OSDN Git Service

android-x86/external-llvm.git
7 years agoStatistic/Timer: Include timers in PrintStatisticsJSON().
Matthias Braun [Fri, 18 Nov 2016 19:43:24 +0000 (19:43 +0000)]
Statistic/Timer: Include timers in PrintStatisticsJSON().

Differential Revision: https://reviews.llvm.org/D25588

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287370 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTimer: Track name and description.
Matthias Braun [Fri, 18 Nov 2016 19:43:18 +0000 (19:43 +0000)]
Timer: Track name and description.

The previously used "names" are rather descriptions (they use multiple
words and contain spaces), use short programming language identifier
like strings for the "names" which should be used when exporting to
machine parseable formats.

Also removed a unused TimerGroup from Hexxagon.

Differential Revision: https://reviews.llvm.org/D25583

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287369 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[MIRPrinter] Print raw branch probabilities as expected by MIRParser
Geoff Berry [Fri, 18 Nov 2016 19:37:24 +0000 (19:37 +0000)]
[MIRPrinter] Print raw branch probabilities as expected by MIRParser

Fixes PR28751.

Reviewers: MatzeB, qcolombet

Subscribers: mcrosier, llvm-commits

Differential Revision: https://reviews.llvm.org/D26775

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287368 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Fix unused variable warning
Matt Arsenault [Fri, 18 Nov 2016 18:33:36 +0000 (18:33 +0000)]
AMDGPU: Fix unused variable warning

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287362 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix test from r287353: don't use /dev/null
Hans Wennborg [Fri, 18 Nov 2016 18:27:31 +0000 (18:27 +0000)]
Fix test from r287353: don't use /dev/null

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287360 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[LTO] Add option to generate optimization records
Adam Nemet [Fri, 18 Nov 2016 18:06:28 +0000 (18:06 +0000)]
[LTO] Add option to generate optimization records

It is used to drive this from the clang driver via -mllvm.

Same option name is used as in opt.

Differential Revision: https://reviews.llvm.org/D26832

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287356 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[DebugInfo] Fix some Clang-tidy modernize-use-default, modernize-use-equal-delete...
Eugene Zelenko [Fri, 18 Nov 2016 18:00:19 +0000 (18:00 +0000)]
[DebugInfo] Fix some Clang-tidy modernize-use-default, modernize-use-equal-delete and Include What You Use warnings; other minor fixes (NFC).

Per Zachary Turner and Mehdi Amini suggestion to make only post-commit reviews.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287355 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoIRMover: Avoid accidentally mapping types from the destination module (PR30799)
Hans Wennborg [Fri, 18 Nov 2016 17:33:05 +0000 (17:33 +0000)]
IRMover: Avoid accidentally mapping types from the destination module (PR30799)

During Module linking, it's possible for SrcM->getIdentifiedStructTypes();
to return types that are actually defined in the destination module
(DstM). Depending on how the bitcode file was read,
getIdentifiedStructTypes() might do a walk over all values, including
metadata nodes, looking for types. In my case, a debug info metadata
node was shared between the two modules, and it referred to a type
defined in the destination module (see test case).

Differential Revision: https://reviews.llvm.org/D26212

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287353 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd link-time detection of LLVM_ABI_BREAKING_CHECKS mismatch
Mehdi Amini [Fri, 18 Nov 2016 17:28:10 +0000 (17:28 +0000)]
Add link-time detection of LLVM_ABI_BREAKING_CHECKS mismatch

Summary:
LLVM will define a symbol, either EnableABIBreakingChecks or
DisableABIBreakingChecks depending on the configuration setting for
LLVM_ABI_BREAKING_CHECKS.

The llvm-config.h header will add weak references to these symbols in
every clients that includes this header. This should ensure that
a mismatch triggers a link failure (or a load time failure for DSO).

On MSVC, the pragma "detect_mismatch" is used instead.

Reviewers: rnk, jroelofs

Subscribers: llvm-commits, mgorny

Differential Revision: https://reviews.llvm.org/D26841

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287352 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[PPC] limit line width to 80 characters
Ehsan Amiri [Fri, 18 Nov 2016 16:24:27 +0000 (16:24 +0000)]
[PPC] limit line width to 80 characters

NFC. Forgot to fix this in the original commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287350 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[mips][msa] Implement f16 support
Simon Dardis [Fri, 18 Nov 2016 16:17:44 +0000 (16:17 +0000)]
[mips][msa] Implement f16 support

The MIPS MSA ASE provides instructions to convert to and from half precision
floating point. This patch teaches the MIPS backend to treat f16 as a legal
type and how to promote such values to f32 for the usual set of operations.

As a result of this, the fexup[lr].w intrinsics no longer crash LLVM during
type legalization.

