Commit Graph

7177 Commits

Author SHA1 Message Date
Chris Lattner 330f0f2567 Fix a regression I introduced in r54107:
http://llvm.org/viewvc/llvm-project?view=rev&revision=54107

llvm-svn: 54242
2008-07-31 04:58:58 +00:00
Mon P Wang b160a0d78d Updated to match new atomic names (to enable overloading for pointers to
different address spaces) for llvm r54195.

llvm-svn: 54241
2008-07-31 03:28:23 +00:00
Daniel Dunbar b39b1cd7c3 Fix implicit initialization of structures.
<rdar://problem/6113085>

llvm-svn: 54237
2008-07-30 21:00:24 +00:00
Daniel Dunbar e49df9b58f Change CodeGenModule GlobalDeclMap to directly reference globals
instead of mapping the decl to a bitcast of the global to the correct
type.
 
 - GetAddrOf{Function,GlobalVar} introduce the bitcast on every use now.

 - This solves a problem where a dangling pointer could be introduced
   by the RAUW done when replacing a forward or tentative
   definition. See testcase for more details.

 - Fixes <rdar://problem/6108358>

llvm-svn: 54211
2008-07-30 16:32:24 +00:00
Daniel Dunbar 9c426524f8 Rework codegen emission of globals
- No (intended) functionality change.

 - Primary purpose is to clearly separate (lazy) construction of
   globals that are a forward declaration or tentative definition from
   those that are the final definition.

 - Lazy construction is now encapsulated in
   GetAddrOf{Function,GlobalVar} while final definitions are
   constructed in EmitGlobal{Function,Var}Definition.
 
 - External interface for dealing with globals is now limited to
   EmitGlobal and GetAddrOf{Function,GlobalVar}.

 - Also updated helper functions dealing with statics, annotations,
   and ctors to be private.

llvm-svn: 54179
2008-07-29 23:18:29 +00:00
Daniel Dunbar 837fd272f8 Fix codegen of chained declarations
- Killed useless CodeGenModule::EmitGlobalVarDeclarator, instead just
   recurse on any ScopedDecl.
 - Fix for <rdar://problem/6093838>

llvm-svn: 54162
2008-07-29 17:47:36 +00:00
Chris Lattner d9fa5d6a23 Fix a bug in my previous patch, CGF is not valid for constants outside a function
llvm-svn: 54108
2008-07-26 22:39:33 +00:00
Chris Lattner 0f398c4472 remove uses of QualType::getCanonicalType() from codegen for PR2189
llvm-svn: 54107
2008-07-26 22:37:01 +00:00
Chris Lattner 574dee6cac change more instances of QualType::getCanonicalType to call
ASTContext::getCanonicalType instead (PR2189)

llvm-svn: 54105
2008-07-26 22:17:49 +00:00
Chris Lattner 7e80097f6e implement codegen support for labels at the end of stmtexprs.
llvm-svn: 54100
2008-07-26 20:23:23 +00:00
Chris Lattner 6284378fcd minor tidying, no functionality change.
llvm-svn: 54099
2008-07-26 20:15:14 +00:00
Chris Lattner 4591fd0aac Fix rdar://6103124, the only dummy blocks are ones without names. Named blocks
are important part of control flow structures.

llvm-svn: 54078
2008-07-25 23:40:10 +00:00
Nate Begeman 448b0cb033 X86 builtin fixes
llvm-svn: 54049
2008-07-25 20:17:24 +00:00
Nate Begeman fe79ca2ab9 Generate vector compares in codegen
llvm-svn: 54048
2008-07-25 20:16:05 +00:00
Nate Begeman e7f45b2977 Move comment to match style
llvm-svn: 54047
2008-07-25 20:15:41 +00:00
Daniel Dunbar 0e5845c13a Rework codegen of case ranges
- Fix multiple issues with the way case ranges were emitted, see test
   cases details about the specific issues.

 - The root issue was not being careful about how basic blocks were
   emitted which led to them being chained together incorrectly,
   resulting in improper control flow.

 - Fixes <rdar://problem/6098585>

llvm-svn: 54006
2008-07-25 01:11:38 +00:00
Daniel Dunbar 2e0f1cd00c Fix EmitCaseStmtRange to ignore empty ranges
- Also cleaned up emission slightly 
 - Inspection of the code revealed several other bugs, however. Case
   ranges are not properly wired and can result in switch cases being
   dropped or even infinite loops. See: <rdar://problem/6098585>

Completes: <rdar://problem/6094119>
llvm-svn: 53975
2008-07-24 01:18:41 +00:00
Chris Lattner 28ee5b3d29 don't depend on the result of CreateNeg to be a binop
llvm-svn: 53948
2008-07-23 06:53:34 +00:00
Chris Lattner d258325c2b Fix a codegen crash on:
int foo(void) {
  float x[2];
  return x;
}

rdar://6093986

llvm-svn: 53946
2008-07-23 06:31:27 +00:00
Daniel Dunbar a2fbefc159 Add note to check what __alignof__ a function type should be.
llvm-svn: 53928
2008-07-22 19:44:18 +00:00
Daniel Dunbar ee8bbb3ccc Prevent codegen crash on sizeof(<function type>), a gcc extension.
llvm-svn: 53899
2008-07-22 01:35:47 +00:00
Daniel Dunbar 327acd7a6c Implement bzero, memset, memmove builtins.
This subsumes Nico Weber's patch and implements bzero in terms
of llvm.memset to match llvm-gcc.

