Chris Lattner [Thu, 26 Jun 2008 04:19:03 +0000 (04:19 +0000)]
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.
Note that Parser::ParseCXXMemberSpecification is temporarily disabled until the Sema support is in place.
Once ParseCXXMemberSpecification is enabled, the Parser/cxx-class.cpp test will pass.
Chris Lattner [Tue, 24 Jun 2008 17:04:18 +0000 (17:04 +0000)]
"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)."
Chris Lattner [Tue, 24 Jun 2008 17:01:28 +0000 (17:01 +0000)]
"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)."
Ted Kremenek [Tue, 24 Jun 2008 15:50:53 +0000 (15:50 +0000)]
ObjCMessageExpr objects that represent messages to class methods now can contain the ObjCInterfaceDecl* of the target class if it was available when the ObjCMessageExpr object was constructed. The original interfaces of the class has been preserved (requiring no functionality changes from clients), but now a "getClasSInfo" method returns both the ObjCInterfaceDecl* and IdentifierInfo* of the target class.
Ted Kremenek [Tue, 24 Jun 2008 03:33:47 +0000 (03:33 +0000)]
Patch by Anders Bergh:
'There's not much to say about this patch, it just adds the Arch Linux
gcc 4.3.1 header paths for i686 and amd64. The patch was generated
using "svn diff" with clang at revision 52660. The paths aren't
distribution-specific, so they should work for all Linux distributions
using the default(?) names like "i686-pc-linux-gnu".'
Ted Kremenek [Mon, 23 Jun 2008 23:30:29 +0000 (23:30 +0000)]
Added ObjCSummaryCache, a new summary cache object to cache summaries for Objective-C methods. Instead of mapping from Selectors -> Summaries, we will now map from (ObjCInterfaceDecl*,Selectors) -> Summaries. This will allow more nuanced summary generation. This patch just swaps in the new data structure; the rest of the code works as before by allowing the ObjCInterfaceDecl* to be null.
Ted Kremenek [Mon, 23 Jun 2008 18:02:52 +0000 (18:02 +0000)]
The CF retain/release checker now assumes that allocations do not fail. Eventually we will add a flag to the driver to enable allocation failures (documented as a FIXME).
Chris Lattner [Sat, 21 Jun 2008 21:44:18 +0000 (21:44 +0000)]
"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."
Chris Lattner [Sat, 21 Jun 2008 18:04:54 +0000 (18:04 +0000)]
Switch 'super' from being a weird cast thing to being a predefined expr node.
Patch by David Chisnall with objc rewriter and stmtdumper updates from me.
Chris Lattner [Sat, 21 Jun 2008 17:46:11 +0000 (17:46 +0000)]
"This maps the -pthread option to -lpthread - this isn't 100% correct,
since handling this correctly is quite complex, and on some platforms
requires additional -D options and on some implies linking against a
different libc, but this works better than just ignoring the option.
The other change passes the -x option across to clang, which allows
compiling .c files as Objective-C and so on. For some reason a lot of
configure scripts seem to be under the misguided impression that this
is a sensible thing to do."
Ted Kremenek [Fri, 20 Jun 2008 21:45:25 +0000 (21:45 +0000)]
Modified the dead stores checker to...
1) Check if a dead store appears as a subexpression. For such cases, we emit
a verbose diagnostic so that users aren't confused. This addresses:
<rdar://problem/5968508> checker gives misleading report for dead store in loop
2) Don't emit a dead store warning when assigning a null value to a pointer.
This is a common form of defensive programming. We may wish to make
this an option to the the checker one day.
This addresses the feature request in the following email:
Ted Kremenek [Fri, 20 Jun 2008 21:39:47 +0000 (21:39 +0000)]
Added "Decl::getCodyBody()", a virtual method that returns the root AST node (Stmt*) that the Decl wraps (if any). Currently this only returns a non-null value for FunctionDecl and ObjCMethodDecl.
Ted Kremenek [Thu, 19 Jun 2008 17:55:38 +0000 (17:55 +0000)]
Introduce initial transfer function support for __imag__ and __real__. We don't
have complex RValues yet, so this logic is only fully implemented when __imag__
and __real__ are used on non-complex types.
Ted Kremenek [Wed, 18 Jun 2008 05:34:07 +0000 (05:34 +0000)]
Added a new ProgramPoint: PostPurgeDeadSymbols. This new program point distinguishes between the cases when we just evaluated the transfer function of a Stmt* (PostStmt) or performed a load (PostLoad). This solves a caching bug observed in a recent bug report.
Ted Kremenek [Tue, 17 Jun 2008 02:43:46 +0000 (02:43 +0000)]
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.
Ted Kremenek [Mon, 16 Jun 2008 18:34:22 +0000 (18:34 +0000)]
Moved CFDate.m from test/Analysis-Apple to test/Analysis, and added the necessary declarations from Foundation.h to CFDate.m so that the test case can be exercised on all platforms.
Eli Friedman [Fri, 13 Jun 2008 23:01:12 +0000 (23:01 +0000)]
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.
projects / clang / log