn: 118638
2010-11-10 06:37:10 +08:00
return dst.getAsOpaquePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Modified the lldb_private::Type clang type resolving code to handle three cases when getting the clang type: - need only a forward declaration - need a clang type that can be used for layout (members and args/return types) - need a full clang type This allows us to partially parse the clang types and be as lazy as possible. The first case is when we just need to declare a type and we will complete it later. The forward declaration happens only for class/union/structs and enums. The layout type allows us to resolve the full clang type _except_ if we have any modifiers on a pointer or reference (both R and L value). In this case when we are adding members or function args or return types, we only need to know how the type will be laid out and we can defer completing the pointee type until we later need it. The last type means we need a full definition for the clang type. Did some renaming of some enumerations to get rid of the old "DC" prefix (which stands for DebugCore which is no longer around). Modified the clang namespace support to be almost ready to be fed to the expression parser. I made a new ClangNamespaceDecl class that can carry around the AST and the namespace decl so we can copy it into the expression AST. I modified the symbol vendor and symbol file plug-ins to use this new class. llvm-svn: 118976
2010-11-13 11:52:47 +08:00
clang::Decl *
Updated to latest LLVM/Clang for external AST source changes that allow TagDecl subclasses and Objective C interfaces to complete themselves through the ExternalASTSource class. llvm-svn: 120749
2010-12-03 07:20:03 +08:00
ClangASTContext::CopyDecl (ASTContext *dst_ast,
ASTContext *src_ast,
Modified the lldb_private::Type clang type resolving code to handle three cases when getting the clang type: - need only a forward declaration - need a clang type that can be used for layout (members and args/return types) - need a full clang type This allows us to partially parse the clang types and be as lazy as possible. The first case is when we just need to declare a type and we will complete it later. The forward declaration happens only for class/union/structs and enums. The layout type allows us to resolve the full clang type _except_ if we have any modifiers on a pointer or reference (both R and L value). In this case when we are adding members or function args or return types, we only need to know how the type will be laid out and we can defer completing the pointee type until we later need it. The last type means we need a full definition for the clang type. Did some renaming of some enumerations to get rid of the old "DC" prefix (which stands for DebugCore which is no longer around). Modified the clang namespace support to be almost ready to be fed to the expression parser. I made a new ClangNamespaceDecl class that can carry around the AST and the namespace decl so we can copy it into the expression AST. I modified the symbol vendor and symbol file plug-ins to use this new class. llvm-svn: 118976
2010-11-13 11:52:47 +08:00
clang::Decl *source_decl)
Made all LLDB-generated ASTContexts have valid DiagnosticClients, and removed code that was patching over the original problem. llvm-svn: 121601
2010-12-11 08:08:56 +08:00
{
Updated to the LLVM/Clang of 2010-11-17 at 3:30pm. llvm-svn: 119677
2010-11-18 10:56:27 +08:00
FileSystemOptions file_system_options;
Updated to latest LLVM/Clang for external AST source changes that allow TagDecl subclasses and Objective C interfaces to complete themselves through the ExternalASTSource class. llvm-svn: 120749
2010-12-03 07:20:03 +08:00
FileManager file_manager (file_system_options);
ASTImporter importer(*dst_ast, file_manager,
Updated to revision 123723 of LLVM, to bring in support for minimal type import functionality. llvm-svn: 123787
2011-01-19 07:32:05 +08:00
*src_ast, file_manager,
false);
Modified the lldb_private::Type clang type resolving code to handle three cases when getting the clang type: - need only a forward declaration - need a clang type that can be used for layout (members and args/return types) - need a full clang type This allows us to partially parse the clang types and be as lazy as possible. The first case is when we just need to declare a type and we will complete it later. The forward declaration happens only for class/union/structs and enums. The layout type allows us to resolve the full clang type _except_ if we have any modifiers on a pointer or reference (both R and L value). In this case when we are adding members or function args or return types, we only need to know how the type will be laid out and we can defer completing the pointee type until we later need it. The last type means we need a full definition for the clang type. Did some renaming of some enumerations to get rid of the old "DC" prefix (which stands for DebugCore which is no longer around). Modified the clang namespace support to be almost ready to be fed to the expression parser. I made a new ClangNamespaceDecl class that can carry around the AST and the namespace decl so we can copy it into the expression AST. I modified the symbol vendor and symbol file plug-ins to use this new class. llvm-svn: 118976
2010-11-13 11:52:47 +08:00
return importer.Import(source_decl);
}
Oops, didn't define AreTypesSame correctly. llvm-svn: 108483
2010-07-16 08:00:27 +08:00
bool
<rdar://problem/11113279> Fixed type lookups to "do the right thing". Prior to this fix, looking up a type using "foo::bar" would result in a type list that contains all types that had "bar" as a basename unless the symbol file was able to match fully qualified names (which our DWARF parser does not). This fix will allow type matches to be made based on the basename and then have the types that don't match filtered out. Types by name can be fully qualified, or partially qualified with the new "bool exact_match" parameter to the Module::FindTypes() method. This fixes some issue that we discovered with dynamic type resolution as well as improves the overall type lookups in LLDB. llvm-svn: 153482
2012-03-27 07:03:23 +08:00
ClangASTContext::AreTypesSame (ASTContext *ast,
clang_type_t type1,
clang_type_t type2,
bool ignore_qualifiers)
Added a function to test whether two types are the same. llvm-svn: 108467
2010-07-16 06:30:52 +08:00
{
Check if the two clang opaque type pointers are equal before doing anything more exhaustive comparison. llvm-svn: 154181
2012-04-07 01:38:55 +08:00
if (type1 == type2)
return true;
Ignore the constness of the object pointer when fetching it. llvm-svn: 150861
2012-02-18 10:01:03 +08:00
QualType type1_qual = QualType::getFromOpaquePtr(type1);
QualType type2_qual = QualType::getFromOpaquePtr(type2);
if (ignore_qualifiers)
{
type1_qual = type1_qual.getUnqualifiedType();
type2_qual = type2_qual.getUnqualifiedType();
}
return ast->hasSameType (type1_qual,
type2_qual);
Added a function to test whether two types are the same. llvm-svn: 108467
2010-07-16 06:30:52 +08:00
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
#pragma mark CVR modifiers
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
ClangASTContext::AddConstModifier (clang_type_t clang_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type)
{
QualType result(QualType::getFromOpaquePtr(clang_type));
result.addConst();
return result.getAsOpaquePtr();
}
return NULL;
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
ClangASTContext::AddRestrictModifier (clang_type_t clang_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type)
{
QualType result(QualType::getFromOpaquePtr(clang_type));
result.getQualifiers().setRestrict (true);
return result.getAsOpaquePtr();
}
return NULL;
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
ClangASTContext::AddVolatileModifier (clang_type_t clang_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type)
{
QualType result(QualType::getFromOpaquePtr(clang_type));
result.getQualifiers().setVolatile (true);
return result.getAsOpaquePtr();
}
return NULL;
}
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
clang_type_t
ClangASTContext::GetTypeForDecl (TagDecl *decl)
{
// No need to call the getASTContext() accessor (which can create the AST
// if it isn't created yet, because we can't have created a decl in this
// AST if our AST didn't already exist...
if (m_ast_ap.get())
return m_ast_ap->getTagDeclType(decl).getAsOpaquePtr();
return NULL;
}
clang_type_t
ClangASTContext::GetTypeForDecl (ObjCInterfaceDecl *decl)
{
// No need to call the getASTContext() accessor (which can create the AST
// if it isn't created yet, because we can't have created a decl in this
// AST if our AST didn't already exist...
if (m_ast_ap.get())
return m_ast_ap->getObjCInterfaceType(decl).getAsOpaquePtr();
return NULL;
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
#pragma mark Structure, Unions, Classes
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
We now record metadata for Objective-C interfaces, Objective-C methods, and Objective-C properties. llvm-svn: 154972
2012-04-18 09:06:17 +08:00
ClangASTContext::CreateRecordType (DeclContext *decl_ctx, AccessType access_type, const char *name, int kind, LanguageType language, uint64_t metadata)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast = getASTContext();
assert (ast != NULL);
We now record metadata for Objective-C interfaces, Objective-C methods, and Objective-C properties. llvm-svn: 154972
2012-04-18 09:06:17 +08:00
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (decl_ctx == NULL)
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
decl_ctx = ast->getTranslationUnitDecl();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
Fix so we don't create C++ classes for all objective C classes when we have a objective C++ source file. llvm-svn: 138527
2011-08-25 07:50:00 +08:00
if (language == eLanguageTypeObjC || language == eLanguageTypeObjC_plus_plus)
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
{
Fixed an issue where objc types weren't ever getting fully resolved (beyond forward declarations). llvm-svn: 116181
2010-10-11 10:25:34 +08:00
bool isForwardDecl = true;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
bool isInternal = false;
We now record metadata for Objective-C interfaces, Objective-C methods, and Objective-C properties. llvm-svn: 154972
2012-04-18 09:06:17 +08:00
return CreateObjCClass (name, decl_ctx, isForwardDecl, isInternal, metadata);
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
// NOTE: Eventually CXXRecordDecl will be merged back into RecordDecl and
// we will need to update this code. I was told to currently always use
// the CXXRecordDecl class since we often don't know from debug information
// if something is struct or a class, so we default to always use the more
// complete definition just in case.
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
CXXRecordDecl *decl = CXXRecordDecl::Create (*ast,
(TagDecl::TagKind)kind,
decl_ctx,
SourceLocation(),
SourceLocation(),
name && name[0] ? &ast->Idents.get(name) : NULL);
If the name of a struct or union is NULL in the debug info, call it anonymous. This isn't perfect, because Clang actually considers the following struct not to be anonymous: – struct { int x; int y; } g_foo; - but DWARF doesn't make the distinction. llvm-svn: 148145
2012-01-14 06:10:18 +08:00
We now record metadata for Objective-C interfaces, Objective-C methods, and Objective-C properties. llvm-svn: 154972
2012-04-18 09:06:17 +08:00
if (decl)
SetMetadata(ast, (uintptr_t)decl, metadata);
Added a mechanism for keeping track of where in the debug information individual Decls came from. We've had a metadata infrastructure for a while, which was intended to solve a problem we've since dealt with in a different way. (It was meant to keep track of which definition of an Objective-C class was the "true" definition, but we now find it by searching the symbols for the class symbol.) The metadata is attached to the ExternalASTSource, which means it has a one-to-one correspondence with AST contexts. I've repurposed the metadata infrastructure to hold the object file and DIE offset for the DWARF information corresponding to a Decl. There are methods in ClangASTContext that get and set this metadata, and the ClangASTImporter is capable of tracking down the metadata for Decls that have been copied out of the debug information into the parser's AST context without using any additional memory. To see the metadata, you just have to enable the expression log: - (lldb) log enable lldb expr - and watch the import messages. The high 32 bits of the metadata indicate the index of the object file in its containing DWARFDebugMap; I have also added a log which you can use to track that mapping: - (lldb) log enable dwarf map - This adds 64 bits per Decl, which in my testing hasn't turned out to be very much (debugging Clang produces around 6500 Decls in my tests). To track how much data is being consumed, I've also added a global variable g_TotalSizeOfMetadata which tracks the total number of Decls that have metadata in all active AST contexts. Right now this metadata is enormously useful for tracking down bugs in the debug info parser. In the future I also want to use this information to provide more intelligent error messages instead of printing empty source lines wherever Clang refers to the location where something is defined. llvm-svn: 154634
2012-04-13 08:10:03 +08:00
If the name of a struct or union is NULL in the debug info, call it anonymous. This isn't perfect, because Clang actually considers the following struct not to be anonymous: – struct { int x; int y; } g_foo; - but DWARF doesn't make the distinction. llvm-svn: 148145
2012-01-14 06:10:18 +08:00
if (!name)
decl->setAnonymousStructOrUnion(true);
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
Fixed an issue where a class that resides inside another class wasn't getting an access specifier set on it, causing an assertion to fire when building with a Debug+Asserts build of clang. llvm-svn: 143010
2011-10-26 11:31:36 +08:00
if (decl_ctx)
{
if (access_type != eAccessNone)
decl->setAccess (ConvertAccessTypeToAccessSpecifier (access_type));
decl_ctx->addDecl (decl);
}
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return ast->getTagDeclType(decl).getAsOpaquePtr();
}
Almost have templatized functions working (templatized classes are already working, but not functions). I need to check on a few things to make sure I am registering everything correctly in the right order and in the right contexts. llvm-svn: 149858
2012-02-06 14:42:51 +08:00
static TemplateParameterList *
CreateTemplateParameterList (ASTContext *ast,
const ClangASTContext::TemplateParameterInfos &template_param_infos,
llvm::SmallVector<NamedDecl *, 8> &template_param_decls)
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
{
const bool parameter_pack = false;
const bool is_typename = false;
const unsigned depth = 0;
const size_t num_template_params = template_param_infos.GetSize();
for (size_t i=0; i<num_template_params; ++i)
{
const char *name = template_param_infos.names[i];
if (template_param_infos.args[i].getAsIntegral())
{
template_param_decls.push_back (NonTypeTemplateParmDecl::Create (*ast,
ast->getTranslationUnitDecl(), // Is this the right decl context?, SourceLocation StartLoc,
SourceLocation(),
SourceLocation(),
depth,
i,
&ast->Idents.get(name),
<rdar://problem/10394517> Fixed templates with NonTypeTemplateParmDecl objects. For example: template <unsigned N> .... This fixes SmallVector and all of the other classes that have template params that are non types. llvm-svn: 145667
2011-12-02 10:09:28 +08:00
template_param_infos.args[i].getIntegralType(),
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
parameter_pack,
NULL));
Almost have templatized functions working (templatized classes are already working, but not functions). I need to check on a few things to make sure I am registering everything correctly in the right order and in the right contexts. llvm-svn: 149858
2012-02-06 14:42:51 +08:00
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
}
else
{
template_param_decls.push_back (TemplateTypeParmDecl::Create (*ast,
ast->getTranslationUnitDecl(), // Is this the right decl context?
SourceLocation(),
SourceLocation(),
depth,
i,
&ast->Idents.get(name),
is_typename,
parameter_pack));
}
}
Almost have templatized functions working (templatized classes are already working, but not functions). I need to check on a few things to make sure I am registering everything correctly in the right order and in the right contexts. llvm-svn: 149858
2012-02-06 14:42:51 +08:00
TemplateParameterList *template_param_list = TemplateParameterList::Create (*ast,
SourceLocation(),
SourceLocation(),
&template_param_decls.front(),
template_param_decls.size(),
SourceLocation());
return template_param_list;
}
clang::FunctionTemplateDecl *
ClangASTContext::CreateFunctionTemplateDecl (clang::DeclContext *decl_ctx,
clang::FunctionDecl *func_decl,
const char *name,
const TemplateParameterInfos &template_param_infos)
{
// /// \brief Create a function template node.
ASTContext *ast = getASTContext();
llvm::SmallVector<NamedDecl *, 8> template_param_decls;
TemplateParameterList *template_param_list = CreateTemplateParameterList (ast,
template_param_infos,
template_param_decls);
FunctionTemplateDecl *func_tmpl_decl = FunctionTemplateDecl::Create (*ast,
decl_ctx,
func_decl->getLocation(),
func_decl->getDeclName(),
template_param_list,
func_decl);
for (size_t i=0, template_param_decl_count = template_param_decls.size();
i < template_param_decl_count;
++i)
{
// TODO: verify which decl context we should put template_param_decls into..
template_param_decls[i]->setDeclContext (func_decl);
}
return func_tmpl_decl;
}
void
ClangASTContext::CreateFunctionTemplateSpecializationInfo (FunctionDecl *func_decl,
clang::FunctionTemplateDecl *func_tmpl_decl,
const TemplateParameterInfos &infos)
{
TemplateArgumentList template_args (TemplateArgumentList::OnStack,
infos.args.data(),
infos.args.size());
func_decl->setFunctionTemplateSpecialization (func_tmpl_decl,
&template_args,
NULL);
}
ClassTemplateDecl *
ClangASTContext::CreateClassTemplateDecl (DeclContext *decl_ctx,
lldb::AccessType access_type,
const char *class_name,
int kind,
const TemplateParameterInfos &template_param_infos)
{
ASTContext *ast = getASTContext();
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
Almost have templatized functions working (templatized classes are already working, but not functions). I need to check on a few things to make sure I am registering everything correctly in the right order and in the right contexts. llvm-svn: 149858
2012-02-06 14:42:51 +08:00
ClassTemplateDecl *class_template_decl = NULL;
if (decl_ctx == NULL)
decl_ctx = ast->getTranslationUnitDecl();
IdentifierInfo &identifier_info = ast->Idents.get(class_name);
DeclarationName decl_name (&identifier_info);
clang::DeclContext::lookup_result result = decl_ctx->lookup(decl_name);
for (clang::DeclContext::lookup_iterator pos = result.first, end = result.second; pos != end; ++pos)
{
class_template_decl = dyn_cast<clang::ClassTemplateDecl>(*pos);
if (class_template_decl)
return class_template_decl;
}
llvm::SmallVector<NamedDecl *, 8> template_param_decls;
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
Almost have templatized functions working (templatized classes are already working, but not functions). I need to check on a few things to make sure I am registering everything correctly in the right order and in the right contexts. llvm-svn: 149858
2012-02-06 14:42:51 +08:00
TemplateParameterList *template_param_list = CreateTemplateParameterList (ast,
template_param_infos,
template_param_decls);
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
CXXRecordDecl *template_cxx_decl = CXXRecordDecl::Create (*ast,
(TagDecl::TagKind)kind,
decl_ctx, // What decl context do we use here? TU? The actual decl context?
SourceLocation(),
SourceLocation(),
&identifier_info);
<rdar://problem/10394517> Fixed templates with NonTypeTemplateParmDecl objects. For example: template <unsigned N> .... This fixes SmallVector and all of the other classes that have template params that are non types. llvm-svn: 145667
2011-12-02 10:09:28 +08:00
for (size_t i=0, template_param_decl_count = template_param_decls.size();
i < template_param_decl_count;
++i)
{
template_param_decls[i]->setDeclContext (template_cxx_decl);
}
Ensure that the empty RecordDecl generated for templates is properly complete (though still empty). llvm-svn: 144982
2011-11-19 09:35:08 +08:00
// With templated classes, we say that a class is templated with
// specializations, but that the bare class has no functions.
template_cxx_decl->startDefinition();
template_cxx_decl->completeDefinition();
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
class_template_decl = ClassTemplateDecl::Create (*ast,
decl_ctx, // What decl context do we use here? TU? The actual decl context?
SourceLocation(),
decl_name,
template_param_list,
template_cxx_decl,
NULL);
if (class_template_decl)
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
{
Fixed an issue where a class that resides inside another class wasn't getting an access specifier set on it, causing an assertion to fire when building with a Debug+Asserts build of clang. llvm-svn: 143010
2011-10-26 11:31:36 +08:00
if (access_type != eAccessNone)
class_template_decl->setAccess (ConvertAccessTypeToAccessSpecifier (access_type));
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
//if (TagDecl *ctx_tag_decl = dyn_cast<TagDecl>(decl_ctx))
// CompleteTagDeclarationDefinition(GetTypeForDecl(ctx_tag_decl));
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
decl_ctx->addDecl (class_template_decl);
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
#ifdef LLDB_CONFIGURATION_DEBUG
VerifyDecl(class_template_decl);
#endif
}
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
return class_template_decl;
}
ClassTemplateSpecializationDecl *
ClangASTContext::CreateClassTemplateSpecializationDecl (DeclContext *decl_ctx,
ClassTemplateDecl *class_template_decl,
int kind,
const TemplateParameterInfos &template_param_infos)
{
ASTContext *ast = getASTContext();
ClassTemplateSpecializationDecl *class_template_specialization_decl = ClassTemplateSpecializationDecl::Create (*ast,
(TagDecl::TagKind)kind,
decl_ctx,
SourceLocation(),
SourceLocation(),
class_template_decl,
&template_param_infos.args.front(),
template_param_infos.args.size(),
NULL);
return class_template_specialization_decl;
}
lldb::clang_type_t
ClangASTContext::CreateClassTemplateSpecializationType (ClassTemplateSpecializationDecl *class_template_specialization_decl)
{
if (class_template_specialization_decl)
{
ASTContext *ast = getASTContext();
if (ast)
return ast->getTagDeclType(class_template_specialization_decl).getAsOpaquePtr();
}
return NULL;
}
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
bool
ClangASTContext::SetHasExternalStorage (clang_type_t clang_type, bool has_extern)
{
if (clang_type == NULL)
return false;
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
case clang::Type::Record:
{
CXXRecordDecl *cxx_record_decl = qual_type->getAsCXXRecordDecl();
if (cxx_record_decl)
{
cxx_record_decl->setHasExternalLexicalStorage (has_extern);
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
cxx_record_decl->setHasExternalVisibleStorage (has_extern);
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return true;
}
}
break;
case clang::Type::Enum:
{
EnumDecl *enum_decl = cast<EnumType>(qual_type)->getDecl();
if (enum_decl)
{
enum_decl->setHasExternalLexicalStorage (has_extern);
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
enum_decl->setHasExternalVisibleStorage (has_extern);
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return true;
}
}
break;
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(qual_type.getTypePtr());
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
assert (objc_class_type);
if (objc_class_type)
{
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
if (class_interface_decl)
{
class_interface_decl->setHasExternalLexicalStorage (has_extern);
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
class_interface_decl->setHasExternalVisibleStorage (has_extern);
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return true;
}
}
}
break;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
case clang::Type::Typedef:
return ClangASTContext::SetHasExternalStorage (cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(), has_extern);
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated:
return ClangASTContext::SetHasExternalStorage (cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr(), has_extern);
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
default:
break;
}
return false;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
static bool
IsOperator (const char *name, OverloadedOperatorKind &op_kind)
{
if (name == NULL || name[0] == '\0')
return false;
Simplified the code that handles overloaded operator names. llvm-svn: 121511
2010-12-11 03:51:54 +08:00
#define OPERATOR_PREFIX "operator"
Modified LLDB expressions to not have to JIT and run code just to see variable values or persistent expression variables. Now if an expression consists of a value that is a child of a variable, or of a persistent variable only, we will create a value object for it and make a ValueObjectConstResult from it to freeze the value (for program variables only, not persistent variables) and avoid running JITed code. For everything else we still parse up and JIT code and run it in the inferior. There was also a lot of clean up in the expression code. I made the ClangExpressionVariables be stored in collections of shared pointers instead of in collections of objects. This will help stop a lot of copy constructors on these large objects and also cleans up the code considerably. The persistent clang expression variables were moved over to the Target to ensure they persist across process executions. Added the ability for lldb_private::Target objects to evaluate expressions. We want to evaluate expressions at the target level in case we aren't running yet, or we have just completed running. We still want to be able to access the persistent expression variables between runs, and also evaluate constant expressions. Added extra logging to the dynamic loader plug-in for MacOSX. ModuleList objects can now dump their contents with the UUID, arch and full paths being logged with appropriate prefix values. Thread hardened the Communication class a bit by making the connection auto_ptr member into a shared pointer member and then making a local copy of the shared pointer in each method that uses it to make sure another thread can't nuke the connection object while it is being used by another thread. Added a new file to the lldb/test/load_unload test that causes the test a.out file to link to the libd.dylib file all the time. This will allow us to test using the DYLD_LIBRARY_PATH environment variable after moving libd.dylib somewhere else. llvm-svn: 121745
2010-12-14 10:59:59 +08:00
#define OPERATOR_PREFIX_LENGTH (sizeof (OPERATOR_PREFIX) - 1)
Fixed detection of overloaded operators in type information. Previously, "operator[]" was not being recognized because the detector was looking for "operator " as a prefix. Now we handle both cases. llvm-svn: 121456
2010-12-10 10:15:55 +08:00
const char *post_op_name = NULL;
Simplified the code that handles overloaded operator names. llvm-svn: 121511
2010-12-11 03:51:54 +08:00
bool no_space = true;
Fixed detection of overloaded operators in type information. Previously, "operator[]" was not being recognized because the detector was looking for "operator " as a prefix. Now we handle both cases. llvm-svn: 121456
2010-12-10 10:15:55 +08:00
Modified LLDB expressions to not have to JIT and run code just to see variable values or persistent expression variables. Now if an expression consists of a value that is a child of a variable, or of a persistent variable only, we will create a value object for it and make a ValueObjectConstResult from it to freeze the value (for program variables only, not persistent variables) and avoid running JITed code. For everything else we still parse up and JIT code and run it in the inferior. There was also a lot of clean up in the expression code. I made the ClangExpressionVariables be stored in collections of shared pointers instead of in collections of objects. This will help stop a lot of copy constructors on these large objects and also cleans up the code considerably. The persistent clang expression variables were moved over to the Target to ensure they persist across process executions. Added the ability for lldb_private::Target objects to evaluate expressions. We want to evaluate expressions at the target level in case we aren't running yet, or we have just completed running. We still want to be able to access the persistent expression variables between runs, and also evaluate constant expressions. Added extra logging to the dynamic loader plug-in for MacOSX. ModuleList objects can now dump their contents with the UUID, arch and full paths being logged with appropriate prefix values. Thread hardened the Communication class a bit by making the connection auto_ptr member into a shared pointer member and then making a local copy of the shared pointer in each method that uses it to make sure another thread can't nuke the connection object while it is being used by another thread. Added a new file to the lldb/test/load_unload test that causes the test a.out file to link to the libd.dylib file all the time. This will allow us to test using the DYLD_LIBRARY_PATH environment variable after moving libd.dylib somewhere else. llvm-svn: 121745
2010-12-14 10:59:59 +08:00
if (::strncmp(name, OPERATOR_PREFIX, OPERATOR_PREFIX_LENGTH))
Simplified the code that handles overloaded operator names. llvm-svn: 121511
2010-12-11 03:51:54 +08:00
return false;
Modified LLDB expressions to not have to JIT and run code just to see variable values or persistent expression variables. Now if an expression consists of a value that is a child of a variable, or of a persistent variable only, we will create a value object for it and make a ValueObjectConstResult from it to freeze the value (for program variables only, not persistent variables) and avoid running JITed code. For everything else we still parse up and JIT code and run it in the inferior. There was also a lot of clean up in the expression code. I made the ClangExpressionVariables be stored in collections of shared pointers instead of in collections of objects. This will help stop a lot of copy constructors on these large objects and also cleans up the code considerably. The persistent clang expression variables were moved over to the Target to ensure they persist across process executions. Added the ability for lldb_private::Target objects to evaluate expressions. We want to evaluate expressions at the target level in case we aren't running yet, or we have just completed running. We still want to be able to access the persistent expression variables between runs, and also evaluate constant expressions. Added extra logging to the dynamic loader plug-in for MacOSX. ModuleList objects can now dump their contents with the UUID, arch and full paths being logged with appropriate prefix values. Thread hardened the Communication class a bit by making the connection auto_ptr member into a shared pointer member and then making a local copy of the shared pointer in each method that uses it to make sure another thread can't nuke the connection object while it is being used by another thread. Added a new file to the lldb/test/load_unload test that causes the test a.out file to link to the libd.dylib file all the time. This will allow us to test using the DYLD_LIBRARY_PATH environment variable after moving libd.dylib somewhere else. llvm-svn: 121745
2010-12-14 10:59:59 +08:00
post_op_name = name + OPERATOR_PREFIX_LENGTH;
Simplified the code that handles overloaded operator names. llvm-svn: 121511
2010-12-11 03:51:54 +08:00
if (post_op_name[0] == ' ')
Fixed detection of overloaded operators in type information. Previously, "operator[]" was not being recognized because the detector was looking for "operator " as a prefix. Now we handle both cases. llvm-svn: 121456
2010-12-10 10:15:55 +08:00
{
Simplified the code that handles overloaded operator names. llvm-svn: 121511
2010-12-11 03:51:54 +08:00
post_op_name++;
no_space = false;
Fixed detection of overloaded operators in type information. Previously, "operator[]" was not being recognized because the detector was looking for "operator " as a prefix. Now we handle both cases. llvm-svn: 121456
2010-12-10 10:15:55 +08:00
}
Simplified the code that handles overloaded operator names. llvm-svn: 121511
2010-12-11 03:51:54 +08:00
#undef OPERATOR_PREFIX
Modified LLDB expressions to not have to JIT and run code just to see variable values or persistent expression variables. Now if an expression consists of a value that is a child of a variable, or of a persistent variable only, we will create a value object for it and make a ValueObjectConstResult from it to freeze the value (for program variables only, not persistent variables) and avoid running JITed code. For everything else we still parse up and JIT code and run it in the inferior. There was also a lot of clean up in the expression code. I made the ClangExpressionVariables be stored in collections of shared pointers instead of in collections of objects. This will help stop a lot of copy constructors on these large objects and also cleans up the code considerably. The persistent clang expression variables were moved over to the Target to ensure they persist across process executions. Added the ability for lldb_private::Target objects to evaluate expressions. We want to evaluate expressions at the target level in case we aren't running yet, or we have just completed running. We still want to be able to access the persistent expression variables between runs, and also evaluate constant expressions. Added extra logging to the dynamic loader plug-in for MacOSX. ModuleList objects can now dump their contents with the UUID, arch and full paths being logged with appropriate prefix values. Thread hardened the Communication class a bit by making the connection auto_ptr member into a shared pointer member and then making a local copy of the shared pointer in each method that uses it to make sure another thread can't nuke the connection object while it is being used by another thread. Added a new file to the lldb/test/load_unload test that causes the test a.out file to link to the libd.dylib file all the time. This will allow us to test using the DYLD_LIBRARY_PATH environment variable after moving libd.dylib somewhere else. llvm-svn: 121745
2010-12-14 10:59:59 +08:00
#undef OPERATOR_PREFIX_LENGTH
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
// This is an operator, set the overloaded operator kind to invalid
// in case this is a conversion operator...