llvm-svn: 53888
2008-07-22 00:26:45 +00:00
Daniel Dunbar b725726c29 Implement nans, prefetch, and trap builtins.
This closes <rdar://problem/6080720>, support for __builtin_constant_p
has been filed separately.

llvm-svn: 53885
2008-07-21 22:59:13 +00:00
Daniel Dunbar 7ed03c4cdd Kill an unused variable warning
llvm-svn: 53878
2008-07-21 21:43:28 +00:00
Chris Lattner 390d39ac7e move two more lists of protocols over to use ObjCList<ObjCProtocolDecl>,
simplifying code along the way and fixing a problem and memory leak or two.

llvm-svn: 53876
2008-07-21 21:32:27 +00:00
Daniel Dunbar c2f67966f4 Add __builtin_powi[fl] support
llvm-svn: 53866
2008-07-21 18:44:41 +00:00
Chris Lattner d004505b74 introduce a new ObjCList templated class and start moving
various objc lists over to it.  First up, the protocol list 
on ObjCInterfaceDecl.

llvm-svn: 53856
2008-07-21 18:19:38 +00:00
Daniel Dunbar d93abc3bb0 Implement ffs, parity, and popcount builtins.
+ test case

llvm-svn: 53851
2008-07-21 17:19:41 +00:00
Chris Lattner f1c149db9f don't pass in null as the Name for GenerateClassStructure,
it goes and makes std::strings out of them, which is not 
defined.

llvm-svn: 53843
2008-07-21 06:31:05 +00:00
Mon P Wang 194c58f56d Add support shufpd
llvm-svn: 53747
2008-07-18 00:14:09 +00:00
Daniel Dunbar 1f32c759dc Fix for codegen crash on multibit bool bitfield initialization
<rdar://problem/6078606>

llvm-svn: 53737
2008-07-17 21:07:48 +00:00
Chris Lattner 6d1a3c2e2b remove CGObjCEtoile until it compiles.
llvm-svn: 53552
2008-07-14 16:54:05 +00:00
Chris Lattner 0bf679180c Codegen support for __builtin_stdarg_start
llvm-svn: 53329
2008-07-09 17:28:44 +00:00
Sanjiv Gupta 1e8b608751 Fixed bugzilla bug# 2489.
This was broken when the GenerateCode function was splitted to use GenerateFunction.

llvm-svn: 53136
2008-07-04 11:04:26 +00:00
Bill Wendling e7264ec999 Remove removed header.
llvm-svn: 53121
2008-07-03 22:54:30 +00:00
Chris Lattner 9a8d1d9e56 Make a few related changes:
1) add a new ASTContext::getFloatTypeSemantics method.
2) Use it from SemaExpr.cpp, CodeGenTypes.cpp and other places.
3) Change the TargetInfo.h get*Format methods to return their 
   fltSemantics byref instead of by pointer.
4) Change CodeGenFunction::EmitBuiltinExpr to allow builtins which
   sometimes expand specially and othertimes fall back to libm.
5) Add support for __builtin_nan("") to codegen, cases that don't pass
   in an empty string are currently lowered to libm calls.
6) Fix codegen of __builtin_infl.

llvm-svn: 52914
2008-06-30 18:32:54 +00:00
Cedric Venet 3903908794 Add missing include file (due to a file splitting in llvm).
Add new file to Sema Project for VS.
this unbreak the build for VS2005 (with the associated patch on llvm).

llvm-svn: 52830
2008-06-27 17:53:02 +00:00
Chris Lattner bf231a61fd refactor more objc codegen interfaces to pass around selectors so
we don't push strings into LLVM IR and then have to read them back out.

llvm-svn: 52765
2008-06-26 05:08:00 +00:00
Chris Lattner 882034dd99 indenting and other minor things.
llvm-svn: 52764
2008-06-26 04:52:29 +00:00
Chris Lattner da35bc8e53 refactor interface to GenerateClassStructure to avoid converting a
string to LLVM IR and then regenerating the string from IR.

llvm-svn: 52762
2008-06-26 04:47:04 +00:00
Chris Lattner d9b98863c4 remove the old getSelector implementation, which removes some
calls to getStringValue

llvm-svn: 52761
2008-06-26 04:44:19 +00:00
Chris Lattner 6cfec78e48 avoid a lot of unneeded selector processing work by passing around
selectors instead of Value*'s.

llvm-svn: 52760
2008-06-26 04:42:20 +00:00
Chris Lattner 9a05d04a0d use cheaper/simpler getselector call for @selector exprs.
llvm-svn: 52759
2008-06-26 04:38:58 +00:00
Chris Lattner 6d522c0133 start avoid doing lots of unneeded work handling selectors
llvm-svn: 52758
2008-06-26 04:37:12 +00:00
Chris Lattner 42587147d4 improve indentation, avoid thrashing on maps and recalculating strings as much.
llvm-svn: 52757
2008-06-26 04:24:57 +00:00
Chris Lattner 87ab27d42f give CreateObjCRuntime a full CGM so it can get whatever state it needs,
instead of passing in just a couple random things it currently 
happens to use.

