OSDN Git Service

android-x86/external-llvm.git
7 years agoFix -Wunused-but-set-variable warning by removing unused 'aggregateIsPacked' checking
Simon Pilgrim [Wed, 22 Feb 2017 13:37:31 +0000 (13:37 +0000)]
Fix -Wunused-but-set-variable warning by removing unused 'aggregateIsPacked' checking

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295830 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[GlobalISel] Fix compiler warnings and make assert assert something.
Benjamin Kramer [Wed, 22 Feb 2017 12:59:47 +0000 (12:59 +0000)]
[GlobalISel] Fix compiler warnings and make assert assert something.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295827 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[SLP] Remove unused initial value from the variable, NFC.
Alexey Bataev [Wed, 22 Feb 2017 12:57:58 +0000 (12:57 +0000)]
[SLP] Remove unused initial value from the variable, NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295826 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][GlobalISel] Initial implementation , select G_ADD gpr, gpr
Igor Breger [Wed, 22 Feb 2017 12:25:09 +0000 (12:25 +0000)]
[X86][GlobalISel] Initial implementation , select G_ADD gpr, gpr

Summary: Initial implementation for X86InstructionSelector. Handle selection COPY and G_ADD/G_SUB gpr, gpr .

Reviewers: qcolombet, rovka, zvi, ab

Reviewed By: rovka

Subscribers: mgorny, dberris, kristof.beyls, llvm-commits

Differential Revision: https://reviews.llvm.org/D29816

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295824 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Regenerate CSE test with codegen instead of just the instruction count
Simon Pilgrim [Wed, 22 Feb 2017 10:12:46 +0000 (10:12 +0000)]
[X86] Regenerate CSE test with codegen instead of just the instruction count

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295819 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM] Fix constant islands pass.
Roger Ferrer Ibanez [Wed, 22 Feb 2017 09:06:21 +0000 (09:06 +0000)]
[ARM] Fix constant islands pass.

The pass tries to fix a spill of LR that turns out to be unnecessary.
So it removes the tPOP but forgets to remove tPUSH.
This causes the stack be misaligned upon returning the function.

Thus, remove the tPUSH as well in this case.

Differential Revision: https://reviews.llvm.org/D30207

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295816 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoWrite to a temporary file in test instead of random file in the test directory.
Benjamin Kramer [Wed, 22 Feb 2017 09:02:27 +0000 (09:02 +0000)]
Write to a temporary file in test instead of random file in the test directory.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295815 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Fix memory operands definition for some instructions.
Ayman Musa [Wed, 22 Feb 2017 08:06:29 +0000 (08:06 +0000)]
[X86] Fix memory operands definition for some instructions.

Change integer memory operands to FP memory operands to some FP instructions.

Differential Revision: https://reviews.llvm.org/D30201

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295813 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoOptDiag: Add const to some interfaces that don't modify anything. NFC
Justin Bogner [Wed, 22 Feb 2017 07:38:17 +0000 (07:38 +0000)]
OptDiag: Add const to some interfaces that don't modify anything. NFC

This needed a const_cast for the dominator tree recalculation in
OptimizationRemarkEmitter, but we do that all over the place already
and it's safe.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295812 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM] Classification Improvements to ARM Sched-Models. NFCI.
Javed Absar [Wed, 22 Feb 2017 07:22:57 +0000 (07:22 +0000)]
[ARM] Classification Improvements to ARM Sched-Models. NFCI.

This patch adds missing sched classes for Thumb2 instructions.
This has been missing so far, and as a consequence, machine
scheduler models for individual sub-targets have tended to
be larger than they needed to be. These patches should help
write schedulers better and faster in the future
for ARM sub-targets.

Reviewer: Diana Picus
Differential Revision: https://reviews.llvm.org/D29953

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295811 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVX-512] Allow legacy scalar min/max intrinsics to select EVEX instructions when...
Craig Topper [Wed, 22 Feb 2017 06:54:18 +0000 (06:54 +0000)]
[AVX-512] Allow legacy scalar min/max intrinsics to select EVEX instructions when available

This patch introduces new X86ISD::FMAXS and X86ISD::FMINS opcodes. The legacy intrinsics now lower to this node. As do the AVX-512 masked intrinsics when the rounding mode is CUR_DIRECTION.

I've merged a copy of the tablegen multiclass avx512_fp_scalar into avx512_fp_scalar_sae. avx512_fp_scalar still needs to support CUR_DIRECTION appearing as a rounding mode for X86ISD::FADD_ROUND and others.

Differential revision: https://reviews.llvm.org/D30186

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295810 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ValueTracking] Make poison propagation more aggressive
Sanjoy Das [Wed, 22 Feb 2017 06:52:32 +0000 (06:52 +0000)]
[ValueTracking] Make poison propagation more aggressive

Summary:
Motivation: fix PR31181 without regression (the actual fix is still in
progress).  However, the actual content of PR31181 is not relevant
here.