Reviewers: zoran.jovanvoic, vkalintiris

Differential Revision: https://reviews.llvm.org/D26398

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287349 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX512] Split AVX512F/AVX512VL tests to demonstrate missed int2fp opportunities...
Simon Pilgrim [Fri, 18 Nov 2016 15:31:36 +0000 (15:31 +0000)]
[X86][AVX512] Split AVX512F/AVX512VL tests to demonstrate missed int2fp opportunities without AVX512VL

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287348 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoGlobalISel: Fix unconditional fallback with global isel abort is disabled
Tom Stellard [Fri, 18 Nov 2016 14:14:35 +0000 (14:14 +0000)]
GlobalISel: Fix unconditional fallback with global isel abort is disabled

Reviewers: t.p.northover, ab, qcolombet

Subscribers: mehdi_amini, vkalintiris, wdng, dberris, llvm-commits, rovka

Differential Revision: https://reviews.llvm.org/D26765

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287344 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU/SI: Remove zero_extend patterns for i16 ops selected to 32-bit insts
Tom Stellard [Fri, 18 Nov 2016 13:53:34 +0000 (13:53 +0000)]
AMDGPU/SI: Remove zero_extend patterns for i16 ops selected to 32-bit insts

Summary:
The 32-bit instructions don't zero the high 16-bits like the 16-bit
instructions do.

Reviewers: arsenm

Subscribers: kzhuravl, wdng, nhaehnle, yaxunl, llvm-commits, tony-tye

Differential Revision: https://reviews.llvm.org/D26828

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287342 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[simplifycfg][loop-simplify] Preserve loop metadata in 2 transformations.
Florian Hahn [Fri, 18 Nov 2016 13:12:07 +0000 (13:12 +0000)]
[simplifycfg][loop-simplify] Preserve loop metadata in 2 transformations.

insertUniqueBackedgeBlock in lib/Transforms/Utils/LoopSimplify.cpp now
propagates existing llvm.loop metadata to newly the added backedge.

llvm::TryToSimplifyUncondBranchFromEmptyBlock in lib/Transforms/Utils/Local.cpp
now propagates existing llvm.loop metadata to the branch instructions in the
predecessor blocks of the empty block that is removed.

Differential Revision: https://reviews.llvm.org/D26495

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287341 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoCleanup function with clang-format. NFCI.
Simon Pilgrim [Fri, 18 Nov 2016 12:16:18 +0000 (12:16 +0000)]
Cleanup function with clang-format. NFCI.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287340 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Fix legalization of MUBUF instructions in shaders
Nicolai Haehnle [Fri, 18 Nov 2016 11:55:52 +0000 (11:55 +0000)]
AMDGPU: Fix legalization of MUBUF instructions in shaders

Summary:
The addr64-based legalization is incorrect for MUBUF instructions with idxen
set as well as for BUFFER_LOAD/STORE_FORMAT_* instructions.  This affects
e.g.  shaders that access buffer textures.

Since we never actually need the addr64-legalization in shaders, this patch
takes the easy route and keys off the calling convention.  If this ever
affects (non-OpenGL) compute, the type of legalization needs to be chosen
based on some TSFlag.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98664

Reviewers: arsenm, tstellarAMD

Subscribers: kzhuravl, wdng, yaxunl, tony-tye, llvm-commits

Differential Revision: https://reviews.llvm.org/D26747

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287339 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix spelling mistakes in MIPS target comments. NFC.
Simon Pilgrim [Fri, 18 Nov 2016 11:53:36 +0000 (11:53 +0000)]
Fix spelling mistakes in MIPS target comments. NFC.

Identified by Pedro Giffuni in PR27636.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287338 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Power9] Add patterns for vnegd, vnegw
Ehsan Amiri [Fri, 18 Nov 2016 11:05:55 +0000 (11:05 +0000)]
[Power9] Add patterns for vnegd, vnegw

Exploit new instructions by adding patterns to .td file.
https://reviews.llvm.org/D26551

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287334 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix spelling mistakes in AMDGPU target comments. NFC.
Simon Pilgrim [Fri, 18 Nov 2016 11:04:02 +0000 (11:04 +0000)]
Fix spelling mistakes in AMDGPU target comments. NFC.

Identified by Pedro Giffuni in PR27636.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287333 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX2] Add v8i32->v8i64 mul test (PR30845)
Simon Pilgrim [Fri, 18 Nov 2016 11:00:36 +0000 (11:00 +0000)]
[X86][AVX2] Add v8i32->v8i64 mul test (PR30845)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287332 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix typo in comment. NFC.
Simon Pilgrim [Fri, 18 Nov 2016 10:52:12 +0000 (10:52 +0000)]
Fix typo in comment. NFC.

Identified by Pedro Giffuni in PR27636.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287331 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[PPC][DAGCombine] Convert SETCC to subtract when the result is zero extended
Ehsan Amiri [Fri, 18 Nov 2016 10:41:44 +0000 (10:41 +0000)]
[PPC][DAGCombine] Convert SETCC to subtract when the result is zero extended

When we see a SETCC whose only users are zero extend operations, we can replace
it with a subtraction. This results in doing all calculations in GPRs and
avoids CR use.

Currently we do this only for ULT, ULE, UGT and UGE condition codes. There are
ways that this can be extended. For example for signed condition codes. In that
case we will be introducing additional sign extend instructions, so more careful
profitability analysis may be required.