llvm-svn: 52756
2008-06-26 04:19:03 +00:00
Chris Lattner 63dd337fc2 Fix 80 col violations, assert on assumptions.
llvm-svn: 52755
2008-06-26 04:10:42 +00:00
Chris Lattner d46944d639 a temporary minimal hack to get clang building after the getStringValue changes in llvm mainline.
llvm-svn: 52753
2008-06-26 04:05:20 +00:00
Mon P Wang 28898b2888 Updated atomic intrinsic name from llvm r52706. Fixed dropped bit in shufps.
llvm-svn: 52707
2008-06-25 08:21:36 +00:00
Chris Lattner b1d329da07 "Support for Objective-C message sends which return structures. Also includes a small fix for constant string handling that should have been in the last patch (sorry!) and a hook for generating selectors (rest of this implementation to follow in the next patch)."
Patch by David Chisnall!

llvm-svn: 52681
2008-06-24 17:04:18 +00:00
Chris Lattner 6447ed3a1f "This is a small fix for a bug where static object instances were being incorrectly generated. The bug was caused by my inability to read the GNU libobjc source and is only apparent when JITing code (static compilation does not expose the bug due to the data layout of other globals)."
Patch by David Chisnall!

llvm-svn: 52680
2008-06-24 17:01:28 +00:00
Chris Lattner 091f698d29 "this patch adds code generation hooks for Objective-C constant strings. It also modifies Sema so that Objective-C constant strings are treated as untyped objects if the interface for the constant string class can not be found. This is consistent with Apple GCC. I thought it was consistent with GNU GCC, since this was causing failures when trying to compile GNUstep with (GNU) GCC, but it appears that this is not the case when attempting to produce a simple test case to demonstrate it. Possibly there is a way of making the error go away, but I haven't found out what it is yet."
Patch by David Chisnall!

llvm-svn: 52599
2008-06-21 21:44:18 +00:00
Chris Lattner c4f38851e6 Don't add isa with @defs only to work around it in the code generator, patch
by David Chisnall. 

llvm-svn: 52597
2008-06-21 21:37:25 +00:00
Mike Stump d7667bd8e0 Remove tabs.
llvm-svn: 52507
2008-06-19 20:57:50 +00:00
Chris Lattner 5696e7badf Change self/_cmd to be instances of ImplicitParamDecl instead of ParmVarDecl.
Patch by David Chisnall!

llvm-svn: 52422
2008-06-17 18:05:57 +00:00
Ted Kremenek 08e171183f This patch is motivated by numerous strict-aliasing warnings when compiling
clang as a Release build.

The big change is that all AST nodes (subclasses of Stmt) whose children are
Expr* store their children as Stmt* or arrays of Stmt*. This is to remove
strict-aliasing warnings when using StmtIterator. None of the interfaces of any
of the classes have changed (except those with arg_iterators, see below), as the
accessor methods introduce the needed casts (via cast<>). While this extra
casting may seem cumbersome, it actually adds some important sanity checks
throughout the codebase, as clients using StmtIterator can potentially overwrite
children that are expected to be Expr* with Stmt* (that aren't Expr*). The casts
provide extra sanity checks that are operational in debug builds to catch
invariant violations such as these.

For classes that have arg_iterators (e.g., CallExpr), the definition of
arg_iterator has been replaced. Instead of it being Expr**, it is an actual
class (called ExprIterator) that wraps a Stmt**, and provides the necessary
operators for iteration. The nice thing about this class is that it also uses
cast<> to type-checking, which introduces extra sanity checks throughout the
codebase that are useful for debugging.

A few of the CodeGen functions that use arg_iterator (especially from
OverloadExpr) have been modified to take begin and end iterators instead of a
base Expr** and the number of arguments. This matches more with the abstraction
of iteration. This still needs to be cleaned up a little bit, as clients expect
that ExprIterator is a RandomAccessIterator (which we may or may not wish to
allow for efficiency of representation).

This is a fairly large patch. It passes the tests (except CodeGen/bitfield.c,
which was already broken) on both a Debug and Release build, but it should
obviously be reviewed.

llvm-svn: 52378
2008-06-17 02:43:46 +00:00
Chris Lattner 22b9ff4131 force size of alloca to i32, which is currently required by LLVM IR.
This fixes use of alloca on 64-bit systems.

llvm-svn: 52334
2008-06-16 17:15:14 +00:00
Eli Friedman 327944b343 Basic support for volatile loads and stores. Stores the volatile
qualifier in the lvalue, and changes lvalue loads/stores to honor 
the volatile flag.  Places which need some further attention are marked 
with FIXMEs.

Patch by Cédric Venet.

llvm-svn: 52264
2008-06-13 23:01:12 +00:00
Argyrios Kyrtzidis d0c1df4c84 Make CodeGenFunction::EmitDecl() recognise CXXClass/CXXStruct/CXXUnion.
llvm-svn: 52162
2008-06-09 23:42:47 +00:00
Argyrios Kyrtzidis 554a07bab3 -Changes to TagDecl:
Added TagKind enum.
  Added getTagKind() method.
  Added convenience methods: isEnum(), isStruct(), isUnion(), isClass().
-RecordDecl/CXXRecordDecl::Create() accept a TagKind enum instead of a DeclKind one.

llvm-svn: 52160
2008-06-09 23:19:58 +00:00
Sanjiv Gupta 224e8ea2bb Generate debug descriptors for array types while generating the debug info.
llvm-svn: 52140
2008-06-09 10:47:41 +00:00
Nuno Lopes b6f7953818 implement the alias attirbute (in both Sema and Codegen)
llvm-svn: 52092
2008-06-08 15:45:52 +00:00
Nuno Lopes cc041ea68e fix compiler warnings
llvm-svn: 52089
2008-06-08 10:16:34 +00:00
Eli Friedman c98a7add43 Don't crash emitting an initializer for a static local with union type.
This fix just makes sure to construct the global with the appropriate 
type, and fixes up the one user this affects to compensate.

llvm-svn: 52084
2008-06-08 01:23:18 +00:00
Eli Friedman 3ce7b20590 Fix crash with conversion to an address-space-qualified pointer. Bug
reported on cfe-dev by Cédric Venet.