This change makes poison propagation more aggressive in the following
cases:

 1. poision * Val == poison, for any Val.  In particular, this changes
    existing intentional and documented behavior in these two cases:
     a. Val is 0
     b. Val is 2^k * N
 2. poison << Val == poison, for any Val
 3. getelementptr is poison if any input is poison

I think all of these are justified (and are axiomatically true in the
new poison / undef model):

1a: we need poison * 0 to be poison to allow transforms like these:

  A * (B + C) ==> A * B + A * C

If poison * 0 were 0 then the above transform could not be allowed
since e.g. we could have A = poison, B = 1, C = -1, making the LHS

  poison * (1 + -1) = poison * 0 = 0

and the RHS

  poison * 1 + poison * -1 = poison + poison = poison

1b: we need e.g. poison * 4 to be poison since we want to allow

  A * 4 ==> A + A + A + A

If poison * 4 were a value with all of their bits poison except the
last four; then we'd not be able to do this transform since then if A
were poison the LHS would only be "partially" poison while the RHS
would be "full" poison.

2: Same reasoning as (1b), we'd like have the following kinds
transforms be legal:

  A << 1 ==> A + A

Reviewers: majnemer, efriedma

Subscribers: mcrosier, llvm-commits

Differential Revision: https://reviews.llvm.org/D30185

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295809 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUse const-ref in range-loop for to avoid copying pairs of std::string
Sean Silva [Wed, 22 Feb 2017 06:34:04 +0000 (06:34 +0000)]
Use const-ref in range-loop for to avoid copying pairs of std::string

No reason to create temporaries.

Differential Revision: https://reviews.llvm.org/D29871

Patch by sergio.martins!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295807 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[WebAssembly] Add skeleton MC support for the Wasm container format
Dan Gohman [Wed, 22 Feb 2017 01:23:18 +0000 (01:23 +0000)]
[WebAssembly] Add skeleton MC support for the Wasm container format

This just adds the basic skeleton for supporting a new object file format.
All of the actual encoding will be implemented in followup patches.

Differential Revision: https://reviews.llvm.org/D26722

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295803 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix -Wcovered-switch-default.
Rui Ueyama [Wed, 22 Feb 2017 01:01:45 +0000 (01:01 +0000)]
Fix -Wcovered-switch-default.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295799 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Add cvt.pkrtz intrinsic
Matt Arsenault [Wed, 22 Feb 2017 00:27:34 +0000 (00:27 +0000)]
AMDGPU: Add cvt.pkrtz intrinsic

Convert llvm.SI.packf16 test uses

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295797 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[LoopUnroll] Enable PGO-based loop peeling by default.
Michael Kuperstein [Wed, 22 Feb 2017 00:27:34 +0000 (00:27 +0000)]
[LoopUnroll] Enable PGO-based loop peeling by default.

This enables peeling of loops with low dynamic iteration count by default,
when profile information is available.

Differential Revision: https://reviews.llvm.org/D27734

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295796 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Remove some uses of llvm.SI.export in tests
Matt Arsenault [Wed, 22 Feb 2017 00:02:21 +0000 (00:02 +0000)]
AMDGPU: Remove some uses of llvm.SI.export in tests

Merge some of the old, smaller tests into more complete versions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295792 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Remove llvm.AMDGPU.clamp intrinsic
Matt Arsenault [Tue, 21 Feb 2017 23:46:04 +0000 (23:46 +0000)]
AMDGPU: Remove llvm.AMDGPU.clamp intrinsic

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295789 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Redefine clamp node as clamp 0.0-1.0
Matt Arsenault [Tue, 21 Feb 2017 23:35:48 +0000 (23:35 +0000)]
AMDGPU: Redefine clamp node as clamp 0.0-1.0

Change implementation to use max instead of add.
min/max/med3 do not flush denormals regardless of the mode,
so it is OK to use it whether or not they are enabled.

Also allow using clamp with f16, and use knowledge
of dx10_clamp.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295788 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[NVPTX] Unify vectorization of load/stores of aggregate arguments and return values.
Artem Belevich [Tue, 21 Feb 2017 22:56:05 +0000 (22:56 +0000)]
[NVPTX] Unify vectorization of load/stores of aggregate arguments and return values.

Original code only used vector loads/stores for explicit vector arguments.
It could also do more loads/stores than necessary (e.g v5f32 would
touch 8 f32 values). Aggregate types were loaded one element at a time,
even the vectors contained within.

This change attempts to generalize (and simplify) parameter space
loads/stores so that vector loads/stores can be used more broadly.
Functionality of the patch has been verified by compiling thrust
test suite and manually checking the differences between PTX
generated by llvm with and without the patch.

General algorithm:
* ComputePTXValueVTs() flattens input/output argument into a flat list
  of scalars to load/store and returns their types and offsets.
* VectorizePTXValueVTs() uses that data to create vectorization plan
  which returns an array of flags marking boundaries of vectorized
  load/stores. Scalars are represented as 1-element vectors.
* Code that generates loads/stores implements a simple state machine
  that constructs a vector according to the plan.