Another direction to extend this is for equal, not equal conditions. Also when
users of SETCC are any_ext or sign_ext, we might be able to do something
similar.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287329 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix go binding to adapt the new attribute API
Amaury Sechet [Fri, 18 Nov 2016 10:11:02 +0000 (10:11 +0000)]
Fix go binding to adapt the new attribute API

https://reviews.llvm.org/D26339

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287328 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[InstCombine][AVX-512] Teach InstCombineCalls how to handle the intrinsics for variab...
Craig Topper [Fri, 18 Nov 2016 06:04:33 +0000 (06:04 +0000)]
[InstCombine][AVX-512] Teach InstCombineCalls how to handle the intrinsics for variable shift with 16-bit elements.

This is a straightforward extension of the existing support for 32/64-bit element types. Just needed to add the additional instrinsics to the switches.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287316 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVX-512] Replace masked 16-bit element variable shift intrinsics with new unmasked...
Craig Topper [Fri, 18 Nov 2016 05:04:44 +0000 (05:04 +0000)]
[AVX-512] Replace masked 16-bit element variable shift intrinsics with new unmasked versions and selects.

The same thing was done to 32-bit and 64-bit element sizes previously.

This will allow us to support these shuffls in InstCombineCalls along with the other variable shift intrinsics.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287312 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Move redundant setting of inst properties
Matt Arsenault [Fri, 18 Nov 2016 04:42:59 +0000 (04:42 +0000)]
AMDGPU: Move redundant setting of inst properties

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287311 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Fix crash on illegal type for inlineasm
Matt Arsenault [Fri, 18 Nov 2016 04:42:57 +0000 (04:42 +0000)]
AMDGPU: Fix crash on illegal type for inlineasm

There are still crashes on non-MVT types in other
places.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287310 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoObject: Simplify; remove unnecessary use of unique_ptr.
Peter Collingbourne [Fri, 18 Nov 2016 03:20:36 +0000 (03:20 +0000)]
Object: Simplify; remove unnecessary use of unique_ptr.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287305 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMachineOperand: Add dump() method
Matthias Braun [Fri, 18 Nov 2016 02:40:40 +0000 (02:40 +0000)]
MachineOperand: Add dump() method

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287302 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoconvert bpf assembler to look like kernel verifier output
Alexei Starovoitov [Fri, 18 Nov 2016 02:32:35 +0000 (02:32 +0000)]
convert bpf assembler to look like kernel verifier output

since bpf instruction set was introduced people learned to
read and understand kernel verifier output whereas llvm asm
output stayed obscure and unknown. Convert llvm to emit
assembler text similar to kernel to avoid this discrepancy

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287300 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Docs][TableGen] Remove reference to tablegen supporting octal integers. It doesn...
Craig Topper [Fri, 18 Nov 2016 02:28:50 +0000 (02:28 +0000)]
[Docs][TableGen] Remove reference to tablegen supporting octal integers. It doesn't and hasn't for at least 9 years.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287299 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVX-512] Support FCOPYSIGN for v16f32 and v8f64
Craig Topper [Fri, 18 Nov 2016 02:25:34 +0000 (02:25 +0000)]
[AVX-512] Support FCOPYSIGN for v16f32 and v8f64

Summary:
This extends FCOPYSIGN support to 512-bit vectors.

I've also added tests to show what the 128-bit and 256-bit cases look like with broadcast loads.

Reviewers: delena, zvi, RKSimon, spatel

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D26791

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287298 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd an option to disable libedit
Yichao Yu [Fri, 18 Nov 2016 01:25:49 +0000 (01:25 +0000)]
Add an option to disable libedit

Summary: This should provide the function similar to `--disable-libedit` with the autotools build system, which seems to be missing from the commit (r200595) that adds this.

Reviewers: pcc, beanz

Subscribers: mgorny, llvm-commits

Differential Revision: https://reviews.llvm.org/D26550

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287293 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CUDA] Update docs to indicate that MacOS is now supported.
Justin Lebar [Fri, 18 Nov 2016 00:42:00 +0000 (00:42 +0000)]
[CUDA] Update docs to indicate that MacOS is now supported.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287290 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CUDA] Update docs; CUDA 8.0 is supported as of a while ago.
Justin Lebar [Fri, 18 Nov 2016 00:41:40 +0000 (00:41 +0000)]
[CUDA] Update docs; CUDA 8.0 is supported as of a while ago.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287289 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[lli] Prefer `exit(1)` to `return 1` for consistency.
Davide Italiano [Thu, 17 Nov 2016 22:59:13 +0000 (22:59 +0000)]
[lli] Prefer `exit(1)` to `return 1` for consistency.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287277 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[lli] Factor out error handling. NFCI.
Davide Italiano [Thu, 17 Nov 2016 22:58:13 +0000 (22:58 +0000)]
[lli] Factor out error handling. NFCI.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287276 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ReleaseNotes] Mention the completion of the upstreaming of the AVR backend
Dylan McKay [Thu, 17 Nov 2016 22:26:09 +0000 (22:26 +0000)]
[ReleaseNotes] Mention the completion of the upstreaming of the AVR backend

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287273 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CMake] Error when LTO and lld are enabled on Darwin
Petr Hosek [Thu, 17 Nov 2016 20:22:49 +0000 (20:22 +0000)]
[CMake] Error when LTO and lld are enabled on Darwin

lld on Darwin does not currently support LTO.

