we default to "i386-apple-darwin". This is an interim solution.
Removed processing of "linux" triples from Targets.cpp, since we don't have
any sensical Linux target support (yet).
Cleaned up error processing of targets; added better diagnostics.
llvm-svn: 44560
replaces the functionality previously provided by just "-arch" (which is still
supported but has different semantics).
The new behavior is as follows:
(1) If the user does not specify -triple:
(a) If no -arch options are specified, the target triple used is the host
triple (in llvm/Config/config.h).
(b) If one or more -arch's are specified (and no -triple), then there is
one triple for each -arch, where the specified arch is substituted
for the arch in the host triple. Example:
host triple = i686-apple-darwin9
command: clang -arch ppc -arch ppc64 ...
triples used: ppc-apple-darwin9 ppc64-apple-darwin9
(2) The user does specify a -triple (only one allowed):
(a) If no -arch options are specified, the triple specified by -triple
is used. E.g clang -triple i686-apple-darwin9
(b) If one or more -arch options are specified, then the triple specified
by -triple is used as the primary target, and the arch's specified
by -arch are used to create secondary targets. For example:
clang -triple i686-apple-darwin9 -arch ppc -arch ppc64
has the following targets:
i686-apple-darwin9 (primary target)
ppc-apple-darwin9
ppc64-apple-darwin9
Other changes related to the changes to the driver:
- TargetInfoImpl now includes the triple string.
- TargetInfo::getTargetTriple returns the triple for its primary target.
- test case test/Parser/portability.c has been updated because "-arch linux" is
no longer valid ("linux" is an OS, not an arch); instead we use a bogus
architecture "bogusW16W16" where WCharWidth=16 and WCharAlign=16.
llvm-svn: 44551
provide handy accessors to the subexpressions of ConditionalOperator that
automatically take into account the GCC extension where the "LHS" expression is
omitted: e.g x ?: y;. When the LHS expression is available, getTrueExpr() is the
same as getLHS(); when LHS is NULL, getTrueExpr() returns the condition expression.
llvm-svn: 44536
rewriter emit this error if it fails to rewrite an @encode:
t.m:17:9: error: rewriter could not replace sub-expression due to macros
c = ENC(char *)[2] + 4;
^~~~~~~~~~~
... where ENC is: #define ENC @encode
llvm-svn: 44498
We now use the CodeGenModule logic for generating the constant
initialiser expression, so happily further initialiser fixes should
automatically work for statics as well.
llvm-svn: 44495
name if a tag type name is not available for the type. This matches how llvm-gcc
chooses CodeGen type names.
This means that "typedef struct {...} foo" now results in a CodeGen name of
"struct.foo" rather than "struct."
llvm-svn: 44489
pointer of MultiKeywordSelectors.
Added optimization to the serialization of SelectorTable where we only serialize
out MultiKeywordSelectors that are ever referenced by an object other than the
SelectorTable.
llvm-svn: 44483
that are builtin and those that are aren't. This is a bunch
of API refactoring that will make this possible, but there is
no functionality change yet.
llvm-svn: 44473
After a discussion with Ted, we both came to the conclusion that adding a "HandleTopLevelDeclaration" hook to ASConsumer is far more elegant. The default implementation of HandleTopLevelDeclaration will be responsible for iterating over the ScopedDecl (which has a chain of the decls:-).
TODO: Once Ted adds HandleTopLevelDeclaration, make sure TagDecls are chainged appropriately...
llvm-svn: 44445
be the new hook that ASTStreamer calls to feed top-level Decls to
ASTConsumers.
The difference between "HandleTopLevelDeclaration" and "HandleTopLevelDecl" is
that "HandleTopLevelDecl" is currently called by ASTStreamer for every
top-level declaration, including those that appear within a Decl chain. Using
the new interface, ASTStreamer would only call HandleTopLevelDeclaration for
Decls that appear that the beginning of a Decl chain (i.e., a group of related
decls).