op_kind = NUM_OVERLOADED_OPERATORS;
switch (post_op_name[0])
{
Fixed detection of overloaded operators in type information. Previously, "operator[]" was not being recognized because the detector was looking for "operator " as a prefix. Now we handle both cases. llvm-svn: 121456
2010-12-10 10:15:55 +08:00
default:
if (no_space)
return false;
break;
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
case 'n':
Fixed detection of overloaded operators in type information. Previously, "operator[]" was not being recognized because the detector was looking for "operator " as a prefix. Now we handle both cases. llvm-svn: 121456
2010-12-10 10:15:55 +08:00
if (no_space)
return false;
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
if (strcmp (post_op_name, "new") == 0)
op_kind = OO_New;
else if (strcmp (post_op_name, "new[]") == 0)
op_kind = OO_Array_New;
break;
case 'd':
Fixed detection of overloaded operators in type information. Previously, "operator[]" was not being recognized because the detector was looking for "operator " as a prefix. Now we handle both cases. llvm-svn: 121456
2010-12-10 10:15:55 +08:00
if (no_space)
return false;
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
if (strcmp (post_op_name, "delete") == 0)
op_kind = OO_Delete;
else if (strcmp (post_op_name, "delete[]") == 0)
op_kind = OO_Array_Delete;
break;
case '+':
if (post_op_name[1] == '\0')
op_kind = OO_Plus;
else if (post_op_name[2] == '\0')
{
if (post_op_name[1] == '=')
op_kind = OO_PlusEqual;
else if (post_op_name[1] == '+')
op_kind = OO_PlusPlus;
}
break;
case '-':
if (post_op_name[1] == '\0')
op_kind = OO_Minus;
else if (post_op_name[2] == '\0')
{
switch (post_op_name[1])
{
case '=': op_kind = OO_MinusEqual; break;
case '-': op_kind = OO_MinusMinus; break;
case '>': op_kind = OO_Arrow; break;
}
}
else if (post_op_name[3] == '\0')
{
if (post_op_name[2] == '*')
op_kind = OO_ArrowStar; break;
}
break;
case '*':
if (post_op_name[1] == '\0')
op_kind = OO_Star;
else if (post_op_name[1] == '=' && post_op_name[2] == '\0')
op_kind = OO_StarEqual;
break;
case '/':
if (post_op_name[1] == '\0')
op_kind = OO_Slash;
else if (post_op_name[1] == '=' && post_op_name[2] == '\0')
op_kind = OO_SlashEqual;
break;
case '%':
if (post_op_name[1] == '\0')
op_kind = OO_Percent;
else if (post_op_name[1] == '=' && post_op_name[2] == '\0')
op_kind = OO_PercentEqual;
break;
case '^':
if (post_op_name[1] == '\0')
op_kind = OO_Caret;
else if (post_op_name[1] == '=' && post_op_name[2] == '\0')
op_kind = OO_CaretEqual;
break;
case '&':
if (post_op_name[1] == '\0')
op_kind = OO_Amp;
else if (post_op_name[2] == '\0')
{
switch (post_op_name[1])
{
case '=': op_kind = OO_AmpEqual; break;
case '&': op_kind = OO_AmpAmp; break;
}
}
break;
case '|':
if (post_op_name[1] == '\0')
op_kind = OO_Pipe;
else if (post_op_name[2] == '\0')
{
switch (post_op_name[1])
{
case '=': op_kind = OO_PipeEqual; break;
case '|': op_kind = OO_PipePipe; break;
}
}
break;
case '~':
if (post_op_name[1] == '\0')
op_kind = OO_Tilde;
break;
case '!':
if (post_op_name[1] == '\0')
op_kind = OO_Exclaim;
else if (post_op_name[1] == '=' && post_op_name[2] == '\0')
op_kind = OO_ExclaimEqual;
break;
case '=':
if (post_op_name[1] == '\0')
op_kind = OO_Equal;
else if (post_op_name[1] == '=' && post_op_name[2] == '\0')
op_kind = OO_EqualEqual;
break;
case '<':
if (post_op_name[1] == '\0')
op_kind = OO_Less;
else if (post_op_name[2] == '\0')
{
switch (post_op_name[1])
{
case '<': op_kind = OO_LessLess; break;
case '=': op_kind = OO_LessEqual; break;
}
}
else if (post_op_name[3] == '\0')
{
if (post_op_name[2] == '=')
op_kind = OO_LessLessEqual;
}
break;
case '>':
if (post_op_name[1] == '\0')
op_kind = OO_Greater;
else if (post_op_name[2] == '\0')
{
switch (post_op_name[1])
{
case '>': op_kind = OO_GreaterGreater; break;
case '=': op_kind = OO_GreaterEqual; break;
}
}
else if (post_op_name[1] == '>' &&
post_op_name[2] == '=' &&
post_op_name[3] == '\0')
{
op_kind = OO_GreaterGreaterEqual;
}
break;
case ',':
if (post_op_name[1] == '\0')
op_kind = OO_Comma;
break;
case '(':
if (post_op_name[1] == ')' && post_op_name[2] == '\0')
op_kind = OO_Call;
break;
case '[':
if (post_op_name[1] == ']' && post_op_name[2] == '\0')
op_kind = OO_Subscript;
break;
}
return true;
}
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
Fixed a case where LLDB would crash if we get C++ operators with invalid operator counts due to bad debug DWARF debug info. We now verify the operator has a valid number of params using the clang operator tables. llvm-svn: 133375
2011-06-19 11:43:27 +08:00
static inline bool
Handled the call operator properly. llvm-svn: 142033
2011-10-15 09:15:07 +08:00
check_op_param (uint32_t op_kind, bool unary, bool binary, uint32_t num_params)
Fixed a case where LLDB would crash if we get C++ operators with invalid operator counts due to bad debug DWARF debug info. We now verify the operator has a valid number of params using the clang operator tables. llvm-svn: 133375
2011-06-19 11:43:27 +08:00
{
Handled the call operator properly. llvm-svn: 142033
2011-10-15 09:15:07 +08:00
// Special-case call since it can take any number of operands
if(op_kind == OO_Call)
return true;
Fixed a case where LLDB would crash if we get C++ operators with invalid operator counts due to bad debug DWARF debug info. We now verify the operator has a valid number of params using the clang operator tables. llvm-svn: 133375
2011-06-19 11:43:27 +08:00
// The parameter count doens't include "this"
if (num_params == 0)
return unary;
if (num_params == 1)
return binary;
Handled the call operator properly. llvm-svn: 142033
2011-10-15 09:15:07 +08:00
else
Fixed a case where LLDB would crash if we get C++ operators with invalid operator counts due to bad debug DWARF debug info. We now verify the operator has a valid number of params using the clang operator tables. llvm-svn: 133375
2011-06-19 11:43:27 +08:00
return false;
}
warnings: Get rid of spurious semicolon. llvm-svn: 143384
2011-11-01 06:50:57 +08:00
Fixed a case where LLDB would crash if we get C++ operators with invalid operator counts due to bad debug DWARF debug info. We now verify the operator has a valid number of params using the clang operator tables. llvm-svn: 133375
2011-06-19 11:43:27 +08:00
bool
ClangASTContext::CheckOverloadedOperatorKindParameterCount (uint32_t op_kind, uint32_t num_params)
{
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
switch (op_kind)
{
default:
break;
// C++ standard allows any number of arguments to new/delete
case OO_New:
case OO_Array_New:
case OO_Delete:
case OO_Array_Delete:
return true;
}
Handled the call operator properly. llvm-svn: 142033
2011-10-15 09:15:07 +08:00
#define OVERLOADED_OPERATOR(Name,Spelling,Token,Unary,Binary,MemberOnly) case OO_##Name: return check_op_param (op_kind, Unary, Binary, num_params);
Fixed a case where LLDB would crash if we get C++ operators with invalid operator counts due to bad debug DWARF debug info. We now verify the operator has a valid number of params using the clang operator tables. llvm-svn: 133375
2011-06-19 11:43:27 +08:00
switch (op_kind)
{
#include "clang/Basic/OperatorKinds.def"
default: break;
}
return false;
}
Added motheds to C++ classes as we parse them to keep clang happy. llvm-svn: 114616
2010-09-23 09:09:21 +08:00
CXXMethodDecl *
Re-committed AddMethodToCXXRecordType, now that the bug I introduced to ClangASTContext is resolved. llvm-svn: 114157
2010-09-17 10:58:26 +08:00
ClangASTContext::AddMethodToCXXRecordType
(
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t record_opaque_type,
Added motheds to C++ classes as we parse them to keep clang happy. llvm-svn: 114616
2010-09-23 09:09:21 +08:00
const char *name,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t method_opaque_type,
Added motheds to C++ classes as we parse them to keep clang happy. llvm-svn: 114616
2010-09-23 09:09:21 +08:00
lldb::AccessType access,
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
bool is_virtual,
bool is_static,
Make C++ constructors and destructors correctly within the clang types we generate from DWARF. llvm-svn: 115268
2010-10-01 10:31:07 +08:00
bool is_inline,
Added the capability (turned off for now) to mark a method as __attribute__ ((used)) when adding it to a class. This functionality is useful when stopped in anonymous namespaces: expressions attached to classes in anonymous namespaces are typically elided by Clang's CodeGen because they have no namespaces are intended not to be externally visible. __attribute__ ((used)) forces CodeGen to emit the function. Right now, __attribute__ ((used)) causes the JIT not to emit the function, so we're not enabling it until we fix that. llvm-svn: 143469
2011-11-02 02:07:13 +08:00
bool is_explicit,
Sometimes the debug information includes artifically- generated special member functions (constructors, destructors, etc.) for classes that don't really have them. We needed to mark these as artificial to reflect the debug information; this bug does that for constructors and destructors. The "etc." case (certain assignment operators, mostly) remains to be fixed. llvm-svn: 143526
2011-11-02 09:38:59 +08:00
bool is_attr_used,
bool is_artificial
Added motheds to C++ classes as we parse them to keep clang happy. llvm-svn: 114616
2010-09-23 09:09:21 +08:00
)
Re-committed AddMethodToCXXRecordType, now that the bug I introduced to ClangASTContext is resolved. llvm-svn: 114157
2010-09-17 10:58:26 +08:00
{
Removed the hacky "#define this ___clang_this" handler for C++ classes. Replaced it with a less hacky approach: - If an expression is defined in the context of a method of class A, then that expression is wrapped as ___clang_class::___clang_expr(void*) { ... } instead of ___clang_expr(void*) { ... }. - ___clang_class is resolved as the type of the target of the "this" pointer in the method the expression is defined in. - When reporting the type of ___clang_class, a method with the signature ___clang_expr(void*) is added to that class, so that Clang doesn't complain about a method being defined without a corresponding declaration. - Whenever the expression gets called, "this" gets looked up, type-checked, and then passed in as the first argument. This required the following changes: - The ABIs were changed to support passing of the "this" pointer as part of trivial calls. - ThreadPlanCallFunction and ClangFunction were changed to support passing of an optional "this" pointer. - ClangUserExpression was extended to perform the wrapping described above. - ClangASTSource was changed to revert the changes required by the hack. - ClangExpressionParser, IRForTarget, and ClangExpressionDeclMap were changed to handle different manglings of ___clang_expr flexibly. This meant no longer searching for a function called ___clang_expr, but rather looking for a function whose name *contains* ___clang_expr. - ClangExpressionParser and ClangExpressionDeclMap now remember whether "this" is required, and know how to look it up as necessary. A few inheritance bugs remain, and I'm trying to resolve these. But it is now possible to use "this" as well as refer implicitly to member variables, when in the proper context. llvm-svn: 114384
2010-09-21 08:44:12 +08:00
if (!record_opaque_type || !method_opaque_type || !name)
Fix from Jean-Daniel. Thanks. Error in object runtime language detection code (spurious '; ') Also replace false by NULL in a place where the compiler expects a pointer instead of a bool. llvm-svn: 114957
2010-09-29 00:10:54 +08:00
return NULL;
Re-committed AddMethodToCXXRecordType, now that the bug I introduced to ClangASTContext is resolved. llvm-svn: 114157
2010-09-17 10:58:26 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
assert(ast);
Re-committed AddMethodToCXXRecordType, now that the bug I introduced to ClangASTContext is resolved. llvm-svn: 114157
2010-09-17 10:58:26 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
IdentifierTable *identifier_table = &ast->Idents;
Re-committed AddMethodToCXXRecordType, now that the bug I introduced to ClangASTContext is resolved. llvm-svn: 114157
2010-09-17 10:58:26 +08:00
assert(identifier_table);
Removed the hacky "#define this ___clang_this" handler for C++ classes. Replaced it with a less hacky approach: - If an expression is defined in the context of a method of class A, then that expression is wrapped as ___clang_class::___clang_expr(void*) { ... } instead of ___clang_expr(void*) { ... }. - ___clang_class is resolved as the type of the target of the "this" pointer in the method the expression is defined in. - When reporting the type of ___clang_class, a method with the signature ___clang_expr(void*) is added to that class, so that Clang doesn't complain about a method being defined without a corresponding declaration. - Whenever the expression gets called, "this" gets looked up, type-checked, and then passed in as the first argument. This required the following changes: - The ABIs were changed to support passing of the "this" pointer as part of trivial calls. - ThreadPlanCallFunction and ClangFunction were changed to support passing of an optional "this" pointer. - ClangUserExpression was extended to perform the wrapping described above. - ClangASTSource was changed to revert the changes required by the hack. - ClangExpressionParser, IRForTarget, and ClangExpressionDeclMap were changed to handle different manglings of ___clang_expr flexibly. This meant no longer searching for a function called ___clang_expr, but rather looking for a function whose name *contains* ___clang_expr. - ClangExpressionParser and ClangExpressionDeclMap now remember whether "this" is required, and know how to look it up as necessary. A few inheritance bugs remain, and I'm trying to resolve these. But it is now possible to use "this" as well as refer implicitly to member variables, when in the proper context. llvm-svn: 114384
2010-09-21 08:44:12 +08:00
QualType record_qual_type(QualType::getFromOpaquePtr(record_opaque_type));
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
CXXRecordDecl *cxx_record_decl = record_qual_type->getAsCXXRecordDecl();
Re-committed AddMethodToCXXRecordType, now that the bug I introduced to ClangASTContext is resolved. llvm-svn: 114157
2010-09-17 10:58:26 +08:00
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
if (cxx_record_decl == NULL)
Added motheds to C++ classes as we parse them to keep clang happy. llvm-svn: 114616
2010-09-23 09:09:21 +08:00
return NULL;
Re-committed AddMethodToCXXRecordType, now that the bug I introduced to ClangASTContext is resolved. llvm-svn: 114157
2010-09-17 10:58:26 +08:00
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
QualType method_qual_type (QualType::getFromOpaquePtr (method_opaque_type));
Make C++ constructors and destructors correctly within the clang types we generate from DWARF. llvm-svn: 115268
2010-10-01 10:31:07 +08:00
CXXMethodDecl *cxx_method_decl = NULL;
Removed the hacky "#define this ___clang_this" handler for C++ classes. Replaced it with a less hacky approach: - If an expression is defined in the context of a method of class A, then that expression is wrapped as ___clang_class::___clang_expr(void*) { ... } instead of ___clang_expr(void*) { ... }. - ___clang_class is resolved as the type of the target of the "this" pointer in the method the expression is defined in. - When reporting the type of ___clang_class, a method with the signature ___clang_expr(void*) is added to that class, so that Clang doesn't complain about a method being defined without a corresponding declaration. - Whenever the expression gets called, "this" gets looked up, type-checked, and then passed in as the first argument. This required the following changes: - The ABIs were changed to support passing of the "this" pointer as part of trivial calls. - ThreadPlanCallFunction and ClangFunction were changed to support passing of an optional "this" pointer. - ClangUserExpression was extended to perform the wrapping described above. - ClangASTSource was changed to revert the changes required by the hack. - ClangExpressionParser, IRForTarget, and ClangExpressionDeclMap were changed to handle different manglings of ___clang_expr flexibly. This meant no longer searching for a function called ___clang_expr, but rather looking for a function whose name *contains* ___clang_expr. - ClangExpressionParser and ClangExpressionDeclMap now remember whether "this" is required, and know how to look it up as necessary. A few inheritance bugs remain, and I'm trying to resolve these. But it is now possible to use "this" as well as refer implicitly to member variables, when in the proper context. llvm-svn: 114384
2010-09-21 08:44:12 +08:00
Make C++ constructors and destructors correctly within the clang types we generate from DWARF. llvm-svn: 115268
2010-10-01 10:31:07 +08:00
DeclarationName decl_name (&identifier_table->get(name));
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::FunctionType *function_Type = dyn_cast<FunctionType>(method_qual_type.getTypePtr());
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
One final patch for all of this class constructor, destructor, and conversion operator naming stuff. We now get the constructor and destructor names right after passing in the type, and we get the correct conversion operator name after passing in the return type when getting the DeclarationNameInfo. llvm-svn: 115398
2010-10-02 09:40:05 +08:00
if (function_Type == NULL)
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
return NULL;
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const FunctionProtoType *method_function_prototype (dyn_cast<FunctionProtoType>(function_Type));
Make C++ constructors and destructors correctly within the clang types we generate from DWARF. llvm-svn: 115268
2010-10-01 10:31:07 +08:00
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
if (!method_function_prototype)
return NULL;
unsigned int num_params = method_function_prototype->getNumArgs();
Sometimes the debug information includes artifically- generated special member functions (constructors, destructors, etc.) for classes that don't really have them. We needed to mark these as artificial to reflect the debug information; this bug does that for constructors and destructors. The "etc." case (certain assignment operators, mostly) remains to be fixed. llvm-svn: 143526
2011-11-02 09:38:59 +08:00
CXXDestructorDecl *cxx_dtor_decl(NULL);
CXXConstructorDecl *cxx_ctor_decl(NULL);
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
if (name[0] == '~')
{
Sometimes the debug information includes artifically- generated special member functions (constructors, destructors, etc.) for classes that don't really have them. We needed to mark these as artificial to reflect the debug information; this bug does that for constructors and destructors. The "etc." case (certain assignment operators, mostly) remains to be fixed. llvm-svn: 143526
2011-11-02 09:38:59 +08:00
cxx_dtor_decl = CXXDestructorDecl::Create (*ast,
cxx_record_decl,
SourceLocation(),
DeclarationNameInfo (ast->DeclarationNames.getCXXDestructorName (ast->getCanonicalType (record_qual_type)), SourceLocation()),
method_qual_type,
NULL,
is_inline,
is_artificial);
cxx_method_decl = cxx_dtor_decl;
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
}
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
else if (decl_name == cxx_record_decl->getDeclName())
Make C++ constructors and destructors correctly within the clang types we generate from DWARF. llvm-svn: 115268
2010-10-01 10:31:07 +08:00
{
Sometimes the debug information includes artifically- generated special member functions (constructors, destructors, etc.) for classes that don't really have them. We needed to mark these as artificial to reflect the debug information; this bug does that for constructors and destructors. The "etc." case (certain assignment operators, mostly) remains to be fixed. llvm-svn: 143526
2011-11-02 09:38:59 +08:00
cxx_ctor_decl = CXXConstructorDecl::Create (*ast,
cxx_record_decl,
SourceLocation(),
DeclarationNameInfo (ast->DeclarationNames.getCXXConstructorName (ast->getCanonicalType (record_qual_type)), SourceLocation()),
method_qual_type,
NULL, // TypeSourceInfo *
is_explicit,
is_inline,
is_artificial,
false /*is_constexpr*/);
cxx_method_decl = cxx_ctor_decl;
Make C++ constructors and destructors correctly within the clang types we generate from DWARF. llvm-svn: 115268
2010-10-01 10:31:07 +08:00
}
else
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
{
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
OverloadedOperatorKind op_kind = NUM_OVERLOADED_OPERATORS;
if (IsOperator (name, op_kind))
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
{
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
if (op_kind != NUM_OVERLOADED_OPERATORS)
{
Fixed a case where LLDB would crash if we get C++ operators with invalid operator counts due to bad debug DWARF debug info. We now verify the operator has a valid number of params using the clang operator tables. llvm-svn: 133375
2011-06-19 11:43:27 +08:00
// Check the number of operator parameters. Sometimes we have
// seen bad DWARF that doesn't correctly describe operators and
// if we try to create a methed and add it to the class, clang
// will assert and crash, so we need to make sure things are
// acceptable.
if (!ClangASTContext::CheckOverloadedOperatorKindParameterCount (op_kind, num_params))
return NULL;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
cxx_method_decl = CXXMethodDecl::Create (*ast,
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
cxx_record_decl,
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
SourceLocation(),
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
DeclarationNameInfo (ast->DeclarationNames.getCXXOperatorName (op_kind), SourceLocation()),
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
method_qual_type,
NULL, // TypeSourceInfo *
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
is_static,
SC_None,
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
is_inline,
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
false /*is_constexpr*/,
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
SourceLocation());
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
}
else if (num_params == 0)
{
// Conversion operators don't take params...
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
cxx_method_decl = CXXConversionDecl::Create (*ast,
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
cxx_record_decl,
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
SourceLocation(),
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
DeclarationNameInfo (ast->DeclarationNames.getCXXConversionFunctionName (ast->getCanonicalType (function_Type->getResultType())), SourceLocation()),
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
method_qual_type,
NULL, // TypeSourceInfo *
is_inline,
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
is_explicit,
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
false /*is_constexpr*/,
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
SourceLocation());
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
}
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
}
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
if (cxx_method_decl == NULL)
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
cxx_method_decl = CXXMethodDecl::Create (*ast,
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
cxx_record_decl,
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
SourceLocation(),
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
DeclarationNameInfo (decl_name, SourceLocation()),
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
method_qual_type,
NULL, // TypeSourceInfo *
is_static,
SC_None,
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
is_inline,
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
false /*is_constexpr*/,
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
SourceLocation());
Added correct C++ method declarations for destructors and for conversion operators. llvm-svn: 115290
2010-10-01 11:45:20 +08:00
}
Make C++ constructors and destructors correctly within the clang types we generate from DWARF. llvm-svn: 115268
2010-10-01 10:31:07 +08:00
}
Get the declaration names for C++ ctors, dtors, and operators from the right name pools. llvm-svn: 115371
2010-10-02 07:13:49 +08:00
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
AccessSpecifier access_specifier = ConvertAccessTypeToAccessSpecifier (access);
Updated to latest LLVM. Major LLVM changes: - Sema is now exported (and there was much rejoicing.) - Storage classes are now centrally defined. Also fixed some bugs that the new LLVM picked up. llvm-svn: 114622
2010-09-23 11:01:22 +08:00
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
cxx_method_decl->setAccess (access_specifier);
cxx_method_decl->setVirtualAsWritten (is_virtual);
Updated to latest LLVM. Major LLVM changes: - Sema is now exported (and there was much rejoicing.) - Storage classes are now centrally defined. Also fixed some bugs that the new LLVM picked up. llvm-svn: 114622
2010-09-23 11:01:22 +08:00
Added the capability (turned off for now) to mark a method as __attribute__ ((used)) when adding it to a class. This functionality is useful when stopped in anonymous namespaces: expressions attached to classes in anonymous namespaces are typically elided by Clang's CodeGen because they have no namespaces are intended not to be externally visible. __attribute__ ((used)) forces CodeGen to emit the function. Right now, __attribute__ ((used)) causes the JIT not to emit the function, so we're not enabling it until we fix that. llvm-svn: 143469
2011-11-02 02:07:13 +08:00
if (is_attr_used)
cxx_method_decl->addAttr(::new (*ast) UsedAttr(SourceRange(), *ast));
Removed the hacky "#define this ___clang_this" handler for C++ classes. Replaced it with a less hacky approach: - If an expression is defined in the context of a method of class A, then that expression is wrapped as ___clang_class::___clang_expr(void*) { ... } instead of ___clang_expr(void*) { ... }. - ___clang_class is resolved as the type of the target of the "this" pointer in the method the expression is defined in. - When reporting the type of ___clang_class, a method with the signature ___clang_expr(void*) is added to that class, so that Clang doesn't complain about a method being defined without a corresponding declaration. - Whenever the expression gets called, "this" gets looked up, type-checked, and then passed in as the first argument. This required the following changes: - The ABIs were changed to support passing of the "this" pointer as part of trivial calls. - ThreadPlanCallFunction and ClangFunction were changed to support passing of an optional "this" pointer. - ClangUserExpression was extended to perform the wrapping described above. - ClangASTSource was changed to revert the changes required by the hack. - ClangExpressionParser, IRForTarget, and ClangExpressionDeclMap were changed to handle different manglings of ___clang_expr flexibly. This meant no longer searching for a function called ___clang_expr, but rather looking for a function whose name *contains* ___clang_expr. - ClangExpressionParser and ClangExpressionDeclMap now remember whether "this" is required, and know how to look it up as necessary. A few inheritance bugs remain, and I'm trying to resolve these. But it is now possible to use "this" as well as refer implicitly to member variables, when in the proper context. llvm-svn: 114384
2010-09-21 08:44:12 +08:00
// Populate the method decl with parameter decls
Use a SmallVector here instead of a VLA. llvm-svn: 131698
2011-05-20 07:33:46 +08:00
llvm::SmallVector<ParmVarDecl *, 12> params;
Removed the hacky "#define this ___clang_this" handler for C++ classes. Replaced it with a less hacky approach: - If an expression is defined in the context of a method of class A, then that expression is wrapped as ___clang_class::___clang_expr(void*) { ... } instead of ___clang_expr(void*) { ... }. - ___clang_class is resolved as the type of the target of the "this" pointer in the method the expression is defined in. - When reporting the type of ___clang_class, a method with the signature ___clang_expr(void*) is added to that class, so that Clang doesn't complain about a method being defined without a corresponding declaration. - Whenever the expression gets called, "this" gets looked up, type-checked, and then passed in as the first argument. This required the following changes: - The ABIs were changed to support passing of the "this" pointer as part of trivial calls. - ThreadPlanCallFunction and ClangFunction were changed to support passing of an optional "this" pointer. - ClangUserExpression was extended to perform the wrapping described above. - ClangASTSource was changed to revert the changes required by the hack. - ClangExpressionParser, IRForTarget, and ClangExpressionDeclMap were changed to handle different manglings of ___clang_expr flexibly. This meant no longer searching for a function called ___clang_expr, but rather looking for a function whose name *contains* ___clang_expr. - ClangExpressionParser and ClangExpressionDeclMap now remember whether "this" is required, and know how to look it up as necessary. A few inheritance bugs remain, and I'm trying to resolve these. But it is now possible to use "this" as well as refer implicitly to member variables, when in the proper context. llvm-svn: 114384
2010-09-21 08:44:12 +08:00
for (int param_index = 0;
param_index < num_params;
++param_index)
{
Use a SmallVector here instead of a VLA. llvm-svn: 131698
2011-05-20 07:33:46 +08:00
params.push_back (ParmVarDecl::Create (*ast,
cxx_method_decl,
SourceLocation(),
SourceLocation(),
NULL, // anonymous
method_function_prototype->getArgType(param_index),
NULL,
SC_None,
SC_None,
NULL));
Removed the hacky "#define this ___clang_this" handler for C++ classes. Replaced it with a less hacky approach: - If an expression is defined in the context of a method of class A, then that expression is wrapped as ___clang_class::___clang_expr(void*) { ... } instead of ___clang_expr(void*) { ... }. - ___clang_class is resolved as the type of the target of the "this" pointer in the method the expression is defined in. - When reporting the type of ___clang_class, a method with the signature ___clang_expr(void*) is added to that class, so that Clang doesn't complain about a method being defined without a corresponding declaration. - Whenever the expression gets called, "this" gets looked up, type-checked, and then passed in as the first argument. This required the following changes: - The ABIs were changed to support passing of the "this" pointer as part of trivial calls. - ThreadPlanCallFunction and ClangFunction were changed to support passing of an optional "this" pointer. - ClangUserExpression was extended to perform the wrapping described above. - ClangASTSource was changed to revert the changes required by the hack. - ClangExpressionParser, IRForTarget, and ClangExpressionDeclMap were changed to handle different manglings of ___clang_expr flexibly. This meant no longer searching for a function called ___clang_expr, but rather looking for a function whose name *contains* ___clang_expr. - ClangExpressionParser and ClangExpressionDeclMap now remember whether "this" is required, and know how to look it up as necessary. A few inheritance bugs remain, and I'm trying to resolve these. But it is now possible to use "this" as well as refer implicitly to member variables, when in the proper context. llvm-svn: 114384
2010-09-21 08:44:12 +08:00
}
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
cxx_method_decl->setParams (ArrayRef<ParmVarDecl*>(params));
Removed the hacky "#define this ___clang_this" handler for C++ classes. Replaced it with a less hacky approach: - If an expression is defined in the context of a method of class A, then that expression is wrapped as ___clang_class::___clang_expr(void*) { ... } instead of ___clang_expr(void*) { ... }. - ___clang_class is resolved as the type of the target of the "this" pointer in the method the expression is defined in. - When reporting the type of ___clang_class, a method with the signature ___clang_expr(void*) is added to that class, so that Clang doesn't complain about a method being defined without a corresponding declaration. - Whenever the expression gets called, "this" gets looked up, type-checked, and then passed in as the first argument. This required the following changes: - The ABIs were changed to support passing of the "this" pointer as part of trivial calls. - ThreadPlanCallFunction and ClangFunction were changed to support passing of an optional "this" pointer. - ClangUserExpression was extended to perform the wrapping described above. - ClangASTSource was changed to revert the changes required by the hack. - ClangExpressionParser, IRForTarget, and ClangExpressionDeclMap were changed to handle different manglings of ___clang_expr flexibly. This meant no longer searching for a function called ___clang_expr, but rather looking for a function whose name *contains* ___clang_expr. - ClangExpressionParser and ClangExpressionDeclMap now remember whether "this" is required, and know how to look it up as necessary. A few inheritance bugs remain, and I'm trying to resolve these. But it is now possible to use "this" as well as refer implicitly to member variables, when in the proper context. llvm-svn: 114384
2010-09-21 08:44:12 +08:00
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
cxx_record_decl->addDecl (cxx_method_decl);
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
<rdar://problem/10020849> Fixed an issue where the DWARF might mention that a class has a constructor (default, copy or move), destructor, or an assignment operator (copy or move) and it might not have an actual implementation in your code. Then you try and use this struct or class in an expression and the JIT would ask for the address of these methods that were in the declaration, yet there are none. We now "do the right thing" for trivial ctors, dtors and assignment operators by telling the methods that they are are defaulted and trivial, and clang will then just do all of the work with builtins! llvm-svn: 143528
2011-11-02 10:06:20 +08:00
// Sometimes the debug info will mention a constructor (default/copy/move),
// destructor, or assignment operator (copy/move) but there won't be any
// version of this in the code. So we check if the function was artificially
// generated and if it is trivial and this lets the compiler/backend know
// that it can inline the IR for these when it needs to and we can avoid a
// "missing function" error when running expressions.
Sometimes the debug information includes artifically- generated special member functions (constructors, destructors, etc.) for classes that don't really have them. We needed to mark these as artificial to reflect the debug information; this bug does that for constructors and destructors. The "etc." case (certain assignment operators, mostly) remains to be fixed. llvm-svn: 143526
2011-11-02 09:38:59 +08:00
if (is_artificial)
{
<rdar://problem/10020849> Fixed an issue where the DWARF might mention that a class has a constructor (default, copy or move), destructor, or an assignment operator (copy or move) and it might not have an actual implementation in your code. Then you try and use this struct or class in an expression and the JIT would ask for the address of these methods that were in the declaration, yet there are none. We now "do the right thing" for trivial ctors, dtors and assignment operators by telling the methods that they are are defaulted and trivial, and clang will then just do all of the work with builtins! llvm-svn: 143528
2011-11-02 10:06:20 +08:00
if (cxx_ctor_decl &&
((cxx_ctor_decl->isDefaultConstructor() && cxx_record_decl->hasTrivialDefaultConstructor ()) ||
(cxx_ctor_decl->isCopyConstructor() && cxx_record_decl->hasTrivialCopyConstructor ()) ||
(cxx_ctor_decl->isMoveConstructor() && cxx_record_decl->hasTrivialMoveConstructor ()) ))
Sometimes the debug information includes artifically- generated special member functions (constructors, destructors, etc.) for classes that don't really have them. We needed to mark these as artificial to reflect the debug information; this bug does that for constructors and destructors. The "etc." case (certain assignment operators, mostly) remains to be fixed. llvm-svn: 143526
2011-11-02 09:38:59 +08:00
{
cxx_ctor_decl->setDefaulted();
cxx_ctor_decl->setTrivial(true);
}
<rdar://problem/10020849> Fixed an issue where the DWARF might mention that a class has a constructor (default, copy or move), destructor, or an assignment operator (copy or move) and it might not have an actual implementation in your code. Then you try and use this struct or class in an expression and the JIT would ask for the address of these methods that were in the declaration, yet there are none. We now "do the right thing" for trivial ctors, dtors and assignment operators by telling the methods that they are are defaulted and trivial, and clang will then just do all of the work with builtins! llvm-svn: 143528
2011-11-02 10:06:20 +08:00
else if (cxx_dtor_decl)
{
if (cxx_record_decl->hasTrivialDestructor())
{
cxx_dtor_decl->setDefaulted();
cxx_dtor_decl->setTrivial(true);
}
}
else if ((cxx_method_decl->isCopyAssignmentOperator() && cxx_record_decl->hasTrivialCopyAssignment()) ||
(cxx_method_decl->isMoveAssignmentOperator() && cxx_record_decl->hasTrivialMoveAssignment()))
Sometimes the debug information includes artifically- generated special member functions (constructors, destructors, etc.) for classes that don't really have them. We needed to mark these as artificial to reflect the debug information; this bug does that for constructors and destructors. The "etc." case (certain assignment operators, mostly) remains to be fixed. llvm-svn: 143526
2011-11-02 09:38:59 +08:00
{
<rdar://problem/10020849> Fixed an issue where the DWARF might mention that a class has a constructor (default, copy or move), destructor, or an assignment operator (copy or move) and it might not have an actual implementation in your code. Then you try and use this struct or class in an expression and the JIT would ask for the address of these methods that were in the declaration, yet there are none. We now "do the right thing" for trivial ctors, dtors and assignment operators by telling the methods that they are are defaulted and trivial, and clang will then just do all of the work with builtins! llvm-svn: 143528
2011-11-02 10:06:20 +08:00
cxx_method_decl->setDefaulted();
cxx_method_decl->setTrivial(true);
Sometimes the debug information includes artifically- generated special member functions (constructors, destructors, etc.) for classes that don't really have them. We needed to mark these as artificial to reflect the debug information; this bug does that for constructors and destructors. The "etc." case (certain assignment operators, mostly) remains to be fixed. llvm-svn: 143526
2011-11-02 09:38:59 +08:00
}
}
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
#ifdef LLDB_CONFIGURATION_DEBUG
VerifyDecl(cxx_method_decl);
#endif
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
// printf ("decl->isPolymorphic() = %i\n", cxx_record_decl->isPolymorphic());
// printf ("decl->isAggregate() = %i\n", cxx_record_decl->isAggregate());
// printf ("decl->isPOD() = %i\n", cxx_record_decl->isPOD());
// printf ("decl->isEmpty() = %i\n", cxx_record_decl->isEmpty());
// printf ("decl->isAbstract() = %i\n", cxx_record_decl->isAbstract());
// printf ("decl->hasTrivialConstructor() = %i\n", cxx_record_decl->hasTrivialConstructor());
// printf ("decl->hasTrivialCopyConstructor() = %i\n", cxx_record_decl->hasTrivialCopyConstructor());
// printf ("decl->hasTrivialCopyAssignment() = %i\n", cxx_record_decl->hasTrivialCopyAssignment());
// printf ("decl->hasTrivialDestructor() = %i\n", cxx_record_decl->hasTrivialDestructor());
Added motheds to C++ classes as we parse them to keep clang happy. llvm-svn: 114616
2010-09-23 09:09:21 +08:00
return cxx_method_decl;
Re-committed AddMethodToCXXRecordType, now that the bug I introduced to ClangASTContext is resolved. llvm-svn: 114157
2010-09-17 10:58:26 +08:00
}
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
clang::FieldDecl *
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
ClangASTContext::AddFieldToRecordType
(
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t record_clang_type,
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
const char *name,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t field_type,
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
AccessType access,
uint32_t bitfield_bit_size
)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (record_clang_type == NULL || field_type == NULL)
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
return NULL;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
FieldDecl *field = NULL;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
IdentifierTable *identifier_table = &ast->Idents;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
assert (ast != NULL);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
assert (identifier_table != NULL);
QualType record_qual_type(QualType::getFromOpaquePtr(record_clang_type));
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::Type *clang_type = record_qual_type.getTypePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (clang_type)
{
const RecordType *record_type = dyn_cast<RecordType>(clang_type);
if (record_type)
{
RecordDecl *record_decl = record_type->getDecl();
clang::Expr *bit_width = NULL;
if (bitfield_bit_size != 0)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
APInt bitfield_bit_size_apint(ast->getTypeSize(ast->IntTy), bitfield_bit_size);
bit_width = new (*ast)IntegerLiteral (*ast, bitfield_bit_size_apint, ast->IntTy, SourceLocation());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
field = FieldDecl::Create (*ast,
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
record_decl,
SourceLocation(),
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
SourceLocation(),
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
name ? &identifier_table->get(name) : NULL, // Identifier
QualType::getFromOpaquePtr(field_type), // Field type
This change brings in the latest LLVM/Clang, and completes the support in the LLDB expression parser for incomplete types. Clang now imports types lazily, and we complete those types as necessary. Changes include: - ClangASTSource now supports three APIs which it passes to ClangExpressionDeclMap. CompleteType completes a TagDecl or an ObjCInterfaceDecl when needed; FindExternalVisibleDecls finds named entities that are visible in the expression's scope; and FindExternalLexicalDecls performs a (potentially restricted) search for entities inside a lexical scope like a namespace. These changes mean that entities in namespaces should work normally. - The SymbolFileDWARF code for searching a context for a specific name is now more general, and can search arbitrary contexts. - We are continuing to adapt our calls into LLVM from interfaces that take start and end iterators when accepting multiple items to interfaces that use ArrayRef. - I have cleaned up some code, especially our use of namespaces. This change is neutral for our testsuite and greatly improves correctness for large programs (like Clang) with complicated type systems. It should also lay the groundwork for improving the expression parser's performance as we are lazier and lazier about providing type information. llvm-svn: 136555
2011-07-30 10:42:06 +08:00
NULL, // TInfo *
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
bit_width, // BitWidth
This change brings in the latest LLVM/Clang, and completes the support in the LLDB expression parser for incomplete types. Clang now imports types lazily, and we complete those types as necessary. Changes include: - ClangASTSource now supports three APIs which it passes to ClangExpressionDeclMap. CompleteType completes a TagDecl or an ObjCInterfaceDecl when needed; FindExternalVisibleDecls finds named entities that are visible in the expression's scope; and FindExternalLexicalDecls performs a (potentially restricted) search for entities inside a lexical scope like a namespace. These changes mean that entities in namespaces should work normally. - The SymbolFileDWARF code for searching a context for a specific name is now more general, and can search arbitrary contexts. - We are continuing to adapt our calls into LLVM from interfaces that take start and end iterators when accepting multiple items to interfaces that use ArrayRef. - I have cleaned up some code, especially our use of namespaces. This change is neutral for our testsuite and greatly improves correctness for large programs (like Clang) with complicated type systems. It should also lay the groundwork for improving the expression parser's performance as we are lazier and lazier about providing type information. llvm-svn: 136555
2011-07-30 10:42:06 +08:00
false, // Mutable
false); // HasInit
Improved the type's handling of anonymous structs, so that the expression parser can look up members of anonymous structs correctly. This meant creating all the proper IndirectFieldDecls in each Record after it has been completely populated with members. llvm-svn: 151868
2012-03-02 09:03:45 +08:00
if (!name)
field->setImplicit();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
field->setAccess (ConvertAccessTypeToAccessSpecifier (access));
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (field)
{
record_decl->addDecl(field);
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
#ifdef LLDB_CONFIGURATION_DEBUG
VerifyDecl(field);
#endif
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
}
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
else
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(clang_type);
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
if (objc_class_type)
{
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
bool is_synthesized = false;
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
field = ClangASTContext::AddObjCClassIVar (ast,
Made AddFieldToRecordType a static method on ClangASTContext. llvm-svn: 114110
2010-09-17 04:01:08 +08:00
record_clang_type,
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
name,
field_type,
access,
bitfield_bit_size,
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
is_synthesized);
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
}
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
return field;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Improved the type's handling of anonymous structs, so that the expression parser can look up members of anonymous structs correctly. This meant creating all the proper IndirectFieldDecls in each Record after it has been completely populated with members. llvm-svn: 151868
2012-03-02 09:03:45 +08:00
static clang::AccessSpecifier UnifyAccessSpecifiers (clang::AccessSpecifier lhs,
clang::AccessSpecifier rhs)
{
clang::AccessSpecifier ret = lhs;
// Make the access equal to the stricter of the field and the nested field's access
switch (ret)
{
case clang::AS_none:
break;
case clang::AS_private:
break;
case clang::AS_protected:
if (rhs == AS_private)
ret = AS_private;
break;
case clang::AS_public:
ret = rhs;
break;
}
return ret;
}
void
ClangASTContext::BuildIndirectFields (clang::ASTContext *ast,
lldb::clang_type_t record_clang_type)
{
QualType record_qual_type(QualType::getFromOpaquePtr(record_clang_type));
const RecordType *record_type = record_qual_type->getAs<RecordType>();
if (!record_type)
return;
RecordDecl *record_decl = record_type->getDecl();
if (!record_decl)
return;
typedef llvm::SmallVector <IndirectFieldDecl *, 1> IndirectFieldVector;
IndirectFieldVector indirect_fields;
for (RecordDecl::field_iterator fi = record_decl->field_begin(), fe = record_decl->field_end();
fi != fe;
++fi)
{
if (fi->isAnonymousStructOrUnion())
{
QualType field_qual_type = fi->getType();
const RecordType *field_record_type = field_qual_type->getAs<RecordType>();
if (!field_record_type)
continue;
RecordDecl *field_record_decl = field_record_type->getDecl();
if (!field_record_decl)
continue;
for (RecordDecl::decl_iterator di = field_record_decl->decls_begin(), de = field_record_decl->decls_end();
di != de;
++di)
{
if (FieldDecl *nested_field_decl = dyn_cast<FieldDecl>(*di))
{
NamedDecl **chain = new (*ast) NamedDecl*[2];
chain[0] = *fi;
chain[1] = nested_field_decl;
IndirectFieldDecl *indirect_field = IndirectFieldDecl::Create(*ast,
record_decl,
SourceLocation(),
nested_field_decl->getIdentifier(),
nested_field_decl->getType(),
chain,
2);
indirect_field->setAccess(UnifyAccessSpecifiers(fi->getAccess(),
nested_field_decl->getAccess()));
indirect_fields.push_back(indirect_field);
}
else if (IndirectFieldDecl *nested_indirect_field_decl = dyn_cast<IndirectFieldDecl>(*di))
{
int nested_chain_size = nested_indirect_field_decl->getChainingSize();
NamedDecl **chain = new (*ast) NamedDecl*[nested_chain_size + 1];
chain[0] = *fi;
int chain_index = 1;
for (IndirectFieldDecl::chain_iterator nci = nested_indirect_field_decl->chain_begin(),
nce = nested_indirect_field_decl->chain_end();
nci < nce;
++nci)
{
chain[chain_index] = *nci;
chain_index++;
}
IndirectFieldDecl *indirect_field = IndirectFieldDecl::Create(*ast,
record_decl,
SourceLocation(),
nested_indirect_field_decl->getIdentifier(),
nested_indirect_field_decl->getType(),
chain,
nested_chain_size + 1);
indirect_field->setAccess(UnifyAccessSpecifiers(fi->getAccess(),
nested_indirect_field_decl->getAccess()));
indirect_fields.push_back(indirect_field);
}
}
}
}
for (IndirectFieldVector::iterator ifi = indirect_fields.begin(), ife = indirect_fields.end();
ifi < ife;
++ifi)
{
record_decl->addDecl(*ifi);
}
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
bool
ClangASTContext::FieldIsBitfield (FieldDecl* field, uint32_t& bitfield_bit_size)
{
return FieldIsBitfield(getASTContext(), field, bitfield_bit_size);
}
bool
ClangASTContext::FieldIsBitfield
(
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
FieldDecl* field,
uint32_t& bitfield_bit_size
)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (ast == NULL || field == NULL)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return false;
if (field->isBitField())
{
Expr* bit_width_expr = field->getBitWidth();
if (bit_width_expr)
{
llvm::APSInt bit_width_apsint;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (bit_width_expr->isIntegerConstantExpr(bit_width_apsint, *ast))
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
bitfield_bit_size = bit_width_apsint.getLimitedValue(UINT32_MAX);
return true;
}
}
}
return false;
}
bool
ClangASTContext::RecordHasFields (const RecordDecl *record_decl)
{
if (record_decl == NULL)
return false;
if (!record_decl->field_empty())
return true;
// No fields, lets check this is a CXX record and check the base classes
const CXXRecordDecl *cxx_record_decl = dyn_cast<CXXRecordDecl>(record_decl);
if (cxx_record_decl)
{
CXXRecordDecl::base_class_const_iterator base_class, base_class_end;
for (base_class = cxx_record_decl->bases_begin(), base_class_end = cxx_record_decl->bases_end();
base_class != base_class_end;
++base_class)
{
const CXXRecordDecl *base_class_decl = cast<CXXRecordDecl>(base_class->getType()->getAs<RecordType>()->getDecl());
if (RecordHasFields(base_class_decl))
return true;
}
}
return false;
}
void
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ClangASTContext::SetDefaultAccessForRecordFields (clang_type_t clang_type, int default_accessibility, int *assigned_accessibilities, size_t num_assigned_accessibilities)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (clang_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const RecordType *record_type = dyn_cast<RecordType>(qual_type.getTypePtr());
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (record_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
RecordDecl *record_decl = record_type->getDecl();
if (record_decl)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
uint32_t field_idx;
RecordDecl::field_iterator field, field_end;
for (field = record_decl->field_begin(), field_end = record_decl->field_end(), field_idx = 0;
field != field_end;
++field, ++field_idx)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
// If no accessibility was assigned, assign the correct one
if (field_idx < num_assigned_accessibilities && assigned_accessibilities[field_idx] == clang::AS_none)
field->setAccess ((AccessSpecifier)default_accessibility);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
}
}
}
}
#pragma mark C++ Base Classes
CXXBaseSpecifier *
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::CreateBaseClassSpecifier (clang_type_t base_class_type, AccessType access, bool is_virtual, bool base_of_class)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (base_class_type)
Updated to llvm/clang from July 30, 2010 at 08:00. llvm-svn: 109887
2010-07-31 04:30:44 +08:00
return new CXXBaseSpecifier (SourceRange(),
is_virtual,
base_of_class,
ConvertAccessTypeToAccessSpecifier (access),
Updated to revision 123723 of LLVM, to bring in support for minimal type import functionality. llvm-svn: 123787
2011-01-19 07:32:05 +08:00
getASTContext()->CreateTypeSourceInfo (QualType::getFromOpaquePtr(base_class_type)),
SourceLocation());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return NULL;
}
More leaks detection: - fixed 3 posix spawn attributes leaks - fixed us always leaking CXXBaseSpecifier objects when we create class base classes. Clang apparently copies the base classes we pass in. Fixed some code formatting in ClangASTContext.cpp. llvm-svn: 107459
2010-07-02 09:29:13 +08:00
void
ClangASTContext::DeleteBaseClassSpecifiers (CXXBaseSpecifier **base_classes, unsigned num_base_classes)
{
for (unsigned i=0; i<num_base_classes; ++i)
{
delete base_classes[i];
base_classes[i] = NULL;
}
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::SetBaseClassesForClassType (clang_type_t class_clang_type, CXXBaseSpecifier const * const *base_classes, unsigned num_base_classes)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (class_clang_type)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
CXXRecordDecl *cxx_record_decl = QualType::getFromOpaquePtr(class_clang_type)->getAsCXXRecordDecl();
if (cxx_record_decl)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
cxx_record_decl->setBases(base_classes, num_base_classes);
return true;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
}
return false;
}
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
#pragma mark Objective C Classes
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
We now record metadata for Objective-C interfaces, Objective-C methods, and Objective-C properties. llvm-svn: 154972
2012-04-18 09:06:17 +08:00
ClangASTContext::CreateObjCClass
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
(
const char *name,
DeclContext *decl_ctx,
bool isForwardDecl,
We now record metadata for Objective-C interfaces, Objective-C methods, and Objective-C properties. llvm-svn: 154972
2012-04-18 09:06:17 +08:00
bool isInternal,
uint64_t metadata
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast = getASTContext();
assert (ast != NULL);
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
assert (name && name[0]);
if (decl_ctx == NULL)
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
decl_ctx = ast->getTranslationUnitDecl();
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
// NOTE: Eventually CXXRecordDecl will be merged back into RecordDecl and
// we will need to update this code. I was told to currently always use
// the CXXRecordDecl class since we often don't know from debug information
// if something is struct or a class, so we default to always use the more
// complete definition just in case.
