mallocs.

llvm-svn: 135366

llvm-svn: 135364

llvm-svn: 135362

llvm-svn: 135360

llvm-svn: 135358

llvm-svn: 135354

llvm-svn: 135353

llvm-svn: 135352

llvm-svn: 135343

llvm-svn: 135339

assertion I added in r135333. Check for the existence of a preheader
before expanding a recurrence.

llvm-svn: 135335

llvm-svn: 135334

related bug fixes and corresponding assertions for uninitialized data
and missing NULL check. Test cases will be included with the new LFTR.

llvm-svn: 135333

llvm-svn: 135332

This should unbreak the build-self-4-mingw32 tester. I have a very
complicated test case that I will try to clean up.

llvm-svn: 135329

fixing some objc llvm-test crashes with LTO.

llvm-svn: 135324

llvm-svn: 135323

llvm-svn: 135320

Get rid of the separate opcodes for the Darwin versions of tBL, tBLXi, and tBLXr, using pseudo-instructions to lower to the single final opcode.  Update the ARM disassembler for this change.

llvm-svn: 135319

and just use the ones from TargetLowering directly.

llvm-svn: 135318

1) Make non-legal 256-bit loads to be promoted to v4i64. This lets us
canonize the loads and handle things the same way we use to handle
for 128-bit registers. Despite of what one of the removed comments
explained, the load promotion would not mess with VPERM, it's only a
matter of doing the appropriate bitcasts when this instructions comes
to be introduced. Also make LOAD v8i32 legal.

2) Doing 1) exposed two bugs:
- v4i64 was being promoted to itself for several opcodes (introduced
in r124447 by David Greene) causing endless recursion and the stack to
explode.
- there was no support for allOnes BUILD_VECTORs and ANDNP would fail to
match because it was generating early target constant pools during
lowering.

3) The testcases are already checked-in, doing 1) exposed the
bugs in the current testcases.

4) Tidy up code to be more clear and explicit about AVX.

llvm-svn: 135313

comming together with other tests.

llvm-svn: 135312

llvm-svn: 135311

This gets rid of some of the gory splitting details in RAGreedy and
makes them available to future SplitKit clients.

Slightly generalize the functionality to support multi-way splitting.
Specifically, SplitEditor::splitLiveThroughBlock() supports switching
between different register intervals in a block.

llvm-svn: 135307

llvm-svn: 135305

llvm-svn: 135303

llvm-svn: 135302

not isEquality().

llvm-svn: 135296

Remove VMOVDneon and VMOVQ, which are just aliases for VORR.  This continues to simplify the path towards an auto-generated disassembler.

llvm-svn: 135290

- No ELF or COFF implementation yet, I don't have a way to test that.
  Should be straightforward to add though.

llvm-svn: 135288

is named after a common idiom (i.e., memset/memcpy).  Otherwise, we can run into 
infinite recursion.  Ideally, the user should use the correct -fno-builtin flag,
but in case they don't we should play nicely.
rdar://9763412

llvm-svn: 135286

Remove unnecessary duplicate instruction definitions that simply overloaded the type of VEXT.  This can be achieved with a Pat definition, and is much more disassembler friendly.

llvm-svn: 135283

llvm-svn: 135282

llvm-svn: 135281

llvm-svn: 135278

MSVC decorates (and distinguishes) "const" in mangler. It brought linkage error between "extern const" declarations and definitions.

llvm-svn: 135269

In Twine::str(), if the Twine stores only a std::string, just return a direct copy of that instead of first copying to a SmallString and converting that to a std::string. Also fix some indentation.

llvm-svn: 135267

llvm-svn: 135265

desired overload.

This is a bit of a hackish workaround to fix the compile after r135259.
Let me know if there is a better approach.

llvm-svn: 135261

memory for the result.

llvm-svn: 135259