To preserve the behavior that all subclasses of ASTConsumer currently expect,
the default implementation of HandleTopLevelDeclaration simply calls
HandleTopLevelDecl, and for decl chains it calls HandleTopLevelDecl for each
Decl* in a chain of Decls.
The advantage of this interface is that some subclasses of ASTConsumer only
really want the Decl chain, and not each individual Decl passed to them. This
extra level of indirection allows subclasses to override the default behavior
if they so desire.
llvm-svn: 44444
both before and after serialization/deserialization. If the CFGs between the pre-
and post- serialized/deserialized ASTs differ, the serialization has failed.
llvm-svn: 44429
floating-point literals that are represented exactly by the APFloat in
FloatingLiteral. For such literals, we do not emit a warning since such checks are
often performed in real code to see if a variable has changed from its original
value. This heuristic clearly can lead to false negatives, but the hope is it will
significantly reduce false positives to help make the compiler flag more useful.
llvm-svn: 44424
the APFloat representing the parsed literal can represent the literal value
exactly. This is useful when performing various semantic checks on the code,
and issuing appropriate warnings to users.
llvm-svn: 44423
converts a parsed literal into an APFloat. We are still performing a copy of the
string, which hopefully will be removed eventually for performance reasons. This
version now is at least safe.
Changed rounding in APFloat construction in NumericLiteralParser from rmTowardsZero
to rmNearestTiesToEven.
llvm-svn: 44422
a bug where the statement pretty-printer used iostreams but the AST printer
did not. This was an issue when dumping ASTs to something other than stderr.
Updated SerializationTest to use the new iostreams interface for the AST printer.
llvm-svn: 44417
(1) Parsed ASTs are pretty-printed to a text file.
(2) The ASTs are serialized to disk.
(3) The ASTs are deserialized from disk.
(4) The deserialized ASTs are pretty-printed to a text file.
(5) The two pretty-printed files are compared. If they are different, the test
fails.
llvm-svn: 44411
subdirectories mirroring where the test case file is located
For example, for the test case "Sema/stmt_exprs.c", instead of the files
"Output/stmt_exprs.c.out" and "Output/stmt_exprs.c.out.script" being created, the
files "Output/Sema/stmt_exprs.c.out" and "Output/Sema/stmt_exprs.c.out.script" are
created. This prevents any collisions from different test directories that have the
same file name for a test case, and also makes it clear where the test case was
drawn from.
llvm-svn: 44410
which is now used (or subclasssed) by the ASTConsumers. This new class
stores a FILE* that is used for writing, instead of just hardwiring output
to stderr (it defaults to stderr if no FILE* is provided).
Modified CreateASTPrinter() to accept a FILE* for printing.
llvm-svn: 44377
diagnostic without a location. This produces:
simpleTest.c:2:18: error: initializer element is not constant
int *myPointer = &(myArray[2]);
^~~~~~~~~~~~~
instead of:
error: initializer element is not constant
llvm-svn: 44375
NumericLiteralParser::GetFloatValue(). Upon method return, this flag has the value
true if the returned APFloat can exactly represent the number in the parsed text,
and false otherwise.
Modified the implementation of GetFloatValue() to parse literals using APFloat's
convertFromString method (which allows us to set the value of isExact).
llvm-svn: 44339
extension "?:" for the ternary operator, e.g.: x ?: y; This expression is
represented in the clang ASTs as a ConditionalOperator whose LHS expression is
NULL. Now we handle this special case, causing the block containing the
condition to be a predecessor to the block that "merges" the values of the
ternary operator.
Thanks to Nuno Lopes for identifying and diagnosing this bug!
llvm-svn: 44327
missing middle expression, and fix a codegen bug where
we didn't correctly promote the condition to the right
result type. This fixes PR1824.
llvm-svn: 44322
Moved utility functions IgnoreParen and friends to be static inline functions
defined in SemaUtil.h.