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ObjCInterfaceDecl *decl = ObjCInterfaceDecl::Create (*ast,
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
decl_ctx,
SourceLocation(),
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
&ast->Idents.get(name),
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
NULL,
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
SourceLocation(),
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
/*isForwardDecl,*/
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
isInternal);
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
We now record metadata for Objective-C interfaces, Objective-C methods, and Objective-C properties. llvm-svn: 154972
2012-04-18 09:06:17 +08:00
if (decl)
SetMetadata(ast, (uintptr_t)decl, metadata);
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return ast->getObjCInterfaceType(decl).getAsOpaquePtr();
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::SetObjCSuperClass (clang_type_t class_opaque_type, clang_type_t super_opaque_type)
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
{
if (class_opaque_type && super_opaque_type)
{
QualType class_qual_type(QualType::getFromOpaquePtr(class_opaque_type));
QualType super_qual_type(QualType::getFromOpaquePtr(super_opaque_type));
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::Type *class_type = class_qual_type.getTypePtr();
const clang::Type *super_type = super_qual_type.getTypePtr();
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
if (class_type && super_type)
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(class_type);
const ObjCObjectType *objc_super_type = dyn_cast<ObjCObjectType>(super_type);
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
if (objc_class_type && objc_super_type)
{
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
ObjCInterfaceDecl *super_interface_decl = objc_super_type->getInterface();
if (class_interface_decl && super_interface_decl)
{
class_interface_decl->setSuperClass(super_interface_decl);
return true;
}
}
}
}
return false;
}
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
FieldDecl *
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
ClangASTContext::AddObjCClassIVar
(
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t class_opaque_type,
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
const char *name,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t ivar_opaque_type,
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
AccessType access,
uint32_t bitfield_bit_size,
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
bool is_synthesized
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
)
{
if (class_opaque_type == NULL || ivar_opaque_type == NULL)
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
return NULL;
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
ObjCIvarDecl *field = NULL;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
IdentifierTable *identifier_table = &ast->Idents;
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
assert (ast != NULL);
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
assert (identifier_table != NULL);
QualType class_qual_type(QualType::getFromOpaquePtr(class_opaque_type));
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::Type *class_type = class_qual_type.getTypePtr();
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
if (class_type)
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(class_type);
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
if (objc_class_type)
{
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
if (class_interface_decl)
{
clang::Expr *bit_width = NULL;
if (bitfield_bit_size != 0)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
APInt bitfield_bit_size_apint(ast->getTypeSize(ast->IntTy), bitfield_bit_size);
bit_width = new (*ast)IntegerLiteral (*ast, bitfield_bit_size_apint, ast->IntTy, SourceLocation());
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
}
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
field = ObjCIvarDecl::Create (*ast,
class_interface_decl,
SourceLocation(),
SourceLocation(),
&identifier_table->get(name), // Identifier
QualType::getFromOpaquePtr(ivar_opaque_type), // Field type
NULL, // TypeSourceInfo *
ConvertAccessTypeToObjCIvarAccessControl (access),
bit_width,
is_synthesized);
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
if (field)
{
class_interface_decl->addDecl(field);
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
#ifdef LLDB_CONFIGURATION_DEBUG
VerifyDecl(field);
#endif
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
return field;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
}
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
}
}
}
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
return NULL;
Added a new enumeration named "ClangASTContext::AccessType" that abstracts the type creation from the various access enumerations in Clang. Currently there are clang::AccessSpecifier and the objective C ivars have their own enumeration. So I added a new enumeration that will allow a consistent interface when creating types through ClangASTContext. I also added new functions to create an Objective C class, ivar and set an objective C superclass. They aren't hooked up in the DWARF parser yet. That is the next step, though I am unsure if I will do this in the DWARF parser or try and do it generically in the existing Record manipulation functions. llvm-svn: 109130
2010-07-23 02:30:50 +08:00
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
bool
ClangASTContext::AddObjCClassProperty
(
ASTContext *ast,
clang_type_t class_opaque_type,
const char *property_name,
clang_type_t property_opaque_type,
ObjCIvarDecl *ivar_decl,
const char *property_setter_name,
const char *property_getter_name,
We now record metadata for Objective-C interfaces, Objective-C methods, and Objective-C properties. llvm-svn: 154972
2012-04-18 09:06:17 +08:00
uint32_t property_attributes,
uint64_t metadata
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
)
{
Fixed a problem with properties where LLDB was not creating appropriate setter/getter methods for property definitions. llvm-svn: 146295
2011-12-10 07:24:26 +08:00
if (class_opaque_type == NULL || property_name == NULL || property_name[0] == '\0')
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
return false;
IdentifierTable *identifier_table = &ast->Idents;
assert (ast != NULL);
assert (identifier_table != NULL);
QualType class_qual_type(QualType::getFromOpaquePtr(class_opaque_type));
const clang::Type *class_type = class_qual_type.getTypePtr();
if (class_type)
{
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(class_type);
if (objc_class_type)
{
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
Fixed a problem with properties where LLDB was not creating appropriate setter/getter methods for property definitions. llvm-svn: 146295
2011-12-10 07:24:26 +08:00
clang_type_t property_opaque_type_to_access;
if (property_opaque_type)
property_opaque_type_to_access = property_opaque_type;
else if (ivar_decl)
property_opaque_type_to_access = ivar_decl->getType().getAsOpaquePtr();
if (class_interface_decl && property_opaque_type_to_access)
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
{
clang::TypeSourceInfo *prop_type_source;
if (ivar_decl)
prop_type_source = ast->CreateTypeSourceInfo (ivar_decl->getType());
else
prop_type_source = ast->CreateTypeSourceInfo (QualType::getFromOpaquePtr(property_opaque_type));
ObjCPropertyDecl *property_decl = ObjCPropertyDecl::Create(*ast,
class_interface_decl,
SourceLocation(), // Source Location
&identifier_table->get(property_name),
SourceLocation(), //Source Location for AT
Updated LLVM to take a new MC JIT that supports allocations by section. We install these sections in the target process and inform the JIT of their new locations. Also removed some unused variable warnings. llvm-svn: 151789
2012-03-01 10:03:47 +08:00
SourceLocation(), //Source location for (
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
prop_type_source
);
We now record metadata for Objective-C interfaces, Objective-C methods, and Objective-C properties. llvm-svn: 154972
2012-04-18 09:06:17 +08:00
Fixed a problem with properties where LLDB was not creating appropriate setter/getter methods for property definitions. llvm-svn: 146295
2011-12-10 07:24:26 +08:00
if (property_decl)
{
We now record metadata for Objective-C interfaces, Objective-C methods, and Objective-C properties. llvm-svn: 154972
2012-04-18 09:06:17 +08:00
SetMetadata(ast, (uintptr_t)property_decl, metadata);
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
class_interface_decl->addDecl (property_decl);
Fixed a problem with properties where LLDB was not creating appropriate setter/getter methods for property definitions. llvm-svn: 146295
2011-12-10 07:24:26 +08:00
Selector setter_sel, getter_sel;
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
if (property_setter_name != NULL)
{
std::string property_setter_no_colon(property_setter_name, strlen(property_setter_name) - 1);
clang::IdentifierInfo *setter_ident = &identifier_table->get(property_setter_no_colon.c_str());
Fixed a problem with properties where LLDB was not creating appropriate setter/getter methods for property definitions. llvm-svn: 146295
2011-12-10 07:24:26 +08:00
setter_sel = ast->Selectors.getSelector(1, &setter_ident);
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
}
Fixed a problem with properties where LLDB was not creating appropriate setter/getter methods for property definitions. llvm-svn: 146295
2011-12-10 07:24:26 +08:00
else if (!(property_attributes & DW_APPLE_PROPERTY_readonly))
{
std::string setter_sel_string("set");
setter_sel_string.push_back(::toupper(property_name[0]));
setter_sel_string.append(&property_name[1]);
clang::IdentifierInfo *setter_ident = &identifier_table->get(setter_sel_string.c_str());
setter_sel = ast->Selectors.getSelector(1, &setter_ident);
}
Fixed a problem where we did not read properties correctly if the setter/getter were not present in the debug information. The fixes are as follows: - We not only look for the method by its full name, but also look for automatically-generated methods when searching for a selector in an Objective-C interface. This is necessary to find accessors. - Extract the getter and setter name from the DW_TAG_APPLE_Property declaration in the DWARF if they are present; generate them if not. llvm-svn: 154067
2012-04-05 08:12:52 +08:00
property_decl->setSetterName(setter_sel);
property_decl->setPropertyAttributes (clang::ObjCPropertyDecl::OBJC_PR_setter);
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
if (property_getter_name != NULL)
{
clang::IdentifierInfo *getter_ident = &identifier_table->get(property_getter_name);
Fixed a problem with properties where LLDB was not creating appropriate setter/getter methods for property definitions. llvm-svn: 146295
2011-12-10 07:24:26 +08:00
getter_sel = ast->Selectors.getSelector(0, &getter_ident);
}
else
{
clang::IdentifierInfo *getter_ident = &identifier_table->get(property_name);
getter_sel = ast->Selectors.getSelector(0, &getter_ident);
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
}
Fixed a problem where we did not read properties correctly if the setter/getter were not present in the debug information. The fixes are as follows: - We not only look for the method by its full name, but also look for automatically-generated methods when searching for a selector in an Objective-C interface. This is necessary to find accessors. - Extract the getter and setter name from the DW_TAG_APPLE_Property declaration in the DWARF if they are present; generate them if not. llvm-svn: 154067
2012-04-05 08:12:52 +08:00
property_decl->setGetterName(getter_sel);
property_decl->setPropertyAttributes (clang::ObjCPropertyDecl::OBJC_PR_getter);
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
if (ivar_decl)
property_decl->setPropertyIvarDecl (ivar_decl);
if (property_attributes & DW_APPLE_PROPERTY_readonly)
property_decl->setPropertyAttributes (clang::ObjCPropertyDecl::OBJC_PR_readonly);
if (property_attributes & DW_APPLE_PROPERTY_readwrite)
property_decl->setPropertyAttributes (clang::ObjCPropertyDecl::OBJC_PR_readwrite);
if (property_attributes & DW_APPLE_PROPERTY_assign)
property_decl->setPropertyAttributes (clang::ObjCPropertyDecl::OBJC_PR_assign);
if (property_attributes & DW_APPLE_PROPERTY_retain)
property_decl->setPropertyAttributes (clang::ObjCPropertyDecl::OBJC_PR_retain);
if (property_attributes & DW_APPLE_PROPERTY_copy)
property_decl->setPropertyAttributes (clang::ObjCPropertyDecl::OBJC_PR_copy);
if (property_attributes & DW_APPLE_PROPERTY_nonatomic)
property_decl->setPropertyAttributes (clang::ObjCPropertyDecl::OBJC_PR_nonatomic);
Fixed a problem with properties where LLDB was not creating appropriate setter/getter methods for property definitions. llvm-svn: 146295
2011-12-10 07:24:26 +08:00
if (!getter_sel.isNull() && !class_interface_decl->lookupInstanceMethod(getter_sel))
{
QualType result_type = QualType::getFromOpaquePtr(property_opaque_type_to_access);
const bool isInstance = true;
const bool isVariadic = false;
const bool isSynthesized = false;
const bool isImplicitlyDeclared = true;
const bool isDefined = false;
const ObjCMethodDecl::ImplementationControl impControl = ObjCMethodDecl::None;
const bool HasRelatedResultType = false;
ObjCMethodDecl *getter = ObjCMethodDecl::Create(*ast,
SourceLocation(),
SourceLocation(),
getter_sel,
result_type,
NULL,
class_interface_decl,
isInstance,
isVariadic,
isSynthesized,
isImplicitlyDeclared,
isDefined,
impControl,
HasRelatedResultType);
We now record metadata for Objective-C interfaces, Objective-C methods, and Objective-C properties. llvm-svn: 154972
2012-04-18 09:06:17 +08:00
if (getter)
SetMetadata(ast, (uintptr_t)getter, metadata);
Fixed a problem with properties where LLDB was not creating appropriate setter/getter methods for property definitions. llvm-svn: 146295
2011-12-10 07:24:26 +08:00
getter->setMethodParams(*ast, ArrayRef<ParmVarDecl*>(), ArrayRef<SourceLocation>());
class_interface_decl->addDecl(getter);
}
if (!setter_sel.isNull() && !class_interface_decl->lookupInstanceMethod(setter_sel))
{
QualType result_type = ast->VoidTy;
const bool isInstance = true;
const bool isVariadic = false;
const bool isSynthesized = false;
const bool isImplicitlyDeclared = true;
const bool isDefined = false;
const ObjCMethodDecl::ImplementationControl impControl = ObjCMethodDecl::None;
const bool HasRelatedResultType = false;
ObjCMethodDecl *setter = ObjCMethodDecl::Create(*ast,
SourceLocation(),
SourceLocation(),
setter_sel,
result_type,
NULL,
class_interface_decl,
isInstance,
isVariadic,
isSynthesized,
isImplicitlyDeclared,
isDefined,
impControl,
HasRelatedResultType);
We now record metadata for Objective-C interfaces, Objective-C methods, and Objective-C properties. llvm-svn: 154972
2012-04-18 09:06:17 +08:00
if (setter)
SetMetadata(ast, (uintptr_t)setter, metadata);
Fixed a problem with properties where LLDB was not creating appropriate setter/getter methods for property definitions. llvm-svn: 146295
2011-12-10 07:24:26 +08:00
llvm::SmallVector<ParmVarDecl *, 1> params;
params.push_back (ParmVarDecl::Create (*ast,
setter,
SourceLocation(),
SourceLocation(),
NULL, // anonymous
QualType::getFromOpaquePtr(property_opaque_type_to_access),
NULL,
SC_Auto,
SC_Auto,
NULL));
setter->setMethodParams(*ast, ArrayRef<ParmVarDecl*>(params), ArrayRef<SourceLocation>());
class_interface_decl->addDecl(setter);
}
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
return true;
Fixed a problem with properties where LLDB was not creating appropriate setter/getter methods for property definitions. llvm-svn: 146295
2011-12-10 07:24:26 +08:00
}
Add code that reads the APPLE_property debug info, and makes up properties from them. llvm-svn: 144440
2011-11-12 09:36:43 +08:00
}
}
}
return false;
}
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::ObjCTypeHasIVars (clang_type_t class_opaque_type, bool check_superclass)
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
{
QualType class_qual_type(QualType::getFromOpaquePtr(class_opaque_type));
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::Type *class_type = class_qual_type.getTypePtr();
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
if (class_type)
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(class_type);
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
if (objc_class_type)
return ObjCDeclHasIVars (objc_class_type->getInterface(), check_superclass);
}
return false;
}
bool
ClangASTContext::ObjCDeclHasIVars (ObjCInterfaceDecl *class_interface_decl, bool check_superclass)
{
while (class_interface_decl)
{
if (class_interface_decl->ivar_size() > 0)
return true;
if (check_superclass)
class_interface_decl = class_interface_decl->getSuperClass();
else
break;
}
return false;
}
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ObjCMethodDecl *
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
ClangASTContext::AddMethodToObjCObjectType
(
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t class_opaque_type,
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
const char *name, // the full symbol name as seen in the symbol table ("-[NString stringWithCString:]")
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t method_opaque_type,
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
lldb::AccessType access
)
{
if (class_opaque_type == NULL || method_opaque_type == NULL)
return NULL;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
IdentifierTable *identifier_table = &ast->Idents;
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
assert (ast != NULL);
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
assert (identifier_table != NULL);
QualType class_qual_type(QualType::getFromOpaquePtr(class_opaque_type));
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::Type *class_type = class_qual_type.getTypePtr();
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
if (class_type == NULL)
return NULL;
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(class_type);
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
if (objc_class_type == NULL)
return NULL;
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
if (class_interface_decl == NULL)
return NULL;
const char *selector_start = ::strchr (name, ' ');
if (selector_start == NULL)
return NULL;
selector_start++;
if (!(::isalpha (selector_start[0]) || selector_start[0] == '_'))
return NULL;
llvm::SmallVector<IdentifierInfo *, 12> selector_idents;
Fixed the Objective C method prototypes to be correct (the selectors weren't being chopped up correctly). The DWARF plug-in also keeps a map of the ObjC class names to selectors for easy parsing of all class selectors when we parse the class type. llvm-svn: 116290
2010-10-12 10:24:53 +08:00
size_t len = 0;
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
const char *start;
Fixed the Objective C method prototypes to be correct (the selectors weren't being chopped up correctly). The DWARF plug-in also keeps a map of the ObjC class names to selectors for easy parsing of all class selectors when we parse the class type. llvm-svn: 116290
2010-10-12 10:24:53 +08:00
//printf ("name = '%s'\n", name);
unsigned num_selectors_with_args = 0;
for (start = selector_start;
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
start && *start != '\0' && *start != ']';
Fixed the Objective C method prototypes to be correct (the selectors weren't being chopped up correctly). The DWARF plug-in also keeps a map of the ObjC class names to selectors for easy parsing of all class selectors when we parse the class type. llvm-svn: 116290
2010-10-12 10:24:53 +08:00
start += len)
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
{
Fixed the Objective C method prototypes to be correct (the selectors weren't being chopped up correctly). The DWARF plug-in also keeps a map of the ObjC class names to selectors for easy parsing of all class selectors when we parse the class type. llvm-svn: 116290
2010-10-12 10:24:53 +08:00
len = ::strcspn(start, ":]");
Don't include the ":" character when building the selector identifiers when making objective C classes from the DWARF. llvm-svn: 117438
2010-10-27 12:01:14 +08:00
bool has_arg = (start[len] == ':');
if (has_arg)
Fixed the Objective C method prototypes to be correct (the selectors weren't being chopped up correctly). The DWARF plug-in also keeps a map of the ObjC class names to selectors for easy parsing of all class selectors when we parse the class type. llvm-svn: 116290
2010-10-12 10:24:53 +08:00
++num_selectors_with_args;
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
selector_idents.push_back (&identifier_table->get (StringRef (start, len)));
Don't include the ":" character when building the selector identifiers when making objective C classes from the DWARF. llvm-svn: 117438
2010-10-27 12:01:14 +08:00
if (has_arg)
len += 1;
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
}
if (selector_idents.size() == 0)
return 0;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
clang::Selector method_selector = ast->Selectors.getSelector (num_selectors_with_args ? selector_idents.size() : 0,
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
selector_idents.data());
QualType method_qual_type (QualType::getFromOpaquePtr (method_opaque_type));
// Populate the method decl with parameter decls
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::Type *method_type(method_qual_type.getTypePtr());
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
if (method_type == NULL)
return NULL;
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const FunctionProtoType *method_function_prototype (dyn_cast<FunctionProtoType>(method_type));
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
if (!method_function_prototype)
return NULL;
bool is_variadic = false;
bool is_synthesized = false;
bool is_defined = false;
ObjCMethodDecl::ImplementationControl imp_control = ObjCMethodDecl::None;
const unsigned num_args = method_function_prototype->getNumArgs();
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ObjCMethodDecl *objc_method_decl = ObjCMethodDecl::Create (*ast,
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
SourceLocation(), // beginLoc,
SourceLocation(), // endLoc,
method_selector,
method_function_prototype->getResultType(),
NULL, // TypeSourceInfo *ResultTInfo,
GetDeclContextForType (class_opaque_type),
name[0] == '-',
is_variadic,
is_synthesized,
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
true, // is_implicitly_declared
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
is_defined,
imp_control,
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
false /*has_related_result_type*/);
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
if (objc_method_decl == NULL)
return NULL;
if (num_args > 0)
{
llvm::SmallVector<ParmVarDecl *, 12> params;
for (int param_index = 0; param_index < num_args; ++param_index)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
params.push_back (ParmVarDecl::Create (*ast,
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
objc_method_decl,
SourceLocation(),
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
SourceLocation(),
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
NULL, // anonymous
method_function_prototype->getArgType(param_index),
NULL,
SC_Auto,
SC_Auto,
NULL));
}
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
objc_method_decl->setMethodParams(*ast, ArrayRef<ParmVarDecl*>(params), ArrayRef<SourceLocation>());
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
}
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
class_interface_decl->addDecl (objc_method_decl);
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
#ifdef LLDB_CONFIGURATION_DEBUG
VerifyDecl(objc_method_decl);
#endif
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
return objc_method_decl;
}
Added support to SBType for getting template arguments from a SBType: uint32_t SBType::GetNumberOfTemplateArguments (); lldb::SBType SBType::GetTemplateArgumentType (uint32_t idx); lldb::TemplateArgumentKind SBType::GetTemplateArgumentKind (uint32_t idx); Some lldb::TemplateArgumentKind values don't have a corresponding SBType that will be returned from SBType::GetTemplateArgumentType(). This will help our data formatters do their job by being able to find out the type of template params and do smart things with those. llvm-svn: 149658
2012-02-03 09:30:30 +08:00
size_t
ClangASTContext::GetNumTemplateArguments (clang::ASTContext *ast, clang_type_t clang_type)
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
case clang::Type::Record:
if (GetCompleteQualType (ast, qual_type))
{
const CXXRecordDecl *cxx_record_decl = qual_type->getAsCXXRecordDecl();
if (cxx_record_decl)
{
const ClassTemplateSpecializationDecl *template_decl = dyn_cast<ClassTemplateSpecializationDecl>(cxx_record_decl);
if (template_decl)
return template_decl->getTemplateArgs().size();
}
}
break;
case clang::Type::Typedef:
return ClangASTContext::GetNumTemplateArguments (ast, cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr());
default:
break;
}
}
return 0;
}
clang_type_t
ClangASTContext::GetTemplateArgument (clang::ASTContext *ast, clang_type_t clang_type, size_t arg_idx, lldb::TemplateArgumentKind &kind)
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
case clang::Type::Record:
if (GetCompleteQualType (ast, qual_type))
{
const CXXRecordDecl *cxx_record_decl = qual_type->getAsCXXRecordDecl();
if (cxx_record_decl)
{
const ClassTemplateSpecializationDecl *template_decl = dyn_cast<ClassTemplateSpecializationDecl>(cxx_record_decl);
if (template_decl && arg_idx < template_decl->getTemplateArgs().size())
{
const TemplateArgument &template_arg = template_decl->getTemplateArgs()[arg_idx];
switch (template_arg.getKind())
{
case clang::TemplateArgument::Null:
kind = eTemplateArgumentKindNull;
return NULL;
case clang::TemplateArgument::Type:
kind = eTemplateArgumentKindType;
return template_arg.getAsType().getAsOpaquePtr();
case clang::TemplateArgument::Declaration:
kind = eTemplateArgumentKindDeclaration;
return NULL;
case clang::TemplateArgument::Integral:
kind = eTemplateArgumentKindIntegral;
return template_arg.getIntegralType().getAsOpaquePtr();
case clang::TemplateArgument::Template:
kind = eTemplateArgumentKindTemplate;
return NULL;
case clang::TemplateArgument::TemplateExpansion:
kind = eTemplateArgumentKindTemplateExpansion;
return NULL;
case clang::TemplateArgument::Expression:
kind = eTemplateArgumentKindExpression;
return NULL;
case clang::TemplateArgument::Pack:
kind = eTemplateArgumentKindPack;
return NULL;
default:
assert (!"Unhandled TemplateArgument::ArgKind");
kind = eTemplateArgumentKindNull;
return NULL;
}
}
}
}
break;
case clang::Type::Typedef:
return ClangASTContext::GetTemplateArgument (ast, cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(), arg_idx, kind);
default:
break;
}
}
kind = eTemplateArgumentKindNull;
return NULL;
}
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
uint32_t
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
ClangASTContext::GetTypeInfo
(
clang_type_t clang_type,
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
clang::ASTContext *ast,
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
clang_type_t *pointee_or_element_clang_type
)
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
{
if (clang_type == NULL)
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
return 0;
if (pointee_or_element_clang_type)
*pointee_or_element_clang_type = NULL;
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
Fixes to Objective-C built-in type handling. Specifically, we fixed handling of the objc_class built-in type, which allowed us to pass named Objective-C objects to functions, call variable list -t on objects safely, etc. llvm-svn: 117249
2010-10-25 08:29:48 +08:00
case clang::Type::Builtin:
switch (cast<clang::BuiltinType>(qual_type)->getKind())
{
case clang::BuiltinType::ObjCId:
case clang::BuiltinType::ObjCClass:
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (ast && pointee_or_element_clang_type)
*pointee_or_element_clang_type = ast->ObjCBuiltinClassTy.getAsOpaquePtr();
Fixes to Objective-C built-in type handling. Specifically, we fixed handling of the objc_class built-in type, which allowed us to pass named Objective-C objects to functions, call variable list -t on objects safely, etc. llvm-svn: 117249
2010-10-25 08:29:48 +08:00
return eTypeIsBuiltIn | eTypeIsPointer | eTypeHasValue;
named summaries: - a new --name option for "type summary add" lets you give a name to a summary - a new --summary option for "frame variable" lets you bind a named summary to one or more variables ${var%s} now works for printing the value of 0-terminated CStrings type format test case now tests for cascading - this is disabled on GCC because GCC may end up stripping typedef chains, basically breaking cascading new design for the FormatNavigator class new template class CleanUp2 meant to support cleanup routines with 1 additional parameter beyond resource handle llvm-svn: 134943
2011-07-12 08:18:11 +08:00
break;
case clang::BuiltinType::Bool:
case clang::BuiltinType::Char_U:
case clang::BuiltinType::UChar:
case clang::BuiltinType::WChar_U:
case clang::BuiltinType::Char16:
case clang::BuiltinType::Char32:
case clang::BuiltinType::UShort:
case clang::BuiltinType::UInt:
case clang::BuiltinType::ULong:
case clang::BuiltinType::ULongLong:
case clang::BuiltinType::UInt128:
case clang::BuiltinType::Char_S:
case clang::BuiltinType::SChar:
case clang::BuiltinType::WChar_S:
case clang::BuiltinType::Short:
case clang::BuiltinType::Int:
case clang::BuiltinType::Long:
case clang::BuiltinType::LongLong:
case clang::BuiltinType::Int128:
case clang::BuiltinType::Float:
case clang::BuiltinType::Double:
case clang::BuiltinType::LongDouble:
return eTypeIsBuiltIn | eTypeHasValue | eTypeIsScalar;
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
default:
break;
Fixes to Objective-C built-in type handling. Specifically, we fixed handling of the objc_class built-in type, which allowed us to pass named Objective-C objects to functions, call variable list -t on objects safely, etc. llvm-svn: 117249
2010-10-25 08:29:48 +08:00
}
return eTypeIsBuiltIn | eTypeHasValue;
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
case clang::Type::BlockPointer:
if (pointee_or_element_clang_type)
*pointee_or_element_clang_type = qual_type->getPointeeType().getAsOpaquePtr();
return eTypeIsPointer | eTypeHasChildren | eTypeIsBlock;
Added complete complex support for displaying and parsing complex types. llvm-svn: 123509
2011-01-15 10:52:14 +08:00
case clang::Type::Complex: return eTypeIsBuiltIn | eTypeHasValue;
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
case clang::Type::ConstantArray:
case clang::Type::DependentSizedArray:
case clang::Type::IncompleteArray:
case clang::Type::VariableArray:
if (pointee_or_element_clang_type)
*pointee_or_element_clang_type = cast<ArrayType>(qual_type.getTypePtr())->getElementType().getAsOpaquePtr();
return eTypeHasChildren | eTypeIsArray;
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
case clang::Type::DependentName: return 0;
case clang::Type::DependentSizedExtVector: return eTypeHasChildren | eTypeIsVector;
case clang::Type::DependentTemplateSpecialization: return eTypeIsTemplate;
case clang::Type::Decltype: return 0;
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
case clang::Type::Enum:
if (pointee_or_element_clang_type)
*pointee_or_element_clang_type = cast<EnumType>(qual_type)->getDecl()->getIntegerType().getAsOpaquePtr();
return eTypeIsEnumeration | eTypeHasValue;
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated:
return ClangASTContext::GetTypeInfo (cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr(),
ast,
pointee_or_element_clang_type);
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
case clang::Type::ExtVector: return eTypeHasChildren | eTypeIsVector;
case clang::Type::FunctionProto: return eTypeIsFuncPrototype | eTypeHasValue;
case clang::Type::FunctionNoProto: return eTypeIsFuncPrototype | eTypeHasValue;
case clang::Type::InjectedClassName: return 0;
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
case clang::Type::LValueReference:
case clang::Type::RValueReference:
if (pointee_or_element_clang_type)
*pointee_or_element_clang_type = cast<ReferenceType>(qual_type.getTypePtr())->getPointeeType().getAsOpaquePtr();
return eTypeHasChildren | eTypeIsReference | eTypeHasValue;
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
case clang::Type::MemberPointer: return eTypeIsPointer | eTypeIsMember | eTypeHasValue;
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
case clang::Type::ObjCObjectPointer:
if (pointee_or_element_clang_type)
*pointee_or_element_clang_type = qual_type->getPointeeType().getAsOpaquePtr();
return eTypeHasChildren | eTypeIsObjC | eTypeIsClass | eTypeIsPointer | eTypeHasValue;
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
case clang::Type::ObjCObject: return eTypeHasChildren | eTypeIsObjC | eTypeIsClass;
case clang::Type::ObjCInterface: return eTypeHasChildren | eTypeIsObjC | eTypeIsClass;
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
case clang::Type::Pointer:
if (pointee_or_element_clang_type)
*pointee_or_element_clang_type = qual_type->getPointeeType().getAsOpaquePtr();
return eTypeHasChildren | eTypeIsPointer | eTypeHasValue;
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
case clang::Type::Record:
if (qual_type->getAsCXXRecordDecl())
return eTypeHasChildren | eTypeIsClass | eTypeIsCPlusPlus;
else
return eTypeHasChildren | eTypeIsStructUnion;
break;
case clang::Type::SubstTemplateTypeParm: return eTypeIsTemplate;
case clang::Type::TemplateTypeParm: return eTypeIsTemplate;
case clang::Type::TemplateSpecialization: return eTypeIsTemplate;
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
case clang::Type::Typedef:
Updated to latest Clang revision. This involved very minor changes, changing how we get the target type from a TypedefType, adding a parameter to EnumDecl::Create(), and other minor tweaks. llvm-svn: 121663
2010-12-13 09:26:27 +08:00
return eTypeIsTypedef | ClangASTContext::GetTypeInfo (cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(),
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ast,
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
pointee_or_element_clang_type);
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
case clang::Type::TypeOfExpr: return 0;
case clang::Type::TypeOf: return 0;
case clang::Type::UnresolvedUsing: return 0;
case clang::Type::Vector: return eTypeHasChildren | eTypeIsVector;
default: return 0;
}
return 0;
}
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
#pragma mark Aggregate Types
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::IsAggregateType (clang_type_t clang_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type == NULL)
return false;
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::IncompleteArray:
case clang::Type::VariableArray:
case clang::Type::ConstantArray:
case clang::Type::ExtVector:
case clang::Type::Vector:
case clang::Type::Record:
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return true;
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated:
return ClangASTContext::IsAggregateType (cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr());
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Typedef:
Updated to latest Clang revision. This involved very minor changes, changing how we get the target type from a TypedefType, adding a parameter to EnumDecl::Create(), and other minor tweaks. llvm-svn: 121663
2010-12-13 09:26:27 +08:00
return ClangASTContext::IsAggregateType (cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
default:
break;
}
// The clang type does have a value
return false;
}
uint32_t
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ClangASTContext::GetNumChildren (clang::ASTContext *ast, clang_type_t clang_type, bool omit_empty_base_classes)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (clang_type == NULL)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return 0;
uint32_t num_children = 0;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
case clang::Type::Builtin:
switch (cast<clang::BuiltinType>(qual_type)->getKind())
{
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
case clang::BuiltinType::ObjCId: // child is Class
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
case clang::BuiltinType::ObjCClass: // child is Class
num_children = 1;
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
break;
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
default:
break;
}
break;
Fixed the "expression" command object to use the StackFrame::GetValueForExpressionPath() function and also hooked up better error reporting for when things fail. Fixed issues with trying to display children of pointers when none are supposed to be shown (no children for function pointers, and more like this). This was causing child value objects to be made that were correctly firing an assertion. llvm-svn: 121841
2010-12-15 13:08:08 +08:00
Added complete complex support for displaying and parsing complex types. llvm-svn: 123509
2011-01-15 10:52:14 +08:00
case clang::Type::Complex: return 0;
Fixed the "expression" command object to use the StackFrame::GetValueForExpressionPath() function and also hooked up better error reporting for when things fail. Fixed issues with trying to display children of pointers when none are supposed to be shown (no children for function pointers, and more like this). This was causing child value objects to be made that were correctly firing an assertion. llvm-svn: 121841
2010-12-15 13:08:08 +08:00
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Record:
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
if (GetCompleteQualType (ast, qual_type))
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
const RecordType *record_type = cast<RecordType>(qual_type.getTypePtr());
const RecordDecl *record_decl = record_type->getDecl();
assert(record_decl);
const CXXRecordDecl *cxx_record_decl = dyn_cast<CXXRecordDecl>(record_decl);
if (cxx_record_decl)
{
if (omit_empty_base_classes)
{
// Check each base classes to see if it or any of its
// base classes contain any fields. This can help
// limit the noise in variable views by not having to
// show base classes that contain no members.