Differential Revision: https://reviews.llvm.org/D30011

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295784 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Formatting fixes
Matt Arsenault [Tue, 21 Feb 2017 22:50:41 +0000 (22:50 +0000)]
AMDGPU: Formatting fixes

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295783 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDAG: Check if extract_vector_elt is legal or custom
Matt Arsenault [Tue, 21 Feb 2017 22:47:27 +0000 (22:47 +0000)]
DAG: Check if extract_vector_elt is legal or custom

Avoids test regressions in future AMDGPU commits when
more vector types are custom lowered.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295782 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AArch64, X86] Add statistics for the MacroFusion pass
Evandro Menezes [Tue, 21 Feb 2017 22:16:13 +0000 (22:16 +0000)]
[AArch64, X86] Add statistics for the MacroFusion pass

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295777 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AArch64, X86] Guard against both instrs being wild cards
Evandro Menezes [Tue, 21 Feb 2017 22:16:11 +0000 (22:16 +0000)]
[AArch64, X86] Guard against both instrs being wild cards

If both instrs are wild cards, the result can be a crash.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295776 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AArch64] Add test case for fusion of literal generation
Evandro Menezes [Tue, 21 Feb 2017 22:16:09 +0000 (22:16 +0000)]
[AArch64] Add test case for fusion of literal generation

Add test case from https://reviews.llvm.org/D28698 that was somehow lost in
transit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295775 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AArch64] Add test case for fusion of AES crypto operations
Evandro Menezes [Tue, 21 Feb 2017 22:16:06 +0000 (22:16 +0000)]
[AArch64] Add test case for fusion of AES crypto operations

Add test case from https://reviews.llvm.org/D28491 that was somehow lost in
transit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295774 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeGen] Fix some Clang-tidy modernize and Include What You Use warnings; other...
Eugene Zelenko [Tue, 21 Feb 2017 22:07:52 +0000 (22:07 +0000)]
[CodeGen] Fix some Clang-tidy modernize and Include What You Use warnings; other minor fixes (NFC).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295773 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTry to fix the buildbot on OSX.
Zachary Turner [Tue, 21 Feb 2017 21:31:28 +0000 (21:31 +0000)]
Try to fix the buildbot on OSX.

Since I'm only seeing failures on OSX, and it's saying
permission denied, I'm suspecting this is due to the addition
of the MAP_RESILIENT_CODESIGN and/or MAP_RESILIENT_MEDIA flags.
Speculatively trying to remove those to get the bots working.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295770 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTry to fix Android build.
Zachary Turner [Tue, 21 Feb 2017 21:13:10 +0000 (21:13 +0000)]
Try to fix Android build.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295769 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Support] Add a function to check if a file resides locally.
Zachary Turner [Tue, 21 Feb 2017 20:55:47 +0000 (20:55 +0000)]
[Support] Add a function to check if a file resides locally.

Differential Revision: https://reviews.llvm.org/D30010

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295768 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMake default value for disable-licm-promotion in licm explicit.
Xin Tong [Tue, 21 Feb 2017 20:53:48 +0000 (20:53 +0000)]
Make default value for disable-licm-promotion in licm explicit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295767 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDon't modify archive members unless really needed.
Rafael Espindola [Tue, 21 Feb 2017 20:40:54 +0000 (20:40 +0000)]
Don't modify archive members unless really needed.

For whatever reason ld64 requires that member headers (not the member
themselves) should be aligned. The only way to do that is to edit the
previous member so that it ends at an aligned boundary.

Since modifying data put in an archive is an undesirable property,
llvm-ar should only do it when it is absolutely necessary.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295765 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix PR31896.
Evgeniy Stepanov [Tue, 21 Feb 2017 20:17:34 +0000 (20:17 +0000)]
Fix PR31896.

Address of an alias of a global with offset is incorrectly lowered as an address of the global (i.e. ignoring offset).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295762 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTry to fix line endings.
Zachary Turner [Tue, 21 Feb 2017 19:52:57 +0000 (19:52 +0000)]
Try to fix line endings.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295759 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[InstCombine] canonicalize non-obivous forms of integer min/max
Sanjay Patel [Tue, 21 Feb 2017 19:33:53 +0000 (19:33 +0000)]
[InstCombine] canonicalize non-obivous forms of integer min/max

This is part of trying to clean up our handling of min/max patterns in IR.
By converting these to canonical form, we're more likely to recognize them
because there are various places in InstCombine that don't use
matchSelectPattern or m_SMax and friends.

The backend fixups referenced in the now deleted TODO comment were added with:
https://reviews.llvm.org/rL291392
https://reviews.llvm.org/rL289738

If there's any codegen fallout from this change, we should be able to address
it in DAGCombiner or target-specific lowering.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295758 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Remove dead declarations in tests
Matt Arsenault [Tue, 21 Feb 2017 19:31:33 +0000 (19:31 +0000)]
AMDGPU: Remove dead declarations in tests

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295757 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove svn:eol-style property from 2 files.
Zachary Turner [Tue, 21 Feb 2017 19:29:56 +0000 (19:29 +0000)]
Remove svn:eol-style property from 2 files.