Differential Revision: https://reviews.llvm.org/D26715

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287256 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix spelling mistakes in Hexagon target comments. NFC.
Simon Pilgrim [Thu, 17 Nov 2016 19:21:20 +0000 (19:21 +0000)]
Fix spelling mistakes in Hexagon target comments. NFC.

Identified by Pedro Giffuni in PR27636.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287248 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix spelling mistakes in X86 target comments. NFC.
Simon Pilgrim [Thu, 17 Nov 2016 19:03:05 +0000 (19:03 +0000)]
Fix spelling mistakes in X86 target comments. NFC.

Identified by Pedro Giffuni in PR27636.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287247 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeView] Fix some Clang-tidy modernize-use-default, modernize-use-override and...
Eugene Zelenko [Thu, 17 Nov 2016 18:11:21 +0000 (18:11 +0000)]
[CodeView] Fix some Clang-tidy modernize-use-default, modernize-use-override and Include What You Use warnings; other minor fixes (NFC).

Per Zachary Turner and Mehdi Amini suggestion to make only post-commit reviews.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287243 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[libFuzzer] better documentation for -fsanitize-coverage=trace-cmp
Kostya Serebryany [Thu, 17 Nov 2016 17:31:54 +0000 (17:31 +0000)]
[libFuzzer] better documentation for -fsanitize-coverage=trace-cmp

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287240 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[asan] Turn on Mach-O global metadata liveness tracking by default
Anna Zaks [Thu, 17 Nov 2016 16:55:40 +0000 (16:55 +0000)]
[asan] Turn on Mach-O global metadata liveness tracking by default

This patch turns on the metadata liveness tracking since all known issues
have been resolved. The future has been implemented in
https://reviews.llvm.org/D16737 and enables support of dead code stripping
option on Mach-O platforms.

As part of enabling the feature, I also plan on reverting the following
patch to compiler-rt:

http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20160704/369910.html

Differential Revision: https://reviews.llvm.org/D26772

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287235 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "AMDGPU: Enable ConstrainCopy DAG mutation"
Konstantin Zhuravlyov [Thu, 17 Nov 2016 16:41:49 +0000 (16:41 +0000)]
Revert "AMDGPU: Enable ConstrainCopy DAG mutation"

This reverts commit r287146.

This breaks few conformance tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287233 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[SCEV] limit recursion depth of CompareSCEVComplexity
Daniil Fukalov [Thu, 17 Nov 2016 16:07:52 +0000 (16:07 +0000)]
[SCEV] limit recursion depth of CompareSCEVComplexity

Summary:
CompareSCEVComplexity goes too deep (50+ on a quite a big unrolled loop) and runs almost infinite time.

Added cache of "equal" SCEV pairs to earlier cutoff of further estimation. Recursion depth limit was also introduced as a parameter.

Reviewers: sanjoy

Subscribers: mzolotukhin, tstellarAMD, llvm-commits

Differential Revision: https://reviews.llvm.org/D26389

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287232 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoWdocumentation fix
Simon Pilgrim [Thu, 17 Nov 2016 12:21:45 +0000 (12:21 +0000)]
Wdocumentation fix

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287224 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][SSE] Improve lowering of vXi64 multiply with known zero 32-bit halves
Simon Pilgrim [Thu, 17 Nov 2016 12:14:49 +0000 (12:14 +0000)]
[X86][SSE] Improve lowering of vXi64 multiply with known zero 32-bit halves

vXi64 multiplication is lowered into 3 calls of vpmuludq with the upper/lower 32-bit halves.

If any of these halves are zero then we can remove individual calls. Although there was isBuildVectorAllZeros code to do this I don't think it ever worked (maybe just for constant folded cases that don't seem to be tested for any longer).

This requires additional X86ISD support for computeKnownBitsForTargetNode, so far I've just added support for X86ISD::VZEXT (VPMOVZX* - helping the AVX2+ cases).

Partial fix for PR30845

Differential Revision: https://reviews.llvm.org/D26590

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287223 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix spelling in comment. NFC.
Simon Pilgrim [Thu, 17 Nov 2016 12:03:05 +0000 (12:03 +0000)]
Fix spelling in comment. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287222 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[cmake] Move LLVM_BUILD_STATIC check to an earlier point
Pavel Labath [Thu, 17 Nov 2016 11:22:23 +0000 (11:22 +0000)]
[cmake] Move LLVM_BUILD_STATIC check to an earlier point

Summary:
The motivation for this is to enable correct detection of dlopen() on Android.
Android does not provide a static version of libdl, so if we add the -static flag
after performing the check, it will succeed even though subsequent link steps
will fail. With this change we correctly detect the absence of libdl in a
LLVM_BUILD_STATIC build on Android.