CXXRecordDecl::base_class_const_iterator base_class, base_class_end;
for (base_class = cxx_record_decl->bases_begin(), base_class_end = cxx_record_decl->bases_end();
base_class != base_class_end;
++base_class)
{
const CXXRecordDecl *base_class_decl = cast<CXXRecordDecl>(base_class->getType()->getAs<RecordType>()->getDecl());
// Skip empty base classes
if (RecordHasFields(base_class_decl) == false)
continue;
num_children++;
}
}
else
{
// Include all base classes
num_children += cxx_record_decl->getNumBases();
}
}
RecordDecl::field_iterator field, field_end;
for (field = record_decl->field_begin(), field_end = record_decl->field_end(); field != field_end; ++field)
++num_children;
}
break;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
if (GetCompleteQualType (ast, qual_type))
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(qual_type.getTypePtr());
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
assert (objc_class_type);
if (objc_class_type)
{
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
if (class_interface_decl)
{
ObjCInterfaceDecl *superclass_interface_decl = class_interface_decl->getSuperClass();
if (superclass_interface_decl)
{
if (omit_empty_base_classes)
{
if (ClangASTContext::ObjCDeclHasIVars (superclass_interface_decl, true))
++num_children;
}
else
++num_children;
}
num_children += class_interface_decl->ivar_size();
}
}
}
break;
case clang::Type::ObjCObjectPointer:
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ObjCObjectPointerType *pointer_type = cast<ObjCObjectPointerType>(qual_type.getTypePtr());
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
QualType pointee_type = pointer_type->getPointeeType();
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
uint32_t num_pointee_children = ClangASTContext::GetNumChildren (ast,
pointee_type.getAsOpaquePtr(),
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
omit_empty_base_classes);
// If this type points to a simple type, then it has 1 child
if (num_pointee_children == 0)
num_children = 1;
else
num_children = num_pointee_children;
}
break;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::ConstantArray:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
num_children = cast<ConstantArrayType>(qual_type.getTypePtr())->getSize().getLimitedValue();
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Pointer:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const PointerType *pointer_type = cast<PointerType>(qual_type.getTypePtr());
Fixed the "expression" command object to use the StackFrame::GetValueForExpressionPath() function and also hooked up better error reporting for when things fail. Fixed issues with trying to display children of pointers when none are supposed to be shown (no children for function pointers, and more like this). This was causing child value objects to be made that were correctly firing an assertion. llvm-svn: 121841
2010-12-15 13:08:08 +08:00
QualType pointee_type (pointer_type->getPointeeType());
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
uint32_t num_pointee_children = ClangASTContext::GetNumChildren (ast,
pointee_type.getAsOpaquePtr(),
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
omit_empty_base_classes);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (num_pointee_children == 0)
Fixed the "expression" command object to use the StackFrame::GetValueForExpressionPath() function and also hooked up better error reporting for when things fail. Fixed issues with trying to display children of pointers when none are supposed to be shown (no children for function pointers, and more like this). This was causing child value objects to be made that were correctly firing an assertion. llvm-svn: 121841
2010-12-15 13:08:08 +08:00
{
// We have a pointer to a pointee type that claims it has no children.
// We will want to look at
num_children = ClangASTContext::GetNumPointeeChildren (pointee_type.getAsOpaquePtr());
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
else
num_children = num_pointee_children;
}
break;
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
case clang::Type::LValueReference:
case clang::Type::RValueReference:
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ReferenceType *reference_type = cast<ReferenceType>(qual_type.getTypePtr());
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
QualType pointee_type = reference_type->getPointeeType();
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
uint32_t num_pointee_children = ClangASTContext::GetNumChildren (ast,
pointee_type.getAsOpaquePtr(),
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
omit_empty_base_classes);
// If this type points to a simple type, then it has 1 child
if (num_pointee_children == 0)
num_children = 1;
else
num_children = num_pointee_children;
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Typedef:
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
num_children = ClangASTContext::GetNumChildren (ast,
cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(),
omit_empty_base_classes);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
break;
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated:
num_children = ClangASTContext::GetNumChildren (ast,
cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr(),
omit_empty_base_classes);
break;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
default:
break;
}
return num_children;
}
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
uint32_t
ClangASTContext::GetNumDirectBaseClasses (clang::ASTContext *ast, clang_type_t clang_type)
{
if (clang_type == NULL)
return 0;
uint32_t count = 0;
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
case clang::Type::Record:
if (GetCompleteQualType (ast, qual_type))
{
const CXXRecordDecl *cxx_record_decl = qual_type->getAsCXXRecordDecl();
if (cxx_record_decl)
count = cxx_record_decl->getNumBases();
}
break;
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
if (GetCompleteQualType (ast, qual_type))
{
const ObjCObjectType *objc_class_type = qual_type->getAsObjCQualifiedInterfaceType();
if (objc_class_type)
{
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
if (class_interface_decl && class_interface_decl->getSuperClass())
count = 1;
}
}
break;
case clang::Type::Typedef:
count = ClangASTContext::GetNumDirectBaseClasses (ast, cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr());
break;
case clang::Type::Elaborated:
count = ClangASTContext::GetNumDirectBaseClasses (ast, cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr());
break;
default:
break;
}
return count;
}
uint32_t
ClangASTContext::GetNumVirtualBaseClasses (clang::ASTContext *ast,
clang_type_t clang_type)
{
if (clang_type == NULL)
return 0;
uint32_t count = 0;
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
case clang::Type::Record:
if (GetCompleteQualType (ast, qual_type))
{
const CXXRecordDecl *cxx_record_decl = qual_type->getAsCXXRecordDecl();
if (cxx_record_decl)
count = cxx_record_decl->getNumVBases();
}
break;
case clang::Type::Typedef:
count = ClangASTContext::GetNumVirtualBaseClasses (ast, cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr());
break;
case clang::Type::Elaborated:
count = ClangASTContext::GetNumVirtualBaseClasses (ast, cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr());
break;
default:
break;
}
return count;
}
uint32_t
ClangASTContext::GetNumFields (clang::ASTContext *ast, clang_type_t clang_type)
{
if (clang_type == NULL)
return 0;
uint32_t count = 0;
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
case clang::Type::Record:
if (GetCompleteQualType (ast, qual_type))
{
const RecordType *record_type = dyn_cast<RecordType>(qual_type.getTypePtr());
if (record_type)
{
RecordDecl *record_decl = record_type->getDecl();
if (record_decl)
{
uint32_t field_idx = 0;
RecordDecl::field_iterator field, field_end;
for (field = record_decl->field_begin(), field_end = record_decl->field_end(); field != field_end; ++field)
++field_idx;
count = field_idx;
}
}
}
break;
case clang::Type::Typedef:
count = ClangASTContext::GetNumFields (ast, cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr());
break;
case clang::Type::Elaborated:
count = ClangASTContext::GetNumFields (ast, cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr());
break;
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
if (GetCompleteQualType (ast, qual_type))
{
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(qual_type.getTypePtr());
if (objc_class_type)
{
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
if (class_interface_decl)
count = class_interface_decl->ivar_size();
}
}
break;
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
default:
break;
}
return count;
}
clang_type_t
ClangASTContext::GetDirectBaseClassAtIndex (clang::ASTContext *ast,
clang_type_t clang_type,
uint32_t idx,
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
uint32_t *bit_offset_ptr)
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
{
if (clang_type == NULL)
return 0;
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
case clang::Type::Record:
if (GetCompleteQualType (ast, qual_type))
{
const CXXRecordDecl *cxx_record_decl = qual_type->getAsCXXRecordDecl();
if (cxx_record_decl)
{
uint32_t curr_idx = 0;
CXXRecordDecl::base_class_const_iterator base_class, base_class_end;
for (base_class = cxx_record_decl->bases_begin(), base_class_end = cxx_record_decl->bases_end();
base_class != base_class_end;
++base_class, ++curr_idx)
{
if (curr_idx == idx)
{
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
if (bit_offset_ptr)
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
{
const ASTRecordLayout &record_layout = ast->getASTRecordLayout(cxx_record_decl);
const CXXRecordDecl *base_class_decl = cast<CXXRecordDecl>(base_class->getType()->getAs<RecordType>()->getDecl());
// if (base_class->isVirtual())
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
// *bit_offset_ptr = record_layout.getVBaseClassOffset(base_class_decl).getQuantity() * 8;
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
// else
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
*bit_offset_ptr = record_layout.getBaseClassOffset(base_class_decl).getQuantity() * 8;
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
}
return base_class->getType().getAsOpaquePtr();
}
}
}
}
break;
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
if (idx == 0 && GetCompleteQualType (ast, qual_type))
{
const ObjCObjectType *objc_class_type = qual_type->getAsObjCQualifiedInterfaceType();
if (objc_class_type)
{
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
if (class_interface_decl)
{
ObjCInterfaceDecl *superclass_interface_decl = class_interface_decl->getSuperClass();
if (superclass_interface_decl)
{
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
if (bit_offset_ptr)
*bit_offset_ptr = 0;
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
return ast->getObjCInterfaceType(superclass_interface_decl).getAsOpaquePtr();
}
}
}
}
break;
case clang::Type::Typedef:
return ClangASTContext::GetDirectBaseClassAtIndex (ast,
cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(),
idx,
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
bit_offset_ptr);
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
case clang::Type::Elaborated:
return ClangASTContext::GetDirectBaseClassAtIndex (ast,
cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr(),
idx,
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
bit_offset_ptr);
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
default:
break;
}
return NULL;
}
clang_type_t
ClangASTContext::GetVirtualBaseClassAtIndex (clang::ASTContext *ast,
clang_type_t clang_type,
uint32_t idx,
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
uint32_t *bit_offset_ptr)
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
{
if (clang_type == NULL)
return 0;
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
case clang::Type::Record:
if (GetCompleteQualType (ast, qual_type))
{
const CXXRecordDecl *cxx_record_decl = qual_type->getAsCXXRecordDecl();
if (cxx_record_decl)
{
uint32_t curr_idx = 0;
CXXRecordDecl::base_class_const_iterator base_class, base_class_end;
for (base_class = cxx_record_decl->vbases_begin(), base_class_end = cxx_record_decl->vbases_end();
base_class != base_class_end;
++base_class, ++curr_idx)
{
if (curr_idx == idx)
{
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
if (bit_offset_ptr)
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
{
const ASTRecordLayout &record_layout = ast->getASTRecordLayout(cxx_record_decl);
const CXXRecordDecl *base_class_decl = cast<CXXRecordDecl>(base_class->getType()->getAs<RecordType>()->getDecl());
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
*bit_offset_ptr = record_layout.getVBaseClassOffset(base_class_decl).getQuantity() * 8;
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
}
return base_class->getType().getAsOpaquePtr();
}
}
}
}
break;
case clang::Type::Typedef:
return ClangASTContext::GetVirtualBaseClassAtIndex (ast,
cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(),
idx,
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
bit_offset_ptr);
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
case clang::Type::Elaborated:
return ClangASTContext::GetVirtualBaseClassAtIndex (ast,
cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr(),
idx,
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
bit_offset_ptr);
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
default:
break;
}
return NULL;
}
clang_type_t
ClangASTContext::GetFieldAtIndex (clang::ASTContext *ast,
clang_type_t clang_type,
uint32_t idx,
std::string& name,
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
uint32_t *bit_offset_ptr)
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
{
if (clang_type == NULL)
return 0;
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
case clang::Type::Record:
if (GetCompleteQualType (ast, qual_type))
{
const RecordType *record_type = cast<RecordType>(qual_type.getTypePtr());
const RecordDecl *record_decl = record_type->getDecl();
uint32_t field_idx = 0;
RecordDecl::field_iterator field, field_end;
for (field = record_decl->field_begin(), field_end = record_decl->field_end(); field != field_end; ++field, ++field_idx)
{
if (idx == field_idx)
{
// Print the member type if requested
// Print the member name and equal sign
name.assign(field->getNameAsString());
// Figure out the type byte size (field_type_info.first) and
// alignment (field_type_info.second) from the AST context.
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
if (bit_offset_ptr)
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
{
const ASTRecordLayout &record_layout = ast->getASTRecordLayout(record_decl);
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
*bit_offset_ptr = record_layout.getFieldOffset (field_idx);
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
}
return field->getType().getAsOpaquePtr();
}
}
}
break;
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
if (GetCompleteQualType (ast, qual_type))
{
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(qual_type.getTypePtr());
assert (objc_class_type);
if (objc_class_type)
{
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
if (class_interface_decl)
{
if (idx < (class_interface_decl->ivar_size()))
{
ObjCInterfaceDecl::ivar_iterator ivar_pos, ivar_end = class_interface_decl->ivar_end();
uint32_t ivar_idx = 0;
for (ivar_pos = class_interface_decl->ivar_begin(); ivar_pos != ivar_end; ++ivar_pos, ++ivar_idx)
{
if (ivar_idx == idx)
{
const ObjCIvarDecl* ivar_decl = *ivar_pos;
QualType ivar_qual_type(ivar_decl->getType());
name.assign(ivar_decl->getNameAsString());
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
if (bit_offset_ptr)
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
{
const ASTRecordLayout &interface_layout = ast->getASTObjCInterfaceLayout(class_interface_decl);
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
*bit_offset_ptr = interface_layout.getFieldOffset (ivar_idx);
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
}
return ivar_qual_type.getAsOpaquePtr();
}
}
}
}
}
}
break;
case clang::Type::Typedef:
return ClangASTContext::GetFieldAtIndex (ast,
cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(),
idx,
name,
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
bit_offset_ptr);
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
case clang::Type::Elaborated:
return ClangASTContext::GetFieldAtIndex (ast,
cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr(),
idx,
name,
<rdar://problem/10126482> Fixed an issues with the SBType and SBTypeMember classes: - Fixed SBType to be able to dump itself from python - Fixed SBType::GetNumberOfFields() to return the correct value for objective C interfaces - Fixed SBTypeMember to be able to dump itself from python - Fixed the SBTypeMember ability to get a field offset in bytes (the value being returned was wrong) - Added the SBTypeMember ability to get a field offset in bits Cleaned up a lot of the Stream usage in the SB API files. llvm-svn: 144493
2011-11-13 14:57:31 +08:00
bit_offset_ptr);
Added the ability to introspect types thourgh the public SBType interface. Fixed up many API calls to not be "const" as const doesn't mean anything to most of our lldb::SB objects since they contain a shared pointer, auto_ptr, or pointer to the types which circumvent the constness anyway. llvm-svn: 139428
2011-09-10 07:04:00 +08:00
default:
break;
}
return NULL;
}
Fixed the "expression" command object to use the StackFrame::GetValueForExpressionPath() function and also hooked up better error reporting for when things fail. Fixed issues with trying to display children of pointers when none are supposed to be shown (no children for function pointers, and more like this). This was causing child value objects to be made that were correctly firing an assertion. llvm-svn: 121841
2010-12-15 13:08:08 +08:00
// If a pointer to a pointee type (the clang_type arg) says that it has no
// children, then we either need to trust it, or override it and return a
// different result. For example, an "int *" has one child that is an integer,
// but a function pointer doesn't have any children. Likewise if a Record type
// claims it has no children, then there really is nothing to show.
uint32_t
ClangASTContext::GetNumPointeeChildren (clang_type_t clang_type)
{
if (clang_type == NULL)
return 0;
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
Make sure we don't assert if we have a child with zero byte size. Also we now say that "void *" value objects don't have children. llvm-svn: 123092
2011-01-09 06:26:47 +08:00
case clang::Type::Builtin:
switch (cast<clang::BuiltinType>(qual_type)->getKind())
{
Centralized a lot of the status information for processes, threads, and stack frame down in the lldb_private::Process, lldb_private::Thread, lldb_private::StackFrameList and the lldb_private::StackFrame classes. We had some command line commands that had duplicate versions of the process status output ("thread list" and "process status" for example). Removed the "file" command and placed it where it should have been: "target create". Made an alias for "file" to "target create" so we stay compatible with GDB commands. We can now have multple usable targets in lldb at the same time. This is nice for comparing two runs of a program or debugging more than one binary at the same time. The new command is "target select <target-idx>" and also to see a list of the current targets you can use the new "target list" command. The flow in a debug session can be: (lldb) target create /path/to/exe/a.out (lldb) breakpoint set --name main (lldb) run ... hit breakpoint (lldb) target create /bin/ls (lldb) run /tmp Process 36001 exited with status = 0 (0x00000000) (lldb) target list Current targets: target #0: /tmp/args/a.out ( arch=x86_64-apple-darwin, platform=localhost, pid=35999, state=stopped ) * target #1: /bin/ls ( arch=x86_64-apple-darwin, platform=localhost, pid=36001, state=exited ) (lldb) target select 0 Current targets: * target #0: /tmp/args/a.out ( arch=x86_64-apple-darwin, platform=localhost, pid=35999, state=stopped ) target #1: /bin/ls ( arch=x86_64-apple-darwin, platform=localhost, pid=36001, state=exited ) (lldb) bt * thread #1: tid = 0x2d03, 0x0000000100000b9a a.out`main + 42 at main.c:16, stop reason = breakpoint 1.1 frame #0: 0x0000000100000b9a a.out`main + 42 at main.c:16 frame #1: 0x0000000100000b64 a.out`start + 52 Above we created a target for "a.out" and ran and hit a breakpoint at "main". Then we created a new target for /bin/ls and ran it. Then we listed the targest and selected our original "a.out" program, so we showed two concurent debug sessions going on at the same time. llvm-svn: 129695
2011-04-18 16:33:37 +08:00
case clang::BuiltinType::UnknownAny:
Make sure we don't assert if we have a child with zero byte size. Also we now say that "void *" value objects don't have children. llvm-svn: 123092
2011-01-09 06:26:47 +08:00
case clang::BuiltinType::Void:
case clang::BuiltinType::NullPtr:
return 0;
case clang::BuiltinType::Bool:
case clang::BuiltinType::Char_U:
case clang::BuiltinType::UChar:
Updated to revision 123723 of LLVM, to bring in support for minimal type import functionality. llvm-svn: 123787
2011-01-19 07:32:05 +08:00
case clang::BuiltinType::WChar_U:
Make sure we don't assert if we have a child with zero byte size. Also we now say that "void *" value objects don't have children. llvm-svn: 123092
2011-01-09 06:26:47 +08:00
case clang::BuiltinType::Char16:
case clang::BuiltinType::Char32:
case clang::BuiltinType::UShort:
case clang::BuiltinType::UInt:
case clang::BuiltinType::ULong:
case clang::BuiltinType::ULongLong:
case clang::BuiltinType::UInt128:
case clang::BuiltinType::Char_S:
case clang::BuiltinType::SChar:
Updated to revision 123723 of LLVM, to bring in support for minimal type import functionality. llvm-svn: 123787
2011-01-19 07:32:05 +08:00
case clang::BuiltinType::WChar_S:
Make sure we don't assert if we have a child with zero byte size. Also we now say that "void *" value objects don't have children. llvm-svn: 123092
2011-01-09 06:26:47 +08:00
case clang::BuiltinType::Short:
case clang::BuiltinType::Int:
case clang::BuiltinType::Long:
case clang::BuiltinType::LongLong:
case clang::BuiltinType::Int128:
case clang::BuiltinType::Float:
case clang::BuiltinType::Double:
case clang::BuiltinType::LongDouble:
case clang::BuiltinType::Dependent:
case clang::BuiltinType::Overload:
case clang::BuiltinType::ObjCId:
case clang::BuiltinType::ObjCClass:
case clang::BuiltinType::ObjCSel:
Updated to use the latest LLVM/Clang, to pick up JIT changes. llvm-svn: 131391
2011-05-16 06:34:38 +08:00
case clang::BuiltinType::BoundMember:
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
case clang::BuiltinType::Half:
case clang::BuiltinType::ARCUnbridgedCast:
Cleanup some warnings. llvm-svn: 144200
2011-11-10 03:04:58 +08:00
case clang::BuiltinType::PseudoObject:
Make sure we don't assert if we have a child with zero byte size. Also we now say that "void *" value objects don't have children. llvm-svn: 123092
2011-01-09 06:26:47 +08:00
return 1;
}
break;
Added complete complex support for displaying and parsing complex types. llvm-svn: 123509
2011-01-15 10:52:14 +08:00
case clang::Type::Complex: return 1;
Fixed the "expression" command object to use the StackFrame::GetValueForExpressionPath() function and also hooked up better error reporting for when things fail. Fixed issues with trying to display children of pointers when none are supposed to be shown (no children for function pointers, and more like this). This was causing child value objects to be made that were correctly firing an assertion. llvm-svn: 121841
2010-12-15 13:08:08 +08:00
case clang::Type::Pointer: return 1;
case clang::Type::BlockPointer: return 0; // If block pointers don't have debug info, then no children for them
case clang::Type::LValueReference: return 1;
case clang::Type::RValueReference: return 1;
case clang::Type::MemberPointer: return 0;
case clang::Type::ConstantArray: return 0;
case clang::Type::IncompleteArray: return 0;
case clang::Type::VariableArray: return 0;
case clang::Type::DependentSizedArray: return 0;
case clang::Type::DependentSizedExtVector: return 0;
case clang::Type::Vector: return 0;
case clang::Type::ExtVector: return 0;
case clang::Type::FunctionProto: return 0; // When we function pointers, they have no children...
case clang::Type::FunctionNoProto: return 0; // When we function pointers, they have no children...
case clang::Type::UnresolvedUsing: return 0;
case clang::Type::Paren: return 0;
case clang::Type::Typedef: return ClangASTContext::GetNumPointeeChildren (cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr());
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated: return ClangASTContext::GetNumPointeeChildren (cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr());
Fixed the "expression" command object to use the StackFrame::GetValueForExpressionPath() function and also hooked up better error reporting for when things fail. Fixed issues with trying to display children of pointers when none are supposed to be shown (no children for function pointers, and more like this). This was causing child value objects to be made that were correctly firing an assertion. llvm-svn: 121841
2010-12-15 13:08:08 +08:00
case clang::Type::TypeOfExpr: return 0;
case clang::Type::TypeOf: return 0;
case clang::Type::Decltype: return 0;
case clang::Type::Record: return 0;
case clang::Type::Enum: return 1;
case clang::Type::TemplateTypeParm: return 1;
case clang::Type::SubstTemplateTypeParm: return 1;
case clang::Type::TemplateSpecialization: return 1;
case clang::Type::InjectedClassName: return 0;
case clang::Type::DependentName: return 1;
case clang::Type::DependentTemplateSpecialization: return 1;
case clang::Type::ObjCObject: return 0;
case clang::Type::ObjCInterface: return 0;
case clang::Type::ObjCObjectPointer: return 1;
default:
break;
}
return 0;
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
ClangASTContext::GetChildClangTypeAtIndex
(
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
ExecutionContext *exe_ctx,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
const char *parent_name,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t parent_clang_type,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
uint32_t idx,
bool transparent_pointers,
bool omit_empty_base_classes,
Fixed the global and static variables to always be in scope. Made it so that you can create synthetic children of array value objects. This is for creating array members when the array index is out of range. This comes in handy when you have a structure definition like: struct Collection { uint32_t count; Item array[0]; }; "array" has 1 item, but many times in practice there are more items in "item_array". This allows you to do: (lldb) target variable g_collection.array[3] To implement this, the get child at index has been modified to have a "ignore_array_bounds" boolean that can be set to true. llvm-svn: 134846
2011-07-10 04:12:33 +08:00
bool ignore_array_bounds,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
std::string& child_name,
uint32_t &child_byte_size,
int32_t &child_byte_offset,
uint32_t &child_bitfield_bit_size,
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
uint32_t &child_bitfield_bit_offset,
Fixed up the SBValue::GetExpressionPath() to be more correct under more circumstances. llvm-svn: 123957
2011-01-21 09:59:00 +08:00
bool &child_is_base_class,
bool &child_is_deref_of_parent
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
)
{
if (parent_clang_type)
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
return GetChildClangTypeAtIndex (exe_ctx,
getASTContext(),
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
parent_name,
parent_clang_type,
idx,
transparent_pointers,
omit_empty_base_classes,
Fixed the global and static variables to always be in scope. Made it so that you can create synthetic children of array value objects. This is for creating array members when the array index is out of range. This comes in handy when you have a structure definition like: struct Collection { uint32_t count; Item array[0]; }; "array" has 1 item, but many times in practice there are more items in "item_array". This allows you to do: (lldb) target variable g_collection.array[3] To implement this, the get child at index has been modified to have a "ignore_array_bounds" boolean that can be set to true. llvm-svn: 134846
2011-07-10 04:12:33 +08:00
ignore_array_bounds,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
child_name,
child_byte_size,
child_byte_offset,
child_bitfield_bit_size,
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
child_bitfield_bit_offset,
Fixed up the SBValue::GetExpressionPath() to be more correct under more circumstances. llvm-svn: 123957
2011-01-21 09:59:00 +08:00
child_is_base_class,
child_is_deref_of_parent);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return NULL;
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
ClangASTContext::GetChildClangTypeAtIndex
(
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
ExecutionContext *exe_ctx,
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
const char *parent_name,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t parent_clang_type,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
uint32_t idx,
bool transparent_pointers,
bool omit_empty_base_classes,
Fixed the global and static variables to always be in scope. Made it so that you can create synthetic children of array value objects. This is for creating array members when the array index is out of range. This comes in handy when you have a structure definition like: struct Collection { uint32_t count; Item array[0]; }; "array" has 1 item, but many times in practice there are more items in "item_array". This allows you to do: (lldb) target variable g_collection.array[3] To implement this, the get child at index has been modified to have a "ignore_array_bounds" boolean that can be set to true. llvm-svn: 134846
2011-07-10 04:12:33 +08:00
bool ignore_array_bounds,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
std::string& child_name,
uint32_t &child_byte_size,
int32_t &child_byte_offset,
uint32_t &child_bitfield_bit_size,
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
uint32_t &child_bitfield_bit_offset,
Fixed up the SBValue::GetExpressionPath() to be more correct under more circumstances. llvm-svn: 123957
2011-01-21 09:59:00 +08:00
bool &child_is_base_class,
bool &child_is_deref_of_parent
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
)
{
if (parent_clang_type == NULL)
return NULL;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (idx < ClangASTContext::GetNumChildren (ast, parent_clang_type, omit_empty_base_classes))
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
uint32_t bit_offset;
child_bitfield_bit_size = 0;
child_bitfield_bit_offset = 0;
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
child_is_base_class = false;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
QualType parent_qual_type(QualType::getFromOpaquePtr(parent_clang_type));
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
const clang::Type::TypeClass parent_type_class = parent_qual_type->getTypeClass();
switch (parent_type_class)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
case clang::Type::Builtin:
switch (cast<clang::BuiltinType>(parent_qual_type)->getKind())
{
case clang::BuiltinType::ObjCId:
case clang::BuiltinType::ObjCClass:
Fixes to Objective-C built-in type handling. Specifically, we fixed handling of the objc_class built-in type, which allowed us to pass named Objective-C objects to functions, call variable list -t on objects safely, etc. llvm-svn: 117249
2010-10-25 08:29:48 +08:00
child_name = "isa";
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
child_byte_size = ast->getTypeSize(ast->ObjCBuiltinClassTy) / CHAR_BIT;
return ast->ObjCBuiltinClassTy.getAsOpaquePtr();
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
default:
break;
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Record:
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
if (GetCompleteQualType (ast, parent_qual_type))
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
const RecordType *record_type = cast<RecordType>(parent_qual_type.getTypePtr());
const RecordDecl *record_decl = record_type->getDecl();
assert(record_decl);
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
const ASTRecordLayout &record_layout = ast->getASTRecordLayout(record_decl);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
uint32_t child_idx = 0;
const CXXRecordDecl *cxx_record_decl = dyn_cast<CXXRecordDecl>(record_decl);
if (cxx_record_decl)
{
// We might have base classes to print out first
CXXRecordDecl::base_class_const_iterator base_class, base_class_end;
for (base_class = cxx_record_decl->bases_begin(), base_class_end = cxx_record_decl->bases_end();
base_class != base_class_end;
++base_class)
{
const CXXRecordDecl *base_class_decl = NULL;
// Skip empty base classes
if (omit_empty_base_classes)
{
base_class_decl = cast<CXXRecordDecl>(base_class->getType()->getAs<RecordType>()->getDecl());
if (RecordHasFields(base_class_decl) == false)
continue;
}
if (idx == child_idx)
{
if (base_class_decl == NULL)
base_class_decl = cast<CXXRecordDecl>(base_class->getType()->getAs<RecordType>()->getDecl());
if (base_class->isVirtual())
Sped up the shutdown time on MacOSX by quite a bit by making sure any threads that we spawn let us know when they are going away and that we don't timeout waiting for a message from threads that have gone away. We also now don't expect the "k" packet (kill) to send a response. This greatly speeds up debugger shutdown performance. The test suite now runs quite a bit faster. Added a fix to the variable display code that fixes the display of base classes. We were assuming the virtual or normal base class offsets were being given in bit sizes, but they were being given as character sizes, so we needed to multiply the offset by 8. This wasn't affecting the expression parser, but it was affecting the correct display of C++ class base classes and all of their children. llvm-svn: 124024
2011-01-22 15:12:45 +08:00
bit_offset = record_layout.getVBaseClassOffset(base_class_decl).getQuantity() * 8;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
else
Sped up the shutdown time on MacOSX by quite a bit by making sure any threads that we spawn let us know when they are going away and that we don't timeout waiting for a message from threads that have gone away. We also now don't expect the "k" packet (kill) to send a response. This greatly speeds up debugger shutdown performance. The test suite now runs quite a bit faster. Added a fix to the variable display code that fixes the display of base classes. We were assuming the virtual or normal base class offsets were being given in bit sizes, but they were being given as character sizes, so we needed to multiply the offset by 8. This wasn't affecting the expression parser, but it was affecting the correct display of C++ class base classes and all of their children. llvm-svn: 124024
2011-01-22 15:12:45 +08:00
bit_offset = record_layout.getBaseClassOffset(base_class_decl).getQuantity() * 8;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
// Base classes should be a multiple of 8 bits in size
child_byte_offset = bit_offset/8;
Centralize all of the type name code so that we always strip the leading "struct ", "class ", and "union " from the start of any type names that are extracted from clang QualType objects. I had to fix test suite cases that were expecting the struct/union/class prefix to be there. llvm-svn: 134132
2011-06-30 10:28:26 +08:00
<rdar://problem/11113279> Fixed type lookups to "do the right thing". Prior to this fix, looking up a type using "foo::bar" would result in a type list that contains all types that had "bar" as a basename unless the symbol file was able to match fully qualified names (which our DWARF parser does not). This fix will allow type matches to be made based on the basename and then have the types that don't match filtered out. Types by name can be fully qualified, or partially qualified with the new "bool exact_match" parameter to the Module::FindTypes() method. This fixes some issue that we discovered with dynamic type resolution as well as improves the overall type lookups in LLDB. llvm-svn: 153482
2012-03-27 07:03:23 +08:00
child_name = ClangASTType::GetTypeNameForQualType(ast, base_class->getType());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
uint64_t clang_type_info_bit_size = ast->getTypeSize(base_class->getType());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
Fix comment typo. llvm-svn: 129621
2011-04-16 07:42:06 +08:00
// Base classes bit sizes should be a multiple of 8 bits in size
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
assert (clang_type_info_bit_size % 8 == 0);
child_byte_size = clang_type_info_bit_size / 8;
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
child_is_base_class = true;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return base_class->getType().getAsOpaquePtr();
}
// We don't increment the child index in the for loop since we might
// be skipping empty base classes
++child_idx;
}
}
// Make sure index is in range...
uint32_t field_idx = 0;
RecordDecl::field_iterator field, field_end;
for (field = record_decl->field_begin(), field_end = record_decl->field_end(); field != field_end; ++field, ++field_idx, ++child_idx)
{
if (idx == child_idx)
{
// Print the member type if requested
// Print the member name and equal sign
child_name.assign(field->getNameAsString().c_str());
// Figure out the type byte size (field_type_info.first) and
// alignment (field_type_info.second) from the AST context.
