llvm-svn: 136254

Make CodeGen for array delete involving incomplete class work without crashing.  Should fix regression on g++.dg/init/delete1.C.

llvm-svn: 136241

llvm-svn: 136239

llvm-svn: 136238

llvm-svn: 136237

llvm-svn: 136236

llvm-svn: 136235

llvm-svn: 136224

llvm-svn: 136210

llvm-svn: 136209

- Replace calling -zone with 'nil'. -zone is obsolete in ARC.
- Allow removing retain/release on a static global var.
- Fix assertion hit when scanning for name references outside a NSAutoreleasePool scope.
- Automatically add bridged casts for results of objc method calls and when calling CFRetain, for example:

NSString *s;
CFStringRef ref = [s string];   -> CFStringRef ref = (__bridge CFStringRef)([s string]);
ref = s.string;                 -> ref = (__bridge CFStringRef)(s.string);
ref = [NSString new];           -> ref = (__bridge_retained CFStringRef)([NSString new]);
ref = [s newString];            -> ref = (__bridge_retained CFStringRef)([s newString]);
ref = [[NSString alloc] init];  -> ref = (__bridge_retained CFStringRef)([[NSString alloc] init]);
ref = [[s string] retain];      -> ref = (__bridge_retained CFStringRef)([s string]);
ref = CFRetain(s);              -> ref = (__bridge_retained CFTypeRef)(s);
ref = [s retain];               -> ref = (__bridge_retained CFStringRef)(s);

- Emit migrator error when trying to cast to CF type the result of autorelease/release:
  for

CFStringRef f3() {
  return (CFStringRef)[[[NSString alloc] init] autorelease];
}

emits:

t.m:12:10: error: [rewriter] it is not safe to cast to 'CFStringRef' the result of 'autorelease' message; a __bridge cast may result in a pointer to a destroyed object and a __bridge_retained may leak the object
  return (CFStringRef)[[[NSString alloc] init] autorelease];
         ^            ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
t.m:12:3: note: [rewriter] remove the cast and change return type of function to 'NSString *' to have the object automatically autoreleased
  return (CFStringRef)[[[NSString alloc] init] autorelease];
  ^

- Before changing attributes to weak/unsafe_unretained, check if the backing ivar
  is set to a +1 object, in which case use 'strong' instead.

llvm-svn: 136208

for-in statements;  specifically, make sure to close over any
temporaries or cleanups it might require.  In ARC, this has
implications for the lifetime of the collection, so emit it
with a retain and release it upon exit from the loop.

rdar://problem/9817306

llvm-svn: 136204

In Microsoft mode, if we are within a templated function and we can't resolve Identifier during BuildCXXNestedNameSpecifier, then extend the SS with Identifier. This will have  the effect of resolving Identifier during template instantiation.  The goal is to be able to resolve a function call whose nested-name-specifier is located inside a dependent base class.

class C {
public:
    static void foo2() {  }
};

template <class T> class A {
public:
   typedef C D;
};

template <class T> class B : public A<T> {
public:
  void foo() { D::foo2(); }
};

Note that this won't work if the NestedNameSpecifier refers to a type.
This fixes 1 error when parsing the MSVC 2010 standard headers file with clang.

llvm-svn: 136203

This required converting the StringMaps to use a BumpPtrAllocator.  I measured the
compile time and saw no observable regression.

llvm-svn: 136190

llvm-svn: 136189

llvm-svn: 136183

[dcl.init.list] as is possible without generalized initializer lists or full
constant expression support, and adds a c++0x-compat warning in C++98 mode.

The FixIt currently uses a typedef's basename without qualification, which is
likely to be incorrect on some code.  If it's incorrect on too much code, we
should write a function to get the string that refers to a type from a
particular context.

The warning is currently off by default. I'll fix LLVM and clang before turning
it on.

llvm-svn: 136181

destructors of abstract classes. It's undefined behavior to actually
call the destructor (e.g., via delete), but the presence of code that
calls this destructor doesn't make the program
ill-formed. Fixes <rdar://problem/9819242>.

llvm-svn: 136180

I'm not completely sure the standard allows us to reject this, but if it doesn't, it should. :)

llvm-svn: 136172

1. Attempting to delete an expression of incomplete class type should be an error, not a warning.

2. If someone tries to delete a pointer to an incomplete class type, make sure we actually emit
the delete expression after we warn.

llvm-svn: 136161

@interface Foo : NSObject
@property (readonly) id myProp;
@end

@implementation Foo
@synthesize myProp;
@end

t.m:9:13: error: ARC forbids synthesizing a property of an Objective-C object with unspecified storage attribute
@synthesize myProp;
            ^

which is fine, we want the ownership of the synthesized ivar to be explicit. But we should _not_ emit an error
for the following cases, because we can get the ownership either from the declared ivar or from the property type:

@interface Foo : NSObject {
	__weak id _myProp1;
	id myProp2;
}
@property (readonly) id myProp1;
@property (readonly) id myProp2;
@property (readonly) __strong id myProp3;
@end

@implementation Foo
@synthesize myProp1 = _myProp1;
@synthesize myProp2;
@synthesize myProp3;
@end

rdar://9844006.

llvm-svn: 136155

Report more memory using in Preprocessor::getTotalMemory() and PreprocessingRecord::getTotalMemory().

Most of the memory was already reported; but now we report more memory from side data structures.

Fixes <rdar://problem/9379717>.

llvm-svn: 136150

llvm-svn: 136113

This patch extends the previous patch by starting to incorporate more functionality, like lookup-by-name and exporting lookup tables, into the module manager.  Methods now have documentation.  A few more functions have been switched over to the new iterator style and away from manual/explicit iteration.  Ultimately we want to move away from name lookup here, as symlinks make filenames not a safe unique value, but we use it here as a stopgap before better measures are in place (namely instead using FileEntry* from a global FileManager).

llvm-svn: 136107

in few more places and in each instance, fix up
the type to the expected type. // rdar://9603056

llvm-svn: 136103

llvm-svn: 136092

provides the partial Objective-C selector used in a code
completion. From Connor Wakamo!

llvm-svn: 136084

that allocates an array of objects with a non-trivial destructor, be
sure to mark the destructor is "used". Fixes PR10480 /
<rdar://problem/9834317>.

llvm-svn: 136081

lifetime-qualified template parameter, ensure that the deduced
template argument is a lifetime type. Fixes <rdar://problem/9828157>.

llvm-svn: 136078

the preferred alignment.  Thus, revert r135934, r135935, and r135940.

llvm-svn: 136062

refer to 'expansion' instead of 'instantiation'.

llvm-svn: 136060

llvm-svn: 136059

'expansion' rather than 'instantiation' for macro source locations.

llvm-svn: 136058

etc. With this I think essentially all of the SourceManager APIs are
converted. Comments and random other bits of cleanup should be all thats
left.

llvm-svn: 136057

and various other 'expansion' based terms. I've tried to reformat where
appropriate and catch as many references in comments but I'm going to do
several more passes. Also I've tried to expand parameter names to be
more clear where appropriate.

llvm-svn: 136056

llvm-svn: 136054

to isMacroArgExpansion.

llvm-svn: 136053

and to work with pointer arithmetic in addition to array indexing.

The new pointer arithmetic porition of the array bounds checking can be
turned on by -Warray-bounds-pointer-arithmetic (and is off by default).

llvm-svn: 136046

llvm-svn: 136044

to declare a static object. // rdar://9603056

llvm-svn: 135970