There are still over 3400 files remaining with this property set, but there are tens of thousands more with the property not set.  Until we decide what to do on a global scale, this at least unblocks me temporarily.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295756 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Remove dead declarations from MIR tests
Matt Arsenault [Tue, 21 Feb 2017 19:27:36 +0000 (19:27 +0000)]
AMDGPU: Remove dead declarations from MIR tests

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295755 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Remove llvm.AMDGPU.flbit intrinsic
Matt Arsenault [Tue, 21 Feb 2017 19:27:33 +0000 (19:27 +0000)]
AMDGPU: Remove llvm.AMDGPU.flbit intrinsic

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295754 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Don't use stack space for SGPR->VGPR spills
Matt Arsenault [Tue, 21 Feb 2017 19:12:08 +0000 (19:12 +0000)]
AMDGPU: Don't use stack space for SGPR->VGPR spills

Before frame offsets are calculated, try to eliminate the
frame indexes used by SGPR spills. Then we can delete them
after.

I think for now we can be sure that no other instruction
will be re-using the same frame indexes. It should be easy
to notice if this assumption ever breaks since everything
asserts if it tries to use a dead frame index later.

The unused emergency stack slot seems to still be left behind,
so an additional 4 bytes is still wasted.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295753 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[LoopSimplify] Simplify how we compute UniqueExit
Xin Tong [Tue, 21 Feb 2017 19:10:58 +0000 (19:10 +0000)]
[LoopSimplify] Simplify how we compute UniqueExit

Summary: Simplify how we compute UniqueExit. Reuse ExitBlockSet.

Reviewers: sanjoy, efriedma, hfinkel

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D30182

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295751 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMore comments for getUniqueExitBlocks. NFCI
Xin Tong [Tue, 21 Feb 2017 19:08:03 +0000 (19:08 +0000)]
More comments for getUniqueExitBlocks. NFCI

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295750 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTeach the IR verifier to reject conflicting debug info for function arguments.
Adrian Prantl [Tue, 21 Feb 2017 19:03:15 +0000 (19:03 +0000)]
Teach the IR verifier to reject conflicting debug info for function arguments.

Conflicting debug info for function arguments causes hard-to-debug
assertions in the DWARF backend, so the Verifier should reject it.
For performance reasons this only checks function arguments from
non-inlined debug intrinsics for now.

rdar://problem/30520286

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295749 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeGenPrepare] Sink and duplicate more 'and' instructions.
Geoff Berry [Tue, 21 Feb 2017 18:53:14 +0000 (18:53 +0000)]
[CodeGenPrepare] Sink and duplicate more 'and' instructions.

Summary:
Rework the code that was sinking/duplicating (icmp and, 0) sequences
into blocks where they were being used by conditional branches to form
more tbz instructions on AArch64.  The new code is more general in that
it just looks for 'and's that have all icmp 0's as users, with a target
hook used to select which subset of 'and' instructions to consider.
This change also enables 'and' sinking for X86, where it is more widely
beneficial than on AArch64.

The 'and' sinking/duplicating code is moved into the optimizeInst phase
of CodeGenPrepare, where it can take advantage of the fact the
OptimizeCmpExpression has already sunk/duplicated any icmps into the
blocks where they are used.  One minor complication from this change is
that optimizeLoadExt needed to be updated to always mark 'and's it has
determined should be in the same block as their feeding load in the
InsertedInsts set to avoid an infinite loop of hoisting and sinking the
same 'and'.

This change fixes a regression on X86 in the tsan runtime caused by
moving GVNHoist to a later place in the optimization pipeline (see
PR31382).

Reviewers: t.p.northover, qcolombet, MatzeB

Subscribers: aemerson, mcrosier, sebpop, llvm-commits

Differential Revision: https://reviews.llvm.org/D28813

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295746 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU : AMDGPU : Update AMDGPU Trap Handler ABI.
Wei Ding [Tue, 21 Feb 2017 18:48:01 +0000 (18:48 +0000)]
AMDGPU : AMDGPU : Update AMDGPU Trap Handler ABI.

Differential Revision: http://reviews.llvm.org/D29913

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295745 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTest commit
Dmitry Preobrazhensky [Tue, 21 Feb 2017 18:07:07 +0000 (18:07 +0000)]
Test commit

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295740 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] EltsFromConsecutiveLoads SDLoc argument should be const&.
Simon Pilgrim [Tue, 21 Feb 2017 17:42:28 +0000 (17:42 +0000)]
[X86] EltsFromConsecutiveLoads SDLoc argument should be const&.

There appears never to have been a time that the reference was updated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295739 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDo not leak OpenedHandles.
Vassil Vassilev [Tue, 21 Feb 2017 17:30:43 +0000 (17:30 +0000)]
Do not leak OpenedHandles.