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
std::pair<uint64_t, unsigned> field_type_info = ast->getTypeInfo(field->getType());
Merged Eli Friedman's linux build changes where he added Makefile files that enabled LLVM make style building and made this compile LLDB on Mac OS X. We can now iterate on this to make the build work on both linux and macosx. llvm-svn: 108009
2010-07-10 04:39:50 +08:00
assert(field_idx < record_layout.getFieldCount());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
child_byte_size = field_type_info.first / 8;
// Figure out the field offset within the current struct/union/class type
bit_offset = record_layout.getFieldOffset (field_idx);
child_byte_offset = bit_offset / 8;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (ClangASTContext::FieldIsBitfield (ast, *field, child_bitfield_bit_size))
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
child_bitfield_bit_offset = bit_offset % 8;
return field->getType().getAsOpaquePtr();
}
}
}
break;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
if (GetCompleteQualType (ast, parent_qual_type))
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(parent_qual_type.getTypePtr());
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
assert (objc_class_type);
if (objc_class_type)
{
uint32_t child_idx = 0;
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
if (class_interface_decl)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
const ASTRecordLayout &interface_layout = ast->getASTObjCInterfaceLayout(class_interface_decl);
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
ObjCInterfaceDecl *superclass_interface_decl = class_interface_decl->getSuperClass();
if (superclass_interface_decl)
{
if (omit_empty_base_classes)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (ClangASTContext::GetNumChildren(ast, ast->getObjCInterfaceType(superclass_interface_decl).getAsOpaquePtr(), omit_empty_base_classes) > 0)
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
{
if (idx == 0)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
QualType ivar_qual_type(ast->getObjCInterfaceType(superclass_interface_decl));
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
child_name.assign(superclass_interface_decl->getNameAsString().c_str());
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
std::pair<uint64_t, unsigned> ivar_type_info = ast->getTypeInfo(ivar_qual_type.getTypePtr());
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
child_byte_size = ivar_type_info.first / 8;
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
child_byte_offset = 0;
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
child_is_base_class = true;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
return ivar_qual_type.getAsOpaquePtr();
}
++child_idx;
}
}
else
++child_idx;
}
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
const uint32_t superclass_idx = child_idx;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
if (idx < (child_idx + class_interface_decl->ivar_size()))
{
ObjCInterfaceDecl::ivar_iterator ivar_pos, ivar_end = class_interface_decl->ivar_end();
for (ivar_pos = class_interface_decl->ivar_begin(); ivar_pos != ivar_end; ++ivar_pos)
{
if (child_idx == idx)
{
Get the bit-field offset & size for ObjC ivars that are bitfields. <rdar://problem/10535460> lldb expression evaluation doesn't handle bit fields in ObjC classes properly llvm-svn: 146134
2011-12-08 10:53:10 +08:00
ObjCIvarDecl* ivar_decl = *ivar_pos;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
QualType ivar_qual_type(ivar_decl->getType());
child_name.assign(ivar_decl->getNameAsString().c_str());
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
std::pair<uint64_t, unsigned> ivar_type_info = ast->getTypeInfo(ivar_qual_type.getTypePtr());
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
child_byte_size = ivar_type_info.first / 8;
// Figure out the field offset within the current struct/union/class type
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
// For ObjC objects, we can't trust the bit offset we get from the Clang AST, since
// that doesn't account for the space taken up by unbacked properties, or from
// the changing size of base classes that are newer than this class.
// So if we have a process around that we can ask about this object, do so.
child_byte_offset = LLDB_INVALID_IVAR_OFFSET;
Converted the lldb_private::Process over to use the intrusive shared pointers. Changed the ExecutionContext over to use shared pointers for the target, process, thread and frame since these objects can easily go away at any time and any object that was holding onto an ExecutionContext was running the risk of using a bad object. Now that the shared pointers for target, process, thread and frame are just a single pointer (they all use the instrusive shared pointers) the execution context is much safer and still the same size. Made the shared pointers in the the ExecutionContext class protected and made accessors for all of the various ways to get at the pointers, references, and shared pointers. llvm-svn: 140298
2011-09-22 12:58:26 +08:00
Process *process = NULL;
if (exe_ctx)
process = exe_ctx->GetProcessPtr();
if (process)
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
{
Converted the lldb_private::Process over to use the intrusive shared pointers. Changed the ExecutionContext over to use shared pointers for the target, process, thread and frame since these objects can easily go away at any time and any object that was holding onto an ExecutionContext was running the risk of using a bad object. Now that the shared pointers for target, process, thread and frame are just a single pointer (they all use the instrusive shared pointers) the execution context is much safer and still the same size. Made the shared pointers in the the ExecutionContext class protected and made accessors for all of the various ways to get at the pointers, references, and shared pointers. llvm-svn: 140298
2011-09-22 12:58:26 +08:00
ObjCLanguageRuntime *objc_runtime = process->GetObjCLanguageRuntime();
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
if (objc_runtime != NULL)
{
Public API changes: - Completely new implementation of SBType - Various enhancements in several other classes Python synthetic children providers for std::vector<T>, std::list<T> and std::map<K,V>: - these return the actual elements into the container as the children of the container - basic template name parsing that works (hopefully) on both Clang and GCC - find them in examples/synthetic and in the test suite in functionalities/data-formatter/data-formatter-python-synth New summary string token ${svar : - the syntax is just the same as in ${var but this new token lets you read the values coming from the synthetic children provider instead of the actual children - Python providers above provide a synthetic child len that returns the number of elements into the container Full bug fix for the issue in which getting byte size for a non-complete type would crash LLDB Several other fixes, including: - inverted the order of arguments in the ClangASTType constructor - EvaluationPoint now only returns SharedPointer's to Target and Process - the help text for several type subcommands now correctly indicates argument-less options as such llvm-svn: 136504
2011-07-30 03:53:35 +08:00
ClangASTType parent_ast_type (ast, parent_qual_type.getAsOpaquePtr());
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
child_byte_offset = objc_runtime->GetByteOffsetForIvar (parent_ast_type, ivar_decl->getNameAsString().c_str());
}
}
Get the bit-field offset & size for ObjC ivars that are bitfields. <rdar://problem/10535460> lldb expression evaluation doesn't handle bit fields in ObjC classes properly llvm-svn: 146134
2011-12-08 10:53:10 +08:00
// Setting this to UINT32_MAX to make sure we don't compute it twice...
bit_offset = UINT32_MAX;
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
if (child_byte_offset == LLDB_INVALID_IVAR_OFFSET)
{
bit_offset = interface_layout.getFieldOffset (child_idx - superclass_idx);
child_byte_offset = bit_offset / 8;
}
Get the bit-field offset & size for ObjC ivars that are bitfields. <rdar://problem/10535460> lldb expression evaluation doesn't handle bit fields in ObjC classes properly llvm-svn: 146134
2011-12-08 10:53:10 +08:00
// Note, the ObjC Ivar Byte offset is just that, it doesn't account for the bit offset
// of a bitfield within its containing object. So regardless of where we get the byte
// offset from, we still need to get the bit offset for bitfields from the layout.
if (ClangASTContext::FieldIsBitfield (ast, ivar_decl, child_bitfield_bit_size))
{
if (bit_offset == UINT32_MAX)
bit_offset = interface_layout.getFieldOffset (child_idx - superclass_idx);
child_bitfield_bit_offset = bit_offset % 8;
}
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
return ivar_qual_type.getAsOpaquePtr();
}
++child_idx;
}
}
}
}
}
break;
case clang::Type::ObjCObjectPointer:
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ObjCObjectPointerType *pointer_type = cast<ObjCObjectPointerType>(parent_qual_type.getTypePtr());
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
QualType pointee_type = pointer_type->getPointeeType();
if (transparent_pointers && ClangASTContext::IsAggregateType (pointee_type.getAsOpaquePtr()))
{
Fixed up the SBValue::GetExpressionPath() to be more correct under more circumstances. llvm-svn: 123957
2011-01-21 09:59:00 +08:00
child_is_deref_of_parent = false;
bool tmp_child_is_deref_of_parent = false;
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
return GetChildClangTypeAtIndex (exe_ctx,
ast,
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
parent_name,
pointer_type->getPointeeType().getAsOpaquePtr(),
idx,
transparent_pointers,
omit_empty_base_classes,
Fixed the global and static variables to always be in scope. Made it so that you can create synthetic children of array value objects. This is for creating array members when the array index is out of range. This comes in handy when you have a structure definition like: struct Collection { uint32_t count; Item array[0]; }; "array" has 1 item, but many times in practice there are more items in "item_array". This allows you to do: (lldb) target variable g_collection.array[3] To implement this, the get child at index has been modified to have a "ignore_array_bounds" boolean that can be set to true. llvm-svn: 134846
2011-07-10 04:12:33 +08:00
ignore_array_bounds,
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
child_name,
child_byte_size,
child_byte_offset,
child_bitfield_bit_size,
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
child_bitfield_bit_offset,
Fixed up the SBValue::GetExpressionPath() to be more correct under more circumstances. llvm-svn: 123957
2011-01-21 09:59:00 +08:00
child_is_base_class,
tmp_child_is_deref_of_parent);
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
}
else
{
Fixed up the SBValue::GetExpressionPath() to be more correct under more circumstances. llvm-svn: 123957
2011-01-21 09:59:00 +08:00
child_is_deref_of_parent = true;
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
if (parent_name)
{
child_name.assign(1, '*');
child_name += parent_name;
}
// We have a pointer to an simple type
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
if (idx == 0 && GetCompleteQualType(ast, pointee_type))
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
std::pair<uint64_t, unsigned> clang_type_info = ast->getTypeInfo(pointee_type);
Added support for objective C built-in types: id, Class, and SEL. This involved watching for the objective C built-in types in DWARF and making sure when we convert the DWARF types into clang types that we use the appropriate ASTContext types. Added a way to find and dump types in lldb (something equivalent to gdb's "ptype" command): image lookup --type <TYPENAME> This only works for looking up types by name and won't work with variables. It also currently dumps out verbose internal information. I will modify it to dump more appropriate user level info in my next submission. Hookup up the "FindTypes()" functions in the SymbolFile and SymbolVendor so we can lookup types by name in one or more images. Fixed "image lookup --address <ADDRESS>" to be able to correctly show all symbol context information, but it will only show this extra information when the new "--verbose" flag is used. Updated to latest LLVM to get a few needed fixes. llvm-svn: 110089
2010-08-03 08:35:52 +08:00
assert(clang_type_info.first % 8 == 0);
child_byte_size = clang_type_info.first / 8;
child_byte_offset = 0;
return pointee_type.getAsOpaquePtr();
}
}
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::ConstantArray:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
const ConstantArrayType *array = cast<ConstantArrayType>(parent_qual_type.getTypePtr());
const uint64_t element_count = array->getSize().getLimitedValue();
Fixed the global and static variables to always be in scope. Made it so that you can create synthetic children of array value objects. This is for creating array members when the array index is out of range. This comes in handy when you have a structure definition like: struct Collection { uint32_t count; Item array[0]; }; "array" has 1 item, but many times in practice there are more items in "item_array". This allows you to do: (lldb) target variable g_collection.array[3] To implement this, the get child at index has been modified to have a "ignore_array_bounds" boolean that can be set to true. llvm-svn: 134846
2011-07-10 04:12:33 +08:00
if (ignore_array_bounds || idx < element_count)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (GetCompleteQualType (ast, array->getElementType()))
{
std::pair<uint64_t, unsigned> field_type_info = ast->getTypeInfo(array->getElementType());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
char element_name[64];
::snprintf (element_name, sizeof (element_name), "[%u]", idx);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
child_name.assign(element_name);
assert(field_type_info.first % 8 == 0);
child_byte_size = field_type_info.first / 8;
Fixed the global and static variables to always be in scope. Made it so that you can create synthetic children of array value objects. This is for creating array members when the array index is out of range. This comes in handy when you have a structure definition like: struct Collection { uint32_t count; Item array[0]; }; "array" has 1 item, but many times in practice there are more items in "item_array". This allows you to do: (lldb) target variable g_collection.array[3] To implement this, the get child at index has been modified to have a "ignore_array_bounds" boolean that can be set to true. llvm-svn: 134846
2011-07-10 04:12:33 +08:00
child_byte_offset = (int32_t)idx * (int32_t)child_byte_size;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return array->getElementType().getAsOpaquePtr();
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Pointer:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const PointerType *pointer_type = cast<PointerType>(parent_qual_type.getTypePtr());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
QualType pointee_type = pointer_type->getPointeeType();
Make sure we don't assert if we have a child with zero byte size. Also we now say that "void *" value objects don't have children. llvm-svn: 123092
2011-01-09 06:26:47 +08:00
// Don't dereference "void *" pointers
if (pointee_type->isVoidType())
return NULL;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (transparent_pointers && ClangASTContext::IsAggregateType (pointee_type.getAsOpaquePtr()))
{
Fixed up the SBValue::GetExpressionPath() to be more correct under more circumstances. llvm-svn: 123957
2011-01-21 09:59:00 +08:00
child_is_deref_of_parent = false;
bool tmp_child_is_deref_of_parent = false;
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
return GetChildClangTypeAtIndex (exe_ctx,
ast,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
parent_name,
pointer_type->getPointeeType().getAsOpaquePtr(),
idx,
transparent_pointers,
omit_empty_base_classes,
Fixed the global and static variables to always be in scope. Made it so that you can create synthetic children of array value objects. This is for creating array members when the array index is out of range. This comes in handy when you have a structure definition like: struct Collection { uint32_t count; Item array[0]; }; "array" has 1 item, but many times in practice there are more items in "item_array". This allows you to do: (lldb) target variable g_collection.array[3] To implement this, the get child at index has been modified to have a "ignore_array_bounds" boolean that can be set to true. llvm-svn: 134846
2011-07-10 04:12:33 +08:00
ignore_array_bounds,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
child_name,
child_byte_size,
child_byte_offset,
child_bitfield_bit_size,
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
child_bitfield_bit_offset,
Fixed up the SBValue::GetExpressionPath() to be more correct under more circumstances. llvm-svn: 123957
2011-01-21 09:59:00 +08:00
child_is_base_class,
tmp_child_is_deref_of_parent);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
else
{
Fixed up the SBValue::GetExpressionPath() to be more correct under more circumstances. llvm-svn: 123957
2011-01-21 09:59:00 +08:00
child_is_deref_of_parent = true;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (parent_name)
{
child_name.assign(1, '*');
child_name += parent_name;
}
// We have a pointer to an simple type
if (idx == 0)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
std::pair<uint64_t, unsigned> clang_type_info = ast->getTypeInfo(pointee_type);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
assert(clang_type_info.first % 8 == 0);
child_byte_size = clang_type_info.first / 8;
child_byte_offset = 0;
return pointee_type.getAsOpaquePtr();
}
}
}
break;
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
case clang::Type::LValueReference:
case clang::Type::RValueReference:
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ReferenceType *reference_type = cast<ReferenceType>(parent_qual_type.getTypePtr());
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
QualType pointee_type(reference_type->getPointeeType());
clang_type_t pointee_clang_type = pointee_type.getAsOpaquePtr();
if (transparent_pointers && ClangASTContext::IsAggregateType (pointee_clang_type))
{
Fixed up the SBValue::GetExpressionPath() to be more correct under more circumstances. llvm-svn: 123957
2011-01-21 09:59:00 +08:00
child_is_deref_of_parent = false;
bool tmp_child_is_deref_of_parent = false;
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
return GetChildClangTypeAtIndex (exe_ctx,
ast,
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
parent_name,
pointee_clang_type,
idx,
transparent_pointers,
omit_empty_base_classes,
Fixed the global and static variables to always be in scope. Made it so that you can create synthetic children of array value objects. This is for creating array members when the array index is out of range. This comes in handy when you have a structure definition like: struct Collection { uint32_t count; Item array[0]; }; "array" has 1 item, but many times in practice there are more items in "item_array". This allows you to do: (lldb) target variable g_collection.array[3] To implement this, the get child at index has been modified to have a "ignore_array_bounds" boolean that can be set to true. llvm-svn: 134846
2011-07-10 04:12:33 +08:00
ignore_array_bounds,
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
child_name,
child_byte_size,
child_byte_offset,
child_bitfield_bit_size,
child_bitfield_bit_offset,
Fixed up the SBValue::GetExpressionPath() to be more correct under more circumstances. llvm-svn: 123957
2011-01-21 09:59:00 +08:00
child_is_base_class,
tmp_child_is_deref_of_parent);
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
}
else
{
if (parent_name)
{
child_name.assign(1, '&');
child_name += parent_name;
}
// We have a pointer to an simple type
if (idx == 0)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
std::pair<uint64_t, unsigned> clang_type_info = ast->getTypeInfo(pointee_type);
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
assert(clang_type_info.first % 8 == 0);
child_byte_size = clang_type_info.first / 8;
child_byte_offset = 0;
return pointee_type.getAsOpaquePtr();
}
}
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Typedef:
Add support for looking up ivar offset from the ObjC runtime. llvm-svn: 133831
2011-06-25 06:03:24 +08:00
return GetChildClangTypeAtIndex (exe_ctx,
ast,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
parent_name,
Updated to latest Clang revision. This involved very minor changes, changing how we get the target type from a TypedefType, adding a parameter to EnumDecl::Create(), and other minor tweaks. llvm-svn: 121663
2010-12-13 09:26:27 +08:00
cast<TypedefType>(parent_qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(),
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
idx,
transparent_pointers,
omit_empty_base_classes,
Fixed the global and static variables to always be in scope. Made it so that you can create synthetic children of array value objects. This is for creating array members when the array index is out of range. This comes in handy when you have a structure definition like: struct Collection { uint32_t count; Item array[0]; }; "array" has 1 item, but many times in practice there are more items in "item_array". This allows you to do: (lldb) target variable g_collection.array[3] To implement this, the get child at index has been modified to have a "ignore_array_bounds" boolean that can be set to true. llvm-svn: 134846
2011-07-10 04:12:33 +08:00
ignore_array_bounds,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
child_name,
child_byte_size,
child_byte_offset,
child_bitfield_bit_size,
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
child_bitfield_bit_offset,
Fixed up the SBValue::GetExpressionPath() to be more correct under more circumstances. llvm-svn: 123957
2011-01-21 09:59:00 +08:00
child_is_base_class,
child_is_deref_of_parent);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
break;
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated:
return GetChildClangTypeAtIndex (exe_ctx,
ast,
parent_name,
cast<ElaboratedType>(parent_qual_type)->getNamedType().getAsOpaquePtr(),
idx,
transparent_pointers,
omit_empty_base_classes,
ignore_array_bounds,
child_name,
child_byte_size,
child_byte_offset,
child_bitfield_bit_size,
child_bitfield_bit_offset,
child_is_base_class,
child_is_deref_of_parent);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
default:
break;
}
}
Warnings cleanup patch from Jean-Daniel Dupas. llvm-svn: 109226
2010-07-23 23:37:46 +08:00
return NULL;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
static inline bool
BaseSpecifierIsEmpty (const CXXBaseSpecifier *b)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return ClangASTContext::RecordHasFields(b->getType()->getAsCXXRecordDecl()) == false;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
static uint32_t
GetNumBaseClasses (const CXXRecordDecl *cxx_record_decl, bool omit_empty_base_classes)
{
uint32_t num_bases = 0;
if (cxx_record_decl)
{
if (omit_empty_base_classes)
{
CXXRecordDecl::base_class_const_iterator base_class, base_class_end;
for (base_class = cxx_record_decl->bases_begin(), base_class_end = cxx_record_decl->bases_end();
base_class != base_class_end;
++base_class)
{
// Skip empty base classes
if (omit_empty_base_classes)
{
if (BaseSpecifierIsEmpty (base_class))
continue;
}
++num_bases;
}
}
else
num_bases = cxx_record_decl->getNumBases();
}
return num_bases;
}
static uint32_t
GetIndexForRecordBase
(
const RecordDecl *record_decl,
const CXXBaseSpecifier *base_spec,
bool omit_empty_base_classes
)
{
uint32_t child_idx = 0;
const CXXRecordDecl *cxx_record_decl = dyn_cast<CXXRecordDecl>(record_decl);
// const char *super_name = record_decl->getNameAsCString();
// const char *base_name = base_spec->getType()->getAs<RecordType>()->getDecl()->getNameAsCString();
// printf ("GetIndexForRecordChild (%s, %s)\n", super_name, base_name);
//
if (cxx_record_decl)
{
CXXRecordDecl::base_class_const_iterator base_class, base_class_end;
for (base_class = cxx_record_decl->bases_begin(), base_class_end = cxx_record_decl->bases_end();
base_class != base_class_end;
++base_class)
{
if (omit_empty_base_classes)
{
if (BaseSpecifierIsEmpty (base_class))
continue;
}
// printf ("GetIndexForRecordChild (%s, %s) base[%u] = %s\n", super_name, base_name,
// child_idx,
// base_class->getType()->getAs<RecordType>()->getDecl()->getNameAsCString());
//
//
if (base_class == base_spec)
return child_idx;
++child_idx;
}
}
return UINT32_MAX;
}
static uint32_t
GetIndexForRecordChild
(
const RecordDecl *record_decl,
NamedDecl *canonical_decl,
bool omit_empty_base_classes
)
{
uint32_t child_idx = GetNumBaseClasses (dyn_cast<CXXRecordDecl>(record_decl), omit_empty_base_classes);
// const CXXRecordDecl *cxx_record_decl = dyn_cast<CXXRecordDecl>(record_decl);
//
//// printf ("GetIndexForRecordChild (%s, %s)\n", record_decl->getNameAsCString(), canonical_decl->getNameAsCString());
// if (cxx_record_decl)
// {
// CXXRecordDecl::base_class_const_iterator base_class, base_class_end;
// for (base_class = cxx_record_decl->bases_begin(), base_class_end = cxx_record_decl->bases_end();
// base_class != base_class_end;
// ++base_class)
// {
// if (omit_empty_base_classes)
// {
// if (BaseSpecifierIsEmpty (base_class))
// continue;
// }
//
//// printf ("GetIndexForRecordChild (%s, %s) base[%u] = %s\n",
//// record_decl->getNameAsCString(),
//// canonical_decl->getNameAsCString(),
//// child_idx,
//// base_class->getType()->getAs<RecordType>()->getDecl()->getNameAsCString());
//
//
// CXXRecordDecl *curr_base_class_decl = cast<CXXRecordDecl>(base_class->getType()->getAs<RecordType>()->getDecl());
// if (curr_base_class_decl == canonical_decl)
// {
// return child_idx;
// }
// ++child_idx;
// }
// }
//
// const uint32_t num_bases = child_idx;
RecordDecl::field_iterator field, field_end;
for (field = record_decl->field_begin(), field_end = record_decl->field_end();
field != field_end;
++field, ++child_idx)
{
// printf ("GetIndexForRecordChild (%s, %s) field[%u] = %s\n",
// record_decl->getNameAsCString(),
// canonical_decl->getNameAsCString(),
// child_idx - num_bases,
// field->getNameAsCString());
if (field->getCanonicalDecl() == canonical_decl)
return child_idx;
}
return UINT32_MAX;
}
// Look for a child member (doesn't include base classes, but it does include
// their members) in the type hierarchy. Returns an index path into "clang_type"
// on how to reach the appropriate member.
//
// class A
// {
// public:
// int m_a;
// int m_b;
// };
//
// class B
// {
// };
//
// class C :
// public B,
// public A
// {
// };
//
// If we have a clang type that describes "class C", and we wanted to looked
// "m_b" in it:
//
// With omit_empty_base_classes == false we would get an integer array back with:
// { 1, 1 }
// The first index 1 is the child index for "class A" within class C
// The second index 1 is the child index for "m_b" within class A
//
// With omit_empty_base_classes == true we would get an integer array back with:
// { 0, 1 }
// The first index 0 is the child index for "class A" within class C (since class B doesn't have any members it doesn't count)
// The second index 1 is the child index for "m_b" within class A
size_t
ClangASTContext::GetIndexOfChildMemberWithName
(
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t clang_type,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
const char *name,
bool omit_empty_base_classes,
std::vector<uint32_t>& child_indexes
)
{
if (clang_type && name && name[0])
{
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Record:
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
if (GetCompleteQualType (ast, qual_type))
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
const RecordType *record_type = cast<RecordType>(qual_type.getTypePtr());
const RecordDecl *record_decl = record_type->getDecl();
assert(record_decl);
uint32_t child_idx = 0;
const CXXRecordDecl *cxx_record_decl = dyn_cast<CXXRecordDecl>(record_decl);
// Try and find a field that matches NAME
RecordDecl::field_iterator field, field_end;
StringRef name_sref(name);
for (field = record_decl->field_begin(), field_end = record_decl->field_end();
field != field_end;
++field, ++child_idx)
{
if (field->getName().equals (name_sref))
{
// We have to add on the number of base classes to this index!
child_indexes.push_back (child_idx + GetNumBaseClasses (cxx_record_decl, omit_empty_base_classes));
return child_indexes.size();
}
}
if (cxx_record_decl)
{
const RecordDecl *parent_record_decl = cxx_record_decl;
//printf ("parent = %s\n", parent_record_decl->getNameAsCString());
//const Decl *root_cdecl = cxx_record_decl->getCanonicalDecl();
// Didn't find things easily, lets let clang do its thang...
This change brings in the latest LLVM/Clang, and completes the support in the LLDB expression parser for incomplete types. Clang now imports types lazily, and we complete those types as necessary. Changes include: - ClangASTSource now supports three APIs which it passes to ClangExpressionDeclMap. CompleteType completes a TagDecl or an ObjCInterfaceDecl when needed; FindExternalVisibleDecls finds named entities that are visible in the expression's scope; and FindExternalLexicalDecls performs a (potentially restricted) search for entities inside a lexical scope like a namespace. These changes mean that entities in namespaces should work normally. - The SymbolFileDWARF code for searching a context for a specific name is now more general, and can search arbitrary contexts. - We are continuing to adapt our calls into LLVM from interfaces that take start and end iterators when accepting multiple items to interfaces that use ArrayRef. - I have cleaned up some code, especially our use of namespaces. This change is neutral for our testsuite and greatly improves correctness for large programs (like Clang) with complicated type systems. It should also lay the groundwork for improving the expression parser's performance as we are lazier and lazier about providing type information. llvm-svn: 136555
2011-07-30 10:42:06 +08:00
IdentifierInfo & ident_ref = ast->Idents.get(name_sref);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
DeclarationName decl_name(&ident_ref);
CXXBasePaths paths;
if (cxx_record_decl->lookupInBases(CXXRecordDecl::FindOrdinaryMember,
decl_name.getAsOpaquePtr(),
paths))
{
CXXBasePaths::const_paths_iterator path, path_end = paths.end();
for (path = paths.begin(); path != path_end; ++path)
{
const size_t num_path_elements = path->size();
for (size_t e=0; e<num_path_elements; ++e)
{
CXXBasePathElement elem = (*path)[e];
child_idx = GetIndexForRecordBase (parent_record_decl, elem.Base, omit_empty_base_classes);
if (child_idx == UINT32_MAX)
{
child_indexes.clear();
return 0;
}
else
{
child_indexes.push_back (child_idx);
parent_record_decl = cast<RecordDecl>(elem.Base->getType()->getAs<RecordType>()->getDecl());
}
}
DeclContext::lookup_iterator named_decl_pos;
for (named_decl_pos = path->Decls.first;
named_decl_pos != path->Decls.second && parent_record_decl;
++named_decl_pos)
{
//printf ("path[%zu] = %s\n", child_indexes.size(), (*named_decl_pos)->getNameAsCString());
child_idx = GetIndexForRecordChild (parent_record_decl, *named_decl_pos, omit_empty_base_classes);
if (child_idx == UINT32_MAX)
{
child_indexes.clear();
return 0;
}
else
{
child_indexes.push_back (child_idx);
}
}
}
return child_indexes.size();
}
}
}
break;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
if (GetCompleteQualType (ast, qual_type))
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
{
StringRef name_sref(name);
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(qual_type.getTypePtr());
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
assert (objc_class_type);
if (objc_class_type)
{
uint32_t child_idx = 0;
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
if (class_interface_decl)
{
ObjCInterfaceDecl::ivar_iterator ivar_pos, ivar_end = class_interface_decl->ivar_end();
ObjCInterfaceDecl *superclass_interface_decl = class_interface_decl->getSuperClass();
Bug #: 8447030 Fixed an issue with ClangASTContext::GetIndexOfChildMemberWithName() where objective C ivars were not being found correctly if they were the second or higher child. llvm-svn: 114258
2010-09-18 10:11:07 +08:00
for (ivar_pos = class_interface_decl->ivar_begin(); ivar_pos != ivar_end; ++ivar_pos, ++child_idx)
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
{
const ObjCIvarDecl* ivar_decl = *ivar_pos;
if (ivar_decl->getName().equals (name_sref))
{
if ((!omit_empty_base_classes && superclass_interface_decl) ||
( omit_empty_base_classes && ObjCDeclHasIVars (superclass_interface_decl, true)))
++child_idx;
child_indexes.push_back (child_idx);
return child_indexes.size();
}
}
if (superclass_interface_decl)
{
// The super class index is always zero for ObjC classes,
// so we push it onto the child indexes in case we find
// an ivar in our superclass...
child_indexes.push_back (0);
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (GetIndexOfChildMemberWithName (ast,
ast->getObjCInterfaceType(superclass_interface_decl).getAsOpaquePtr(),
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
name,
omit_empty_base_classes,
child_indexes))
{
// We did find an ivar in a superclass so just
// return the results!
return child_indexes.size();
}
// We didn't find an ivar matching "name" in our
// superclass, pop the superclass zero index that
// we pushed on above.
child_indexes.pop_back();
}
}
}
}
break;
case clang::Type::ObjCObjectPointer:
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return GetIndexOfChildMemberWithName (ast,
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
cast<ObjCObjectPointerType>(qual_type.getTypePtr())->getPointeeType().getAsOpaquePtr(),
name,
omit_empty_base_classes,
child_indexes);
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::ConstantArray:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
// const ConstantArrayType *array = cast<ConstantArrayType>(parent_qual_type.getTypePtr());
// const uint64_t element_count = array->getSize().getLimitedValue();
//
// if (idx < element_count)
// {
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
// std::pair<uint64_t, unsigned> field_type_info = ast->getTypeInfo(array->getElementType());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
//
// char element_name[32];
// ::snprintf (element_name, sizeof (element_name), "%s[%u]", parent_name ? parent_name : "", idx);
//
// child_name.assign(element_name);
// assert(field_type_info.first % 8 == 0);
// child_byte_size = field_type_info.first / 8;
// child_byte_offset = idx * child_byte_size;
// return array->getElementType().getAsOpaquePtr();
// }
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
// case clang::Type::MemberPointerType:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
// {
// MemberPointerType *mem_ptr_type = cast<MemberPointerType>(qual_type.getTypePtr());
// QualType pointee_type = mem_ptr_type->getPointeeType();
//
// if (ClangASTContext::IsAggregateType (pointee_type.getAsOpaquePtr()))
// {
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
// return GetIndexOfChildWithName (ast,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
// mem_ptr_type->getPointeeType().getAsOpaquePtr(),
// name);
// }
// }
// break;
//
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::LValueReference:
case clang::Type::RValueReference:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ReferenceType *reference_type = cast<ReferenceType>(qual_type.getTypePtr());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
QualType pointee_type = reference_type->getPointeeType();
if (ClangASTContext::IsAggregateType (pointee_type.getAsOpaquePtr()))
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return GetIndexOfChildMemberWithName (ast,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
reference_type->getPointeeType().getAsOpaquePtr(),
name,
omit_empty_base_classes,
child_indexes);
}
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Pointer:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const PointerType *pointer_type = cast<PointerType>(qual_type.getTypePtr());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
QualType pointee_type = pointer_type->getPointeeType();
if (ClangASTContext::IsAggregateType (pointee_type.getAsOpaquePtr()))
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return GetIndexOfChildMemberWithName (ast,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
pointer_type->getPointeeType().getAsOpaquePtr(),
name,
omit_empty_base_classes,
child_indexes);
}
else
{
// if (parent_name)
// {
// child_name.assign(1, '*');
// child_name += parent_name;
// }
//
// // We have a pointer to an simple type
// if (idx == 0)
// {
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
// std::pair<uint64_t, unsigned> clang_type_info = ast->getTypeInfo(pointee_type);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
// assert(clang_type_info.first % 8 == 0);
// child_byte_size = clang_type_info.first / 8;
// child_byte_offset = 0;
// return pointee_type.getAsOpaquePtr();
// }
}
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Typedef:
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return GetIndexOfChildMemberWithName (ast,
Updated to latest Clang revision. This involved very minor changes, changing how we get the target type from a TypedefType, adding a parameter to EnumDecl::Create(), and other minor tweaks. llvm-svn: 121663
2010-12-13 09:26:27 +08:00
cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(),
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
name,
omit_empty_base_classes,
child_indexes);
default:
break;
}
}
return 0;
}
// Get the index of the child of "clang_type" whose name matches. This function
// doesn't descend into the children, but only looks one level deep and name
// matches can include base class names.
