llvm-svn: 171257

llvm-svn: 171256

llvm-svn: 171255

directly.

This is in preparation for removing the use of the 'Attribute' class as a
collection of attributes. That will shift to the AttributeSet class instead.

llvm-svn: 171253

llvm-svn: 171252

LCSSA PHIs may have undef values. The vectorizer updates values that are used by outside users such as PHIs.
The bug happened because undefs are not loop values. This patch handles these PHIs.

PR14725

llvm-svn: 171251

* One that accepts a single Attribute::AttrKind.

* One that accepts an Attribute::AttrKind plus a list of values. This is for
  attributes defined like this:

    #1 = attributes { align = 4 }

* One that accepts a string, for target-specific attributes like this:

    #2 = attributes { "cpu=cortex-a8" }

llvm-svn: 171249

stored here is of a certain kind. This is in preparation for when an Attribute
object represents a single attribute, instead of a bitmask of attributes.

llvm-svn: 171247

already uniquified.

Note: This will be expanded in the future to add more than just one pointer
value.

llvm-svn: 171245

s/Raw/getBitMask/g to be more in line with current naming conventions. This method won't be sticking around.

llvm-svn: 171244

Remove intrinsic specific instructions for (V)SQRTPS/PD. Instead lower to target-independent ISD nodes and use the existing patterns for those.

llvm-svn: 171237

llvm-svn: 171229

Remove intrinsic specific instructions for SSE/SSE2/AVX floating point max/min instructions. Lower them to target specific nodes and use those patterns instead. This also allows them to be commuted if UnsafeFPMath is enabled.

llvm-svn: 171227

llvm-svn: 171226

llvm-svn: 171225

llvm-svn: 171221

llvm-svn: 171220

deleted this, but apparantly not. Charmingly, Clang didn't warn on it
but GCC did.

llvm-svn: 171197

propagating one of the values it simplified to a constant across
a myriad of instructions. Notably, ptrtoint instructions when we had
a constant pointer (say, 0) didn't propagate that, blocking a massive
number of down-stream optimizations.

This was uncovered when investigating why we fail to inline and delete
the boilerplate in:

  void f() {
    std::vector<int> v;
    v.push_back(1);
  }

It turns out most of the efforts I've made thus far to improve the
analysis weren't making it far purely because of this. After this is
fixed, the store-to-load forwarding patch enables LLVM to optimize the
above to an empty function. We still can't nuke a second push_back, but
for different reasons.

There is a very real chance this will cause somewhat noticable changes
in inlining behavior, so please let me know if you see regressions (or
improvements!) because of this patch.

llvm-svn: 171196

how to propagate constants through insert and extract value
instructions.

With the recent improvements to instsimplify, this allows inline cost
analysis to constant fold through intrinsic functions, including notably
the with.overflow intrinsic math routines which often show up inside of
STL abstractions. This is yet another piece in the puzzle of breaking
down the code for:

  void f() {
    std::vector<int> v;
    v.push_back(1);
  }

But it still isn't enough. There are a pile of bugs in inline cost still
blocking this.

llvm-svn: 171195

constant folding calls. Add the initial tests for this which show that
now instsimplify can simplify blindingly obvious code patterns expressed
with both intrinsics and library calls.

llvm-svn: 171194

are nice and decomposed so that we can simplify synthesized calls as
easily as actually call instructions. The internal utility still has the
same behavior, it just now operates on a more generic interface so that
I can extend the set of call simplifications that instsimplify knows
about.

llvm-svn: 171189

llvm-svn: 171183

llvm-svn: 171180

llvm-svn: 171179

AVX: Move the ZEXT/ANYEXT DAGCo optimizations to the lowering of these optimizations. The old test cases still cover all of these lowering/optimizations. The single change that we have is that now anyext does not need to zero a register, because it does not use the exact code path as the zero_extend.

llvm-svn: 171178

llvm-svn: 171172

llvm-svn: 171171

llvm-svn: 171170

llvm-svn: 171166

getAnalysisUsage implementations.

llvm-svn: 171157

[ASan] Fix lifetime intrinsics handling. Now for each intrinsic we check if it describes one of 'interesting' allocas. Assume that allocas can go through casts and phi-nodes before apperaring as llvm.lifetime arguments

llvm-svn: 171153

register. In most cases we actually compare or select YMM-sized registers
and mixing the two types creates horrible code. This commit optimizes
some of the transition sequences.

PR14657.

llvm-svn: 171148

The vector truncs were scalarized during LegalizeVectorOps, later vectorized again by some DAGCombine optimization
and finally, lowered by a dagcombing optimization. Now, they are properly lowered during LegalizeVectorOps.
No new testcase because the original testcases still work.

llvm-svn: 171146

llvm-svn: 171143

Refactor DAGCombinerInfo. Change the different booleans that indicate if we are before or after different runs of DAGCo, with the CombineLevel enum.
Also, added a new API for checking if we are running before or after the LegalizeVectorOps phase. 

llvm-svn: 171142

Move single letter 'P' prefix out of multiclass now that tablegen allows defm to start with #NAME. This makes instruction names more searchable again.

llvm-svn: 171141

llvm-svn: 171140

Add hasSideEffects=0 to some shift and rotate instructions. None of which are currently used by code generation.

llvm-svn: 171137

llvm-svn: 171136