This currently breaks test/SemaObjC/id-isa-ref.m and issues some spurious warnings when you attempt to assign a struct objc_class* value to a Class variable.  The test case probably should fail as it's written, because without the definition of Class the compiler should not assume struct objc_class* is a valid receiver type, but it's left broken because it would be nice if we could get that passing too for the special case of isa.

Approved by snaroff.

llvm-svn: 79248

llvm-svn: 79200

is found.  Instead of complaining about a generic "unused expr",
emit:
t.m:7:3: warning: property access result unused - getters should not have side effects

While objc property getters *could* have side effects, according to
the language best practices, they *shouldn't*.  Hopefully the
diagnostic now gets this across.

llvm-svn: 79192

AddInitializerToDecl can't take a FullExprArg. Make it take an ExprArg, and create the CXXExprWithTemporaries before setting the initializer on the VarDecl.

llvm-svn: 79176

llvm-svn: 79173

llvm-svn: 79172

llvm-svn: 79171

llvm-svn: 79168

Move builtin call checking out into a separate function, make CheckFunctionCall and CheckBlockCall return bool instead. No intended functionality change.

llvm-svn: 79157

llvm-svn: 79149

initializes all the members; this fixes a valgrind error in the 
regression tests.

llvm-svn: 79148

llvm-svn: 79143

simplifies the AST, and can matter in some rare cases involving 
casts to vector types. Patch by Enea Zaffanella.

llvm-svn: 79126

llvm-svn: 79070

declarations (and not function pointers). This is consistent with GCC. Accepting
this attribute on function pointers means that the attribute should be treated
as a type qualifier, which apparently is not what GCC does. We obviously can
change this later should we desire to enhance the 'malloc' attribute in this
way.

llvm-svn: 79060

declarations of function pointers.

llvm-svn: 79053

a block pointer too.

llvm-svn: 79050

'cast'.

llvm-svn: 79041

attaching to Objective-C methods (which mirrors GCC's behavior) and to allow the
return type of the function to be an Objective-C pointer or Block pointer (which
GCC also accepts).

Along the way, add 'const' to some of the pointer arguments of various utility
functions...

llvm-svn: 79040

llvm-svn: 79013

implement NSCopying protocol in GC mode.

llvm-svn: 79008

still works.

llvm-svn: 78979

function in the order according to c++03. ir-gen 
for copy assignment in the trivial case and the first
test case.

llvm-svn: 78938

llvm-svn: 78876

llvm-svn: 78862

llvm-svn: 78861

WIP.

llvm-svn: 78841

llvm-svn: 78826

llvm-svn: 78780

llvm-svn: 78735

may not implement NSCopying protocol in -fobjc-gc[-only] mode.

llvm-svn: 78726

the AST, and create such declarations.

llvm-svn: 78719

copying in copy constructors and used in
default constructor's initializer list.

llvm-svn: 78700

llvm-svn: 78674

FriendFunctionDecl, and create instances as appropriate.

The design of FriendFunctionDecl is still somewhat up in the air;  you can
befriend arbitrary types of functions --- methods, constructors, etc. ---
and it's not clear that this representation captures that very well.
We'll have a better picture when we start consuming this data in access
control.

llvm-svn: 78653

generic tree transformation (also used for recanonicalization) and a
small amount of template-instantiation-specific logic.

llvm-svn: 78645

llvm-svn: 78644

Fixes PR4704 problems

Addresses Eli's patch feedback re: ugly cast code

Updates all postfix operators to remove ParenListExprs.  While this is awful,
no better solution (say, in the parser) is obvious to me.  Better solutions
welcome.

llvm-svn: 78621

Add a CastInfo struct that will be used for cast information when constructing cast expressions. Right now it only stores the cast kind, but in the future it might store conversion functions and constructors.

llvm-svn: 78599

1. Passing something that isn't a string used to cause:
   "argument to annotate attribute was not a string literal"
  make it say "section attribute" instead.

2. Fix the location of the above message to point to the
   bad argument instead of the section token.

3. Implement rdar://4341926, by diagnosing invalid section
   specifiers in the frontend rather than letting them slip all
   the way to the assembler (a QoI win).

An example of #3 is that we used to produce something like this:

/var/folders/n7/n7Yno9ihEm894640nJdSQU+++TI/-Tmp-//ccFPFGtT.s:2:Expected comma after segment-name
/var/folders/n7/n7Yno9ihEm894640nJdSQU+++TI/-Tmp-//ccFPFGtT.s:2:Rest of line ignored. 1st junk character valued 46 (.).

Daniel improved clang to use llvm_report_error, so now we got:

$ clang t.c -c
fatal error: error in backend: Global variable 'x' has an invalid section specifier 'sadf': mach-o section specifier
      requires a segment and section separated by a comma.

with no loc info.  Now we get:

$ clang t.c -fsyntax-only
t.c:4:30: error: argument to 'section' attribute is not valid for this target: mach-o section specifier requires a segment
      and section separated by a comma
int x __attribute__((section("sadf")));
                             ^

which is nice :)

llvm-svn: 78586