The link itself still does not succeed because the code does not check the result
of this check properly, but I plan to fix that in a separate change.

Reviewers: beanz

Subscribers: danalbert, mgorny, srhines, tberghammer, llvm-commits

Differential Revision: https://reviews.llvm.org/D26463

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287220 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM] Relax restriction on variadic functions for tailcall optimization
Pablo Barrio [Thu, 17 Nov 2016 10:56:58 +0000 (10:56 +0000)]
[ARM] Relax restriction on variadic functions for tailcall optimization

Summary:
Variadic functions can be treated in the same way as normal functions
with respect to the number and types of parameters.

Reviewers: grosbach, olista01, t.p.northover, rengolin

Subscribers: javed.absar, aemerson, llvm-commits

Differential Revision: https://reviews.llvm.org/D26748

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287219 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] RegCall - Handling v64i1 in 32/64 bit target
Oren Ben Simhon [Thu, 17 Nov 2016 09:59:40 +0000 (09:59 +0000)]
[X86] RegCall - Handling v64i1 in 32/64 bit target

Register Calling Convention defines a new behavior for v64i1 types.
This type should be saved in GPR.
However for 32 bit machine we need to split the value into 2 GPRs (because each is 32 bit).

Differential Revision: https://reviews.llvm.org/D26181

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287217 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDelete dead code and add asserts instead; NFC
Sanjoy Das [Thu, 17 Nov 2016 07:29:43 +0000 (07:29 +0000)]
Delete dead code and add asserts instead; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287214 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ImplicitNullCheck] Fix an edge case where we were hoisting incorrectly
Sanjoy Das [Thu, 17 Nov 2016 07:29:40 +0000 (07:29 +0000)]
[ImplicitNullCheck] Fix an edge case where we were hoisting incorrectly

ImplicitNullCheck keeps track of one instruction that the memory
operation depends on that it also hoists with the memory operation.
When hoisting this dependency, it would sometimes clobber a live-in
value to the basic block we were hoisting the two things out of.  Fix
this by explicitly looking for such dependencies.

I also noticed two redundant checks on `MO.isDef()` in IsMIOperandSafe.
They're redundant since register MachineOperands are either Defs or Uses
-- there is no third kind.  I'll change the checks to asserts in a later
commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287213 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Fix formatting. NFC
Craig Topper [Thu, 17 Nov 2016 05:59:55 +0000 (05:59 +0000)]
[X86] Fix formatting. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287211 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Add a test case where, due to a bug in selectScalarSSELoad, we fold the same...
Craig Topper [Thu, 17 Nov 2016 05:37:39 +0000 (05:37 +0000)]
[X86] Add a test case where, due to a bug in selectScalarSSELoad, we fold the same load twice.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287210 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[XRay] Support AArch64 in LLVM
Dean Michael Berris [Thu, 17 Nov 2016 05:15:37 +0000 (05:15 +0000)]
[XRay] Support AArch64 in LLVM

This patch adds XRay support in LLVM for AArch64 targets.
This patch is one of a series:

Clang: https://reviews.llvm.org/D26415
compiler-rt: https://reviews.llvm.org/D26413

Author: rSerge

Reviewers: rengolin, dberris

Subscribers: amehsan, aemerson, llvm-commits, iid_iunknown

Differential Revision: https://reviews.llvm.org/D26412

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287209 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CMake] [Darwin] Add support for debugging tablegen dependencies
Chris Bieneman [Thu, 17 Nov 2016 04:36:59 +0000 (04:36 +0000)]
[CMake] [Darwin] Add support for debugging tablegen dependencies

This patch adds an option to the build system LLVM_DEPENDENCY_DEBUGGING. Over time I plan to extend this to do more complex verifications, but the initial patch causes compile errors wherever there is missing a dependency on intrinsics_gen.

Because intrinsics_gen is a compile-time dependency not a link-time dependency, everything that relies on the headers generated in intrinsics_gen needs an explicit dependency.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287207 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CMake] NFC. Updating CMake dependency specifications
Chris Bieneman [Thu, 17 Nov 2016 04:36:50 +0000 (04:36 +0000)]
[CMake] NFC. Updating CMake dependency specifications

This patch updates a bunch of places where add_dependencies was being explicitly called to add dependencies on intrinsics_gen to instead use the DEPENDS named parameter. This cleanup is needed for a patch I'm working on to add a dependency debugging mode to the build system.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287206 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AMDGPU] Add missing test for rL287203
Konstantin Zhuravlyov [Thu, 17 Nov 2016 04:33:20 +0000 (04:33 +0000)]
[AMDGPU] Add missing test for rL287203

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287204 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AMDGPU] Custom lower f16 = fp_round f64
Konstantin Zhuravlyov [Thu, 17 Nov 2016 04:28:37 +0000 (04:28 +0000)]
[AMDGPU] Custom lower f16 = fp_round f64

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287203 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AMDGPU] Promote f16/i16 conversions to f32/i32
Konstantin Zhuravlyov [Thu, 17 Nov 2016 04:00:46 +0000 (04:00 +0000)]
[AMDGPU] Promote f16/i16 conversions to f32/i32