Reviewed by Vedant Kumar (D30178)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295737 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX512] Update VPBROADCASTQ test to combine from VPERMQ instead of VPERMI2Q.
Simon Pilgrim [Tue, 21 Feb 2017 17:04:11 +0000 (17:04 +0000)]
[X86][AVX512] Update VPBROADCASTQ test to combine from VPERMQ instead of VPERMI2Q.

VPERMI2Q doesn't have shuffle decoding from re-materializable constants.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295736 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX] Rename shuffle combine tests to show combined shuffle type. NFCI.
Simon Pilgrim [Tue, 21 Feb 2017 16:45:31 +0000 (16:45 +0000)]
[X86][AVX] Rename shuffle combine tests to show combined shuffle type. NFCI.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295735 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM] Correct SP/PC handling in t2MOVr
John Brawn [Tue, 21 Feb 2017 16:45:04 +0000 (16:45 +0000)]
[ARM] Correct SP/PC handling in t2MOVr

Add a missing test that I forgot to svn add in my previous commit

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295734 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX2] Fix VPBROADCASTQ folding on 32-bit targets.
Simon Pilgrim [Tue, 21 Feb 2017 16:41:44 +0000 (16:41 +0000)]
[X86][AVX2] Fix VPBROADCASTQ folding on 32-bit targets.

As i64 isn't a value type on 32-bit targets, we need to fold the VZEXT_LOAD into VPBROADCASTQ.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295733 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM] Correct SP/PC handling in t2MOVr
John Brawn [Tue, 21 Feb 2017 16:41:29 +0000 (16:41 +0000)]
[ARM] Correct SP/PC handling in t2MOVr

PC isn't allowed in the source operand of t2MOVr, so change the register class
to one without PC. SP handling is slightly trickier and changes depending on if
we're in ARMv8, so do that in checkTargetMatchPredicate.

Differential Revision: https://reviews.llvm.org/D30199

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295732 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX2] Add AVX512 test targets to AVX2 shuffle combines.
Simon Pilgrim [Tue, 21 Feb 2017 16:29:28 +0000 (16:29 +0000)]
[X86][AVX2] Add AVX512 test targets to AVX2 shuffle combines.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295731 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX] Add tests showing missed VPBROADCASTQ folding on 32-bit targets.
Simon Pilgrim [Tue, 21 Feb 2017 16:05:35 +0000 (16:05 +0000)]
[X86][AVX] Add tests showing missed VPBROADCASTQ folding on 32-bit targets.

As i64 isn't a value type on 32-bit targets, we fail to fold the VZEXT_LOAD into VPBROADCASTQ.

Also shows that we're not decoding VPERMIV3 shuffles very well....

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295729 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][SSE] Prefer to combine shuffles to VZEXT over VZEXT_MOVL.
Simon Pilgrim [Tue, 21 Feb 2017 15:09:00 +0000 (15:09 +0000)]
[X86][SSE] Prefer to combine shuffles to VZEXT over VZEXT_MOVL.

This matches what is already done during shuffle lowering and helps prevent the need for a zero-vector in cases where shuffles match both patterns.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295723 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][SSE] Added SSE41 shuffle combining test file.
Simon Pilgrim [Tue, 21 Feb 2017 14:51:15 +0000 (14:51 +0000)]
[X86][SSE] Added SSE41 shuffle combining test file.

Currently just contains one case where we combine to VZEXT_MOVL instead of VZEXT which would avoid the need for a zero vector to be generated

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295721 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[InstCombine] Do not exercise nested max/min pattern on abs
Anna Thomas [Tue, 21 Feb 2017 14:40:28 +0000 (14:40 +0000)]
[InstCombine] Do not exercise nested max/min pattern on abs

Summary:
This is a fix for assertion failure in
`getInverseMinMaxSelectPattern` when ABS is passed in as a select pattern.

We should not be invoking the simplification rule for
ABS(MIN(~ x,y))) or ABS(MAX(~x,y)) combinations.

Added a test case which would cause an assertion failure without the patch.

Reviewers: sanjoy, majnemer

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D30051

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295719 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVX512] Fix EXTRACT_VECTOR_ELT for v2i1/v4i1/v32i1/v64i1 with variable index.
Igor Breger [Tue, 21 Feb 2017 14:01:25 +0000 (14:01 +0000)]
[AVX512] Fix EXTRACT_VECTOR_ELT for v2i1/v4i1/v32i1/v64i1 with variable index.

Differential Revision: https://reviews.llvm.org/D30189

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295718 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[SLP] Tests for shuffle/blending operations.
Alexey Bataev [Tue, 21 Feb 2017 13:40:55 +0000 (13:40 +0000)]
[SLP] Tests for shuffle/blending operations.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295717 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM] GlobalISel: Lower calls to void() functions
Diana Picus [Tue, 21 Feb 2017 11:33:59 +0000 (11:33 +0000)]
[ARM] GlobalISel: Lower calls to void() functions

For now, we hardcode a BLX instruction, and generate an ADJCALLSTACKDOWN/UP pair
with amount 0.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295716 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agotablegen: Fix android build
Pavel Labath [Tue, 21 Feb 2017 09:19:41 +0000 (09:19 +0000)]
tablegen: Fix android build

use llvm::to_string instead of std:: version.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295711 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Remove ssse3 intrinsic tests from the avx intrinsics test file.
Craig Topper [Tue, 21 Feb 2017 08:06:08 +0000 (08:06 +0000)]
[X86] Remove ssse3 intrinsic tests from the avx intrinsics test file.