uint32_t
ClangASTContext::GetIndexOfChildWithName
(
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t clang_type,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
const char *name,
bool omit_empty_base_classes
)
{
if (clang_type && name && name[0])
{
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
switch (type_class)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Record:
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
if (GetCompleteQualType (ast, qual_type))
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
const RecordType *record_type = cast<RecordType>(qual_type.getTypePtr());
const RecordDecl *record_decl = record_type->getDecl();
assert(record_decl);
uint32_t child_idx = 0;
const CXXRecordDecl *cxx_record_decl = dyn_cast<CXXRecordDecl>(record_decl);
if (cxx_record_decl)
{
CXXRecordDecl::base_class_const_iterator base_class, base_class_end;
for (base_class = cxx_record_decl->bases_begin(), base_class_end = cxx_record_decl->bases_end();
base_class != base_class_end;
++base_class)
{
// Skip empty base classes
CXXRecordDecl *base_class_decl = cast<CXXRecordDecl>(base_class->getType()->getAs<RecordType>()->getDecl());
if (omit_empty_base_classes && RecordHasFields(base_class_decl) == false)
continue;
<rdar://problem/11113279> Fixed type lookups to "do the right thing". Prior to this fix, looking up a type using "foo::bar" would result in a type list that contains all types that had "bar" as a basename unless the symbol file was able to match fully qualified names (which our DWARF parser does not). This fix will allow type matches to be made based on the basename and then have the types that don't match filtered out. Types by name can be fully qualified, or partially qualified with the new "bool exact_match" parameter to the Module::FindTypes() method. This fixes some issue that we discovered with dynamic type resolution as well as improves the overall type lookups in LLDB. llvm-svn: 153482
2012-03-27 07:03:23 +08:00
std::string base_class_type_name (ClangASTType::GetTypeNameForQualType(ast, base_class->getType()));
Centralize all of the type name code so that we always strip the leading "struct ", "class ", and "union " from the start of any type names that are extracted from clang QualType objects. I had to fix test suite cases that were expecting the struct/union/class prefix to be there. llvm-svn: 134132
2011-06-30 10:28:26 +08:00
if (base_class_type_name.compare (name) == 0)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return child_idx;
++child_idx;
}
}
// Try and find a field that matches NAME
RecordDecl::field_iterator field, field_end;
StringRef name_sref(name);
for (field = record_decl->field_begin(), field_end = record_decl->field_end();
field != field_end;
++field, ++child_idx)
{
if (field->getName().equals (name_sref))
return child_idx;
}
}
break;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
if (GetCompleteQualType (ast, qual_type))
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
{
StringRef name_sref(name);
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ObjCObjectType *objc_class_type = dyn_cast<ObjCObjectType>(qual_type.getTypePtr());
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
assert (objc_class_type);
if (objc_class_type)
{
uint32_t child_idx = 0;
ObjCInterfaceDecl *class_interface_decl = objc_class_type->getInterface();
if (class_interface_decl)
{
ObjCInterfaceDecl::ivar_iterator ivar_pos, ivar_end = class_interface_decl->ivar_end();
ObjCInterfaceDecl *superclass_interface_decl = class_interface_decl->getSuperClass();
for (ivar_pos = class_interface_decl->ivar_begin(); ivar_pos != ivar_end; ++ivar_pos)
{
const ObjCIvarDecl* ivar_decl = *ivar_pos;
if (ivar_decl->getName().equals (name_sref))
{
if ((!omit_empty_base_classes && superclass_interface_decl) ||
( omit_empty_base_classes && ObjCDeclHasIVars (superclass_interface_decl, true)))
++child_idx;
return child_idx;
}
}
if (superclass_interface_decl)
{
if (superclass_interface_decl->getName().equals (name_sref))
return 0;
}
}
}
}
break;
case clang::Type::ObjCObjectPointer:
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return GetIndexOfChildWithName (ast,
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
cast<ObjCObjectPointerType>(qual_type.getTypePtr())->getPointeeType().getAsOpaquePtr(),
name,
omit_empty_base_classes);
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::ConstantArray:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
// const ConstantArrayType *array = cast<ConstantArrayType>(parent_qual_type.getTypePtr());
// const uint64_t element_count = array->getSize().getLimitedValue();
//
// if (idx < element_count)
// {
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
// std::pair<uint64_t, unsigned> field_type_info = ast->getTypeInfo(array->getElementType());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
//
// char element_name[32];
// ::snprintf (element_name, sizeof (element_name), "%s[%u]", parent_name ? parent_name : "", idx);
//
// child_name.assign(element_name);
// assert(field_type_info.first % 8 == 0);
// child_byte_size = field_type_info.first / 8;
// child_byte_offset = idx * child_byte_size;
// return array->getElementType().getAsOpaquePtr();
// }
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
// case clang::Type::MemberPointerType:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
// {
// MemberPointerType *mem_ptr_type = cast<MemberPointerType>(qual_type.getTypePtr());
// QualType pointee_type = mem_ptr_type->getPointeeType();
//
// if (ClangASTContext::IsAggregateType (pointee_type.getAsOpaquePtr()))
// {
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
// return GetIndexOfChildWithName (ast,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
// mem_ptr_type->getPointeeType().getAsOpaquePtr(),
// name);
// }
// }
// break;
//
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::LValueReference:
case clang::Type::RValueReference:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ReferenceType *reference_type = cast<ReferenceType>(qual_type.getTypePtr());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
QualType pointee_type = reference_type->getPointeeType();
if (ClangASTContext::IsAggregateType (pointee_type.getAsOpaquePtr()))
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return GetIndexOfChildWithName (ast,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
reference_type->getPointeeType().getAsOpaquePtr(),
name,
omit_empty_base_classes);
}
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Pointer:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const PointerType *pointer_type = cast<PointerType>(qual_type.getTypePtr());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
QualType pointee_type = pointer_type->getPointeeType();
if (ClangASTContext::IsAggregateType (pointee_type.getAsOpaquePtr()))
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return GetIndexOfChildWithName (ast,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
pointer_type->getPointeeType().getAsOpaquePtr(),
name,
omit_empty_base_classes);
}
else
{
// if (parent_name)
// {
// child_name.assign(1, '*');
// child_name += parent_name;
// }
//
// // We have a pointer to an simple type
// if (idx == 0)
// {
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
// std::pair<uint64_t, unsigned> clang_type_info = ast->getTypeInfo(pointee_type);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
// assert(clang_type_info.first % 8 == 0);
// child_byte_size = clang_type_info.first / 8;
// child_byte_offset = 0;
// return pointee_type.getAsOpaquePtr();
// }
}
}
break;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Typedef:
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return GetIndexOfChildWithName (ast,
Updated to latest Clang revision. This involved very minor changes, changing how we get the target type from a TypedefType, adding a parameter to EnumDecl::Create(), and other minor tweaks. llvm-svn: 121663
2010-12-13 09:26:27 +08:00
cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(),
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
name,
omit_empty_base_classes);
default:
break;
}
}
return UINT32_MAX;
}
#pragma mark TagType
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::SetTagTypeKind (clang_type_t tag_clang_type, int kind)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (tag_clang_type)
{
QualType tag_qual_type(QualType::getFromOpaquePtr(tag_clang_type));
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::Type *clang_type = tag_qual_type.getTypePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (clang_type)
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const TagType *tag_type = dyn_cast<TagType>(clang_type);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (tag_type)
{
TagDecl *tag_decl = dyn_cast<TagDecl>(tag_type->getDecl());
if (tag_decl)
{
tag_decl->setTagKind ((TagDecl::TagKind)kind);
return true;
}
}
}
}
return false;
}
#pragma mark DeclContext Functions
DeclContext *
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::GetDeclContextForType (clang_type_t clang_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type == NULL)
return NULL;
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
This change brings in the latest LLVM/Clang, and completes the support in the LLDB expression parser for incomplete types. Clang now imports types lazily, and we complete those types as necessary. Changes include: - ClangASTSource now supports three APIs which it passes to ClangExpressionDeclMap. CompleteType completes a TagDecl or an ObjCInterfaceDecl when needed; FindExternalVisibleDecls finds named entities that are visible in the expression's scope; and FindExternalLexicalDecls performs a (potentially restricted) search for entities inside a lexical scope like a namespace. These changes mean that entities in namespaces should work normally. - The SymbolFileDWARF code for searching a context for a specific name is now more general, and can search arbitrary contexts. - We are continuing to adapt our calls into LLVM from interfaces that take start and end iterators when accepting multiple items to interfaces that use ArrayRef. - I have cleaned up some code, especially our use of namespaces. This change is neutral for our testsuite and greatly improves correctness for large programs (like Clang) with complicated type systems. It should also lay the groundwork for improving the expression parser's performance as we are lazier and lazier about providing type information. llvm-svn: 136555
2011-07-30 10:42:06 +08:00
case clang::Type::UnaryTransform: break;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
case clang::Type::FunctionNoProto: break;
case clang::Type::FunctionProto: break;
case clang::Type::IncompleteArray: break;
case clang::Type::VariableArray: break;
case clang::Type::ConstantArray: break;
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
case clang::Type::DependentSizedArray: break;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
case clang::Type::ExtVector: break;
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
case clang::Type::DependentSizedExtVector: break;
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
case clang::Type::Vector: break;
case clang::Type::Builtin: break;
case clang::Type::BlockPointer: break;
case clang::Type::Pointer: break;
case clang::Type::LValueReference: break;
case clang::Type::RValueReference: break;
case clang::Type::MemberPointer: break;
case clang::Type::Complex: break;
case clang::Type::ObjCObject: break;
case clang::Type::ObjCInterface: return cast<ObjCObjectType>(qual_type.getTypePtr())->getInterface();
case clang::Type::ObjCObjectPointer: return ClangASTContext::GetDeclContextForType (cast<ObjCObjectPointerType>(qual_type.getTypePtr())->getPointeeType().getAsOpaquePtr());
case clang::Type::Record: return cast<RecordType>(qual_type)->getDecl();
case clang::Type::Enum: return cast<EnumType>(qual_type)->getDecl();
Updated to latest Clang revision. This involved very minor changes, changing how we get the target type from a TypedefType, adding a parameter to EnumDecl::Create(), and other minor tweaks. llvm-svn: 121663
2010-12-13 09:26:27 +08:00
case clang::Type::Typedef: return ClangASTContext::GetDeclContextForType (cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr());
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated: return ClangASTContext::GetDeclContextForType (cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr());
Created lldb::LanguageType by moving an enumeration from the lldb_private::Language class into the enumerations header so it can be freely used by other interfaces. Added correct objective C class support to the DWARF symbol parser. Prior to this fix we were parsing objective C classes as C++ classes and now that the expression parser is ready to call functions we need to make sure the objective C classes have correct AST types. llvm-svn: 109574
2010-07-28 10:04:09 +08:00
case clang::Type::TypeOfExpr: break;
case clang::Type::TypeOf: break;
case clang::Type::Decltype: break;
//case clang::Type::QualifiedName: break;
case clang::Type::TemplateSpecialization: break;
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
case clang::Type::DependentTemplateSpecialization: break;
case clang::Type::TemplateTypeParm: break;
case clang::Type::SubstTemplateTypeParm: break;
case clang::Type::SubstTemplateTypeParmPack:break;
case clang::Type::PackExpansion: break;
case clang::Type::UnresolvedUsing: break;
case clang::Type::Paren: break;
case clang::Type::Attributed: break;
case clang::Type::Auto: break;
case clang::Type::InjectedClassName: break;
case clang::Type::DependentName: break;
Added more functionality to Range template classes in RangeMap.h and converted remaining DWARF areas that were using ranges over to this class. Also converted lldb_private::Block to use it. llvm-svn: 141460
2011-10-08 08:49:15 +08:00
case clang::Type::Atomic: break;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
// No DeclContext in this type...
return NULL;
}
#pragma mark Namespace Declarations
NamespaceDecl *
Make sure we create only unique one namespace per AST when parsing the DWARF. llvm-svn: 142005
2011-10-15 05:34:45 +08:00
ClangASTContext::GetUniqueNamespaceDeclaration (const char *name, DeclContext *decl_ctx)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Make sure we create only unique one namespace per AST when parsing the DWARF. llvm-svn: 142005
2011-10-15 05:34:45 +08:00
NamespaceDecl *namespace_decl = NULL;
Fixed some warnings after enabling some stricter warnings in the Xcode project settings. Also fixed an issue where we weren't creating anonymous namepaces correctly: <rdar://problem/10371295> llvm-svn: 143403
2011-11-01 07:51:19 +08:00
ASTContext *ast = getASTContext();
TranslationUnitDecl *translation_unit_decl = ast->getTranslationUnitDecl ();
if (decl_ctx == NULL)
decl_ctx = translation_unit_decl;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (name)
{
Make sure we create only unique one namespace per AST when parsing the DWARF. llvm-svn: 142005
2011-10-15 05:34:45 +08:00
IdentifierInfo &identifier_info = ast->Idents.get(name);
DeclarationName decl_name (&identifier_info);
clang::DeclContext::lookup_result result = decl_ctx->lookup(decl_name);
for (clang::DeclContext::lookup_iterator pos = result.first, end = result.second; pos != end; ++pos)
{
namespace_decl = dyn_cast<clang::NamespaceDecl>(*pos);
if (namespace_decl)
return namespace_decl;
}
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
namespace_decl = NamespaceDecl::Create(*ast,
decl_ctx,
false,
SourceLocation(),
SourceLocation(),
&identifier_info,
NULL);
Make sure we create only unique one namespace per AST when parsing the DWARF. llvm-svn: 142005
2011-10-15 05:34:45 +08:00
Fixed some warnings after enabling some stricter warnings in the Xcode project settings. Also fixed an issue where we weren't creating anonymous namepaces correctly: <rdar://problem/10371295> llvm-svn: 143403
2011-11-01 07:51:19 +08:00
decl_ctx->addDecl (namespace_decl);
}
else
{
if (decl_ctx == translation_unit_decl)
{
namespace_decl = translation_unit_decl->getAnonymousNamespace();
if (namespace_decl)
return namespace_decl;
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
namespace_decl = NamespaceDecl::Create(*ast,
decl_ctx,
false,
SourceLocation(),
SourceLocation(),
NULL,
NULL);
Fixed some warnings after enabling some stricter warnings in the Xcode project settings. Also fixed an issue where we weren't creating anonymous namepaces correctly: <rdar://problem/10371295> llvm-svn: 143403
2011-11-01 07:51:19 +08:00
translation_unit_decl->setAnonymousNamespace (namespace_decl);
translation_unit_decl->addDecl (namespace_decl);
assert (namespace_decl == translation_unit_decl->getAnonymousNamespace());
}
else
{
NamespaceDecl *parent_namespace_decl = cast<NamespaceDecl>(decl_ctx);
if (parent_namespace_decl)
{
namespace_decl = parent_namespace_decl->getAnonymousNamespace();
if (namespace_decl)
return namespace_decl;
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
namespace_decl = NamespaceDecl::Create(*ast,
decl_ctx,
false,
SourceLocation(),
SourceLocation(),
NULL,
NULL);
Fixed some warnings after enabling some stricter warnings in the Xcode project settings. Also fixed an issue where we weren't creating anonymous namepaces correctly: <rdar://problem/10371295> llvm-svn: 143403
2011-11-01 07:51:19 +08:00
parent_namespace_decl->setAnonymousNamespace (namespace_decl);
parent_namespace_decl->addDecl (namespace_decl);
assert (namespace_decl == parent_namespace_decl->getAnonymousNamespace());
}
else
{
// BAD!!!
}
}
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
Fixed some warnings after enabling some stricter warnings in the Xcode project settings. Also fixed an issue where we weren't creating anonymous namepaces correctly: <rdar://problem/10371295> llvm-svn: 143403
2011-11-01 07:51:19 +08:00
if (namespace_decl)
{
// If we make it here, we are creating the anonymous namespace decl
// for the first time, so we need to do the using directive magic
// like SEMA does
UsingDirectiveDecl* using_directive_decl = UsingDirectiveDecl::Create (*ast,
decl_ctx,
SourceLocation(),
SourceLocation(),
NestedNameSpecifierLoc(),
SourceLocation(),
namespace_decl,
decl_ctx);
using_directive_decl->setImplicit();
decl_ctx->addDecl(using_directive_decl);
}
}
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
#ifdef LLDB_CONFIGURATION_DEBUG
Fixed some warnings after enabling some stricter warnings in the Xcode project settings. Also fixed an issue where we weren't creating anonymous namepaces correctly: <rdar://problem/10371295> llvm-svn: 143403
2011-11-01 07:51:19 +08:00
VerifyDecl(namespace_decl);
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
#endif
Make sure we create only unique one namespace per AST when parsing the DWARF. llvm-svn: 142005
2011-10-15 05:34:45 +08:00
return namespace_decl;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
#pragma mark Function Types
FunctionDecl *
Add function decls to their parent decl context. llvm-svn: 142011
2011-10-15 06:47:18 +08:00
ClangASTContext::CreateFunctionDeclaration (DeclContext *decl_ctx, const char *name, clang_type_t function_clang_type, int storage, bool is_inline)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Add function decls to their parent decl context. llvm-svn: 142011
2011-10-15 06:47:18 +08:00
FunctionDecl *func_decl = NULL;
ASTContext *ast = getASTContext();
if (decl_ctx == NULL)
decl_ctx = ast->getTranslationUnitDecl();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
Add function decls to their parent decl context. llvm-svn: 142011
2011-10-15 06:47:18 +08:00
if (name && name[0])
{
func_decl = FunctionDecl::Create (*ast,
decl_ctx,
SourceLocation(),
SourceLocation(),
DeclarationName (&ast->Idents.get(name)),
QualType::getFromOpaquePtr(function_clang_type),
NULL,
(FunctionDecl::StorageClass)storage,
(FunctionDecl::StorageClass)storage,
is_inline);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Add function decls to their parent decl context. llvm-svn: 142011
2011-10-15 06:47:18 +08:00
else
{
func_decl = FunctionDecl::Create (*ast,
decl_ctx,
SourceLocation(),
SourceLocation(),
DeclarationName (),
QualType::getFromOpaquePtr(function_clang_type),
NULL,
(FunctionDecl::StorageClass)storage,
(FunctionDecl::StorageClass)storage,
is_inline);
}
if (func_decl)
decl_ctx->addDecl (func_decl);
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
#ifdef LLDB_CONFIGURATION_DEBUG
VerifyDecl(func_decl);
#endif
Add function decls to their parent decl context. llvm-svn: 142011
2011-10-15 06:47:18 +08:00
return func_decl;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ClangASTContext::CreateFunctionType (ASTContext *ast,
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t result_type,
clang_type_t *args,
Made CreateFunctionType static. Also fixed the spelling for CreateParameterDeclaration. llvm-svn: 114111
2010-09-17 04:40:25 +08:00
unsigned num_args,
bool is_variadic,
unsigned type_quals)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
assert (ast != NULL);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
std::vector<QualType> qual_type_args;
for (unsigned i=0; i<num_args; ++i)
qual_type_args.push_back (QualType::getFromOpaquePtr(args[i]));
// TODO: Detect calling convention in DWARF?
Updated to revision 123723 of LLVM, to bring in support for minimal type import functionality. llvm-svn: 123787
2011-01-19 07:32:05 +08:00
FunctionProtoType::ExtProtoInfo proto_info;
proto_info.Variadic = is_variadic;
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
proto_info.ExceptionSpecType = EST_None;
Updated to revision 123723 of LLVM, to bring in support for minimal type import functionality. llvm-svn: 123787
2011-01-19 07:32:05 +08:00
proto_info.TypeQuals = type_quals;
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
proto_info.RefQualifier = RQ_None;
Updated to revision 123723 of LLVM, to bring in support for minimal type import functionality. llvm-svn: 123787
2011-01-19 07:32:05 +08:00
proto_info.NumExceptions = 0;
proto_info.Exceptions = NULL;
Add function decls to their parent decl context. llvm-svn: 142011
2011-10-15 06:47:18 +08:00
return ast->getFunctionType (QualType::getFromOpaquePtr(result_type),
qual_type_args.empty() ? NULL : &qual_type_args.front(),
qual_type_args.size(),
proto_info).getAsOpaquePtr(); // NoReturn);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
ParmVarDecl *
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::CreateParameterDeclaration (const char *name, clang_type_t param_type, int storage)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast = getASTContext();
assert (ast != NULL);
return ParmVarDecl::Create(*ast,
ast->getTranslationUnitDecl(),
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
SourceLocation(),
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
SourceLocation(),
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
name && name[0] ? &ast->Idents.get(name) : NULL,
Made CreateFunctionType static. Also fixed the spelling for CreateParameterDeclaration. llvm-svn: 114111
2010-09-17 04:40:25 +08:00
QualType::getFromOpaquePtr(param_type),
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
NULL,
(VarDecl::StorageClass)storage,
(VarDecl::StorageClass)storage,
0);
}
void
ClangASTContext::SetFunctionParameters (FunctionDecl *function_decl, ParmVarDecl **params, unsigned num_params)
{
if (function_decl)
Updated LLVM/Clang to pull in the latest ARM disassembler. This involved minor changes to the way we report Objective-C methods, as well as cosmetic changes and added parameters for a variety of Clang APIs. llvm-svn: 141437
2011-10-08 07:18:13 +08:00
function_decl->setParams (ArrayRef<ParmVarDecl*>(params, num_params));
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
#pragma mark Array Types
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
ClangASTContext::CreateArrayType (clang_type_t element_type, size_t element_count, uint32_t bit_stride)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (element_type)
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast = getASTContext();
assert (ast != NULL);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
llvm::APInt ap_element_count (64, element_count);
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return ast->getConstantArrayType(QualType::getFromOpaquePtr(element_type),
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
ap_element_count,
ArrayType::Normal,
0).getAsOpaquePtr(); // ElemQuals
}
return NULL;
}
#pragma mark TagDecl
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::StartTagDeclarationDefinition (clang_type_t clang_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::Type *t = qual_type.getTypePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (t)
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const TagType *tag_type = dyn_cast<TagType>(t);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (tag_type)
{
TagDecl *tag_decl = tag_type->getDecl();
if (tag_decl)
{
tag_decl->startDefinition();
return true;
}
}
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
const ObjCObjectType *object_type = dyn_cast<ObjCObjectType>(t);
if (object_type)
{
ObjCInterfaceDecl *interface_decl = object_type->getInterface();
if (interface_decl)
{
interface_decl->startDefinition();
return true;
}
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
}
return false;
}
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::CompleteTagDeclarationDefinition (clang_type_t clang_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
Updated LLVM to: --revision '{2010-09-28T19:30}' This gets us the new clang::CXXRecordDecl improvments in clang so that when we add fields, methods and other things to the clang::CXXRecordDecl, the correct bits are automatically set by clang::CXXRecordDecl itself instead of having SEMA and our lldb_private::ClangASTContext functions that create types for DWARF do it all manually. This allows the clang::ASTContext deep copying of types to work correctly and it means that the expression parser can now evaluate expressions in the context of a class method correctly. Previously when a class was copied from the DWARF generated ASTContext over into the expression ASTContext, we were losing CXXRecordDecl bits in the conversion which caused all classes to think they were at offset zero because the the bools for empty, POD, and others would end up being incorrect. llvm-svn: 115023
2010-09-29 11:44:17 +08:00
CXXRecordDecl *cxx_record_decl = qual_type->getAsCXXRecordDecl();
if (cxx_record_decl)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated LLVM to: --revision '{2010-09-28T19:30}' This gets us the new clang::CXXRecordDecl improvments in clang so that when we add fields, methods and other things to the clang::CXXRecordDecl, the correct bits are automatically set by clang::CXXRecordDecl itself instead of having SEMA and our lldb_private::ClangASTContext functions that create types for DWARF do it all manually. This allows the clang::ASTContext deep copying of types to work correctly and it means that the expression parser can now evaluate expressions in the context of a class method correctly. Previously when a class was copied from the DWARF generated ASTContext over into the expression ASTContext, we were losing CXXRecordDecl bits in the conversion which caused all classes to think they were at offset zero because the the bools for empty, POD, and others would end up being incorrect. llvm-svn: 115023
2010-09-29 11:44:17 +08:00
cxx_record_decl->completeDefinition();
return true;
}
const EnumType *enum_type = dyn_cast<EnumType>(qual_type.getTypePtr());
if (enum_type)
{
EnumDecl *enum_decl = enum_type->getDecl();
if (enum_decl)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated LLVM to: --revision '{2010-09-28T19:30}' This gets us the new clang::CXXRecordDecl improvments in clang so that when we add fields, methods and other things to the clang::CXXRecordDecl, the correct bits are automatically set by clang::CXXRecordDecl itself instead of having SEMA and our lldb_private::ClangASTContext functions that create types for DWARF do it all manually. This allows the clang::ASTContext deep copying of types to work correctly and it means that the expression parser can now evaluate expressions in the context of a class method correctly. Previously when a class was copied from the DWARF generated ASTContext over into the expression ASTContext, we were losing CXXRecordDecl bits in the conversion which caused all classes to think they were at offset zero because the the bools for empty, POD, and others would end up being incorrect. llvm-svn: 115023
2010-09-29 11:44:17 +08:00
/// TODO This really needs to be fixed.
unsigned NumPositiveBits = 1;
unsigned NumNegativeBits = 0;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast = getASTContext();
Modified the "breakpoint set --name NAME" to be the auto breakpoint set function. It will inspect NAME and do the following: - if the name contains '(' or starts with "-[" or "+[" then a full name search will happen to match full function names with args (C++ demangled names) or full objective C method prototypes. - if the name contains "::" and no '(', then it is assumed to be a qualified function name that is in a namespace or class. For "foo::bar::baz" we will search for any functions with the basename or method name of "baz", then filter the results to only those that contain "foo::bar::baz". This allows setting breakpoint on C++ functions and methods without having to fully qualify all of the types that would appear in C++ mangled names. - if the name contains ":" (not "::"), then NAME is assumed to be an ObjC selector. _ otherwise, we assume just a plain function basename. Now that "--name" is our "auto" mode, I introduced the new "--basename" option ("breakpoint set --basename NAME") to allow for function names that aren't methods or selectors, just basenames. This can also be used to ignore C++ namespaces and class hierarchies for class methods. Fixed clang enumeration promotion types to be correct. llvm-svn: 116293
2010-10-12 12:29:14 +08:00
QualType promotion_qual_type;
// If the enum integer type is less than an integer in bit width,
// then we must promote it to an integer size.
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
if (ast->getTypeSize(enum_decl->getIntegerType()) < ast->getTypeSize(ast->IntTy))
Modified the "breakpoint set --name NAME" to be the auto breakpoint set function. It will inspect NAME and do the following: - if the name contains '(' or starts with "-[" or "+[" then a full name search will happen to match full function names with args (C++ demangled names) or full objective C method prototypes. - if the name contains "::" and no '(', then it is assumed to be a qualified function name that is in a namespace or class. For "foo::bar::baz" we will search for any functions with the basename or method name of "baz", then filter the results to only those that contain "foo::bar::baz". This allows setting breakpoint on C++ functions and methods without having to fully qualify all of the types that would appear in C++ mangled names. - if the name contains ":" (not "::"), then NAME is assumed to be an ObjC selector. _ otherwise, we assume just a plain function basename. Now that "--name" is our "auto" mode, I introduced the new "--basename" option ("breakpoint set --basename NAME") to allow for function names that aren't methods or selectors, just basenames. This can also be used to ignore C++ namespaces and class hierarchies for class methods. Fixed clang enumeration promotion types to be correct. llvm-svn: 116293
2010-10-12 12:29:14 +08:00
{
if (enum_decl->getIntegerType()->isSignedIntegerType())
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
promotion_qual_type = ast->IntTy;
Modified the "breakpoint set --name NAME" to be the auto breakpoint set function. It will inspect NAME and do the following: - if the name contains '(' or starts with "-[" or "+[" then a full name search will happen to match full function names with args (C++ demangled names) or full objective C method prototypes. - if the name contains "::" and no '(', then it is assumed to be a qualified function name that is in a namespace or class. For "foo::bar::baz" we will search for any functions with the basename or method name of "baz", then filter the results to only those that contain "foo::bar::baz". This allows setting breakpoint on C++ functions and methods without having to fully qualify all of the types that would appear in C++ mangled names. - if the name contains ":" (not "::"), then NAME is assumed to be an ObjC selector. _ otherwise, we assume just a plain function basename. Now that "--name" is our "auto" mode, I introduced the new "--basename" option ("breakpoint set --basename NAME") to allow for function names that aren't methods or selectors, just basenames. This can also be used to ignore C++ namespaces and class hierarchies for class methods. Fixed clang enumeration promotion types to be correct. llvm-svn: 116293
2010-10-12 12:29:14 +08:00
else
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
promotion_qual_type = ast->UnsignedIntTy;
Modified the "breakpoint set --name NAME" to be the auto breakpoint set function. It will inspect NAME and do the following: - if the name contains '(' or starts with "-[" or "+[" then a full name search will happen to match full function names with args (C++ demangled names) or full objective C method prototypes. - if the name contains "::" and no '(', then it is assumed to be a qualified function name that is in a namespace or class. For "foo::bar::baz" we will search for any functions with the basename or method name of "baz", then filter the results to only those that contain "foo::bar::baz". This allows setting breakpoint on C++ functions and methods without having to fully qualify all of the types that would appear in C++ mangled names. - if the name contains ":" (not "::"), then NAME is assumed to be an ObjC selector. _ otherwise, we assume just a plain function basename. Now that "--name" is our "auto" mode, I introduced the new "--basename" option ("breakpoint set --basename NAME") to allow for function names that aren't methods or selectors, just basenames. This can also be used to ignore C++ namespaces and class hierarchies for class methods. Fixed clang enumeration promotion types to be correct. llvm-svn: 116293
2010-10-12 12:29:14 +08:00
}
else
promotion_qual_type = enum_decl->getIntegerType();
enum_decl->completeDefinition(enum_decl->getIntegerType(), promotion_qual_type, NumPositiveBits, NumNegativeBits);
Updated LLVM to: --revision '{2010-09-28T19:30}' This gets us the new clang::CXXRecordDecl improvments in clang so that when we add fields, methods and other things to the clang::CXXRecordDecl, the correct bits are automatically set by clang::CXXRecordDecl itself instead of having SEMA and our lldb_private::ClangASTContext functions that create types for DWARF do it all manually. This allows the clang::ASTContext deep copying of types to work correctly and it means that the expression parser can now evaluate expressions in the context of a class method correctly. Previously when a class was copied from the DWARF generated ASTContext over into the expression ASTContext, we were losing CXXRecordDecl bits in the conversion which caused all classes to think they were at offset zero because the the bools for empty, POD, and others would end up being incorrect. llvm-svn: 115023
2010-09-29 11:44:17 +08:00
return true;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
}
}
return false;
}
#pragma mark Enumeration Types
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
Fixed an error in the type map for "char **" that was a bad memory smasher. Anytime we had a valid python list that was trying to go from Python down into our C++ API, it was allocating too little memory and it ended up smashing whatever was next to the allocated memory. Added typemap conversions for "void *, size_t" so we can get SBProcess::ReadMemory() working. Also added a typemap for "const void *, size_t" so we can get SBProcess::WriteMemory() to work. Fixed an issue in the DWARF parser where we weren't correctly calculating the DeclContext for all types and classes. We now should be a lot more accurate. Fixes include: enums should now be setting their parent decl context correctly. We saw a lot of examples where enums in classes were not being properly namespace scoped. Also, classes within classes now get properly scoped. Fixed the objective C runtime pointer checkers to let "nil" pointers through since these are accepted by compiled code. We also now don't call "abort()" when a pointer doesn't validate correctly since this was wreaking havoc on the process due to the way abort() works. We now just dereference memory which should give us an exception from which we can easily and reliably recover. llvm-svn: 123428
2011-01-14 12:54:56 +08:00
ClangASTContext::CreateEnumerationType
(
const char *name,
DeclContext *decl_ctx,
const Declaration &decl,
clang_type_t integer_qual_type
)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
// TODO: Do something intelligent with the Declaration object passed in
// like maybe filling in the SourceLocation with it...
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast = getASTContext();
assert (ast != NULL);
Modified the "breakpoint set --name NAME" to be the auto breakpoint set function. It will inspect NAME and do the following: - if the name contains '(' or starts with "-[" or "+[" then a full name search will happen to match full function names with args (C++ demangled names) or full objective C method prototypes. - if the name contains "::" and no '(', then it is assumed to be a qualified function name that is in a namespace or class. For "foo::bar::baz" we will search for any functions with the basename or method name of "baz", then filter the results to only those that contain "foo::bar::baz". This allows setting breakpoint on C++ functions and methods without having to fully qualify all of the types that would appear in C++ mangled names. - if the name contains ":" (not "::"), then NAME is assumed to be an ObjC selector. _ otherwise, we assume just a plain function basename. Now that "--name" is our "auto" mode, I introduced the new "--basename" option ("breakpoint set --basename NAME") to allow for function names that aren't methods or selectors, just basenames. This can also be used to ignore C++ namespaces and class hierarchies for class methods. Fixed clang enumeration promotion types to be correct. llvm-svn: 116293
2010-10-12 12:29:14 +08:00
// TODO: ask about these...
// const bool IsScoped = false;
// const bool IsFixed = false;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
EnumDecl *enum_decl = EnumDecl::Create (*ast,
Fixed an error in the type map for "char **" that was a bad memory smasher. Anytime we had a valid python list that was trying to go from Python down into our C++ API, it was allocating too little memory and it ended up smashing whatever was next to the allocated memory. Added typemap conversions for "void *, size_t" so we can get SBProcess::ReadMemory() working. Also added a typemap for "const void *, size_t" so we can get SBProcess::WriteMemory() to work. Fixed an issue in the DWARF parser where we weren't correctly calculating the DeclContext for all types and classes. We now should be a lot more accurate. Fixes include: enums should now be setting their parent decl context correctly. We saw a lot of examples where enums in classes were not being properly namespace scoped. Also, classes within classes now get properly scoped. Fixed the objective C runtime pointer checkers to let "nil" pointers through since these are accepted by compiled code. We also now don't call "abort()" when a pointer doesn't validate correctly since this was wreaking havoc on the process due to the way abort() works. We now just dereference memory which should give us an exception from which we can easily and reliably recover. llvm-svn: 123428
2011-01-14 12:54:56 +08:00
decl_ctx,
Modified the "breakpoint set --name NAME" to be the auto breakpoint set function. It will inspect NAME and do the following: - if the name contains '(' or starts with "-[" or "+[" then a full name search will happen to match full function names with args (C++ demangled names) or full objective C method prototypes. - if the name contains "::" and no '(', then it is assumed to be a qualified function name that is in a namespace or class. For "foo::bar::baz" we will search for any functions with the basename or method name of "baz", then filter the results to only those that contain "foo::bar::baz". This allows setting breakpoint on C++ functions and methods without having to fully qualify all of the types that would appear in C++ mangled names. - if the name contains ":" (not "::"), then NAME is assumed to be an ObjC selector. _ otherwise, we assume just a plain function basename. Now that "--name" is our "auto" mode, I introduced the new "--basename" option ("breakpoint set --basename NAME") to allow for function names that aren't methods or selectors, just basenames. This can also be used to ignore C++ namespaces and class hierarchies for class methods. Fixed clang enumeration promotion types to be correct. llvm-svn: 116293
2010-10-12 12:29:14 +08:00
SourceLocation(),
SourceLocation(),
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
name && name[0] ? &ast->Idents.get(name) : NULL,
Updated to latest Clang revision. This involved very minor changes, changing how we get the target type from a TypedefType, adding a parameter to EnumDecl::Create(), and other minor tweaks. llvm-svn: 121663
2010-12-13 09:26:27 +08:00
NULL,
false, // IsScoped
false, // IsScopedUsingClassTag
false); // IsFixed
Quick fixes to set access specifiers for Typedefs and Enums inside classes. llvm-svn: 123710
2011-01-18 09:03:44 +08:00
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (enum_decl)
Fixed a crash that would happen when using "frame variables" on any struct, union, or class that contained an enumeration type. When I was creating the clang enumeration decl, I wasn't calling "EnumDecl::setIntegerType (QualType)" which means that if the enum decl was ever asked to figure out it's bit width (getTypeInfo()) it would crash. We didn't run into this with enum types that weren't inside classes because the DWARF already told us how big the type was and when we printed an enum we would never need to calculate the size, we would use the pre-cached byte size we got from the DWARF. When the enum was in a struct/union/class and we tried to layout the struct, the layout code would attempt to get the type info and segfault. llvm-svn: 113729
2010-09-13 07:17:56 +08:00
{
// TODO: check if we should be setting the promotion type too?
enum_decl->setIntegerType(QualType::getFromOpaquePtr (integer_qual_type));
Quick fixes to set access specifiers for Typedefs and Enums inside classes. llvm-svn: 123710
2011-01-18 09:03:44 +08:00
enum_decl->setAccess(AS_public); // TODO respect what's in the debug info
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return ast->getTagDeclType(enum_decl).getAsOpaquePtr();
Fixed a crash that would happen when using "frame variables" on any struct, union, or class that contained an enumeration type. When I was creating the clang enumeration decl, I wasn't calling "EnumDecl::setIntegerType (QualType)" which means that if the enum decl was ever asked to figure out it's bit width (getTypeInfo()) it would crash. We didn't run into this with enum types that weren't inside classes because the DWARF already told us how big the type was and when we printed an enum we would never need to calculate the size, we would use the pre-cached byte size we got from the DWARF. When the enum was in a struct/union/class and we tried to layout the struct, the layout code would attempt to get the type info and segfault. llvm-svn: 113729
2010-09-13 07:17:56 +08:00
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return NULL;
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
ClangASTContext::GetEnumerationIntegerType (clang_type_t enum_clang_type)
{
QualType enum_qual_type (QualType::getFromOpaquePtr(enum_clang_type));
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::Type *clang_type = enum_qual_type.getTypePtr();
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
if (clang_type)
{
const EnumType *enum_type = dyn_cast<EnumType>(clang_type);
if (enum_type)
{
EnumDecl *enum_decl = enum_type->getDecl();
if (enum_decl)
return enum_decl->getIntegerType().getAsOpaquePtr();
}
}
return NULL;
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
bool
ClangASTContext::AddEnumerationValueToEnumerationType
(
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t enum_clang_type,
clang_type_t enumerator_clang_type,
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
const Declaration &decl,
const char *name,
int64_t enum_value,
uint32_t enum_value_bit_size
)
{
if (enum_clang_type && enumerator_clang_type && name)
{
// TODO: Do something intelligent with the Declaration object passed in
// like maybe filling in the SourceLocation with it...