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287201 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AMDGPU] Expand `br_cc` for f16
Konstantin Zhuravlyov [Thu, 17 Nov 2016 03:49:01 +0000 (03:49 +0000)]
[AMDGPU] Expand `br_cc` for f16

Differential Revision: https://reviews.llvm.org/D26732

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287199 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Orc] Clang-format the recent RPC update (r286620 and related).
Lang Hames [Thu, 17 Nov 2016 02:33:47 +0000 (02:33 +0000)]
[Orc] Clang-format the recent RPC update (r286620 and related).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287195 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUse profile info to adjust loop unroll threshold.
Dehao Chen [Thu, 17 Nov 2016 01:17:02 +0000 (01:17 +0000)]
Use profile info to adjust loop unroll threshold.

Summary:
For flat loop, even if it is hot, it is not a good idea to unroll in runtime, thus we set a lower partial unroll threshold.
For hot loop, we set a higher unroll threshold and allows expensive tripcount computation to allow more aggressive unrolling.

Reviewers: davidxl, mzolotukhin

Subscribers: sanjoy, mehdi_amini, llvm-commits

Differential Revision: https://reviews.llvm.org/D26527

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287186 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CUDA] Update docs to indicate that clang now supports std::complex in CUDA mode.
Justin Lebar [Thu, 17 Nov 2016 01:03:42 +0000 (01:03 +0000)]
[CUDA] Update docs to indicate that clang now supports std::complex in CUDA mode.

The last remaining necessary change was D25403, landed as r287012.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287184 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove a stale test case.
Lang Hames [Thu, 17 Nov 2016 01:02:52 +0000 (01:02 +0000)]
Remove a stale test case.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287183 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agollvm-dis: Remove dead code.
Peter Collingbourne [Thu, 17 Nov 2016 00:42:08 +0000 (00:42 +0000)]
llvm-dis: Remove dead code.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287182 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVR] Remove some accidentally-commited code that broke the bots
Dylan McKay [Thu, 17 Nov 2016 00:09:38 +0000 (00:09 +0000)]
[AVR] Remove some accidentally-commited code that broke the bots

This is a remnant of an on-chip unit testing tool that has since been
moved out-of-tree.

It was accidentally committed in r287162.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287180 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoIntroduce GlobalSplit pass.
Peter Collingbourne [Wed, 16 Nov 2016 23:40:26 +0000 (23:40 +0000)]
Introduce GlobalSplit pass.

This pass splits globals into elements using inrange annotations on
getelementptr indices.

Differential Revision: https://reviews.llvm.org/D22295

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287178 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVR] Wrap all methods in the pseudo expansion pass in an anon namespace
Dylan McKay [Wed, 16 Nov 2016 23:06:14 +0000 (23:06 +0000)]
[AVR] Wrap all methods in the pseudo expansion pass in an anon namespace

The '-fpermissive' compiler flag complains if the template
specializations used in the class are used in a different namespace.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287176 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVR] Fix basic block naming in ctlz and cttz tests
Dylan McKay [Wed, 16 Nov 2016 22:48:38 +0000 (22:48 +0000)]
[AVR] Fix basic block naming in ctlz and cttz tests

The branch selector would change the names.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287174 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVR] Remove unused method from AVRTargetMachine
Dylan McKay [Wed, 16 Nov 2016 22:48:30 +0000 (22:48 +0000)]
[AVR] Remove unused method from AVRTargetMachine

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287173 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVR] Add tests for counting leading/trailing zeros
Dylan McKay [Wed, 16 Nov 2016 22:38:43 +0000 (22:38 +0000)]
[AVR] Add tests for counting leading/trailing zeros

This adds two test files that verify the 'cttz' and 'ctlz' operations.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287172 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[x86] allow FP-logic ops when one operand is FP and result is FP
Sanjay Patel [Wed, 16 Nov 2016 22:34:05 +0000 (22:34 +0000)]
[x86] allow FP-logic ops when one operand is FP and result is FP

We save an inter-register file move this way. If there's any CPU where
the FP logic is slower, we could transform this back to int-logic in
MachineCombiner.

This helps, but doesn't solve, PR6137:
https://llvm.org/bugs/show_bug.cgi?id=6137

The 'andn' test shows that we're missing a pattern match to
recognize the xor with -1 constant as a 'not' op.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287171 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AsmParser] Avoid recursing when lexing ';'. NFC.
Ahmed Bougacha [Wed, 16 Nov 2016 22:25:05 +0000 (22:25 +0000)]
[AsmParser] Avoid recursing when lexing ';'. NFC.

This should prevent stack overflows in non-optimized builds on
.ll files with lots of consecutive commented-out lines.

Instead of recursing into LexToken(), continue into a 'while (true)'.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287170 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeGen] Pass references, not pointers, to MMI helpers. NFC.
Ahmed Bougacha [Wed, 16 Nov 2016 22:25:03 +0000 (22:25 +0000)]
[CodeGen] Pass references, not pointers, to MMI helpers. NFC.