They are all covered by the SSSE3 intrinsics test with SSSE3, AVX, and AVX512 command lines.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295708 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Remove sse4.2 intrinsic tests from the avx intrinsics test file. Fix some other...
Craig Topper [Tue, 21 Feb 2017 08:06:05 +0000 (08:06 +0000)]
[X86] Remove sse4.2 intrinsic tests from the avx intrinsics test file. Fix some other consistency issues.

They are all covered by the SSE4.2 intrinsics test with SSE4.2, AVX, and AVX512 command lines.

Merge sse42.ll into the other intrinsics test. Rename sse42_64.ll to be named like other intrinsic tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295707 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Remove sse4.1 intrinsic tests from the avx intrinsics test file.
Craig Topper [Tue, 21 Feb 2017 08:06:02 +0000 (08:06 +0000)]
[X86] Remove sse4.1 intrinsic tests from the avx intrinsics test file.

They are all covered by the SSE4.1 intrinsics test with SSE4.1, AVX, and AVX512 command lines.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295706 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Remove sse3 intrinsic tests from the avx intrinsics test file.
Craig Topper [Tue, 21 Feb 2017 08:05:59 +0000 (08:05 +0000)]
[X86] Remove sse3 intrinsic tests from the avx intrinsics test file.

They are all covered by the SSE3 intrinsics test with SSE2, AVX, and AVX512 command lines.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295705 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoThe patch introduces new way of narrowing complex (>UINT16 variants) solutions.
Evgeny Stupachenko [Tue, 21 Feb 2017 07:34:40 +0000 (07:34 +0000)]
The patch introduces new way of narrowing complex (>UINT16 variants) solutions.
The new method introduced under "-lsr-exp-narrow" option (currenlty set to true).

Summary:

The method is based on registers number mathematical expectation and should be
 generally closer to optimal solution.
Please see details in comments to
 "LSRInstance::NarrowSearchSpaceByDeletingCostlyFormulas()" function
 (in lib/Transforms/Scalar/LoopStrengthReduce.cpp).

Reviewers: qcolombet

Differential Revision: http://reviews.llvm.org/D29862

From: Evgeny Stupachenko <evstupac@gmail.com>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295704 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Remove aes intrinsic tests from the avx intrinsics test file.
Craig Topper [Tue, 21 Feb 2017 07:32:18 +0000 (07:32 +0000)]
[X86] Remove aes intrinsic tests from the avx intrinsics test file.

They are all covered by the AES intrinsics test with a legacy command line and an AVX command line.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295702 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Add an AVX command line and regenerate AES intrinsics test using the update_llc...
Craig Topper [Tue, 21 Feb 2017 07:32:14 +0000 (07:32 +0000)]
[X86] Add an AVX command line and regenerate AES intrinsics test using the update_llc_test_checks.py

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295701 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Remove sse2 intrinsic tests from the avx intrinsics test file.
Craig Topper [Tue, 21 Feb 2017 07:32:11 +0000 (07:32 +0000)]
[X86] Remove sse2 intrinsic tests from the avx intrinsics test file.

They are all covered by the SSE2 intrinsics test with SSE2, AVX, and AVX512 command lines.

Also remove an unneeded lfence intrinsic test since it was already covered.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295700 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Remove sse1 intrinsic tests from the avx intrinsics test file.
Craig Topper [Tue, 21 Feb 2017 07:32:03 +0000 (07:32 +0000)]
[X86] Remove sse1 intrinsic tests from the avx intrinsics test file.

They are all covered by the SSE intrinsics test with SSE, AVX, and AVX512 command lines.

Also remove an unneeded sfence intrinsic test since it was already covered.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295699 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Use SHLD with both inputs from the same register to implement rotate on Sandy...
Craig Topper [Tue, 21 Feb 2017 06:39:13 +0000 (06:39 +0000)]
[X86] Use SHLD with both inputs from the same register to implement rotate on Sandy Bridge and later Intel CPUs

Summary:
Sandy Bridge and later CPUs have better throughput using a SHLD to implement rotate versus the normal rotate instructions. Additionally it saves one uop and avoids a partial flag update dependency.

This patch implements this change on any Sandy Bridge or later processor without BMI2 instructions. With BMI2 we will use RORX as we currently do.