Added SemaUtil.h to Xcode project.
llvm-svn: 44312
We accidentally were throttling the propagation of uninitialized state across
assignments (e.g. x = y). Thanks to Anders Carlsson for spotting this problem.
Added test cases to test suite to provide regression testing for the
uninitialized values analysis.
llvm-svn: 44306
"block-expressions" when visiting arbitrary expressions (via calls to
"Visit()"). This results in a refactoring where a dataflow analysis no
longer needs to always special case when handling block-expressions versus
non-block expressions.
Updated LiveVariables and UninitializedValues to conform to the slightly
altered interface of these visitor classes.
Thanks to Nuno Lopes for providing a test case that illustrated some
fundamental problems in the current design of the CFGXXXStmtVisitor classes
and how they were used.
llvm-svn: 44246
for correct propagation/update of liveness information within subexpressions
of Block-Level expressions. Test case provided by Nuno Lopes.
llvm-svn: 44225
the StmtIterator referring to the initializers of a chain of Decls was equal
to the "end" iterator. The particular bug manifested when an iterator was
created on a chain of decls with no initializers.
Thanks to Nuno Lopes for reporting this bug and providing a patch.
llvm-svn: 44220
Some FunctionDecls do not appear at the top-level or are owned by a DeclStmt.
In calls to implicitly defined functions, a FunctionDecl is created, but only
the DeclRefExprs reference them. Since an implicitly defined function may be
called multiple times, there is no clear ownership model for such objects.
Temporary solution: when serializing out DeclRefExprs, emit an ownership bit
for the Decl. This bit is determined by querying the serializer to see if the
Decl has already been serialized. If it hasn't, emit the Decl as an owned
pointer.
I repeat: this is a hack. This should be fixed.
llvm-svn: 44176
backpatching. This original was available, but then we removed it. It is back
again to help with deserialization of FieldDecls. Because FieldDecls are
currently owned by RecordDecls, which are owned by a TagType, the type of the
FieldDecl may not be deserialized prior to deserializing the FieldDecl. Thus
backpatching solves the problem of constructing a FieldDecl that references a
type that has not yet been deserialized.
Simplified serialization of TagType to not require passing in the
SerializedPtrID. Registration of the materialized type object is done after
the CreateImpl method returns (as with other types).
llvm-svn: 44143
block separate from ASTContext. This block is serialized out AFTER writing out
ASTContext, but deserialized BEFORE reading in ASTContext. This permits the
optimization of the serialization of the IdentifierTable where we only write
out identifiers that are used.
This was needed because TagDecls are owned by Types, and TagDecls contain
identifiers. Thus types need to be written out first to register with the
serializer any identifiers they refer to (and hence need to be serialized out
with IdentifierTable).
llvm-svn: 44125
have an owning pointer to the referred TagDecl. This should hopefully fix a
bug where TagDecls (including decls from structs, etc.) were not serialized.
llvm-svn: 44106
to declare a function with a typedef:
typedef int unary_int_func(int arg);
unary_int_func add_one;
This patch contributed by Seo Sanghyeon!
llvm-svn: 44100
in TagDecl*. This allows the deserializer to use ASTContext to create the
TagTypes. Deserialize TagTypes then rely on pointer-backpatching to resolve
the decls.
This may not be the interface that we want, but as the implementation of
TagTypes will potentially change significantly in the future, I'm leaving this
for now. An appropriate FIXME is in place.
llvm-svn: 44089
Removed tons of dead code in ASTContext concerning how types use to be
serialized.
Removed serialization methods from QualType that are no longer used.
llvm-svn: 44070
This now exposes the following bug...
******************** TEST 'Sema/message.m' FAILED! ********************
Command:
clang -fsyntax-only -verify Sema/message.m
Output:
Errors seen but not expected:
Line 9: invalid receiver to message expression
******************** TEST 'Sema/message.m' FAILED! ********************
As far as I can tell, all messages to method agruments fail.