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast = getASTContext();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
IdentifierTable *identifier_table = getIdentifierTable();
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
assert (ast != NULL);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
assert (identifier_table != NULL);
QualType enum_qual_type (QualType::getFromOpaquePtr(enum_clang_type));
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const clang::Type *clang_type = enum_qual_type.getTypePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (clang_type)
{
const EnumType *enum_type = dyn_cast<EnumType>(clang_type);
if (enum_type)
{
llvm::APSInt enum_llvm_apsint(enum_value_bit_size, false);
enum_llvm_apsint = enum_value;
EnumConstantDecl *enumerator_decl =
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
EnumConstantDecl::Create (*ast,
enum_type->getDecl(),
SourceLocation(),
name ? &identifier_table->get(name) : NULL, // Identifier
QualType::getFromOpaquePtr(enumerator_clang_type),
NULL,
enum_llvm_apsint);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (enumerator_decl)
{
enum_type->getDecl()->addDecl(enumerator_decl);
Added VerifyDecl, a function that, when LLDB is linked against a debug LLVM, runs a variety of functions -- currently just one -- that verify that the Decls we create are valid. ClangASTContext now calls this verifier whenever it adds a Decl to a DeclContext, and the verifier checks that the AccessSpecifier is sane. llvm-svn: 143000
2011-10-26 09:06:27 +08:00
#ifdef LLDB_CONFIGURATION_DEBUG
VerifyDecl(enumerator_decl);
#endif
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return true;
}
}
}
}
return false;
}
#pragma mark Pointers & References
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
ClangASTContext::CreatePointerType (clang_type_t clang_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Modified LLDB expressions to not have to JIT and run code just to see variable values or persistent expression variables. Now if an expression consists of a value that is a child of a variable, or of a persistent variable only, we will create a value object for it and make a ValueObjectConstResult from it to freeze the value (for program variables only, not persistent variables) and avoid running JITed code. For everything else we still parse up and JIT code and run it in the inferior. There was also a lot of clean up in the expression code. I made the ClangExpressionVariables be stored in collections of shared pointers instead of in collections of objects. This will help stop a lot of copy constructors on these large objects and also cleans up the code considerably. The persistent clang expression variables were moved over to the Target to ensure they persist across process executions. Added the ability for lldb_private::Target objects to evaluate expressions. We want to evaluate expressions at the target level in case we aren't running yet, or we have just completed running. We still want to be able to access the persistent expression variables between runs, and also evaluate constant expressions. Added extra logging to the dynamic loader plug-in for MacOSX. ModuleList objects can now dump their contents with the UUID, arch and full paths being logged with appropriate prefix values. Thread hardened the Communication class a bit by making the connection auto_ptr member into a shared pointer member and then making a local copy of the shared pointer in each method that uses it to make sure another thread can't nuke the connection object while it is being used by another thread. Added a new file to the lldb/test/load_unload test that causes the test a.out file to link to the libd.dylib file all the time. This will allow us to test using the DYLD_LIBRARY_PATH environment variable after moving libd.dylib somewhere else. llvm-svn: 121745
2010-12-14 10:59:59 +08:00
return CreatePointerType (getASTContext(), clang_type);
}
clang_type_t
ClangASTContext::CreatePointerType (clang::ASTContext *ast, clang_type_t clang_type)
{
if (ast && clang_type)
Fixed "void *ClangASTContext::CreatePointerType (void *clang_type);" to return objective C pointers for clang::Type::TypeClass types that are "clang::Type::ObjCObject" and "clang::Type::ObjCInterface" . llvm-svn: 109795
2010-07-30 04:06:32 +08:00
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
Fixed "void *ClangASTContext::CreatePointerType (void *clang_type);" to return objective C pointers for clang::Type::TypeClass types that are "clang::Type::ObjCObject" and "clang::Type::ObjCInterface" . llvm-svn: 109795
2010-07-30 04:06:32 +08:00
{
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
Modified LLDB expressions to not have to JIT and run code just to see variable values or persistent expression variables. Now if an expression consists of a value that is a child of a variable, or of a persistent variable only, we will create a value object for it and make a ValueObjectConstResult from it to freeze the value (for program variables only, not persistent variables) and avoid running JITed code. For everything else we still parse up and JIT code and run it in the inferior. There was also a lot of clean up in the expression code. I made the ClangExpressionVariables be stored in collections of shared pointers instead of in collections of objects. This will help stop a lot of copy constructors on these large objects and also cleans up the code considerably. The persistent clang expression variables were moved over to the Target to ensure they persist across process executions. Added the ability for lldb_private::Target objects to evaluate expressions. We want to evaluate expressions at the target level in case we aren't running yet, or we have just completed running. We still want to be able to access the persistent expression variables between runs, and also evaluate constant expressions. Added extra logging to the dynamic loader plug-in for MacOSX. ModuleList objects can now dump their contents with the UUID, arch and full paths being logged with appropriate prefix values. Thread hardened the Communication class a bit by making the connection auto_ptr member into a shared pointer member and then making a local copy of the shared pointer in each method that uses it to make sure another thread can't nuke the connection object while it is being used by another thread. Added a new file to the lldb/test/load_unload test that causes the test a.out file to link to the libd.dylib file all the time. This will allow us to test using the DYLD_LIBRARY_PATH environment variable after moving libd.dylib somewhere else. llvm-svn: 121745
2010-12-14 10:59:59 +08:00
return ast->getObjCObjectPointerType(qual_type).getAsOpaquePtr();
Fixed "void *ClangASTContext::CreatePointerType (void *clang_type);" to return objective C pointers for clang::Type::TypeClass types that are "clang::Type::ObjCObject" and "clang::Type::ObjCInterface" . llvm-svn: 109795
2010-07-30 04:06:32 +08:00
default:
Modified LLDB expressions to not have to JIT and run code just to see variable values or persistent expression variables. Now if an expression consists of a value that is a child of a variable, or of a persistent variable only, we will create a value object for it and make a ValueObjectConstResult from it to freeze the value (for program variables only, not persistent variables) and avoid running JITed code. For everything else we still parse up and JIT code and run it in the inferior. There was also a lot of clean up in the expression code. I made the ClangExpressionVariables be stored in collections of shared pointers instead of in collections of objects. This will help stop a lot of copy constructors on these large objects and also cleans up the code considerably. The persistent clang expression variables were moved over to the Target to ensure they persist across process executions. Added the ability for lldb_private::Target objects to evaluate expressions. We want to evaluate expressions at the target level in case we aren't running yet, or we have just completed running. We still want to be able to access the persistent expression variables between runs, and also evaluate constant expressions. Added extra logging to the dynamic loader plug-in for MacOSX. ModuleList objects can now dump their contents with the UUID, arch and full paths being logged with appropriate prefix values. Thread hardened the Communication class a bit by making the connection auto_ptr member into a shared pointer member and then making a local copy of the shared pointer in each method that uses it to make sure another thread can't nuke the connection object while it is being used by another thread. Added a new file to the lldb/test/load_unload test that causes the test a.out file to link to the libd.dylib file all the time. This will allow us to test using the DYLD_LIBRARY_PATH environment variable after moving libd.dylib somewhere else. llvm-svn: 121745
2010-12-14 10:59:59 +08:00
return ast->getPointerType(qual_type).getAsOpaquePtr();
Fixed "void *ClangASTContext::CreatePointerType (void *clang_type);" to return objective C pointers for clang::Type::TypeClass types that are "clang::Type::ObjCObject" and "clang::Type::ObjCInterface" . llvm-svn: 109795
2010-07-30 04:06:32 +08:00
}
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return NULL;
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
Implemented a major overhaul of the way variables are handled by LLDB. Instead of being materialized into the input structure passed to the expression, variables are left in place and pointers to them are materialzied into the structure. Variables not resident in memory (notably, registers) get temporary memory regions allocated for them. Persistent variables are the most complex part of this, because they are made in various ways and there are different expectations about their lifetime. Persistent variables now have flags indicating their status and what the expectations for longevity are. They can be marked as residing in target memory permanently -- this is the default for result variables from expressions entered on the command line and for explicitly declared persistent variables (but more on that below). Other result variables have their memory freed. Some major improvements resulting from this include being able to properly take the address of variables, better and cleaner support for functions that return references, and cleaner C++ support in general. One problem that remains is the problem of explicitly declared persistent variables; I have not yet implemented the code that makes references to them into indirect references, so currently materialization and dematerialization of these variables is broken. llvm-svn: 123371
2011-01-13 16:53:35 +08:00
ClangASTContext::CreateLValueReferenceType (clang::ASTContext *ast,
clang_type_t clang_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type)
Implemented a major overhaul of the way variables are handled by LLDB. Instead of being materialized into the input structure passed to the expression, variables are left in place and pointers to them are materialzied into the structure. Variables not resident in memory (notably, registers) get temporary memory regions allocated for them. Persistent variables are the most complex part of this, because they are made in various ways and there are different expectations about their lifetime. Persistent variables now have flags indicating their status and what the expectations for longevity are. They can be marked as residing in target memory permanently -- this is the default for result variables from expressions entered on the command line and for explicitly declared persistent variables (but more on that below). Other result variables have their memory freed. Some major improvements resulting from this include being able to properly take the address of variables, better and cleaner support for functions that return references, and cleaner C++ support in general. One problem that remains is the problem of explicitly declared persistent variables; I have not yet implemented the code that makes references to them into indirect references, so currently materialization and dematerialization of these variables is broken. llvm-svn: 123371
2011-01-13 16:53:35 +08:00
return ast->getLValueReferenceType (QualType::getFromOpaquePtr(clang_type)).getAsOpaquePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return NULL;
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
Implemented a major overhaul of the way variables are handled by LLDB. Instead of being materialized into the input structure passed to the expression, variables are left in place and pointers to them are materialzied into the structure. Variables not resident in memory (notably, registers) get temporary memory regions allocated for them. Persistent variables are the most complex part of this, because they are made in various ways and there are different expectations about their lifetime. Persistent variables now have flags indicating their status and what the expectations for longevity are. They can be marked as residing in target memory permanently -- this is the default for result variables from expressions entered on the command line and for explicitly declared persistent variables (but more on that below). Other result variables have their memory freed. Some major improvements resulting from this include being able to properly take the address of variables, better and cleaner support for functions that return references, and cleaner C++ support in general. One problem that remains is the problem of explicitly declared persistent variables; I have not yet implemented the code that makes references to them into indirect references, so currently materialization and dematerialization of these variables is broken. llvm-svn: 123371
2011-01-13 16:53:35 +08:00
ClangASTContext::CreateRValueReferenceType (clang::ASTContext *ast,
clang_type_t clang_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type)
Implemented a major overhaul of the way variables are handled by LLDB. Instead of being materialized into the input structure passed to the expression, variables are left in place and pointers to them are materialzied into the structure. Variables not resident in memory (notably, registers) get temporary memory regions allocated for them. Persistent variables are the most complex part of this, because they are made in various ways and there are different expectations about their lifetime. Persistent variables now have flags indicating their status and what the expectations for longevity are. They can be marked as residing in target memory permanently -- this is the default for result variables from expressions entered on the command line and for explicitly declared persistent variables (but more on that below). Other result variables have their memory freed. Some major improvements resulting from this include being able to properly take the address of variables, better and cleaner support for functions that return references, and cleaner C++ support in general. One problem that remains is the problem of explicitly declared persistent variables; I have not yet implemented the code that makes references to them into indirect references, so currently materialization and dematerialization of these variables is broken. llvm-svn: 123371
2011-01-13 16:53:35 +08:00
return ast->getRValueReferenceType (QualType::getFromOpaquePtr(clang_type)).getAsOpaquePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
return NULL;
}
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
ClangASTContext::CreateMemberPointerType (clang_type_t clang_pointee_type, clang_type_t clang_class_type)
Anders Carlsson patch for member pointers. Thanks Anders. llvm-svn: 105868
2010-06-12 09:20:30 +08:00
{
if (clang_pointee_type && clang_pointee_type)
return getASTContext()->getMemberPointerType(QualType::getFromOpaquePtr(clang_pointee_type),
QualType::getFromOpaquePtr(clang_class_type).getTypePtr()).getAsOpaquePtr();
return NULL;
}
Enabled extra warnings and fixed a bunch of small issues. llvm-svn: 124250
2011-01-26 07:55:37 +08:00
uint32_t
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
ClangASTContext::GetPointerBitSize ()
{
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast = getASTContext();
return ast->getTypeSize(ast->VoidPtrTy);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Added support for finding and global variables in the SBTarget and SBModule level in the public API. Also modified the ValueObject values to be able to display global variables without having a valid running process. The globals will read themselves from the object file section data if there is no process, and from the process if there is one. Also fixed an issue where modifications for dynamic types could cause child values of ValueObjects to not show up if the value was unable to evaluate itself (children of NULL pointer objects). llvm-svn: 134102
2011-06-30 06:09:02 +08:00
bool
<rdar://problem/11082392> Fixed an issue that could cause circular type parsing that will assert and kill LLDB. Prior to this fix the DWARF parser would always create class types and not start their definitions (for both C++ and ObjC classes) until we were asked to complete the class later. When we had cases like: class A { class B { }; }; We would alway try to complete A before specifying "A" as the decl context for B. Turns out we can just start the definition and still not complete the class since we can check the TagDecl::isCompleteDefinition() function. This only works for C++ types. This means we will not be pulling in the full definition of parent classes all the time and should help with our memory consumption and also reduce the amount of debug info we have to parse. I also reduced redundant code that was checking in a lldb::clang_type_t was a possible C++ dynamic type since it was still completing the type, just to see if it was dynamic. This was fixed in another function that was checking for a type being dynamic as an ObjC or a C++ type, but there was dedicated fucntion for C++ that we missed. llvm-svn: 153713
2012-03-30 08:51:13 +08:00
ClangASTContext::IsPossibleDynamicType (clang::ASTContext *ast,
clang_type_t clang_type,
clang_type_t *dynamic_pointee_type,
bool check_cplusplus,
bool check_objc)
Added support for finding and global variables in the SBTarget and SBModule level in the public API. Also modified the ValueObject values to be able to display global variables without having a valid running process. The globals will read themselves from the object file section data if there is no process, and from the process if there is one. Also fixed an issue where modifications for dynamic types could cause child values of ValueObjects to not show up if the value was unable to evaluate itself (children of NULL pointer objects). llvm-svn: 134102
2011-06-30 06:09:02 +08:00
{
QualType pointee_qual_type;
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
bool success = false;
switch (type_class)
{
case clang::Type::Builtin:
<rdar://problem/11082392> Fixed an issue that could cause circular type parsing that will assert and kill LLDB. Prior to this fix the DWARF parser would always create class types and not start their definitions (for both C++ and ObjC classes) until we were asked to complete the class later. When we had cases like: class A { class B { }; }; We would alway try to complete A before specifying "A" as the decl context for B. Turns out we can just start the definition and still not complete the class since we can check the TagDecl::isCompleteDefinition() function. This only works for C++ types. This means we will not be pulling in the full definition of parent classes all the time and should help with our memory consumption and also reduce the amount of debug info we have to parse. I also reduced redundant code that was checking in a lldb::clang_type_t was a possible C++ dynamic type since it was still completing the type, just to see if it was dynamic. This was fixed in another function that was checking for a type being dynamic as an ObjC or a C++ type, but there was dedicated fucntion for C++ that we missed. llvm-svn: 153713
2012-03-30 08:51:13 +08:00
if (check_objc && cast<clang::BuiltinType>(qual_type)->getKind() == clang::BuiltinType::ObjCId)
Added support for finding and global variables in the SBTarget and SBModule level in the public API. Also modified the ValueObject values to be able to display global variables without having a valid running process. The globals will read themselves from the object file section data if there is no process, and from the process if there is one. Also fixed an issue where modifications for dynamic types could cause child values of ValueObjects to not show up if the value was unable to evaluate itself (children of NULL pointer objects). llvm-svn: 134102
2011-06-30 06:09:02 +08:00
{
if (dynamic_pointee_type)
*dynamic_pointee_type = clang_type;
return true;
}
break;
case clang::Type::ObjCObjectPointer:
<rdar://problem/11082392> Fixed an issue that could cause circular type parsing that will assert and kill LLDB. Prior to this fix the DWARF parser would always create class types and not start their definitions (for both C++ and ObjC classes) until we were asked to complete the class later. When we had cases like: class A { class B { }; }; We would alway try to complete A before specifying "A" as the decl context for B. Turns out we can just start the definition and still not complete the class since we can check the TagDecl::isCompleteDefinition() function. This only works for C++ types. This means we will not be pulling in the full definition of parent classes all the time and should help with our memory consumption and also reduce the amount of debug info we have to parse. I also reduced redundant code that was checking in a lldb::clang_type_t was a possible C++ dynamic type since it was still completing the type, just to see if it was dynamic. This was fixed in another function that was checking for a type being dynamic as an ObjC or a C++ type, but there was dedicated fucntion for C++ that we missed. llvm-svn: 153713
2012-03-30 08:51:13 +08:00
if (check_objc)
{
if (dynamic_pointee_type)
*dynamic_pointee_type = cast<ObjCObjectPointerType>(qual_type)->getPointeeType().getAsOpaquePtr();
return true;
}
break;
Added support for finding and global variables in the SBTarget and SBModule level in the public API. Also modified the ValueObject values to be able to display global variables without having a valid running process. The globals will read themselves from the object file section data if there is no process, and from the process if there is one. Also fixed an issue where modifications for dynamic types could cause child values of ValueObjects to not show up if the value was unable to evaluate itself (children of NULL pointer objects). llvm-svn: 134102
2011-06-30 06:09:02 +08:00
case clang::Type::Pointer:
pointee_qual_type = cast<PointerType>(qual_type)->getPointeeType();
success = true;
break;
case clang::Type::LValueReference:
case clang::Type::RValueReference:
pointee_qual_type = cast<ReferenceType>(qual_type)->getPointeeType();
success = true;
break;
case clang::Type::Typedef:
Fixed a few issues where typedefs weren't passing through to the correct recursive function. Also fixed ClangASTContext::IsPointerType to correctly NULL out the pointee handle if a valid one is provided. llvm-svn: 134715
2011-07-09 02:27:39 +08:00
return ClangASTContext::IsPossibleDynamicType (ast, cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(), dynamic_pointee_type);
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated:
return ClangASTContext::IsPossibleDynamicType (ast, cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr(), dynamic_pointee_type);
Added support for finding and global variables in the SBTarget and SBModule level in the public API. Also modified the ValueObject values to be able to display global variables without having a valid running process. The globals will read themselves from the object file section data if there is no process, and from the process if there is one. Also fixed an issue where modifications for dynamic types could cause child values of ValueObjects to not show up if the value was unable to evaluate itself (children of NULL pointer objects). llvm-svn: 134102
2011-06-30 06:09:02 +08:00
default:
break;
}
if (success)
{
// Check to make sure what we are pointing too is a possible dynamic C++ type
// We currently accept any "void *" (in case we have a class that has been
// watered down to an opaque pointer) and virtual C++ classes.
const clang::Type::TypeClass pointee_type_class = pointee_qual_type->getTypeClass();
switch (pointee_type_class)
{
case clang::Type::Builtin:
switch (cast<clang::BuiltinType>(pointee_qual_type)->getKind())
{
case clang::BuiltinType::UnknownAny:
case clang::BuiltinType::Void:
if (dynamic_pointee_type)
*dynamic_pointee_type = pointee_qual_type.getAsOpaquePtr();
return true;
case clang::BuiltinType::NullPtr:
case clang::BuiltinType::Bool:
case clang::BuiltinType::Char_U:
case clang::BuiltinType::UChar:
case clang::BuiltinType::WChar_U:
case clang::BuiltinType::Char16:
case clang::BuiltinType::Char32:
case clang::BuiltinType::UShort:
case clang::BuiltinType::UInt:
case clang::BuiltinType::ULong:
case clang::BuiltinType::ULongLong:
case clang::BuiltinType::UInt128:
case clang::BuiltinType::Char_S:
case clang::BuiltinType::SChar:
case clang::BuiltinType::WChar_S:
case clang::BuiltinType::Short:
case clang::BuiltinType::Int:
case clang::BuiltinType::Long:
case clang::BuiltinType::LongLong:
case clang::BuiltinType::Int128:
case clang::BuiltinType::Float:
case clang::BuiltinType::Double:
case clang::BuiltinType::LongDouble:
case clang::BuiltinType::Dependent:
case clang::BuiltinType::Overload:
case clang::BuiltinType::ObjCId:
case clang::BuiltinType::ObjCClass:
case clang::BuiltinType::ObjCSel:
case clang::BuiltinType::BoundMember:
Added template support when parsing DWARF into types. We can now use STL classes in the expression parser. llvm-svn: 142717
2011-10-22 11:33:13 +08:00
case clang::BuiltinType::Half:
case clang::BuiltinType::ARCUnbridgedCast:
Cleanup some warnings. llvm-svn: 144200
2011-11-10 03:04:58 +08:00
case clang::BuiltinType::PseudoObject:
Added support for finding and global variables in the SBTarget and SBModule level in the public API. Also modified the ValueObject values to be able to display global variables without having a valid running process. The globals will read themselves from the object file section data if there is no process, and from the process if there is one. Also fixed an issue where modifications for dynamic types could cause child values of ValueObjects to not show up if the value was unable to evaluate itself (children of NULL pointer objects). llvm-svn: 134102
2011-06-30 06:09:02 +08:00
break;
}
break;
case clang::Type::Record:
<rdar://problem/11082392> Fixed an issue that could cause circular type parsing that will assert and kill LLDB. Prior to this fix the DWARF parser would always create class types and not start their definitions (for both C++ and ObjC classes) until we were asked to complete the class later. When we had cases like: class A { class B { }; }; We would alway try to complete A before specifying "A" as the decl context for B. Turns out we can just start the definition and still not complete the class since we can check the TagDecl::isCompleteDefinition() function. This only works for C++ types. This means we will not be pulling in the full definition of parent classes all the time and should help with our memory consumption and also reduce the amount of debug info we have to parse. I also reduced redundant code that was checking in a lldb::clang_type_t was a possible C++ dynamic type since it was still completing the type, just to see if it was dynamic. This was fixed in another function that was checking for a type being dynamic as an ObjC or a C++ type, but there was dedicated fucntion for C++ that we missed. llvm-svn: 153713
2012-03-30 08:51:13 +08:00
if (check_cplusplus)
Added support for finding and global variables in the SBTarget and SBModule level in the public API. Also modified the ValueObject values to be able to display global variables without having a valid running process. The globals will read themselves from the object file section data if there is no process, and from the process if there is one. Also fixed an issue where modifications for dynamic types could cause child values of ValueObjects to not show up if the value was unable to evaluate itself (children of NULL pointer objects). llvm-svn: 134102
2011-06-30 06:09:02 +08:00
{
CXXRecordDecl *cxx_record_decl = pointee_qual_type->getAsCXXRecordDecl();
if (cxx_record_decl)
{
<rdar://problem/11095005> Fixed a performance regression when dynamic types are enable where we would ask a C++ type if it can possibly be dynamic. Previously we would force the type to complete itself and then anwwer the question definitively. Now we ask the type if it is already complete and only definitively answer the question for completed types and just say "yes" for non-complete C++ types. We also always now answer yes for Objective C classes and do not complete those types either. llvm-svn: 153284
2012-03-23 06:23:17 +08:00
// Do NOT complete the type here like we used to do
// otherwise EVERY "class *" variable we have will try
// to fully complete itself and this will take a lot of
// time, memory and slow down debugging. If we have a complete
// type, then answer the question definitively, else we
// just say that a C++ class can possibly be dynamic...
<rdar://problem/11082392> Fixed an issue that could cause circular type parsing that will assert and kill LLDB. Prior to this fix the DWARF parser would always create class types and not start their definitions (for both C++ and ObjC classes) until we were asked to complete the class later. When we had cases like: class A { class B { }; }; We would alway try to complete A before specifying "A" as the decl context for B. Turns out we can just start the definition and still not complete the class since we can check the TagDecl::isCompleteDefinition() function. This only works for C++ types. This means we will not be pulling in the full definition of parent classes all the time and should help with our memory consumption and also reduce the amount of debug info we have to parse. I also reduced redundant code that was checking in a lldb::clang_type_t was a possible C++ dynamic type since it was still completing the type, just to see if it was dynamic. This was fixed in another function that was checking for a type being dynamic as an ObjC or a C++ type, but there was dedicated fucntion for C++ that we missed. llvm-svn: 153713
2012-03-30 08:51:13 +08:00
if (cxx_record_decl->isCompleteDefinition())
Added support for finding and global variables in the SBTarget and SBModule level in the public API. Also modified the ValueObject values to be able to display global variables without having a valid running process. The globals will read themselves from the object file section data if there is no process, and from the process if there is one. Also fixed an issue where modifications for dynamic types could cause child values of ValueObjects to not show up if the value was unable to evaluate itself (children of NULL pointer objects). llvm-svn: 134102
2011-06-30 06:09:02 +08:00
{
success = cxx_record_decl->isDynamicClass();
}
else
{
// We failed to get the complete type, so we have to
// treat this as a void * which we might possibly be
// able to complete
success = true;
}
if (success)
{
if (dynamic_pointee_type)
*dynamic_pointee_type = pointee_qual_type.getAsOpaquePtr();
return true;
}
}
}
break;
case clang::Type::ObjCObject:
case clang::Type::ObjCInterface:
<rdar://problem/11082392> Fixed an issue that could cause circular type parsing that will assert and kill LLDB. Prior to this fix the DWARF parser would always create class types and not start their definitions (for both C++ and ObjC classes) until we were asked to complete the class later. When we had cases like: class A { class B { }; }; We would alway try to complete A before specifying "A" as the decl context for B. Turns out we can just start the definition and still not complete the class since we can check the TagDecl::isCompleteDefinition() function. This only works for C++ types. This means we will not be pulling in the full definition of parent classes all the time and should help with our memory consumption and also reduce the amount of debug info we have to parse. I also reduced redundant code that was checking in a lldb::clang_type_t was a possible C++ dynamic type since it was still completing the type, just to see if it was dynamic. This was fixed in another function that was checking for a type being dynamic as an ObjC or a C++ type, but there was dedicated fucntion for C++ that we missed. llvm-svn: 153713
2012-03-30 08:51:13 +08:00
if (check_objc)
{
if (dynamic_pointee_type)
*dynamic_pointee_type = pointee_qual_type.getAsOpaquePtr();
return true;
}
break;
Added support for finding and global variables in the SBTarget and SBModule level in the public API. Also modified the ValueObject values to be able to display global variables without having a valid running process. The globals will read themselves from the object file section data if there is no process, and from the process if there is one. Also fixed an issue where modifications for dynamic types could cause child values of ValueObjects to not show up if the value was unable to evaluate itself (children of NULL pointer objects). llvm-svn: 134102
2011-06-30 06:09:02 +08:00
default:
break;
}
}
}
if (dynamic_pointee_type)
*dynamic_pointee_type = NULL;
return false;
}
lldb-59. llvm-svn: 132304
2011-05-30 08:49:24 +08:00
bool
ClangASTContext::IsPossibleCPlusPlusDynamicType (clang::ASTContext *ast, clang_type_t clang_type, clang_type_t *dynamic_pointee_type)
{
<rdar://problem/11082392> Fixed an issue that could cause circular type parsing that will assert and kill LLDB. Prior to this fix the DWARF parser would always create class types and not start their definitions (for both C++ and ObjC classes) until we were asked to complete the class later. When we had cases like: class A { class B { }; }; We would alway try to complete A before specifying "A" as the decl context for B. Turns out we can just start the definition and still not complete the class since we can check the TagDecl::isCompleteDefinition() function. This only works for C++ types. This means we will not be pulling in the full definition of parent classes all the time and should help with our memory consumption and also reduce the amount of debug info we have to parse. I also reduced redundant code that was checking in a lldb::clang_type_t was a possible C++ dynamic type since it was still completing the type, just to see if it was dynamic. This was fixed in another function that was checking for a type being dynamic as an ObjC or a C++ type, but there was dedicated fucntion for C++ that we missed. llvm-svn: 153713
2012-03-30 08:51:13 +08:00
return IsPossibleDynamicType (ast,
clang_type,
dynamic_pointee_type,
true, // Check for dynamic C++ types
false); // Check for dynamic ObjC types
lldb-59. llvm-svn: 132304
2011-05-30 08:49:24 +08:00
}
Changed the way the expression parser handles variables of reference types. Previously, such variables were materialized as references to those references, which caused undesried behavior in Clang and was useless anyway (the benefit of using references to variables is that it allows expressions to modify variables in place, but for references that's not required). Now we just materialize the references directly, which fixes a variety of expressions that use references. llvm-svn: 143137
2011-10-28 03:41:13 +08:00
bool
ClangASTContext::IsReferenceType (clang_type_t clang_type, clang_type_t *target_type)
{
if (clang_type == NULL)
return false;
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
case clang::Type::LValueReference:
if (target_type)
*target_type = cast<LValueReferenceType>(qual_type)->desugar().getAsOpaquePtr();
return true;
case clang::Type::RValueReference:
if (target_type)
*target_type = cast<LValueReferenceType>(qual_type)->desugar().getAsOpaquePtr();
return true;
case clang::Type::Typedef:
return ClangASTContext::IsReferenceType (cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr());
case clang::Type::Elaborated:
return ClangASTContext::IsReferenceType (cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr());
default:
break;
}
return false;
}
lldb-59. llvm-svn: 132304
2011-05-30 08:49:24 +08:00
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::IsPointerOrReferenceType (clang_type_t clang_type, clang_type_t*target_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type == NULL)
return false;
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Fixes to Objective-C built-in type handling. Specifically, we fixed handling of the objc_class built-in type, which allowed us to pass named Objective-C objects to functions, call variable list -t on objects safely, etc. llvm-svn: 117249
2010-10-25 08:29:48 +08:00
case clang::Type::Builtin:
switch (cast<clang::BuiltinType>(qual_type)->getKind())
{
default:
break;
case clang::BuiltinType::ObjCId:
case clang::BuiltinType::ObjCClass:
return true;
}
return false;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::ObjCObjectPointer:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (target_type)
*target_type = cast<ObjCObjectPointerType>(qual_type)->getPointeeType().getAsOpaquePtr();
return true;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::BlockPointer:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (target_type)
*target_type = cast<BlockPointerType>(qual_type)->getPointeeType().getAsOpaquePtr();
return true;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Pointer:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (target_type)
*target_type = cast<PointerType>(qual_type)->getPointeeType().getAsOpaquePtr();
return true;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::MemberPointer:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (target_type)
*target_type = cast<MemberPointerType>(qual_type)->getPointeeType().getAsOpaquePtr();
return true;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::LValueReference:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (target_type)
*target_type = cast<LValueReferenceType>(qual_type)->desugar().getAsOpaquePtr();
return true;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::RValueReference:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (target_type)
*target_type = cast<LValueReferenceType>(qual_type)->desugar().getAsOpaquePtr();
return true;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Typedef:
Updated to latest Clang revision. This involved very minor changes, changing how we get the target type from a TypedefType, adding a parameter to EnumDecl::Create(), and other minor tweaks. llvm-svn: 121663
2010-12-13 09:26:27 +08:00
return ClangASTContext::IsPointerOrReferenceType (cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr());
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated:
return ClangASTContext::IsPointerOrReferenceType (cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr());
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
default:
break;
}
return false;
}
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::IsIntegerType (clang_type_t clang_type, bool &is_signed)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (!clang_type)
return false;
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
const BuiltinType *builtin_type = dyn_cast<BuiltinType>(qual_type->getCanonicalTypeInternal());
if (builtin_type)
{
if (builtin_type->isInteger())
Improve the x86_64 return value decoder to handle most structure returns. Switch from GetReturnValue, which was hardly ever used, to GetReturnValueObject which is much more convenient. Return the "return value object" as a persistent variable if requested. llvm-svn: 147157
2011-12-23 03:12:40 +08:00
{
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
is_signed = builtin_type->isSignedInteger();
Improve the x86_64 return value decoder to handle most structure returns. Switch from GetReturnValue, which was hardly ever used, to GetReturnValueObject which is much more convenient. Return the "return value object" as a persistent variable if requested. llvm-svn: 147157
2011-12-23 03:12:40 +08:00
return true;
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
return false;
}
bool
Fixed a few issues where typedefs weren't passing through to the correct recursive function. Also fixed ClangASTContext::IsPointerType to correctly NULL out the pointee handle if a valid one is provided. llvm-svn: 134715
2011-07-09 02:27:39 +08:00
ClangASTContext::IsPointerType (clang_type_t clang_type, clang_type_t *target_type)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Fixed a few issues where typedefs weren't passing through to the correct recursive function. Also fixed ClangASTContext::IsPointerType to correctly NULL out the pointee handle if a valid one is provided. llvm-svn: 134715
2011-07-09 02:27:39 +08:00
if (target_type)
*target_type = NULL;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Fixes to Objective-C built-in type handling. Specifically, we fixed handling of the objc_class built-in type, which allowed us to pass named Objective-C objects to functions, call variable list -t on objects safely, etc. llvm-svn: 117249
2010-10-25 08:29:48 +08:00
case clang::Type::Builtin:
switch (cast<clang::BuiltinType>(qual_type)->getKind())
{
default:
break;
case clang::BuiltinType::ObjCId:
case clang::BuiltinType::ObjCClass:
return true;
}
return false;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::ObjCObjectPointer:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (target_type)
*target_type = cast<ObjCObjectPointerType>(qual_type)->getPointeeType().getAsOpaquePtr();
return true;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::BlockPointer:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (target_type)
*target_type = cast<BlockPointerType>(qual_type)->getPointeeType().getAsOpaquePtr();
return true;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Pointer:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (target_type)
*target_type = cast<PointerType>(qual_type)->getPointeeType().getAsOpaquePtr();
return true;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::MemberPointer:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (target_type)
*target_type = cast<MemberPointerType>(qual_type)->getPointeeType().getAsOpaquePtr();