Reviewers: zvi

Reviewed By: zvi

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D30181

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295697 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Fix formatting. NFC
Craig Topper [Tue, 21 Feb 2017 06:27:13 +0000 (06:27 +0000)]
[X86] Fix formatting. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295695 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVX-512] Use sse_load_f32/f64 in place of scalar_to_vector and scalar load in some...
Craig Topper [Tue, 21 Feb 2017 04:26:10 +0000 (04:26 +0000)]
[AVX-512] Use sse_load_f32/f64 in place of scalar_to_vector and scalar load in some patterns.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295693 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVX-512] Add test cases showing failure to fold zero extending scalar loads in scala...
Craig Topper [Tue, 21 Feb 2017 04:26:07 +0000 (04:26 +0000)]
[AVX-512] Add test cases showing failure to fold zero extending scalar loads in scalar intrinsics without the peephole pass.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295692 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVX-512] Fix the ExeDomain for vcmpss/vcmpsd.
Craig Topper [Tue, 21 Feb 2017 04:26:04 +0000 (04:26 +0000)]
[AVX-512] Fix the ExeDomain for vcmpss/vcmpsd.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295691 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ValueTracking] clang-format a section I'm about to touch; NFC
Sanjoy Das [Tue, 21 Feb 2017 02:42:42 +0000 (02:42 +0000)]
[ValueTracking] clang-format a section I'm about to touch; NFC

(Whitespace only change)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295690 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoScheduleDAG: Cleanup; NFC
Matthias Braun [Tue, 21 Feb 2017 01:27:33 +0000 (01:27 +0000)]
ScheduleDAG: Cleanup; NFC

- Fix doxygen comments (do not repeat documented name, remove definition
    comment if there is already one at the declaration, add \p, ...)
- Add some const modifiers
- Use range based for

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295688 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoSubtargetFeature: Cleanup; NFC
Matthias Braun [Tue, 21 Feb 2017 01:27:29 +0000 (01:27 +0000)]
SubtargetFeature: Cleanup; NFC

- Fix doxygen comments
- Remove duplicated comments
- Remove section comments (which became wrong over time)
- Use more `const` and references but less `auto`

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295687 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd a wrapper around copy_if in STLExtras; NFC
Sanjoy Das [Tue, 21 Feb 2017 00:38:44 +0000 (00:38 +0000)]
Add a wrapper around copy_if in STLExtras; NFC

I will add one more use for this in a later change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295685 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[BranchFolding] Update debug location along with the update of branch instruction.
Taewook Oh [Tue, 21 Feb 2017 00:12:38 +0000 (00:12 +0000)]
[BranchFolding] Update debug location along with the update of branch instruction.

Summary:
Currently, BranchFolder drops DebugLoc for branch instructions in some places. For example, for the test code attached, the branch instruction of 'entry' block has a DILocation of

```
!12 = !DILocation(line: 6, column: 3, scope: !11)
```

, but this information is gone when then block is lowered because BranchFolder misses it. This patch is a fix for this issue.

Reviewers: qcolombet, aprantl, craig.topper, MatzeB

Reviewed By: aprantl

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D29902

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295684 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Add additonal check lines to one of the rotate tests.
Craig Topper [Mon, 20 Feb 2017 23:38:51 +0000 (23:38 +0000)]
[X86] Add additonal check lines to one of the rotate tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295682 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] FileCheckize one of the rotate tests.
Craig Topper [Mon, 20 Feb 2017 23:38:48 +0000 (23:38 +0000)]
[X86] FileCheckize one of the rotate tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295681 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[IndVars] Add an assert
Sanjoy Das [Mon, 20 Feb 2017 23:37:11 +0000 (23:37 +0000)]
[IndVars] Add an assert

We've already checked that the loop is in simplify form before, but a
little paranoia never hurt anyone.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295680 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[IR/Verifier] List the CU we weren't able to find in `llvm.dbg.cu`.
Davide Italiano [Mon, 20 Feb 2017 22:51:42 +0000 (22:51 +0000)]
[IR/Verifier] List the CU we weren't able to find in `llvm.dbg.cu`.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295678 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMemorySSA: Add support for renaming uses in the updater.
Daniel Berlin [Mon, 20 Feb 2017 22:26:03 +0000 (22:26 +0000)]
MemorySSA: Add support for renaming uses in the updater.

Summary:
This lets one add aliasing stores to the updater.
(i'm next going to move the creation/etc functions to the updater)

Reviewers: george.burgess.iv

Subscribers: llvm-commits, Prazek

Differential Revision: https://reviews.llvm.org/D30154

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295677 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] FileCheckize one of the rotate tests.
Craig Topper [Mon, 20 Feb 2017 19:44:10 +0000 (19:44 +0000)]
[X86] FileCheckize one of the rotate tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295676 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix use-after-free found by ASAN
Steven Wu [Mon, 20 Feb 2017 18:33:40 +0000 (18:33 +0000)]
Fix use-after-free found by ASAN

DenseMap::lookup returns copy of the value in the map. Returning the
address of the temporary return value will cause use-after-free.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295675 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVX-512] Add a few more patterns for selecting masked vpternlog with broadcast loads...
Craig Topper [Mon, 20 Feb 2017 17:44:09 +0000 (17:44 +0000)]
[AVX-512] Add a few more patterns for selecting masked vpternlog with broadcast loads where the passthru operand is not operand 0.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295673 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Tidyup combineExtractVectorElt. NFCI.
Simon Pilgrim [Mon, 20 Feb 2017 16:09:45 +0000 (16:09 +0000)]
[X86] Tidyup combineExtractVectorElt. NFCI.