Note that I seriously doubt that this perticular construct is useful, 
though: it's a pointer in an alternate address space pointing into 
unqualified address space.

llvm-svn: 52076
2008-06-07 17:27:51 +00:00
Sanjiv Gupta 1929242a94 Create debug type descriptors for aggregate/enum types.
llvm-svn: 52070
2008-06-07 04:46:53 +00:00
Eli Friedman fd1e32b1f3 Reorganize this loop a bit so it doesn't crash for empty unions. Fixes
PR2419.

llvm-svn: 52060
2008-06-06 20:12:37 +00:00
Nuno Lopes 9b07b7b099 implement constant expr. sub ptr ptr
llvm-svn: 52049
2008-06-06 16:41:08 +00:00
Sanjiv Gupta 158143ad54 Emit debug information for global and static variables when -g is specified.
llvm-svn: 51993
2008-06-05 08:59:10 +00:00
Eli Friedman 2425cd8158 For setting attributes, don't assume there are ParamVarDecls available,
because trying to access non-existent ParamVarDecls can crash. 
Testcase from the original source for PR2414.

llvm-svn: 51960
2008-06-04 19:41:28 +00:00
Nuno Lopes 7bd6e58ddc fix crash when codegening 'cond ? lhs : call-to-void-func()'
llvm-svn: 51958
2008-06-04 19:15:45 +00:00
Eli Friedman c0b54901f4 clang fix to parallel LLVM r51928
(http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080602/063306.html).

llvm-svn: 51944
2008-06-04 16:13:19 +00:00
Argyrios Kyrtzidis 28172cb812 Include <string> header to allow compiling for MSVC.
llvm-svn: 51847
2008-06-01 21:23:24 +00:00
Eli Friedman 412c66839e Don't name the ObjC types if there aren't any ObjC objects in the file.
Patch by David Chisnall.

llvm-svn: 51840
2008-06-01 16:00:02 +00:00
Eli Friedman f6b091f73a First cut at setting attributes for functions and calls; this puts us
much closer to passing the gcc struct layout tests.

It might be possible to refactor this a bit, but I'm not sure there's 
actually enough common code for that to be useful.

To get the calling convention completely correct, a bit of 
platform-specific code is necessary even for x86-Linux. On x86-Linux, the
alignment of function parameters is extremely strange; as far as I can tell,
it's always 4 except for SSE vectors or structs containing SSE vectors.  I'm 
continuing to investigate this.

llvm-svn: 51839
2008-06-01 15:54:49 +00:00
Eli Friedman 10c2417570 Make sure _Bool globals have the correct type.
llvm-svn: 51838
2008-06-01 15:31:44 +00:00
Eli Friedman 149a57f7d4 Minor cleanup to use the ConvertTypeForMem helper.
llvm-svn: 51837
2008-06-01 15:16:01 +00:00
Anton Korobeynikov 3b6dd58ee7 Remove spurious == 0
llvm-svn: 51836
2008-06-01 15:14:46 +00:00
Anton Korobeynikov 1200aca763 Support for code generation of Objective-C top-level language constructs.
Implemented by David Chisnall!

llvm-svn: 51835
2008-06-01 14:13:53 +00:00
Eli Friedman 252e5f1b29 Calculate alignment for local variables.
llvm-svn: 51826
2008-05-31 21:20:41 +00:00
Eli Friedman 6859a1b961 PR1893: Fix up the type of tentative definitions of incomplete array
types so that they end up the correct size.

llvm-svn: 51787
2008-05-30 20:39:54 +00:00
Eli Friedman 34994cbcbf Fix and enable generating general union initializers. Essentially, what
this does is reconstruct the type for structs and arrays if the type 
wouldn't be compatible otherwise.

The assertion about packing in the struct type reconstruction code 
sucks, but I don't see any obvious way to fix it.  Maybe we need a general
utility method to take a list of types and alignments and try to construct an
unpacked type if possible?

llvm-svn: 51785
2008-05-30 19:58:50 +00:00
Eli Friedman c18d9d53bb Allow the type of a global to be different from the type of its
associated declaration. This is a prerequisite to handling
general union initializations; for example, an array of unions involving 
pointers has to be turned into a struct because the elements can have 
incompatible types.

I refactored the code a bit to make it more readable; now, the logic for 
definitions is all in EmitGlobalVarInit.

The second parameter for GetAddrOfGlobalVar is now dead; I'll remove it 
separately.

By itself, this patch should not cause any visible changes.

llvm-svn: 51783
2008-05-30 19:50:47 +00:00
Eli Friedman f7fe41656c Fix constant vector init for initializer lists with an incomplete list
of elements.

llvm-svn: 51769
2008-05-30 12:02:14 +00:00
Eli Friedman 617ba48fd5 Always check that the definition of a function has the correct type.
This fixes a crash on the included testcase (found in NetHack).

llvm-svn: 51767
2008-05-30 11:13:18 +00:00
Sanjiv Gupta 18de624ca0 Emit parameter and local variable debug information with -g.
llvm-svn: 51765
2008-05-30 10:30:31 +00:00
Eli Friedman 711665c309 Add FIXME to CodeGen struct layout.
llvm-svn: 51764
2008-05-30 10:29:32 +00:00
Eli Friedman a2eaffc747 Some additions to constant initializer generation code. Adds
bit-field initialization; ugly code, X86-only, but it works, at least 
for basic stuff.  Separates/adds union initialization; currently disabled,
though, because the struct/array code needs modifications to support 
elements of the wrong type.

