llvm-svn: 174107

llvm-svn: 174106

The Cnt variable is adjusted (incremented) for simplification of
checking logic. The increment should not be stored in the state.

llvm-svn: 174104

catches uses of an extremely minor and widely-available C++ extension (which
every C++ compiler I could find supports, but EDG and Clang reject in strict
mode).

The diagnosed code pattern looks like this:

struct X {
  union {
    struct {
      int a;
      int b;
    } S;
  };
};

llvm-svn: 174103

llvm-svn: 174102

llvm-svn: 174101

llvm-svn: 174100

Patch by: Vincent Lejeune

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 174099

Patch by: Vincent Lejeune

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 174098

Patch by: Vincent Lejeune

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 174097

This matches our behavior for autorelease pools created by +alloc. Some
people like to create autorelease pools in one method and release them
somewhere else.

If you want safe autorelease pool semantics, use the new ARC-compatible
syntax: @autoreleasepool { ... }

<rdar://problem/13121353>

llvm-svn: 174096

Add "thread return -x" to unwind the innermost user called expression (if you happen to have stopped in it due to a crash.)
Make the message when you hit an crash while evaluating an expression a little clearer, and mention "thread return -x".

rdar://problem/13110464

llvm-svn: 174095

llvm-svn: 174094

This is a re-worked version of r174048.
Given source IR:
call void @llvm.dbg.declare(metadata !{i32* %argc.addr}, metadata !14), !dbg !15
we used to generate 
call void @llvm.dbg.declare(metadata !27, metadata !28), !dbg !29
!27 = metadata !{null}

With this patch, we will correctly generate
call void @llvm.dbg.declare(metadata !{i32* %argc.addr}, metadata !27), !dbg !28

Looking up %argc.addr in ValueMap will return null, since %argc.addr is already
correctly set up, we can use identity mapping.

rdar://problem/13089880

llvm-svn: 174093

llvm-svn: 174092

llvm-svn: 174091

Attributes that are strings are typically target-dependent attributes. They are
of this form in the IR:

   "attr"
   "attr" = "val"

llvm-svn: 174090

 - Also, change the exit code to match 'sh'.

llvm-svn: 174089

llvm-svn: 174088

 - Testing finds bugs, who knew.

llvm-svn: 174087

llvm-svn: 174086

past the natural stack alignment.

llvm-svn: 174085

We used to create children DIEs for a scope, then check whether ScopeDIE is
null. If ScopeDIE is null, the children DIEs will be dangling. Other DIEs can
link to those dangling DIEs, which are not emitted at all, causing dwarf error.

The current testing case is 4k lines, from MultiSource/BenchMark/McCat/09-vor.

rdar://problem/13071959

llvm-svn: 174084

Each target implementation was needlessly recomputing the index.
Part of rdar://13076458

llvm-svn: 174083

<rdar://problem/13121442>

llvm-svn: 174082

This allows people to use Unicode in their #pragma mark and in macros
that exist only to be string-ized.

<rdar://problem/13107323&13121362>

llvm-svn: 174081

llvm-svn: 174080

llvm-svn: 174079

llvm-svn: 174078

Compilation always sets this explicitly, but creating a preprocessor
manually should still put the 'IsPreprocessedOutput' flag in a valid state.

llvm-svn: 174077

llvm-svn: 174076

llvm-svn: 174075

llvm-svn: 174074

    
gcc produces false positives for empty braces so turning the warning off.
Instead, turning the warning on for clang so proper warnings aren't missed.
    
Reviewers: dblaikie, chandlerc

llvm-svn: 174073

llvm-svn: 174072

llvm-svn: 174071

 - Also, kill the pointless LitFormats module.

llvm-svn: 174070

It's causing hangs on our internal analyzer buildbot. Will restore after
investigating.

This reverts r173951 / baa7ca1142990e1ad6d4e9d2c73adb749ff50789.

llvm-svn: 174069

llvm-svn: 174068

llvm-svn: 174067