platform/upstream/llvm.git
8 years agoAdd support of the latest Ubuntu (Yakkety Yak - 16.10)
Sylvestre Ledru [Tue, 19 Jul 2016 14:00:57 +0000 (14:00 +0000)]
Add support of the latest Ubuntu (Yakkety Yak - 16.10)

llvm-svn: 275975

8 years agoFix for failing bot sanitizer-x86_64-linux-fast after r275970
Dmitry Polukhin [Tue, 19 Jul 2016 13:35:15 +0000 (13:35 +0000)]
Fix for failing bot sanitizer-x86_64-linux-fast after r275970

More info http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-fast/builds/14774/steps/check-clang%20msan/logs/stdio

llvm-svn: 275974

8 years ago[AARCH64] Enable AARCH64 lit tests on windows dev machines
Simon Pilgrim [Tue, 19 Jul 2016 13:35:11 +0000 (13:35 +0000)]
[AARCH64] Enable AARCH64 lit tests on windows dev machines

As discussed on PR27654, this patch fixes the triples of a lot of aarch64 tests and enables lit tests on windows

This will hopefully help stop cases where windows developers break the aarch64 target

Differential Revision: https://reviews.llvm.org/D22191

llvm-svn: 275973

8 years agoFix build with gcc 6.
Rafael Espindola [Tue, 19 Jul 2016 12:33:46 +0000 (12:33 +0000)]
Fix build with gcc 6.

llvm-svn: 275972

8 years agoGet rid of VS2015 operator precedence warning. NFCI.
Simon Pilgrim [Tue, 19 Jul 2016 12:26:51 +0000 (12:26 +0000)]
Get rid of VS2015 operator precedence warning. NFCI.

llvm-svn: 275971

8 years agoDeprecated (legacy) string literal conversion to 'char *' causes strange overloading...
Dmitry Polukhin [Tue, 19 Jul 2016 11:29:16 +0000 (11:29 +0000)]
Deprecated (legacy) string literal conversion to 'char *' causes strange overloading resolution

It's a patch for PR28050. Seems like overloading resolution wipes out
the first standard conversion sequence (before user-defined conversion)
in case of deprecated string literal conversion.

Differential revision: https://reviews.llvm.org/D21228

Patch by Alexander Makarov

llvm-svn: 275970

8 years agoGPGPU: Disable invariant load hoisting for GPU code generation
Tobias Grosser [Tue, 19 Jul 2016 11:13:58 +0000 (11:13 +0000)]
GPGPU: Disable invariant load hoisting for GPU code generation

This simplifies the upcoming patches to add code generation for ScopStmts. Load
hoisting support will later be added in a separate commit. This commit will
be implicitly tested by the subsequent GPGPU changes.

llvm-svn: 275969

8 years ago[mips][ias] R_MIPS_GOT_(PAGE|OFST) do not need symbols
Daniel Sanders [Tue, 19 Jul 2016 10:58:06 +0000 (10:58 +0000)]
[mips][ias] R_MIPS_GOT_(PAGE|OFST) do not need symbols

Reviewers: sdardis

Subscribers: dsanders, llvm-commits, sdardis

Differential Revision: https://reviews.llvm.org/D22458

llvm-svn: 275968

8 years ago[mips] Correct label prefixes for N32 and N64.
Daniel Sanders [Tue, 19 Jul 2016 10:49:03 +0000 (10:49 +0000)]
[mips] Correct label prefixes for N32 and N64.

Summary:
N32 and N64 follow the standard ELF conventions (.L) whereas O32 uses its own
($).

This fixes the majority of object differences between -fintegrated-as and
-fno-integrated-as.

Reviewers: sdardis

Subscribers: dsanders, sdardis, llvm-commits

Differential Revision: https://reviews.llvm.org/D22412

llvm-svn: 275967

8 years ago[mips] Recognise the triple used by Debian stretch for mips64el.
Daniel Sanders [Tue, 19 Jul 2016 10:22:19 +0000 (10:22 +0000)]
[mips] Recognise the triple used by Debian stretch for mips64el.

Summary:
The triple used for this distribution is mips64el-linux-gnuabi64.