Fixes PR2381 and PR2309 with the bit-field initialization.  And NetHack 
compiles and appears to work with a few tweaks (to work around the lack 
of transparent_union support, and clang being a bit strict about
conflicting declarations).

llvm-svn: 51763
2008-05-30 10:24:46 +00:00
Steve Naroff 021ca18bb5 - Move ObjC Expresssion AST's from Expr.h => ExprObjC.h
- #include ExprObjC.h in many places

llvm-svn: 51703
2008-05-29 21:12:08 +00:00
Eli Friedman 43a6d51494 More complete/correct implementation of the comparison operators for
constant expressions.

llvm-svn: 51682
2008-05-29 15:17:49 +00:00
Eli Friedman 3c28524632 Fix an extremely subtle bug with pointer comparisons: they have to be
unsigned because it's possible (at least in theory) to have
have both positive and negative pointers pointing to the same object.

llvm-svn: 51681
2008-05-29 15:09:15 +00:00
Eli Friedman 133e8040ca Rearrange EmitLValueForField a bit to work properly for _Bool bitfields
in unions (we don't want to do the union-specific bitcast for 
bit-fields).

llvm-svn: 51678
2008-05-29 11:33:25 +00:00
Eli Friedman 045bf4ff82 Add codegen support for a few more kinds of initializer constant
expressions.

llvm-svn: 51677
2008-05-29 11:22:45 +00:00
Eli Friedman 174d9c26f1 Add CodeGen support for alignment on globals, both for unusual natural
alignment and alignment attributes.

llvm-svn: 51676
2008-05-29 11:10:27 +00:00
Eli Friedman b65ff27f53 A couple minor fixes to make debug info usable for arbitrary code: don't
emit incomplete types, because they crash llc, and always use the 
logical location as the current location so we don't crash doing invalid 
queries on CurLoc.

llvm-svn: 51675
2008-05-29 11:08:17 +00:00
Eli Friedman d72a9624a1 Tentative declarations are supposed to have common linkage, not weak.
This change makes clang generate the same thing as llvm-gcc.

llvm-svn: 51674
2008-05-29 11:03:17 +00:00
Eli Friedman 5a6d0ede91 Always use packed structs. This isn't really very nice, but there's
nothing fundamentally wrong with it. Emitting unpacked structs where 
possible is more work for almost no practical benefit.  We'll probably 
want to fix it at some point anyway, but it's low priority.

The issue with long double in particular is that LLVM thinks an X86 long
double is 10 bytes, while clang considers it for all purposes to be
either 12 or 16 bytes, depending on the platform, even in a packed
struct.

llvm-svn: 51673
2008-05-29 10:58:49 +00:00
Devang Patel 1c56ace4df Check first member alignment and uses packed struct if required.
llvm-svn: 51619
2008-05-27 22:45:40 +00:00
Eli Friedman 21911e89d5 A few more cases for aggregate values.
llvm-svn: 51596
2008-05-27 15:51:49 +00:00
Eli Friedman 5e2281ede8 Implementation of __builtin_ctlz.
llvm-svn: 51595
2008-05-27 15:32:46 +00:00
Eli Friedman 70f5a921b7 The cache is working again; re-enable it.
llvm-svn: 51587
2008-05-27 08:43:22 +00:00
Eli Friedman 078c963a31 Add a more reliable check for whether a static declaration has already
been used.  In preparation for the fix to PR2360, but also a minor bug 
in its own right.

llvm-svn: 51583
2008-05-27 04:58:01 +00:00
Eli Friedman 3e1d35b92d Stop leaking the target data. (Leak found with valgrind.)
llvm-svn: 51579
2008-05-27 04:22:24 +00:00
Eli Friedman bdef5df44e Generalize the float type generation code, and specifically fix the
codegen of X86 long double.

llvm-svn: 51578
2008-05-27 04:20:05 +00:00
Eli Friedman 42ed07e10e Rewrite struct/union layout. This is mostly cleanup; this might also fix
a few bugs, but I don't know of any in particular. This has good effects 
besides cleanup, though: it also should make it easier to implement the 
aligned and packed attributes, and also target-specific struct layouts, 
because the code won't have to be duplicated in codegen.

llvm-svn: 51576
2008-05-27 03:14:44 +00:00
Eli Friedman df649f3da5 Emit memmove, not memcpy, for structure copies; this is unfortunately
required for correctness in cases of copying a struct to itself or to 
an overlapping struct (itself for cases like *a = *a, and overlapping 
is possible with unions).

Hopefully, this won't end up being a perf issue; LLVM *should* be able 
to optimize memmove to memcpy in a lot of cases, and for small copies 
the generated code *should* be mostly comparable. (In reality, LLVM 
is currently horrible at optimizing memmove, but that's a bug, not a 
fundamental issue.)

gcc currently generates wrong code; that's 
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32667.

llvm-svn: 51566
2008-05-26 12:59:39 +00:00
Eli Friedman 292e98cc18 Fix for PR2001. I'm not really fond of it, but it is correct (unless
someone tries to make a bitfield volatile?).