Pull out repeated code for extraction index operand and source vector value type.

Use isNullConstant helper to check for zero extraction index.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295670 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][SSE] Regenerate extracted bitcasted constant tests and add 32-bit test target
Simon Pilgrim [Mon, 20 Feb 2017 15:57:14 +0000 (15:57 +0000)]
[X86][SSE] Regenerate extracted bitcasted constant tests and add 32-bit test target

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295669 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[globalisel] OperandPredicateMatcher's shouldn't need to generate the MachineOperand...
Daniel Sanders [Mon, 20 Feb 2017 15:30:43 +0000 (15:30 +0000)]
[globalisel] OperandPredicateMatcher's shouldn't need to generate the MachineOperand expr. NFC

Summary:
Each OperandPredicateMatcher shouldn't need to know how to generate the expression
to reference a MachineOperand. The OperandMatcher should provide it.

In addition to separating responsibilities, this also lays some groundwork for
decoupling source patterns from destination patterns to allow invented operands
or operands provided by GlobalISel's equivalent to the ComplexPattern<> class.

Depends on D29709

Reviewers: t.p.northover, ab, rovka, qcolombet, aditya_nandakumar

Reviewed By: ab

Subscribers: dberris, kristof.beyls, llvm-commits, igorb

Differential Revision: https://reviews.llvm.org/D29710

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295668 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][SSE] Regenerate re-materialized store tests and add 64-bit test target
Simon Pilgrim [Mon, 20 Feb 2017 15:20:37 +0000 (15:20 +0000)]
[X86][SSE] Regenerate re-materialized store tests and add 64-bit test target

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295666 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][SSE] Regenerate vselect widening tests and add 32-bit test target
Simon Pilgrim [Mon, 20 Feb 2017 15:16:43 +0000 (15:16 +0000)]
[X86][SSE] Regenerate vselect widening tests and add 32-bit test target

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295665 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM] GlobalISel: Don't select atomic loads
Diana Picus [Mon, 20 Feb 2017 14:45:58 +0000 (14:45 +0000)]
[ARM] GlobalISel: Don't select atomic loads

There used to be a check in the IRTranslator that prevented us from having to
deal with atomic loads/stores. That check has been removed in r294993 and the
AArch64 backend was updated accordingly. This commit does the same thing for the
ARM backend.

In general, in the ARM backend we introduce fences during the atomic expand
pass, so we don't have to worry about atomics, *except* for the 32-bit ARMv8
target, which handles atomics more like AArch64. Since we don't want to worry
about that yet, just bail out of instruction selection if we find any atomic
loads.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295662 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[globalisel] Separate the SelectionDAG importer from the emitter. NFC
Daniel Sanders [Mon, 20 Feb 2017 14:31:27 +0000 (14:31 +0000)]
[globalisel] Separate the SelectionDAG importer from the emitter. NFC

Summary:
In the near future the rules will be sorted between these two steps to
ensure that more important rules are not prevented by less important ones.

Reviewers: t.p.northover, ab, rovka, qcolombet, aditya_nandakumar

Reviewed By: ab

Subscribers: dberris, kristof.beyls, llvm-commits

Differential Revision: https://reviews.llvm.org/D29709

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295661 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Fix EXTRACT_VECTOR_ELT with variable index from v32i16 and v64i8 vector.
Igor Breger [Mon, 20 Feb 2017 14:16:29 +0000 (14:16 +0000)]
[X86] Fix EXTRACT_VECTOR_ELT with variable index from v32i16 and v64i8 vector.

Its more profitable to go through memory (1 cycles throughput)
than using VMOVD + VPERMV/PSHUFB sequence ( 2/3 cycles throughput) to implement EXTRACT_VECTOR_ELT with variable index.
IACA tool was used to get performace estimation (https://software.intel.com/en-us/articles/intel-architecture-code-analyzer)
For example for var_shuffle_v16i8_v16i8_xxxxxxxxxxxxxxxx_i8 test from vector-shuffle-variable-128.ll I get 26 cycles vs 79 cycles.
Removing the VINSERT node, we don't need it any more.

Differential Revision: https://reviews.llvm.org/D29690

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295660 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[SLP] Additional test for vectorization of cal/invoke args vectorization
Alexey Bataev [Mon, 20 Feb 2017 12:41:16 +0000 (12:41 +0000)]
[SLP] Additional test for vectorization of cal/invoke args vectorization

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295657 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX512] Add support for ASHR v2i64/v4i64 support without VLX
Simon Pilgrim [Mon, 20 Feb 2017 12:16:38 +0000 (12:16 +0000)]
[X86][AVX512] Add support for ASHR v2i64/v4i64 support without VLX

Use v8i64 ASHR instructions if we don't have VLX.

Differential Revision: https://reviews.llvm.org/D28537

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@295656 91177308-0d34-0410-b5e6-96231b3b80d8