Reviewers: sdardis

Subscribers: sdardis, llvm-commits

Differential Revision: https://reviews.llvm.org/D22406

llvm-svn: 275966

8 years ago[ELF] Minimal PHDRS parser and section to segment assignment support
Eugene Leviant [Tue, 19 Jul 2016 09:25:43 +0000 (09:25 +0000)]
[ELF] Minimal PHDRS parser and section to segment assignment support

llvm-svn: 275965

8 years ago[InstCombine] Minor cleanup of cast simplification code [NFC]
Tobias Grosser [Tue, 19 Jul 2016 09:06:08 +0000 (09:06 +0000)]
[InstCombine] Minor cleanup of cast simplification code [NFC]

Summary:
This patch cleans up parts of InstCombine to raise its compliance with the LLVM coding standards and to increase its readability. The changes and according rationale are summarized in the following:

- Rename `ShouldOptimizeCast()` to `shouldOptimizeCast()` since functions should start with a lower case letter.

- Move `shouldOptimizeCast()` from InstCombineCasts.cpp to InstCombineAndOrXor.cpp since it's only used there.

- Simplify interface of `shouldOptimizeCast()`.

- Minor code style adaptions in `shouldOptimizeCast()`.

- Remove the documentation on the function definition of `shouldOptimizeCast()` since it just repeats the documentation on its declaration. Also enhance the documentation on its declaration with more information describing its intended use and make it doxygen-compliant.

- Change a comment in `foldCastedBitwiseLogic()` from `fold (logic (cast A), (cast B)) -> (cast (logic A, B))` to `fold logic(cast(A), cast(B)) -> cast(logic(A, B))` since the surrounding comments use this format.

- Remove comment `Only do this if the casts both really cause code to be generated.` in `foldCastedBitwiseLogic()` since it just repeats parts of the documentation of `shouldOptimizeCast()` and does not help to improve readability.

- Simplify the interface of `isEliminableCastPair()`.

- Removed the documentation on the function definition of `isEliminableCastPair()` which only contained obvious statements about its implementation. Instead added more general doxygen-compliant documentation to its declaration.

- Renamed parameter `DoXform` of `transformZExtIcmp()` to `DoTransform` to make its intention clearer.

- Moved documentation of `transformZExtIcmp()` from its definition to its declaration and made it doxygen-compliant.

Reviewers: vtjnash, grosser

Subscribers: majnemer, llvm-commits

Differential Revision: https://reviews.llvm.org/D22449

Contributed-by: Matthias Reisinger
llvm-svn: 275964

8 years agoStyle: drop some unnecessary ';' [NFC]
Tobias Grosser [Tue, 19 Jul 2016 09:01:46 +0000 (09:01 +0000)]
Style: drop some unnecessary ';' [NFC]

llvm-svn: 275963

8 years agotest: Add missing 'REQUIRES' line
Tobias Grosser [Tue, 19 Jul 2016 07:47:27 +0000 (07:47 +0000)]
test: Add missing 'REQUIRES' line

llvm-svn: 275962

8 years agoReformat comment from 3 to 2 lines. NFC.
George Rimar [Tue, 19 Jul 2016 07:42:07 +0000 (07:42 +0000)]
Reformat comment from 3 to 2 lines. NFC.

llvm-svn: 275961

8 years agotest: Add missing 'REQUIRES' line
Tobias Grosser [Tue, 19 Jul 2016 07:39:54 +0000 (07:39 +0000)]
test: Add missing 'REQUIRES' line

llvm-svn: 275960

8 years agoFixed comment. NFC.
George Rimar [Tue, 19 Jul 2016 07:39:07 +0000 (07:39 +0000)]
Fixed comment. NFC.

llvm-svn: 275959

8 years ago[clang-rename] add support for overridden functions
Kirill Bobyrev [Tue, 19 Jul 2016 07:37:43 +0000 (07:37 +0000)]
[clang-rename] add support for overridden functions

Reviewers: klimek

Differential Revision: https://reviews.llvm.org/D22408

llvm-svn: 275958

8 years agoGPGPU: Emit in-kernel synchronization statements
Tobias Grosser [Tue, 19 Jul 2016 07:33:16 +0000 (07:33 +0000)]
GPGPU: Emit in-kernel synchronization statements

We use this opportunity to further classify the different user statements that
can arise and add TODOs for the ones not yet implemented.