Not sure how to write a test; any suggestions?

llvm-svn: 51558
2008-05-25 14:13:57 +00:00
Sanjiv Gupta 98070578bf Generate subprogram debug info with -g.
Also take care of freeing memory at the right places.

llvm-svn: 51553
2008-05-25 05:15:42 +00:00
Dan Gohman 5d30975e5e Change uses of llvm::Type::isFirstClassType to use the new
llvm::Type::isSingleValueType. Currently these two functions have
the same behavior, but soon isFirstClassType will return true for
struct and array types.

Clang may some day want to use of isFirstClassType for some of
these some day as an optimization, but it'll require some
consideration.

llvm-svn: 51446
2008-05-22 22:12:56 +00:00
Eli Friedman 1763075b9c Make debugging information usable. This is barebones, but it makes -g
actually work (instead of crashing llc), and there's enough info emitted 
to get line number information in gdb. This should hopefully be helpful 
for debugging non-working programs.

I got rid of the begin/endregion calls because the implementation wasn't 
working; someone who knows the debugging info a bit better might try to 
add it. I really have no clue how a compiler is supposed to emit them.

This commit shouldn't have any effect without -g.

llvm-svn: 51404
2008-05-22 01:40:10 +00:00
Eli Friedman c7460b1011 Minor refactoring: compute the return value separately from emitting the
ret.

llvm-svn: 51403
2008-05-22 01:22:33 +00:00
Eli Friedman e53d15386f Make CurFuncDecl more specific; I have some code I'm going to land in a
bit that wants to use CurFuncDecl as a FunctionDecl.

llvm-svn: 51402
2008-05-22 01:16:59 +00:00
Dan Gohman 75d69da5ab Move getAccessedFieldNo out of lib/AST/Expr.cpp into
lib/CodeGen/CGExpr.cpp and to change include/clang/AST/Attr.h to
use its own enum for visibility types instead of using
llvm::GlobalValue::VisibilityTypes. These changes eliminate
dependencies in the AST library on LLVM's VMCore library.

llvm-svn: 51398
2008-05-22 00:50:06 +00:00
Eli Friedman 9a5ffcb14e Remove an unnecessary/buggy if check. Ran into this with some other
changes in my tree, so I don't have a testcase which affects trunk.

llvm-svn: 51371
2008-05-21 13:24:44 +00:00
Eli Friedman 3ed1d5f496 Disable the function address cache; this doesn't have any ill effects,
as far as I can tell, and it fixes code like test/CodeGen/functions.c.

(Whatever performance effect it might have, crashing on a 
construct like this isn't really acceptable; I've run into this 
multiple times.)

llvm-svn: 51312
2008-05-20 09:21:07 +00:00
Eli Friedman 53e38bd6b6 Add codegen support for stack address intrinsics.
llvm-svn: 51309
2008-05-20 08:59:34 +00:00
Eli Friedman 4b0e2a30e0 Implement codegen for comma operator for structs.
llvm-svn: 51304
2008-05-20 07:56:31 +00:00
Eli Friedman 247ee87e26 Restructure constant structure init codegen so that it's possible to
implement bitfield codegen (although I don't envy the person who 
implements it).  This also prevents a crash on code like that from PR2309
(it's still broken, but it fails more gracefully).

llvm-svn: 51285
2008-05-20 00:11:07 +00:00
Eli Friedman 8d0b976e7c Remove code that shouldn't have any effect anymore.
llvm-svn: 51284
2008-05-20 00:04:21 +00:00
Eli Friedman a3a4068949 Implement CodeGen for __builtin_memcpy.
llvm-svn: 51282
2008-05-19 23:27:48 +00:00
Eli Friedman f23b6fa41f Fix the emission of expressions like char a[10] = "asdf"; previously,
they were causing bad code to be emitted.  There are two fixes here: one 
makes sure we emit a string that is long enough, and one makes sure we 
properly handle string initialization in init lists.

llvm-svn: 51259
2008-05-19 17:51:16 +00:00
Eli Friedman f2442dcc8d Fix support for _Bool bitfields. The issue is that the bitfield width
used for _Bool is not the same as the primitive width (which for _Bool 
is 1 bit).  The load and store changes add some casts to make the 
types consistent.  The EmitLValue changes make sure that the pointer is 
of an appropriate type for loading the bitfield.

This isn't perfect, but it's an improvement, and getting everything 
right depends on actually laying out structs in an ABI-compliant way.

llvm-svn: 51224
2008-05-17 20:03:47 +00:00
Eli Friedman d5a4838e3d Backout of codegen-based fix to PR2334; this has been fixed differently
at the Sema layer.

llvm-svn: 51203
2008-05-16 20:38:39 +00:00
Nate Begeman 9800faa578 Fix some prototypes, and implement some builtins until we have our header files completed.
llvm-svn: 51199
2008-05-16 18:59:42 +00:00
Eli Friedman cb92a3f03b Patch for PR2334, and a similar ObjC bug.
llvm-svn: 51193
2008-05-16 17:37:11 +00:00
Nate Begeman 6c59132241 Handle a couple missing builtins that will go away eventually when the various
intrin.h files are finished

llvm-svn: 51137
2008-05-15 07:38:03 +00:00
Eli Friedman a1b4ed8003 Implementation of __builtin_shufflevector, a portable builtin capable of
expressing the full flexibility of the LLVM shufflevector instruction.  
The expected immediate usage is in *mmintrin.h, so that they don't 
depend on the mess of gcc-inherited (and not completely implemented)
shuffle builtins.

