There is no advantage in being different and using the same names simplifies
clang a bit.

llvm-svn: 189141

llvm-svn: 189111

llvm-svn: 189005

llvm-svn: 188899

Synchronize VEX JIT encoding code with the MCJIT version. Fix a bug in the MCJIT code where CurOp was being incremented even if the operand it was pointing at wasn't used. Maybe only matters if there are any EVEX_K instructions that aren't VEX_4V.

llvm-svn: 188868

In LLVM FMA3 operands are dst, src1, src2, src3, however dst is not encoded as it is always src1. This was causing the encoding of the operands to be off by one.

Patch by Chris Bieneman.

llvm-svn: 188866

Rename mattr names for AVX-512 to from avx-512 -> avx512f, avx-512-pfi -> av512pf, avx-512-cdi -> avx512cd, avx-512-eri->avx512er. This matches better with official docs and what gcc patches appearto be using. I didn't touch the has* functions or the feature flag names to avoid change the td and lowering file while commits are still happening.

llvm-svn: 188859

I suppose all "lli -use-mcjit i686-*" should require GOT, (and to fail.)

llvm-svn: 188856

llvm-svn: 188786

llvm-svn: 188746

llvm-svn: 188745

Move AVX and non-AVX replication inside a couple multiclasses to avoid repeating each instruction for both individually.

llvm-svn: 188743

ADD, SUB, MUL integer and FP types. OR, AND, XOR.
Added embeded broadcast form for these instructions.

llvm-svn: 188673

llvm-svn: 188637

llvm-svn: 188629

llvm-svn: 188534

llvm-svn: 188469

Revert r188449 as it turns out we're just missing the instructions that need the v16i32/v16f32 matching.

llvm-svn: 188454

Don't let isPermImmMask handle v16i32 since VPERMI doesn't match on that type. Remove 128-bit vector handling from isPermImmMask too, it's covered by isPSHUFDMask.

llvm-svn: 188449

llvm-svn: 188446

llvm-svn: 188445

Replace getValueType().getSimpleVT() with getSimpleValueType(). Also remove one weird cast from MVT->EVT just to call getSimpleVT().

llvm-svn: 188441

llvm-svn: 188366

llvm-svn: 188365

llvm-svn: 188364

llvm-svn: 188363

Replace EVT with MVT in isVectorShift. Keeps compiler from generating unneeded checks and handling for extended types.

llvm-svn: 188362

Replace EVT with MVT in many of the shuffle lowering functions. Keeps compiler from generating unneeded checks and handling for extended types.

llvm-svn: 188361

../lib/Target/X86/X86ISelLowering.cpp:9715:7: error: unused variable 'OpVT' [-Werror,-Wunused-variable]
  EVT OpVT = Op0.getValueType();
      ^
../lib/Target/X86/X86ISelLowering.cpp:9763:14: error: unused variable 'NumElems' [-Werror,-Wunused-variable]
    unsigned NumElems = VT.getVectorNumElements();

llvm-svn: 188269

Lowering for SETCC.

llvm-svn: 188265

undefined and produce an error message instead as this is a non-relocatable
expression with X86 Mach-O.

rdar://8920876

llvm-svn: 188218

llvm-svn: 188148

Added a test for shuffles using VPERM.

llvm-svn: 188147

Use it to avoid repeating ourselves too often. Also store MVT::SimpleValueType
in the TTI tables so they can be statically initialized, MVT's constructors
create bloated initialization code otherwise.

llvm-svn: 188095

* ELFTypes.h contains template magic for defining types based on endianess, size, and alignment.
* ELFFile.h defines the ELFFile class which provides low level ELF specific access.
* ELFObjectFile.h contains ELFObjectFile which uses ELFFile to implement the ObjectFile interface.

llvm-svn: 188022

llvm-svn: 187986

llvm-svn: 187984

with lowering logic and a test.

llvm-svn: 187884

llvm-svn: 187870

This change came about primarily because of two issues in the existing code.
Niether of:

define i64 @test1(i64 %val) {
  %in = trunc i64 %val to i32
  tail call i32 @ret32(i32 returned %in)
  ret i64 %val
}

define i64 @test2(i64 %val) {
  tail call i32 @ret32(i32 returned undef)
  ret i32 42
}

should be tail calls, and the function sameNoopInput is responsible. The main
problem is that it is completely symmetric in the "tail call" and "ret" value,
but in reality different things are allowed on each side.

For these cases:
1. Any truncation should lead to a larger value being generated by "tail call"
   than needed by "ret".
2. Undef should only be allowed as a source for ret, not as a result of the
   call.

Along the way I noticed that a mismatch between what this function treats as a
valid truncation and what the backends see can lead to invalid calls as well
(see x86-32 test case).

This patch refactors the code so that instead of being based primarily on
values which it recurses into when necessary, it starts by inspecting the type
and considers each fundamental slot that the backend will see in turn. For
example, given a pathological function that returned {{}, {{}, i32, {}}, i32}
we would consider each "real" i32 in turn, and ask if it passes through
unchanged. This is much closer to what the backend sees as a result of
ComputeValueVTs.

Aside from the bug fixes, this eliminates the recursion that's going on and, I
believe, makes the bulk of the code significantly easier to understand. The
trade-off is the nasty iterators needed to find the real types inside a
returned value.

llvm-svn: 187787