llvm-svn: 275957

8 years agoGPGPU: generate control flow within the kernel
Tobias Grosser [Tue, 19 Jul 2016 07:33:11 +0000 (07:33 +0000)]
GPGPU: generate control flow within the kernel

llvm-svn: 275956

8 years agoGPGPU: add scop parameters to kernel arguments
Tobias Grosser [Tue, 19 Jul 2016 07:33:06 +0000 (07:33 +0000)]
GPGPU: add scop parameters to kernel arguments

llvm-svn: 275955

8 years agoGPGPU: add host iterators to kernel arguments
Tobias Grosser [Tue, 19 Jul 2016 07:32:55 +0000 (07:32 +0000)]
GPGPU: add host iterators to kernel arguments

llvm-svn: 275954

8 years agoGPGPU: add intrinsic functions to obtain a kernels thread and block ids
Tobias Grosser [Tue, 19 Jul 2016 07:32:44 +0000 (07:32 +0000)]
GPGPU: add intrinsic functions to obtain a kernels thread and block ids

llvm-svn: 275953

8 years agoGPGPU: create kernel function skeleton
Tobias Grosser [Tue, 19 Jul 2016 07:32:38 +0000 (07:32 +0000)]
GPGPU: create kernel function skeleton

Create for each kernel a separate LLVM-IR module containing a single function
marked as kernel function and taking one pointer for each array referenced
by this kernel. Add debugging output to verify the kernels are generated
correctly.

llvm-svn: 275952

8 years ago[driver][mips] Remove empty folder from test inputs
Simon Atanasyan [Tue, 19 Jul 2016 07:23:15 +0000 (07:23 +0000)]
[driver][mips] Remove empty folder from test inputs

llvm-svn: 275951

8 years agoAVX-512: Fixed BT instruction selection.
Elena Demikhovsky [Tue, 19 Jul 2016 07:14:21 +0000 (07:14 +0000)]
AVX-512: Fixed BT instruction selection.

The following condition expression ( a >> n) & 1 is converted to "bt a, n" instruction. It works on all intel targets.
But on AVX-512 it was broken because the expression is modified to (truncate (a >>n) to i1).

I added the new sequence (truncate (a >>n) to i1) to the BT pattern.

Differential Revision: https://reviews.llvm.org/D22354

llvm-svn: 275950

8 years ago[driver][mips] Support MIPS targets in modern Android NDK
Simon Atanasyan [Tue, 19 Jul 2016 07:09:48 +0000 (07:09 +0000)]
[driver][mips] Support MIPS targets in modern Android NDK

Initial patch provided by Duane Sand.

llvm-svn: 275949

8 years ago[esan|wset] Fix flaky sampling tests
Derek Bruening [Tue, 19 Jul 2016 05:06:48 +0000 (05:06 +0000)]
[esan|wset] Fix flaky sampling tests

Adds a new esan public interface routine __esan_get_sample_count() and uses
it to ensure that tests of sampling receive the minimum number of samples.

llvm-svn: 275948

8 years ago[OPENMP] Removed loop statement as its body executes at most once, NFC.
Alexey Bataev [Tue, 19 Jul 2016 05:06:39 +0000 (05:06 +0000)]
[OPENMP] Removed loop statement as its body executes at most once, NFC.

Removed not required loop statement, addressing comments from Richard
Smith.

llvm-svn: 275947

8 years ago[esan] Fix sideline thread flaky assert
Derek Bruening [Tue, 19 Jul 2016 05:03:38 +0000 (05:03 +0000)]
[esan] Fix sideline thread flaky assert

Fixes an esan sideline thread CHECK that failed to account for the sideline
thread reaching its code before the internal_clone() return value was
assigned in the parent.

llvm-svn: 275946

8 years ago[OPENMP] Improved processing of 'priority' clause, NFC.
Alexey Bataev [Tue, 19 Jul 2016 04:21:09 +0000 (04:21 +0000)]
[OPENMP] Improved processing of 'priority' clause, NFC.