llvm-svn: 51113
2008-05-14 19:38:39 +00:00
Eli Friedman 9fd8b68568 Add codegen support for block-level compound literals.
llvm-svn: 51081
2008-05-13 23:18:27 +00:00
Eli Friedman dd7406e65c Add support for init lists for complex variables.
llvm-svn: 51080
2008-05-13 23:11:35 +00:00
Nate Begeman d386215916 Remove AST dependency on VMCore by switching ExtVectorElementExpr off Constant.
llvm-svn: 51068
2008-05-13 21:03:02 +00:00
Eli Friedman 1a5754a26a Detabify.
llvm-svn: 51042
2008-05-13 14:40:48 +00:00
Eli Friedman 6a7087e455 Emit basic block for switch body; fixes PR2307.
llvm-svn: 50968
2008-05-12 16:08:04 +00:00
Eli Friedman 6e31321043 Don't try to take the address of a bitfield; fixes PR2310.
llvm-svn: 50966
2008-05-12 15:06:05 +00:00
Mon P Wang b84407d6ec Added support to generate some atomic operators (add, sub, and, or etc..)
llvm-svn: 50919
2008-05-09 22:40:52 +00:00
Nate Begeman f322eabbce Extend vector member references to include {.hi, .lo, .e, .o} which return a
vector of the same element type and half the width, with the high, low, even,
and odd elements respectively.

Allow member references to member references, so that .hi.hi gives you the high
quarter of a vector.  This is fairly convenient syntax for some insert/extract
operations.

Remove some unnecessary methods/types in the ExtVectorElementExpr class.

llvm-svn: 50892
2008-05-09 06:41:27 +00:00
Sanjiv Gupta 15cb669039 Added -g command line options to clang for generating source level debug information. This patch currently enables generation of line number debug information (stoppoints) and region begin/end debug information. The new files CGDebugInfo.h and CGDebugInfo.cpp implements the debug info manager class CGDebugInfo.
llvm-svn: 50848
2008-05-08 08:54:20 +00:00
Chris Lattner b781dc79aa Add basic support for the pic-* target triples and add support for
targets that do not support recursion (and thus codegen stack variables
as globals).

Patch contributed by Alireza Moshtaghi!

llvm-svn: 50844
2008-05-08 05:58:21 +00:00
Devang Patel ac25fc6f2e Begin handling union bitfields.
Note, this is just beginning.

llvm-svn: 50835
2008-05-07 22:28:29 +00:00
Devang Patel 243d5cd283 Fix PR2101 - Codegen crash during bitfield initialization.
llvm-svn: 50769
2008-05-06 19:57:59 +00:00
Chris Lattner dbcc2ca6b2 simplify some builder calls.
llvm-svn: 50694
2008-05-06 00:56:42 +00:00
Chris Lattner 222c6123df remove a stray printout
llvm-svn: 50691
2008-05-06 00:44:00 +00:00
Chris Lattner 1e151816b2 use simplified API for making fp constants.
llvm-svn: 50623
2008-05-04 18:23:51 +00:00
Chris Lattner 7b9293ba20 Simplify FunctionDecl::AddRedeclaration a bit by using std::swap.
Fix 'swapping' of attributes to not insert null values into the 
DeclAttrs map.

llvm-svn: 50612
2008-05-04 02:29:49 +00:00
Chris Lattner bc22b5b327 if a decl is both 'static' and weak or static and inline, its linkage
type should be internal, not weak/linkonce.

llvm-svn: 50611
2008-05-04 01:44:26 +00:00
Chris Lattner dd477ce32b Verify the whole module after codegen to catch silly IR bugs.
llvm-svn: 50485
2008-04-30 16:05:42 +00:00
Chris Lattner 73669defed remove the 'unused static' diagnostic from codegen. This should be
reimplemented in Sema someday.  This is tracked in PR2236.  Until then
disabling it removes some bogus diagnostics (see 2236).

llvm-svn: 50263
2008-04-25 15:47:13 +00:00
Nate Begeman 1c906206b8 Fix PR2236
Add test
Implement feedback from Chris re: PR1998

llvm-svn: 50001
2008-04-20 20:38:08 +00:00
Nate Begeman 8e8d498d34 Only generate code for static functions and global variables that are actually used.
Warn about unused static functions and variables.
This fixes PR1998.

llvm-svn: 49987
2008-04-20 06:29:50 +00:00
Chris Lattner ff2367c0c9 implement a fixme, add support for ++/-- on long double.
llvm-svn: 49980
2008-04-20 00:50:39 +00:00
Chris Lattner 1886e71e47 use simplified ConstantFP:get method.
llvm-svn: 49979
2008-04-20 00:45:53 +00:00
Nate Begeman faae081131 Generate code for annotation attributes.
llvm-svn: 49951
2008-04-19 04:17:09 +00:00
Nate Begeman 7fab578b0d Add support to module codegen for adding and emitting annotations
llvm-svn: 49944
2008-04-18 23:43:57 +00:00
Nate Begeman ce4d7fce6b OCUVector -> ExtVector, shorthand for extended vector, per feedback from Chris.
llvm-svn: 49942
2008-04-18 23:10:10 +00:00
Nate Begeman 26ed25253e Allow targets to override double size, alignment, and semantics, and tie this
information into the IR type codegen.  This allows a target to codegen the C
'double' type as either double or float.