While there, rename them to follow the coding style.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287169 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "Get GlobalISel to build on Linux after r286407"
Ahmed Bougacha [Wed, 16 Nov 2016 22:24:59 +0000 (22:24 +0000)]
Revert "Get GlobalISel to build on Linux after r286407"

This reverts commit r286962.

We want to avoid depending on SelectionDAG, and AddLandingPadInfo
lives in CodeGen now.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287168 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeGen] Pull MMI helpers from FunctionLoweringInfo to MMI. NFC.
Ahmed Bougacha [Wed, 16 Nov 2016 22:24:56 +0000 (22:24 +0000)]
[CodeGen] Pull MMI helpers from FunctionLoweringInfo to MMI. NFC.

They're not SelectionDAG- or FunctionLoweringInfo-specific.  They
are, however, specific to building MMI from IR.
We could make them members, but it's nice having MMI be a "simple" data
structure and this logic kept separate.

This also lets us reuse them from GlobalISel.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287167 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeGen] Cleanup MachineModuleInfo doxygen comments. NFC.
Ahmed Bougacha [Wed, 16 Nov 2016 22:24:53 +0000 (22:24 +0000)]
[CodeGen] Cleanup MachineModuleInfo doxygen comments. NFC.

Remove redundant names and only keep header comments.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287166 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeGen] Sort MMI forward declarations. NFC.
Ahmed Bougacha [Wed, 16 Nov 2016 22:24:46 +0000 (22:24 +0000)]
[CodeGen] Sort MMI forward declarations. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287165 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoGeneral clean up of error handling in llvm-objdump to remove its use of report_fatal_...
Kevin Enderby [Wed, 16 Nov 2016 22:17:38 +0000 (22:17 +0000)]
General clean up of error handling in llvm-objdump to remove its use of report_fatal_error().
No real functional change with this commit.

The problem with report_fatal_error() is it does not include the tool name
and the file name the for which the error message was generated.

Uses of report_fatal_error() were change to report_error() or error()
to get a better error and to make the code smaller and cleaner.

Also changed things like error(errorToErrorCode(SOrErr.takeError())) to
use report_error() with a file name and the llvm::Error (as well as the
ArchitectureName if available) so the error message is printed.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287163 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVR] Add the pseudo instruction expansion pass
Dylan McKay [Wed, 16 Nov 2016 21:58:04 +0000 (21:58 +0000)]
[AVR] Add the pseudo instruction expansion pass

Summary:
A lot of the pseudo instructions are required because LLVM assumes that
all integers of the same size as the pointer size are legal. This means
that it will not currently expand 16-bit instructions to their 8-bit
variants because it thinks 16-bit types are legal for the operations.

This also adds all of the CodeGen tests that required the pass to run.

Reviewers: arsenm, kparzysz

Subscribers: wdng, mgorny, modocache, llvm-commits

Differential Revision: https://reviews.llvm.org/D26577

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287162 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix "isn't a prototype" warning
Vitaly Buka [Wed, 16 Nov 2016 21:51:39 +0000 (21:51 +0000)]
Fix "isn't a prototype" warning

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287161 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoX86: Simplify X86ISD::Wrapper operand checks. NFCI.
Peter Collingbourne [Wed, 16 Nov 2016 21:48:59 +0000 (21:48 +0000)]
X86: Simplify X86ISD::Wrapper operand checks. NFCI.

We only ever create TargetConstantPool, TargetJumpTable, TargetExternalSymbol,
TargetGlobalAddress, TargetGlobalTLSAddress, MCSymbol and TargetBlockAddress
nodes as operands of X86ISD::Wrapper nodes, so we can remove one check and
invert the other.

Also update the documentation comment for X86ISD::Wrapper.

Differential Revision: https://reviews.llvm.org/D26731

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287160 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ImplicitNullChecks] Do not not handle call MachineInstrs
Sanjoy Das [Wed, 16 Nov 2016 21:45:22 +0000 (21:45 +0000)]
[ImplicitNullChecks] Do not not handle call MachineInstrs

We don't track callee clobbered registers correctly, so avoid hoisting
across calls.

Note: for this bug to trigger we need a `readonly` call target, since we
already have logic to not hoist across potentially storing instructions
either.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287159 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoBitcode: Introduce initial multi-module reader API.
Peter Collingbourne [Wed, 16 Nov 2016 21:44:45 +0000 (21:44 +0000)]
Bitcode: Introduce initial multi-module reader API.

Implement getLazyBitcodeModule() and parseBitcodeFile() in terms of it.

Differential Revision: https://reviews.llvm.org/D26719

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287156 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoARM: fix CodeGen for 64-bit shifts.
Tim Northover [Wed, 16 Nov 2016 20:54:28 +0000 (20:54 +0000)]
ARM: fix CodeGen for 64-bit shifts.