Removed some old comments + improved handling of 'priority' clause value
during codegen after comments from Richard Smith.

llvm-svn: 275945

8 years agoIgnore clang-module-cache directories that may be created
Jason Molenda [Tue, 19 Jul 2016 02:37:07 +0000 (02:37 +0000)]
Ignore clang-module-cache directories that may be created
in the testsuite directory while it runs.

llvm-svn: 275944

8 years agoclang-rename: fix referenced variable in vim-script
Saleem Abdulrasool [Tue, 19 Jul 2016 02:13:08 +0000 (02:13 +0000)]
clang-rename: fix referenced variable in vim-script

llvm-svn: 275943

8 years ago[AVX512] Give priority to EVEX encoded PSHUFB over the VEX versions.
Craig Topper [Tue, 19 Jul 2016 02:00:38 +0000 (02:00 +0000)]
[AVX512] Give priority to EVEX encoded PSHUFB over the VEX versions.

llvm-svn: 275942

8 years ago[X86] Remove superfluous parameter from a multiclass. All instantiations passed the...
Craig Topper [Tue, 19 Jul 2016 02:00:35 +0000 (02:00 +0000)]
[X86] Remove superfluous parameter from a multiclass. All instantiations passed the same value.

llvm-svn: 275941

8 years ago[MemorySSA] Update to the new shiny walker.
George Burgess IV [Tue, 19 Jul 2016 01:29:15 +0000 (01:29 +0000)]
[MemorySSA] Update to the new shiny walker.

This patch updates MemorySSA's use-optimizing walker to be more
accurate and, in some cases, faster.

Essentially, this changed our core walking algorithm from a
cache-as-you-go DFS to an iteratively expanded DFS, with all of the
caching happening at the end. Said expansion happens when we hit a Phi,
P; we'll try to do the smallest amount of work possible to see if
optimizing above that Phi is legal in the first place. If so, we'll
expand the search to see if we can optimize to the next phi, etc.

An iteratively expanded DFS lets us potentially quit earlier (because we
don't assume that we can optimize above all phis) than our old walker.
Additionally, because we don't cache as we go, we can now optimize above
loops.

As an added bonus, this patch adds a ton of verification (if
EXPENSIVE_CHECKS are enabled), so finding bugs is easier.

Differential Revision: https://reviews.llvm.org/D21777

llvm-svn: 275940

8 years ago[X86] Rename VINSERTzrr to use a capital Z to match other instructions. NFC
Craig Topper [Tue, 19 Jul 2016 01:26:19 +0000 (01:26 +0000)]
[X86] Rename VINSERTzrr to use a capital Z to match other instructions. NFC

llvm-svn: 275939

8 years agoRetry: [llvm-profdata] Speed up merging by using a thread pool
Vedant Kumar [Tue, 19 Jul 2016 01:17:20 +0000 (01:17 +0000)]
Retry: [llvm-profdata] Speed up merging by using a thread pool

Add a "-j" option to llvm-profdata to control the number of threads used.
Auto-detect NumThreads when it isn't specified, and avoid spawning threads when
they wouldn't be beneficial.

I tested this patch using a raw profile produced by clang (147MB). Here is the
time taken to merge 4 copies together on my laptop:

  No thread pool: 112.87s user 5.92s system 97% cpu 2:01.08 total
  With 2 threads: 134.99s user 26.54s system 164% cpu 1:33.31 total

Changes since the initial commit:

  - When handling odd-length inputs, call ThreadPool::wait() before merging the
    last profile. Should fix a race/off-by-one (see r275937).

Differential Revision: https://reviews.llvm.org/D22438

llvm-svn: 275938

8 years agoRevert "[llvm-profdata] Speed up merging by using a thread pool"
Vedant Kumar [Tue, 19 Jul 2016 00:57:09 +0000 (00:57 +0000)]
Revert "[llvm-profdata] Speed up merging by using a thread pool"

This reverts commit r275921. It broke the ppc64be bot:

  http://lab.llvm.org:8011/builders/clang-ppc64be-linux-multistage/builds/3537

I'm not sure why it broke, but based on the output, it looks like an
off-by-one (one profile left un-merged).

llvm-svn: 275937

8 years agoRecommit the patch "Use uniforms set to populate VecValuesToIgnore".
Wei Mi [Tue, 19 Jul 2016 00:50:43 +0000 (00:50 +0000)]
Recommit the patch "Use uniforms set to populate VecValuesToIgnore".