llvm-svn: 49898
2008-04-18 05:41:31 +00:00
Steve Naroff 08899ff85d Remove FileVarDecl and BlockVarDecl. They are replaced by VarDecl::isBlockVarDecl() and VarDecl::isFileVarDecl().
This is a fairly mechanical/large change. As a result, I avoided making any changes/simplifications that weren't directly related. I did break two Analysis tests. I also have a couple FIXME's in UninitializedValues.cpp. Ted, can you take a look? If the bug isn't obvious, I am happy to dig in and fix it (since I broke it).

llvm-svn: 49748
2008-04-15 22:42:06 +00:00
Nate Begeman 91f40e3680 Enable clang to codegen emmintrin.h until we have our own emmintrin.h
llvm-svn: 49633
2008-04-14 04:49:57 +00:00
Chris Lattner 61705c1a6e update to follow mainline llvm API change.
llvm-svn: 49607
2008-04-13 07:32:11 +00:00
Ted Kremenek 43e0633048 The dtor CGObjCRuntime::~CGObjCRuntime() was implemented twice, once
in CGObjCGNU.cpp and once in CGObjCEtoile.cpp.  Moved its definition to
CGObjC.cpp.  This resolves a build error in Xcode, and also cleans things up.

llvm-svn: 49429
2008-04-09 15:51:31 +00:00
Chris Lattner aa9c7aed0f Add support for C++ default arguments, and rework Parse-Sema
interaction for function parameters, fixing PR2046.

Patch by Doug Gregor!

llvm-svn: 49369
2008-04-08 04:40:51 +00:00
Chris Lattner fc4379f916 remove the Decl::getCanonicalType() method.
llvm-svn: 49295
2008-04-06 23:10:54 +00:00
Gabor Greif d36afd7d65 tracking API changes arising from r49277
llvm-svn: 49279
2008-04-06 20:42:52 +00:00
Chris Lattner 579a05d767 Fix PR2049, updating Eli's patch that fixes to mainline. This produces incorrect
code, but the codegen doesn't crash.  I'll file a bugzilla for the AST being wrong.

llvm-svn: 49226
2008-04-04 18:42:16 +00:00
Chris Lattner f3bc75afcf Since isComplexType() no longer returns true for _Complex integers, the code
generator needs to call isAnyComplexType().  This fixes PR1960.

llvm-svn: 49220
2008-04-04 16:54:41 +00:00
Chris Lattner 019193acea updates from David.
llvm-svn: 49217
2008-04-04 15:59:59 +00:00
Chris Lattner 15c6b7c948 various bug fixes, add to xcode project.
llvm-svn: 49216
2008-04-04 15:47:22 +00:00
Chris Lattner 6175bbbee2 initial checkin of Étoilé runtime support, patch by David Chisnall
llvm-svn: 49215
2008-04-04 15:45:52 +00:00
Torok Edwin 602647aa70 fix build error: add missing braces.
llvm-svn: 49209
2008-04-04 06:16:25 +00:00
Chris Lattner 5506f8cf4c Codegen assignment to self correctly, patch by David Chisnall!
llvm-svn: 49201
2008-04-04 04:07:35 +00:00
Nate Begeman 11b6bb408e Teach clang how to codegen punpcklbw and punpcklwd
llvm-svn: 49186
2008-04-03 19:58:06 +00:00
Chris Lattner e799eb5daa Fix PR2081 (problems codegen'ing some recursive structures) patch
by Lauro Venancio!

llvm-svn: 49144
2008-04-03 05:50:42 +00:00
Chris Lattner 182f660d8d simplify some code by using PointerLikeType.
llvm-svn: 49101
2008-04-02 17:45:06 +00:00
Chris Lattner db5f1fa3d8 add a common base class "PointerLikeType" for PointerType and ReferenceType,
allowing them to be treated the same in some contexts.  A suggestion for a
better name is welcome :)

llvm-svn: 49100
2008-04-02 17:35:06 +00:00
Chris Lattner c00c35a857 some cleanups on top of David's patch. There are still two
remaining open issues I've communicated to him:

1) self can be assigned to, and his patch didn't handle it correctly.
2) CollectObjCIvarTypes is N^2 (because each subclass reprocesses
   all parent class ivars) and flattens classes.  If A derives from B, 
   and both have an int, I'd expect to get { {i32}, i32}, not { i32, i32}.

David, please review.

llvm-svn: 48970
2008-03-30 23:25:33 +00:00
Chris Lattner 4bd5596d08 Add initial support for objc codegen for methods, ivars, and the
etoile runtime, patch by David Chisnall!

llvm-svn: 48969
2008-03-30 23:03:07 +00:00
Chris Lattner 554221e9fb simplify code by using the variadic StructType::get method.
llvm-svn: 48535
2008-03-19 05:24:56 +00:00
Chris Lattner 3e593cdb98 simplify the clang codegen by using the new Builder.CreateStructGEP method.
llvm-svn: 48534
2008-03-19 05:19:41 +00:00
Chris Lattner 7a51313d8a Make a major restructuring of the clang tree: introduce a top-level
lib dir and move all the libraries into it.  This follows the main
llvm tree, and allows the libraries to be built in parallel.  The
top level now enforces that all the libs are built before Driver,
but we don't care what order the libs are built in.  This speeds
up parallel builds, particularly incremental ones.

llvm-svn: 48402
2008-03-15 23:59:48 +00:00