llvm-svn: 44686

type.

llvm-svn: 44685

Have Sema::CheckConstantInitList skip semantic analysis when the element type is a record (until we implement the FIXME). This removes a bogus error for the following code...

snarofflocal% cat bug.m

#import <Foundation/NSGeometry.h>

#define NUMHELICOPTERRECTS 5

static NSRect helicopterRects[NUMHELICOPTERRECTS] = {
    {{27, 0}, {18, 11}},	// Bottom
    {{0, 8}, {4, 11}},		// Tail
    {{0, 16}, {50, 1}},
    {{22, 5}, {18, 12}},	// Body
    {{0, 10}, {42, 3}}
};

llvm-svn: 44684

Lookup methods in the global pool even when a statically typed object's class interface isn't in scope!

As the comment in the code indicates, I'm not fond of this. Nevertheless, gcc compat is a goal.

Here is the case I'm talking about...

#import <Foundation/Foundation.h>

@interface AnyClass : NSObject
- (NSRect)rect;
@end

@class Helicopter;

static void func(Helicopter *obj) {
  NSRect r = [obj rect];
}

...before this patch, we would warn/error. With this patch, everything "just works".

llvm-svn: 44682

llvm-svn: 44681

llvm-svn: 44676

llvm-svn: 44674

Rewrite 'super' within a class method. This required some minor tweaks to the front-end.

llvm-svn: 44673

llvm-svn: 44672

llvm-svn: 44671

"@class Protocol;"

llvm-svn: 44670

Remove a bogus optimization. It's not possible to do a move to low element to a <8 x i16> or <16 x i8> vector.

llvm-svn: 44669

llvm-svn: 44668

llvm-svn: 44666

to rewriter (my previous patch).

llvm-svn: 44665

struct STestB1 {char a; char b:2; } stb1;
struct STestB2 {char a; char b:5; char c:4} stb2;

llvm-svn: 44664

Simpler and safer.

llvm-svn: 44663

llvm-svn: 44662

llvm-svn: 44661

llvm-svn: 44660

only disable it if we don't know it will be obviously profitable.
Still fixme, but less so. :)

llvm-svn: 44658

the X86 backend are needed before this should be enabled by default.

llvm-svn: 44657

_foo:
	movl	$12, %eax
	andl	4(%esp), %eax
	movl	_array(%eax), %eax
	ret

instead of:

_foo:
	movl	4(%esp), %eax
	shrl	$2, %eax
	andl	$3, %eax
	movl	_array(,%eax,4), %eax
	ret

As it turns out, this triggers all the time, in a wide variety of
situations, for example, I see diffs like this in various programs:

-       movl    8(%eax), %eax
-       shll    $2, %eax
-       andl    $1020, %eax
-       movl    (%esi,%eax), %eax
+       movzbl  8(%eax), %eax
+       movl    (%esi,%eax,4), %eax


-       shll    $2, %edx
-       andl    $1020, %edx
-       movl    (%edi,%edx), %edx
+       andl    $255, %edx
+       movl    (%edi,%edx,4), %edx

Unfortunately, I also see stuff like this, which can be fixed in the
X86 backend:

-       andl    $85, %ebx
-       addl    _bit_count(,%ebx,4), %ebp
+       shll    $2, %ebx
+       andl    $340, %ebx
+       addl    _bit_count(%ebx), %ebp

llvm-svn: 44656

llvm-svn: 44655

llvm-svn: 44654

llvm-svn: 44653

llvm-svn: 44651

llvm-svn: 44650

llvm-svn: 44649

crash because of this.

llvm-svn: 44648

to create a JIT.  This lets you specify JIT-specific configuration items
like the JITMemoryManager to use.

llvm-svn: 44647

delete one ExecutionEngine ctor, minor cleanup.

llvm-svn: 44646

llvm-svn: 44645

Fix for PR1831: if all defs of an interval are re-materializable, then it's a preferred spill candiate.

llvm-svn: 44644

If both result of the {s|z}xt and its source are live out, rewrite all uses of the source with result of extension.

llvm-svn: 44643

GetLanguage, and InitializeLangOptions.  The goal is to break up this logic
into atomic units of functionality that can later be refactored into better
driver logic that is capable of handling a mixture of source files of
different languages.

llvm-svn: 44642

llvm-svn: 44641

own JITMemoryManager interface.  There is no functionality change with 
this patch.

llvm-svn: 44640

llvm-svn: 44639

llvm-svn: 44638