return true;
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::Typedef:
Fixed a few issues where typedefs weren't passing through to the correct recursive function. Also fixed ClangASTContext::IsPointerType to correctly NULL out the pointee handle if a valid one is provided. llvm-svn: 134715
2011-07-09 02:27:39 +08:00
return ClangASTContext::IsPointerType (cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(), target_type);
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated:
return ClangASTContext::IsPointerType (cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr(), target_type);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
default:
break;
}
}
return false;
}
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::IsFloatingPointType (clang_type_t clang_type, uint32_t &count, bool &is_complex)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
if (const BuiltinType *BT = dyn_cast<BuiltinType>(qual_type->getCanonicalTypeInternal()))
{
clang::BuiltinType::Kind kind = BT->getKind();
if (kind >= BuiltinType::Float && kind <= BuiltinType::LongDouble)
{
count = 1;
is_complex = false;
return true;
}
}
else if (const ComplexType *CT = dyn_cast<ComplexType>(qual_type->getCanonicalTypeInternal()))
{
if (IsFloatingPointType(CT->getElementType().getAsOpaquePtr(), count, is_complex))
{
count = 2;
is_complex = true;
return true;
}
}
else if (const VectorType *VT = dyn_cast<VectorType>(qual_type->getCanonicalTypeInternal()))
{
if (IsFloatingPointType(VT->getElementType().getAsOpaquePtr(), count, is_complex))
{
count = VT->getNumElements();
is_complex = false;
return true;
}
}
}
return false;
}
new syntax for summary strings: - ${*expr} now simply means to dereference expr before actually using it - bitfields, array ranges and pointer ranges now work in a (hopefully) more natural and language-compliant way a new class TypeHierarchyNavigator replicates the behavior of the FormatManager in going through type hierarchies when one-lining summary strings, children's summaries can be used as well as values llvm-svn: 134458
2011-07-06 10:13:41 +08:00
bool
ClangASTContext::IsScalarType (lldb::clang_type_t clang_type)
{
bool is_signed;
if (ClangASTContext::IsIntegerType(clang_type, is_signed))
return true;
uint32_t count;
bool is_complex;
return ClangASTContext::IsFloatingPointType(clang_type, count, is_complex) && !is_complex;
}
bool
ClangASTContext::IsPointerToScalarType (lldb::clang_type_t clang_type)
{
if (!IsPointerType(clang_type))
return false;
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
lldb::clang_type_t pointee_type = qual_type.getTypePtr()->getPointeeType().getAsOpaquePtr();
return IsScalarType(pointee_type);
}
bool
ClangASTContext::IsArrayOfScalarType (lldb::clang_type_t clang_type)
{
Made IsArrayOfScalarType handle typedefs correctly. We should ultimately introduce GetAs...Type functions in all cases where we have Is...Type functions that know how to look inside typedefs. llvm-svn: 148512
2012-01-20 07:54:24 +08:00
clang_type = GetAsArrayType(clang_type);
if (clang_type == 0)
new syntax for summary strings: - ${*expr} now simply means to dereference expr before actually using it - bitfields, array ranges and pointer ranges now work in a (hopefully) more natural and language-compliant way a new class TypeHierarchyNavigator replicates the behavior of the FormatManager in going through type hierarchies when one-lining summary strings, children's summaries can be used as well as values llvm-svn: 134458
2011-07-06 10:13:41 +08:00
return false;
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
lldb::clang_type_t item_type = cast<ArrayType>(qual_type.getTypePtr())->getElementType().getAsOpaquePtr();
return IsScalarType(item_type);
}
Fixed an expression parsing issue where if you were stopped somewhere without debug information and you evaluated an expression, a crash would occur as a result of an unchecked pointer. Added the ability to get the expression path for a ValueObject. For a rectangle point child "x" the expression path would be something like: "rect.top_left.x". This will allow GUI and command lines to get ahold of the expression path for a value object without having to explicitly know about the hierarchy. This means the ValueObject base class now has a "ValueObject *m_parent;" member. All ValueObject subclasses now correctly track their lineage and are able to provide value expression paths as well. Added a new "--flat" option to the "frame variable" to allow for flat variable output. An example of the current and new outputs: (lldb) frame variable argc = 1 argv = 0x00007fff5fbffe80 pt = { x = 2 y = 3 } rect = { bottom_left = { x = 1 y = 2 } top_right = { x = 3 y = 4 } } (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffe80 pt.x = 2 pt.y = 3 rect.bottom_left.x = 1 rect.bottom_left.y = 2 rect.top_right.x = 3 rect.top_right.y = 4 As you can see when there is a lot of hierarchy it can help flatten things out. Also if you want to use a member in an expression, you can copy the text from the "--flat" output and not have to piece it together manually. This can help when you want to use parts of the STL in expressions: (lldb) frame variable --flat argc = 1 argv = 0x00007fff5fbffea8 hello_world._M_dataplus._M_p = 0x0000000000000000 (lldb) expr hello_world._M_dataplus._M_p[0] == '\0' llvm-svn: 116532
2010-10-15 06:52:14 +08:00
bool
ClangASTContext::GetCXXClassName (clang_type_t clang_type, std::string &class_name)
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
CXXRecordDecl *cxx_record_decl = qual_type->getAsCXXRecordDecl();
if (cxx_record_decl)
{
class_name.assign (cxx_record_decl->getIdentifier()->getNameStart());
return true;
}
}
class_name.clear();
return false;
}
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::IsCXXClassType (clang_type_t clang_type)
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
if (qual_type->getAsCXXRecordDecl() != NULL)
return true;
}
return false;
}
Fixed a case where we might end up trying to parse a type in the DWARF parser for a method whose class isn't currently in the process of completing itself. Currently, methods of a class, must be parsed when the class type that contains the method is asked to complete itself through the clang::ExternalASTSource virtual functions. Now we "do the right thing" by checking if the class is being defined, and if so we parse it, else we tell the class to complete itself so everything happens correctly. llvm-svn: 141908
2011-10-14 07:13:20 +08:00
bool
ClangASTContext::IsBeingDefined (lldb::clang_type_t clang_type)
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
const clang::TagType *tag_type = dyn_cast<clang::TagType>(qual_type);
if (tag_type)
return tag_type->isBeingDefined();
}
return false;
}
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::IsObjCClassType (clang_type_t clang_type)
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
if (qual_type->isObjCObjectOrInterfaceType())
return true;
}
return false;
}
Added support for looking up the complete type for Objective-C classes. This allows LLDB to find ivars declared in class extensions in modules other than where the debugger is currently stopped (we already supported this when the debugger was stopped in the same module as the definition). This involved the following main changes: - The ObjCLanguageRuntime now knows how to hunt for the authoritative version of an Objective-C type. It looks for the symbol indicating a definition, and then gets the type from the module containing that symbol. - ValueObjects now report their type with a potential override, and the override is set if the type of the ValueObject is an Objective-C class or pointer type that is defined somewhere other than the original reported type. This means that "frame variable" will always use the complete type if one is available. - The ClangASTSource now looks for the complete type when looking for ivars. This means that "expr" will always use the complete type if one is available. - I added a testcase that verifies that both "frame variable" and "expr" work. llvm-svn: 151214
2012-02-23 07:57:45 +08:00
bool
ClangASTContext::IsObjCObjectPointerType (lldb::clang_type_t clang_type, clang_type_t *class_type)
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
if (qual_type->isObjCObjectPointerType())
{
if (class_type)
{
*class_type = NULL;
if (!qual_type->isObjCClassType() &&
!qual_type->isObjCIdType())
{
const ObjCObjectPointerType *obj_pointer_type = dyn_cast<ObjCObjectPointerType>(qual_type);
Hardened isObjCObjectPointerType() against NULLs. llvm-svn: 152471
2012-03-10 09:59:11 +08:00
if (!obj_pointer_type)
*class_type = NULL;
...And finished the job. llvm-svn: 152472
2012-03-10 10:00:32 +08:00
else
*class_type = QualType(obj_pointer_type->getInterfaceType(), 0).getAsOpaquePtr();
Added support for looking up the complete type for Objective-C classes. This allows LLDB to find ivars declared in class extensions in modules other than where the debugger is currently stopped (we already supported this when the debugger was stopped in the same module as the definition). This involved the following main changes: - The ObjCLanguageRuntime now knows how to hunt for the authoritative version of an Objective-C type. It looks for the symbol indicating a definition, and then gets the type from the module containing that symbol. - ValueObjects now report their type with a potential override, and the override is set if the type of the ValueObject is an Objective-C class or pointer type that is defined somewhere other than the original reported type. This means that "frame variable" will always use the complete type if one is available. - The ClangASTSource now looks for the complete type when looking for ivars. This means that "expr" will always use the complete type if one is available. - I added a testcase that verifies that both "frame variable" and "expr" work. llvm-svn: 151214
2012-02-23 07:57:45 +08:00
}
}
return true;
}
}
return false;
}
bool
ClangASTContext::GetObjCClassName (lldb::clang_type_t clang_type,
std::string &class_name)
{
if (!clang_type)
return false;
const ObjCObjectType *object_type = dyn_cast<ObjCObjectType>(QualType::getFromOpaquePtr(clang_type));
if (!object_type)
return false;
const ObjCInterfaceDecl *interface = object_type->getInterface();
if (!interface)
return false;
class_name = interface->getNameAsString();
return true;
}
Added the ability to create an objective C method for an objective C interface in ClangASTContext. Also added two bool returning functions that indicated if an opaque clang qual type is a CXX class type, and if it is an ObjC class type. Objective C classes now will get their methods added lazily as they are encountered. The reason for this is currently, unlike C++, the DW_TAG_structure_type and owns the ivars, doesn't not also contain the member functions. This means when we parse the objective C class interface we either need to find all functions whose names start with "+[CLASS_NAME" or "-[CLASS_NAME" and add them all to the class, or when we parse each objective C function, we slowly add it to the class interface definition. Since objective C's class doesn't change internal bits according to whether it has certain types of member functions (like C++ does if it has virtual functions, or if it has user ctors/dtors), I currently chose to lazily populate the class when each functions is parsed. Another issue we run into with ObjC method declarations is the "self" and "_cmd" implicit args are not marked as artificial in the DWARF (DW_AT_artifical), so we currently have to look for the parameters by name if we are trying to omit artificial function args if the language of the compile unit is ObjC or ObjC++. llvm-svn: 114722
2010-09-24 13:15:53 +08:00
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
bool
ClangASTContext::IsCharType (clang_type_t clang_type)
{
if (clang_type)
return QualType::getFromOpaquePtr(clang_type)->isCharType();
return false;
}
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::IsCStringType (clang_type_t clang_type, uint32_t &length)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
clang_type_t pointee_or_element_clang_type = NULL;
Flags type_flags (ClangASTContext::GetTypeInfo (clang_type, NULL, &pointee_or_element_clang_type));
if (pointee_or_element_clang_type == NULL)
return false;
if (type_flags.AnySet (eTypeIsArray | eTypeIsPointer))
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
QualType pointee_or_element_qual_type (QualType::getFromOpaquePtr (pointee_or_element_clang_type));
if (pointee_or_element_qual_type.getUnqualifiedType()->isCharType())
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
if (type_flags.Test (eTypeIsArray))
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
// We know the size of the array and it could be a C string
// since it is an array of characters
length = cast<ConstantArrayType>(qual_type.getTypePtr())->getSize().getLimitedValue();
return true;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
else
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
length = 0;
return true;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
}
}
return false;
}
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
bool
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::IsFunctionPointerType (clang_type_t clang_type)
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
if (qual_type->isFunctionPointerType())
return true;
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
default:
break;
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
case clang::Type::Typedef:
Updated to latest Clang revision. This involved very minor changes, changing how we get the target type from a TypedefType, adding a parameter to EnumDecl::Create(), and other minor tweaks. llvm-svn: 121663
2010-12-13 09:26:27 +08:00
return ClangASTContext::IsFunctionPointerType (cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr());
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated:
return ClangASTContext::IsFunctionPointerType (cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr());
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
case clang::Type::LValueReference:
case clang::Type::RValueReference:
{
Updated Clang to a version that supports propagating the "virtual" flag when importing a C++ function declaration. Made changes to LLDB to support other changes in Clang. llvm-svn: 124355
2011-01-27 12:42:51 +08:00
const ReferenceType *reference_type = cast<ReferenceType>(qual_type.getTypePtr());
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
if (reference_type)
return ClangASTContext::IsFunctionPointerType (reference_type->getPointeeType().getAsOpaquePtr());
}
break;
}
}
return false;
}
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
size_t
ClangASTContext::GetArraySize (clang_type_t clang_type)
{
if (clang_type)
{
Fixed an issue with getting the size of an array from an opaque clang_type_t variable where we were using a static cast when we shouldn't have been. Fixed another array related issue where we weren't looking through a typedef. llvm-svn: 134823
2011-07-10 01:12:27 +08:00
QualType qual_type(QualType::getFromOpaquePtr(clang_type));
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
{
case clang::Type::ConstantArray:
{
const ConstantArrayType *array = cast<ConstantArrayType>(QualType::getFromOpaquePtr(clang_type).getTypePtr());
if (array)
return array->getSize().getLimitedValue();
}
break;
case clang::Type::Typedef:
return ClangASTContext::GetArraySize(cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr());
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated:
return ClangASTContext::GetArraySize(cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr());
named summaries: - a new --name option for "type summary add" lets you give a name to a summary - a new --summary option for "frame variable" lets you bind a named summary to one or more variables ${var%s} now works for printing the value of 0-terminated CStrings type format test case now tests for cascading - this is disabled on GCC because GCC may end up stripping typedef chains, basically breaking cascading new design for the FormatNavigator class new template class CleanUp2 meant to support cleanup routines with 1 additional parameter beyond resource handle llvm-svn: 134943
2011-07-12 08:18:11 +08:00
default:
break;
Fixed an issue with getting the size of an array from an opaque clang_type_t variable where we were using a static cast when we shouldn't have been. Fixed another array related issue where we weren't looking through a typedef. llvm-svn: 134823
2011-07-10 01:12:27 +08:00
}
Updated the lldb_private::Flags class to have better method names and made all of the calls inlined in the header file for better performance. Fixed the summary for C string types (array of chars (with any combo if modifiers), and pointers to chars) work in all cases. Fixed an issue where a forward declaration to a clang type could cause itself to resolve itself more than once if, during the resolving of the type itself it caused something to try and resolve itself again. We now remove the clang type from the forward declaration map in the DWARF parser when we start to resolve it and avoid this additional call. This should stop any duplicate members from appearing and throwing all the alignment of structs, unions and classes. llvm-svn: 117437
2010-10-27 11:32:59 +08:00
}
return 0;
}
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
Made IsArrayOfScalarType handle typedefs correctly. We should ultimately introduce GetAs...Type functions in all cases where we have Is...Type functions that know how to look inside typedefs. llvm-svn: 148512
2012-01-20 07:54:24 +08:00
clang_type_t
ClangASTContext::GetAsArrayType (clang_type_t clang_type, clang_type_t*member_type, uint64_t *size)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (!clang_type)
Made IsArrayOfScalarType handle typedefs correctly. We should ultimately introduce GetAs...Type functions in all cases where we have Is...Type functions that know how to look inside typedefs. llvm-svn: 148512
2012-01-20 07:54:24 +08:00
return 0;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
Added the summary values for function pointers so we can show where they point to. llvm-svn: 113735
2010-09-13 11:32:57 +08:00
const clang::Type::TypeClass type_class = qual_type->getTypeClass();
switch (type_class)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
default:
break;
Fixed an issue with getting the size of an array from an opaque clang_type_t variable where we were using a static cast when we shouldn't have been. Fixed another array related issue where we weren't looking through a typedef. llvm-svn: 134823
2011-07-10 01:12:27 +08:00
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::ConstantArray:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (member_type)
*member_type = cast<ConstantArrayType>(qual_type)->getElementType().getAsOpaquePtr();
if (size)
Get rid of LONG_LONG_MAX and ULONG_LONG_MAX, and use LLONG_MAX and ULLONG_MAX respectively. llvm-svn: 128720
2011-04-02 02:14:08 +08:00
*size = cast<ConstantArrayType>(qual_type)->getSize().getLimitedValue(ULLONG_MAX);
Made IsArrayOfScalarType handle typedefs correctly. We should ultimately introduce GetAs...Type functions in all cases where we have Is...Type functions that know how to look inside typedefs. llvm-svn: 148512
2012-01-20 07:54:24 +08:00
return clang_type;
Fixed an issue with getting the size of an array from an opaque clang_type_t variable where we were using a static cast when we shouldn't have been. Fixed another array related issue where we weren't looking through a typedef. llvm-svn: 134823
2011-07-10 01:12:27 +08:00
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::IncompleteArray:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (member_type)
*member_type = cast<IncompleteArrayType>(qual_type)->getElementType().getAsOpaquePtr();
if (size)
*size = 0;
Made IsArrayOfScalarType handle typedefs correctly. We should ultimately introduce GetAs...Type functions in all cases where we have Is...Type functions that know how to look inside typedefs. llvm-svn: 148512
2012-01-20 07:54:24 +08:00
return clang_type;
Fixed an issue with getting the size of an array from an opaque clang_type_t variable where we were using a static cast when we shouldn't have been. Fixed another array related issue where we weren't looking through a typedef. llvm-svn: 134823
2011-07-10 01:12:27 +08:00
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::VariableArray:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (member_type)
*member_type = cast<VariableArrayType>(qual_type)->getElementType().getAsOpaquePtr();
if (size)
*size = 0;
Made IsArrayOfScalarType handle typedefs correctly. We should ultimately introduce GetAs...Type functions in all cases where we have Is...Type functions that know how to look inside typedefs. llvm-svn: 148512
2012-01-20 07:54:24 +08:00
return clang_type;
Fixed an issue with getting the size of an array from an opaque clang_type_t variable where we were using a static cast when we shouldn't have been. Fixed another array related issue where we weren't looking through a typedef. llvm-svn: 134823
2011-07-10 01:12:27 +08:00
Change over to using the definitions for mach-o types and defines to the defines that are in "llvm/Support/MachO.h". This should allow ObjectFileMachO and ObjectContainerUniversalMachO to be able to be cross compiled in Linux. Also did some cleanup on the ASTType by renaming it to ClangASTType and renaming the header file. Moved a lot of "AST * + opaque clang type *" functionality from lldb_private::Type over into ClangASTType. llvm-svn: 109046
2010-07-22 06:12:05 +08:00
case clang::Type::DependentSizedArray:
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
if (member_type)
*member_type = cast<DependentSizedArrayType>(qual_type)->getElementType().getAsOpaquePtr();
if (size)
*size = 0;
Made IsArrayOfScalarType handle typedefs correctly. We should ultimately introduce GetAs...Type functions in all cases where we have Is...Type functions that know how to look inside typedefs. llvm-svn: 148512
2012-01-20 07:54:24 +08:00
return clang_type;
Fixed an issue with getting the size of an array from an opaque clang_type_t variable where we were using a static cast when we shouldn't have been. Fixed another array related issue where we weren't looking through a typedef. llvm-svn: 134823
2011-07-10 01:12:27 +08:00
case clang::Type::Typedef:
Made IsArrayOfScalarType handle typedefs correctly. We should ultimately introduce GetAs...Type functions in all cases where we have Is...Type functions that know how to look inside typedefs. llvm-svn: 148512
2012-01-20 07:54:24 +08:00
return ClangASTContext::GetAsArrayType (cast<TypedefType>(qual_type)->getDecl()->getUnderlyingType().getAsOpaquePtr(),
member_type,
size);
Fixed LLDB's handling of ElaboratedTypes, which was causing problems with printing the values of persistent variables with struct types. llvm-svn: 137392
2011-08-12 07:56:13 +08:00
case clang::Type::Elaborated:
Made IsArrayOfScalarType handle typedefs correctly. We should ultimately introduce GetAs...Type functions in all cases where we have Is...Type functions that know how to look inside typedefs. llvm-svn: 148512
2012-01-20 07:54:24 +08:00
return ClangASTContext::GetAsArrayType (cast<ElaboratedType>(qual_type)->getNamedType().getAsOpaquePtr(),
member_type,
size);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
Made IsArrayOfScalarType handle typedefs correctly. We should ultimately introduce GetAs...Type functions in all cases where we have Is...Type functions that know how to look inside typedefs. llvm-svn: 148512
2012-01-20 07:54:24 +08:00
return 0;
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
#pragma mark Typedefs
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
clang_type_t
ClangASTContext::CreateTypedefType (const char *name, clang_type_t clang_type, DeclContext *decl_ctx)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ASTContext *ast = getASTContext();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
IdentifierTable *identifier_table = getIdentifierTable();
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
assert (ast != NULL);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
assert (identifier_table != NULL);
if (decl_ctx == NULL)
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
decl_ctx = ast->getTranslationUnitDecl();
TypedefDecl *decl = TypedefDecl::Create (*ast,
decl_ctx,
SourceLocation(),
Updated to LLVM/Clang revision 127600. llvm-svn: 127634
2011-03-15 08:17:19 +08:00
SourceLocation(),
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
name ? &identifier_table->get(name) : NULL, // Identifier
ast->CreateTypeSourceInfo(qual_type));
Quick fixes to set access specifiers for Typedefs and Enums inside classes. llvm-svn: 123710
2011-01-18 09:03:44 +08:00
Add function decls to their parent decl context. llvm-svn: 142011
2011-10-15 06:47:18 +08:00
//decl_ctx->addDecl (decl);
Quick fixes to set access specifiers for Typedefs and Enums inside classes. llvm-svn: 123710
2011-01-18 09:03:44 +08:00
decl->setAccess(AS_public); // TODO respect proper access specifier
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
// Get a uniqued QualType for the typedef decl type
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
return ast->getTypedefType (decl).getAsOpaquePtr();
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
}
return NULL;
}
// Disable this for now since I can't seem to get a nicely formatted float
// out of the APFloat class without just getting the float, double or quad
// and then using a formatted print on it which defeats the purpose. We ideally
// would like to get perfect string values for any kind of float semantics
// so we can support remote targets. The code below also requires a patch to
// llvm::APInt.
//bool
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
//ClangASTContext::ConvertFloatValueToString (ASTContext *ast, clang_type_t clang_type, const uint8_t* bytes, size_t byte_size, int apint_byte_order, std::string &float_str)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
//{
// uint32_t count = 0;
// bool is_complex = false;
// if (ClangASTContext::IsFloatingPointType (clang_type, count, is_complex))
// {
// unsigned num_bytes_per_float = byte_size / count;
// unsigned num_bits_per_float = num_bytes_per_float * 8;
//
// float_str.clear();
// uint32_t i;
// for (i=0; i<count; i++)
// {
// APInt ap_int(num_bits_per_float, bytes + i * num_bytes_per_float, (APInt::ByteOrder)apint_byte_order);
// bool is_ieee = false;
// APFloat ap_float(ap_int, is_ieee);
// char s[1024];
// unsigned int hex_digits = 0;
// bool upper_case = false;
//
// if (ap_float.convertToHexString(s, hex_digits, upper_case, APFloat::rmNearestTiesToEven) > 0)
// {
// if (i > 0)
// float_str.append(", ");
// float_str.append(s);
// if (i == 1 && is_complex)
// float_str.append(1, 'i');
// }
// }
// return !float_str.empty();
// }
// return false;
//}
size_t
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
ClangASTContext::ConvertStringToFloatValue (ASTContext *ast, clang_type_t clang_type, const char *s, uint8_t *dst, size_t dst_size)
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
{
if (clang_type)
{
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
uint32_t count = 0;
bool is_complex = false;
if (ClangASTContext::IsFloatingPointType (clang_type, count, is_complex))
{
// TODO: handle complex and vector types
if (count != 1)
return false;
StringRef s_sref(s);
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
APFloat ap_float(ast->getFloatTypeSemantics(qual_type), s_sref);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
const uint64_t bit_size = ast->getTypeSize (qual_type);
Initial checkin of lldb code from internal Apple repo. llvm-svn: 105619
2010-06-09 00:52:24 +08:00
const uint64_t byte_size = bit_size / 8;
if (dst_size >= byte_size)
{
if (bit_size == sizeof(float)*8)
{
float float32 = ap_float.convertToFloat();
::memcpy (dst, &float32, byte_size);
return byte_size;
}
else if (bit_size >= 64)
{
llvm::APInt ap_int(ap_float.bitcastToAPInt());
::memcpy (dst, ap_int.getRawData(), byte_size);
return byte_size;
}
}
}
}
return 0;
}
Added a static function to get the void type for an ASTContext; also added a function to get the Clang-style CVR qualifiers for a type as an unsigned int. llvm-svn: 114152
2010-09-17 10:24:29 +08:00
unsigned
Fixed the forward declaration issue that was present in the DWARF parser after adding methods to C++ and objective C classes. In order to make methods, we need the function prototype which means we need the arguments. Parsing these could cause a circular reference that caused an assertion. Added a new typedef for the clang opaque types which are just void pointers: lldb::clang_type_t. This appears in lldb-types.h. This was fixed by enabling struct, union, class, and enum types to only get a forward declaration when we make the clang opaque qual type for these types. When they need to actually be resolved, lldb_private::Type will call a new function in the SymbolFile protocol to resolve a clang type when it is not fully defined (clang::TagDecl::getDefinition() returns NULL). This allows us to be a lot more lazy when parsing clang types and keeps down the amount of data that gets parsed into the ASTContext for each module. Getting the clang type from a "lldb_private::Type" object now takes a boolean that indicates if a forward declaration is ok: clang_type_t lldb_private::Type::GetClangType (bool forward_decl_is_ok); So function prototypes that define parameters that are "const T&" can now just parse the forward declaration for type 'T' and we avoid circular references in the type system. llvm-svn: 115012
2010-09-29 09:12:09 +08:00
ClangASTContext::GetTypeQualifiers(clang_type_t clang_type)
Added a static function to get the void type for an ASTContext; also added a function to get the Clang-style CVR qualifiers for a type as an unsigned int. llvm-svn: 114152
2010-09-17 10:24:29 +08:00
{
assert (clang_type);
QualType qual_type (QualType::getFromOpaquePtr(clang_type));
return qual_type.getQualifiers().getCVRQualifiers();
}
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
bool
ClangASTContext::GetCompleteType (clang::ASTContext *ast, lldb::clang_type_t clang_type)
{
if (clang_type == NULL)
return false;
Fixed the auto completion of objective C types with the new ExternalASTSource changes that recently happened. llvm-svn: 123881
2011-01-20 12:18:48 +08:00
return GetCompleteQualType (ast, clang::QualType::getFromOpaquePtr(clang_type));
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 11:46:26 +08:00
}
bool
ClangASTContext::GetCompleteType (clang_type_t clang_type)
{
return ClangASTContext::GetCompleteType (getASTContext(), clang_type);
}
Adding a new API call IsTypeComplete() to SBType. This call is meant to check if the type has been previously completed or not (which is mostly interesting from a performance point of view) Adding a test case that checks that we do not complete types before due time. This should help us track cases similar to the cascading data formatters. llvm-svn: 153363
2012-03-24 09:11:14 +08:00
bool
ClangASTContext::IsCompleteType (clang::ASTContext *ast, lldb::clang_type_t clang_type)
{
if (clang_type == NULL)
return false;
return GetCompleteQualType (ast, clang::QualType::getFromOpaquePtr(clang_type), false); // just check but don't let it actually complete
}
bool
ClangASTContext::IsCompleteType (clang_type_t clang_type)
{
return ClangASTContext::IsCompleteType (getASTContext(), clang_type);
}
This commit adds broad architectural support for hierarchical inspection of namespaces in the expression parser. ClangExpressionDeclMap hitherto reported that namespaces had been completely imported, even though the namespaces are returned empty. To deal with this situation, ClangASTSource was recently extended with an API to complete incomplete type definitions, and, for greater efficiency, to complete these definitions partially, returning only those objects that have a given name. This commit supports these APIs on LLDB's side, and uses it to provide information on types resident in namespaces. Namespaces are now imported as they were -- that is to say, empty -- but with minimal import mode on. This means that Clang will come back and request their contents by name as needed. We now respond with information on the contained types; this will be followed soon by information on functions and variables. llvm-svn: 133852
2011-06-25 08:44:06 +08:00
bool
ClangASTContext::GetCompleteDecl (clang::ASTContext *ast,
clang::Decl *decl)
{
if (!decl)
return false;
ExternalASTSource *ast_source = ast->getExternalSource();
if (!ast_source)
return false;
if (clang::TagDecl *tag_decl = llvm::dyn_cast<clang::TagDecl>(decl))
{
<rdar://problem/11082392> Fixed an issue that could cause circular type parsing that will assert and kill LLDB. Prior to this fix the DWARF parser would always create class types and not start their definitions (for both C++ and ObjC classes) until we were asked to complete the class later. When we had cases like: class A { class B { }; }; We would alway try to complete A before specifying "A" as the decl context for B. Turns out we can just start the definition and still not complete the class since we can check the TagDecl::isCompleteDefinition() function. This only works for C++ types. This means we will not be pulling in the full definition of parent classes all the time and should help with our memory consumption and also reduce the amount of debug info we have to parse. I also reduced redundant code that was checking in a lldb::clang_type_t was a possible C++ dynamic type since it was still completing the type, just to see if it was dynamic. This was fixed in another function that was checking for a type being dynamic as an ObjC or a C++ type, but there was dedicated fucntion for C++ that we missed. llvm-svn: 153713
2012-03-30 08:51:13 +08:00
if (tag_decl->isCompleteDefinition())
This commit adds broad architectural support for hierarchical inspection of namespaces in the expression parser. ClangExpressionDeclMap hitherto reported that namespaces had been completely imported, even though the namespaces are returned empty. To deal with this situation, ClangASTSource was recently extended with an API to complete incomplete type definitions, and, for greater efficiency, to complete these definitions partially, returning only those objects that have a given name. This commit supports these APIs on LLDB's side, and uses it to provide information on types resident in namespaces. Namespaces are now imported as they were -- that is to say, empty -- but with minimal import mode on. This means that Clang will come back and request their contents by name as needed. We now respond with information on the contained types; this will be followed soon by information on functions and variables. llvm-svn: 133852
2011-06-25 08:44:06 +08:00
return true;
if (!tag_decl->hasExternalLexicalStorage())
return false;
ast_source->CompleteType(tag_decl);
return !tag_decl->getTypeForDecl()->isIncompleteType();
}
else if (clang::ObjCInterfaceDecl *objc_interface_decl = llvm::dyn_cast<clang::ObjCInterfaceDecl>(decl))
{
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
if (objc_interface_decl->getDefinition())
This commit adds broad architectural support for hierarchical inspection of namespaces in the expression parser. ClangExpressionDeclMap hitherto reported that namespaces had been completely imported, even though the namespaces are returned empty. To deal with this situation, ClangASTSource was recently extended with an API to complete incomplete type definitions, and, for greater efficiency, to complete these definitions partially, returning only those objects that have a given name. This commit supports these APIs on LLDB's side, and uses it to provide information on types resident in namespaces. Namespaces are now imported as they were -- that is to say, empty -- but with minimal import mode on. This means that Clang will come back and request their contents by name as needed. We now respond with information on the contained types; this will be followed soon by information on functions and variables. llvm-svn: 133852
2011-06-25 08:44:06 +08:00
return true;
if (!objc_interface_decl->hasExternalLexicalStorage())
return false;
ast_source->CompleteType(objc_interface_decl);
I have brought LLDB up-to-date with top of tree LLVM/Clang. This brings in several fixes, including: - Improvements in the Just-In-Time compiler's allocation of memory: the JIT now allocates memory in chunks of sections, improving its ability to generate relocations. I have revamped the RecordingMemoryManager to reflect these changes, as well as to get the memory allocation and data copying out fo the ClangExpressionParser code. Jim Grosbach wrote the updates to the JIT on the LLVM side. - A new ExternalASTSource interface to allow LLDB to report accurate structure layout information to Clang. Previously we could only report the sizes of fields, not their offsets. This meant that if data structures included field alignment directives, we could not communicate the necessary alignment to Clang and accesses to the data would fail. Now we can (and I have update the relevant test case). Thanks to Doug Gregor for implementing the Clang side of this fix. - The way Objective-C interfaces are completed by Clang has been made consistent with RecordDecls; with help from Doug Gregor and Greg Clayton I have ensured that this still works. - I have eliminated all local LLVM and Clang patches, committing the ones that are still relevant to LLVM and Clang as needed. I have tested the changes extensively locally, but please let me know if they cause any trouble for you. llvm-svn: 149775
2012-02-04 16:49:35 +08:00
return !objc_interface_decl->getTypeForDecl()->isIncompleteType();
This commit adds broad architectural support for hierarchical inspection of namespaces in the expression parser. ClangExpressionDeclMap hitherto reported that namespaces had been completely imported, even though the namespaces are returned empty. To deal with this situation, ClangASTSource was recently extended with an API to complete incomplete type definitions, and, for greater efficiency, to complete these definitions partially, returning only those objects that have a given name. This commit supports these APIs on LLDB's side, and uses it to provide information on types resident in namespaces. Namespaces are now imported as they were -- that is to say, empty -- but with minimal import mode on. This means that Clang will come back and request their contents by name as needed. We now respond with information on the contained types; this will be followed soon by information on functions and variables. llvm-svn: 133852
2011-06-25 08:44:06 +08:00
}
else
{
return false;
}
}
Added a mechanism for keeping track of where in the debug information individual Decls came from. We've had a metadata infrastructure for a while, which was intended to solve a problem we've since dealt with in a different way. (It was meant to keep track of which definition of an Objective-C class was the "true" definition, but we now find it by searching the symbols for the class symbol.) The metadata is attached to the ExternalASTSource, which means it has a one-to-one correspondence with AST contexts. I've repurposed the metadata infrastructure to hold the object file and DIE offset for the DWARF information corresponding to a Decl. There are methods in ClangASTContext that get and set this metadata, and the ClangASTImporter is capable of tracking down the metadata for Decls that have been copied out of the debug information into the parser's AST context without using any additional memory. To see the metadata, you just have to enable the expression log: - (lldb) log enable lldb expr - and watch the import messages. The high 32 bits of the metadata indicate the index of the object file in its containing DWARFDebugMap; I have also added a log which you can use to track that mapping: - (lldb) log enable dwarf map - This adds 64 bits per Decl, which in my testing hasn't turned out to be very much (debugging Clang produces around 6500 Decls in my tests). To track how much data is being consumed, I've also added a global variable g_TotalSizeOfMetadata which tracks the total number of Decls that have metadata in all active AST contexts. Right now this metadata is enormously useful for tracking down bugs in the debug info parser. In the future I also want to use this information to provide more intelligent error messages instead of printing empty source lines wherever Clang refers to the location where something is defined. llvm-svn: 154634
2012-04-13 08:10:03 +08:00
void
ClangASTContext::SetMetadata (clang::ASTContext *ast,
uintptr_t object,
uint64_t metadata)
{
ClangExternalASTSourceCommon *external_source =
static_cast<ClangExternalASTSourceCommon*>(ast->getExternalSource());
if (external_source)
external_source->SetMetadata(object, metadata);
}
uint64_t
ClangASTContext::GetMetadata (clang::ASTContext *ast,
uintptr_t object)
{
ClangExternalASTSourceCommon *external_source =
static_cast<ClangExternalASTSourceCommon*>(ast->getExternalSource());
if (external_source && external_source->HasMetadata(object))
return external_source->GetMetadata(object);
else
return 0;
}
Make sure we track CXX and objc method decls. llvm-svn: 136920
2011-08-05 05:02:57 +08:00
clang::DeclContext *
ClangASTContext::GetAsDeclContext (clang::CXXMethodDecl *cxx_method_decl)
{
Refined the rollback to LLVM, picking up a newer revision and adding a patch that fixes an AsmParser crash on ARM. One feature that we unfortunately lost (for the moment!) is the ability to cast unknown code symbols to arbitrary function types and put the resulting function pointer in a result variable. This feature will be back, though. llvm-svn: 138036
2011-08-19 14:19:25 +08:00
return llvm::dyn_cast<clang::DeclContext>(cxx_method_decl);
Make sure we track CXX and objc method decls. llvm-svn: 136920
2011-08-05 05:02:57 +08:00
}
clang::DeclContext *
ClangASTContext::GetAsDeclContext (clang::ObjCMethodDecl *objc_method_decl)
{
Refined the rollback to LLVM, picking up a newer revision and adding a patch that fixes an AsmParser crash on ARM. One feature that we unfortunately lost (for the moment!) is the ability to cast unknown code symbols to arbitrary function types and put the resulting function pointer in a result variable. This feature will be back, though. llvm-svn: 138036
2011-08-19 14:19:25 +08:00
return llvm::dyn_cast<clang::DeclContext>(objc_method_decl);
Make sure we track CXX and objc method decls. llvm-svn: 136920
2011-08-05 05:02:57 +08:00
}