llvm-svn: 126188

llvm-svn: 126187

mingw-w64's i128 tweak should be done with x86_64-mingw32.

llvm-svn: 126186

llvm-svn: 126185

underlying type: we weren't parsing unnamed enumeration types with a
fixed underlying type.

llvm-svn: 126184

enumeration types. Fixes <rdar://problem/8559831>.

llvm-svn: 126183

fixed in r126147, where a field in the X86 decode
structure was being read as bits, not bytes.

llvm-svn: 126182

operand.evaluate as an error code, not as the
contents of the operand.

llvm-svn: 126181

edis name as needing expansion.  This tool was
been replaced by llvm-mc long ago.

llvm-svn: 126180

llvm-svn: 126179

Add ARM/Thumb encoding entries for "CMN (immediate)" and "CMN (register)" operations,
with the EmulateCMNImm()/Reg() methods not implemented yet for now.

llvm-svn: 126178

Make clang -cc1 disable Objective-C exceptions by default, and add a -fobjc-exceptions flag to turn them on.

Update all tests accordingly.

llvm-svn: 126177

up by the dynamic linker, but it's better to use the correct instruction
to begin with.

Fixes rdar://9011034

llvm-svn: 126176

llvm-svn: 126175

In Objective-C, there are no trailing return types, so don't produce diagnostics suggesting they are missing.

llvm-svn: 126174

llvm-svn: 126172

llvm-svn: 126171

at phis. This enables us to eliminate a lot of pointless zexts during the DAGCombine
phase. This fixes <rdar://problem/8760114>.

llvm-svn: 126170

allows for the information propagated across basic blocks to be merged at phis.

llvm-svn: 126169

From Dimitry Andric.

llvm-svn: 126168

It works like -isystem and the search path keeps -isystem and
-cxx-isystem in order relative to each other. -cxx-isystem is only used
for C++ sources though. Drop the existing -cxx-system-include option for
cc1 as it is now redundant.

llvm-svn: 126167

 * 'auto' was being rejected on abstract-declarators with trailing return
types and on typedefs with trailing return types. 'auto' is always
allowed in these cases. This was found while testing the fix for PR 9278.

 * A very poor diagnostic was being issued for auto (f() -> int): "return
type must be 'auto', not 'auto'". This is closely related to PR 9060.

 * Trailing return type handling was happening slightly too late,
resulting in the checks for functions returning arrays and functions
returning functions being missed.

llvm-svn: 126166

Patch by Jai Menon.

llvm-svn: 126165

llvm-svn: 126164

llvm-svn: 126163

protocols do not match with method implementation.
// rdar://7076235

llvm-svn: 126162

The instance size of a metaclass should be the size of a class.  This is not, in fact, 0, even for very small classes.  (GNU runtime)

llvm-svn: 126161

"TEQ (immediate)", and "TEQ (register)" operations.

llvm-svn: 126160

Skipping over debugvalue instructions to determine whether the split spot is in a IT block. rdar://9030770

llvm-svn: 126159

llvm-svn: 126158

llvm-svn: 126157

In other words, do not keep track of argument's location.  The debugger (gdb) is not prepared to see line table entries for arguments. For the debugger, "second" line table entry marks beginning of function body.
This requires some coordination with debugger to get this working. 
 - The debugger needs to be aware of prolog_end attribute attached with line table entries.
 - The compiler needs to accurately mark prolog_end in line table entries (at -O0 and at -O1+)

llvm-svn: 126155

This actually rules out too much, since it also catches typedefs for pointers to functions with trailing return types:

  typedef auto (*F)() -> int;

Fix for that (and the same issue in all abstract-declarators) to follow shortly.

llvm-svn: 126153

llvm-svn: 126152

An original endpoint is an instruction that killed or defined the original live
range before any live ranges were split.

When splitting global live ranges, avoid creating local live ranges without any
original endpoints. We may still create global live ranges without original
endpoints, but such a range won't be split again, and live range splitting still
terminates.

llvm-svn: 126151

llvm-svn: 126149

When the mismatch is due to a larger input operand that is
a constant, truncate it down to the size of the output.  This
allows us to accept some cases in the linux kernel and elsewhere.
Pedantically speaking, we generate different code than GCC, though
I can't imagine how it would matter:

Clang:
	movb	$-1, %al
	frob %al

GCC:

	movl	$255, %eax
	frob %al

llvm-svn: 126148

X86 instruction decode structure was being interpreted as
being in units of bits, although it is actually stored in
units of bytes.

llvm-svn: 126147

llvm-svn: 126146

llvm-svn: 126145