One half of the shifts obviously needed conditional selection based on whether
the shift amount is more than 32-bits, but leaving the other half as the
natural shift isn't acceptable either: it's undefined behaviour to shift a
32-bit value by more than 31.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287149 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMake block placement deterministic
Rong Xu [Wed, 16 Nov 2016 20:50:06 +0000 (20:50 +0000)]
Make block placement deterministic

We fail to produce bit-to-bit matching stage2 and stage3 compiler in PGO
bootstrap build. The reason is because LoopBlockSet is of SmallPtrSet type
whose iterating order depends on the pointer value.

This patch fixes this issue by changing to use SmallSetVector.

Differential Revision: http://reviews.llvm.org/D26634

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287148 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[InstCombine] replace unreachable with assert and remove unreachable code; NFCI
Sanjay Patel [Wed, 16 Nov 2016 20:40:02 +0000 (20:40 +0000)]
[InstCombine] replace unreachable with assert and remove unreachable code; NFCI

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287147 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Enable ConstrainCopy DAG mutation
Matt Arsenault [Wed, 16 Nov 2016 20:35:23 +0000 (20:35 +0000)]
AMDGPU: Enable ConstrainCopy DAG mutation

This fixes a probably unintended divergence from the default
scheduler behavior.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287146 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[InstCombine] fix formatting and add FIXMEs to foldOperationIntoSelectOperand(); NFC
Sanjay Patel [Wed, 16 Nov 2016 20:18:34 +0000 (20:18 +0000)]
[InstCombine] fix formatting and add FIXMEs to foldOperationIntoSelectOperand(); NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287145 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AArch64] Handle vector types in replaceZeroVectorStore.
Geoff Berry [Wed, 16 Nov 2016 19:35:19 +0000 (19:35 +0000)]
[AArch64] Handle vector types in replaceZeroVectorStore.

Summary:
Extend replaceZeroVectorStore to handle more vector type stores,
floating point zero vectors and set alignment more accurately on split
stores.

This is a follow-up change to r286875.

This change fixes PR31038.

Reviewers: MatzeB

Subscribers: mcrosier, aemerson, llvm-commits, rengolin

Differential Revision: https://reviews.llvm.org/D26682

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287142 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[LoopVectorize] Fix for non-determinism in codegen
Mandeep Singh Grang [Wed, 16 Nov 2016 18:53:17 +0000 (18:53 +0000)]
[LoopVectorize] Fix for non-determinism in codegen

Summary: This patch fixes issues in codegen uncovered due to https://reviews.llvm.org/D26718

Reviewers: mssimpso

Subscribers: llvm-commits, mzolotukhin

Differential Revision: https://reviews.llvm.org/D26727

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287135 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU/SI: Avoid creating unnecessary copies in the SIFixSGPRCopies pass
Tom Stellard [Wed, 16 Nov 2016 18:42:17 +0000 (18:42 +0000)]
AMDGPU/SI: Avoid creating unnecessary copies in the SIFixSGPRCopies pass

Summary:
1. Don't try to copy values to and from the same register class.
2. Replace copies with of registers with immediate values with v_mov/s_mov
   instructions.

The main purpose of this change is to make MachineSink do a better job of
determining when it is beneficial to split a critical edge, since the pass
assumes that copies will become move instructions.

This prevents a regression in uniform-cfg.ll if we enable critical edge
splitting for AMDGPU.

Reviewers: arsenm

Subscribers: arsenm, kzhuravl, llvm-commits

Differential Revision: https://reviews.llvm.org/D23408

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287131 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ExecutionEngine] Fix examples build broken in r287126 and other Include What You...
Eugene Zelenko [Wed, 16 Nov 2016 18:32:58 +0000 (18:32 +0000)]
[ExecutionEngine] Fix examples build broken in r287126 and other Include What You Use warnings.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287130 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agofix comment formatting; NFC
Sanjay Patel [Wed, 16 Nov 2016 18:09:44 +0000 (18:09 +0000)]
fix comment formatting; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287127 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ExecutionEngine] Fix some Clang-tidy modernize-use-default, modernize-use-equals...
Eugene Zelenko [Wed, 16 Nov 2016 18:07:33 +0000 (18:07 +0000)]
[ExecutionEngine] Fix some Clang-tidy modernize-use-default, modernize-use-equals-delete and Include What You Use warnings; other minor fixes.

Differential revision: https://reviews.llvm.org/D26729

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287126 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[x86] add fake scalar FP logic instructions to ReplaceableInstrs to save some bytes
Sanjay Patel [Wed, 16 Nov 2016 17:42:40 +0000 (17:42 +0000)]
[x86] add fake scalar FP logic instructions to ReplaceableInstrs to save some bytes

We can replace "scalar" FP-bitwise-logic with other forms of bitwise-logic instructions.
Scalar SSE/AVX FP-logic instructions only exist in your imagination and/or the bowels of
compilers, but logically equivalent int, float, and double variants of bitwise-logic
instructions are reality in x86, and the float variant may be a shorter instruction
depending on which flavor (SSE or AVX) of vector ISA you have...so just prefer float all
the time.

This is a preliminary step towards solving PR6137:
https://llvm.org/bugs/show_bug.cgi?id=6137

Differential Revision:
https://reviews.llvm.org/D26712

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@287122 91177308-0d34-0410-b5e6-96231b3b80d8