The method arguments are built by Sema::ObjcActOnStartOfMethodDef().
llvm-svn: 44022
- Add Sema::CurMethodDecl, in preparation for adding ObjcIvarRefExpr.
- Add ObjcInterfaceDecl::lookupInstanceVariable(), in prep for adding ivars.
- A couple renames in ObjcInterfaceDecl, while I was in the vicinity:-)
llvm-svn: 44015
- It is not an "action" - it is never called by the parser.
- It was only used by one method, Sema::ObjcActOnStartOfMethodDef().
As a result, the logic it embodied is now directly implemented in Sema::ObjcActOnStartOfMethodDef().
llvm-svn: 44008
While the diff is large, the idea is very simple. When we parse method definitions (in an @implementation), we need to add them incrementally (rather than wait until the @end).
Other details...
- Renamed Sema::ActOnAddMethodsToObjcDecl() to Sema::ActOnAtEnd(). The methods are now optional arguments.
- Removed Parser::AllImplMethods (a nice cleanup).
- Added location info to ObjcImplementationDecl (since we will need it very soon:-)
- Modified message.m test to no longer allow the bogus diagnostic.
llvm-svn: 43995
This commit is only 95% of the bug fix. The last piece to this puzzle is to add the method decls to the implementation incrementally (as we encounter them). At the moment, the methods aren't added until we see an @end (which is too late).
I will complete this later...
llvm-svn: 43989
that are referenced in the ASTs. This assumes that we serialize out the
decls/stmts first, and use the pointer-tracking logic in the Serializer to
determine if an IdentifierInfo (or its string key) is ever referenced.
This is a significant space optimization for serialized ASTs.
Consider the following program:
void foo(int x,int y) {
return x > y ? x : y+1;
}
Here are the sizes of the files for the serialized ASTs:
Full IdentifierTable: 23676 bytes
Only-referenced Identifiers: 304 bytes.
For this simple program, this is a 77% reduction in the file size of the
serialized ASTs.
llvm-svn: 43975
(1) serialize out top-level decls BEFORE serializing out translation unit
structures like ASTContext.
(2) deserialize out translation unit structures like ASTContext before
top-level decls by first skipping the decls in the bitstream, deserializing
ASTContext and friends, and then jumping back to the bitstream block with the
decls and then deserializing them.
Change (1) allows us to utilize the pointer-tracking system in the Serializer
to only serialize out metadata that is actually referenced by the ASTS.
Change (2) allows us to deserialize the metadata first as before, which
signficantly reduces the amount of pointer backpatching the deserializer
would have to do if the decls were deserialized first.
llvm-svn: 43974
and Materialize/Read instead of using specializations of SerializeTrait<>. The
resulting code is much cleaner. We are also setting the stage so that only the
parts of the IdentifierTable that are ever referenced within the ASTs are
serialized, and not the whole table.
llvm-svn: 43904
Implemented serialization of ConditionalOperator.
Implemented serialization of StmtExpr.
Fixed bug in serialization of IndirectGoto (did not properly serialize subexpression).
llvm-svn: 43891
backing a rewrite buffer than using an std::vector<char>. This class was hacked
together very quickly and needs to be cleaned up, but it seems to work. It speeds
up rewriting a a 7M file from 6.43s to 0.24s on my machine. The impl could also
be made to be a lot more algorithmically sound.
This produces identical output to using vector on this testcase, if it causes a
problems or bugs are encountered, it can be disabled by changing the
RewriteBuffer::Buffer typedef back.
llvm-svn: 43884
Disable rewrite-tabs. This speeds up processing of the commentified huge
crazy testcase steve gave me from 20s to 6.6s in a release build.
llvm-svn: 43880
first thing in the file.
The trick is that text replacement should go after the insert point,
not before it, because it will be replacing text after the point, not before
it.
llvm-svn: 43879
Ted Kremenek