For instructions in uniform set, they will not have vector versions so
add them to VecValuesToIgnore.
For induction vars, those only used in uniform instructions or consecutive
ptrs instructions have already been added to VecValuesToIgnore above. For
those induction vars which are only used in uniform instructions or
non-consecutive/non-gather scatter ptr instructions, the related phi and
update will also be added into VecValuesToIgnore set.

The change will make the vector RegUsages estimation less conservative.

Differential Revision: https://reviews.llvm.org/D20474

The recommit fixed the testcase global_alias.ll.

llvm-svn: 275936

8 years agoAMDGPU/SI: Fix SI scheduler refcount issue
Matt Arsenault [Tue, 19 Jul 2016 00:35:22 +0000 (00:35 +0000)]
AMDGPU/SI: Fix SI scheduler refcount issue

Without this fix, releaseSuccessors when InOrOutBlock is
false could release SUs outside the schedule BasicBlock.

Patch by Axel Davy

llvm-svn: 275935

8 years agoAMDGPU: Expand register indexing pseudos in custom inserter
Matt Arsenault [Tue, 19 Jul 2016 00:35:03 +0000 (00:35 +0000)]
AMDGPU: Expand register indexing pseudos in custom inserter

This is to help moveSILowerControlFlow to before regalloc.
There are a couple of tradeoffs with this. The complete CFG
is visible to more passes, the loop body avoids an extra copy of m0,
vcc isn't required, and immediate offsets can be shrunk into s_movk_i32.

The disadvantage is the register allocator doesn't understand that
the single lane's vector is dead within the loop body, so an extra
register is used to outlive the loop block when expanding the
VGPR -> m0 loop. This also now results in worse waitcnt insertion
before the loop instead of after for pending operations at the point
of the indexing, but that should be fixed by future improvements to
cross block waitcnt insertion.

v_movreld_b32's operands are now modeled more correctly since vdst
is not a true output. This is kind of a hack to treat vdst as a
use operand. Extra checking is required in the verifier since
I can't seem to get tablegen to emit an implicit operand for a
virtual register.

llvm-svn: 275934

8 years ago[Kaleidoscope][BuildingAJIT] More work on the text for Chapter 3.
Lang Hames [Tue, 19 Jul 2016 00:25:52 +0000 (00:25 +0000)]
[Kaleidoscope][BuildingAJIT] More work on the text for Chapter 3.

Add an overview of stubs and compile callbacks before the discussion of the
source changes.

-- This line, and those below, will be ignored--

M    docs/tutorial/BuildingAJIT3.rst

llvm-svn: 275933

8 years ago[LoopReroll] Reroll loops with unordered atomic memory accesses
Sanjoy Das [Tue, 19 Jul 2016 00:23:54 +0000 (00:23 +0000)]
[LoopReroll] Reroll loops with unordered atomic memory accesses

Reviewers: hfinkel, jfb, reames

Subscribers: mcrosier, mzolotukhin, llvm-commits

Differential Revision: https://reviews.llvm.org/D22385

llvm-svn: 275932

8 years agoAppend clang system include path for offloading tool chains.
Samuel Antao [Tue, 19 Jul 2016 00:01:12 +0000 (00:01 +0000)]
Append clang system include path for offloading tool chains.

Summary:
This patch adds clang system include path when offloading tool chains, e.g. CUDA, are used in the current compilation.

This fixes an issue detected by @rsmith in response to r275645.

Reviewers: rsmith, tra

Subscribers: rsmith, cfe-commits

Differential Revision: https://reviews.llvm.org/D22490

llvm-svn: 275931

8 years ago[OpenMP] Remove dead code in conditional of mappable expressions SEMA.
Samuel Antao [Mon, 18 Jul 2016 23:22:11 +0000 (23:22 +0000)]
[OpenMP] Remove dead code in conditional of mappable expressions SEMA.

llvm-svn: 275930

8 years agoTableGen: Allow custom register operand decoder method
Matt Arsenault [Mon, 18 Jul 2016 23:20:46 +0000 (23:20 +0000)]
TableGen: Allow custom register operand decoder method

This is for a situation where the encoding for a register may be
different depending on the specific operand. For some instructions,
we want to apply additional restrictions beyond the encoding's
constraints.

In AMDGPU some operands are VSrc_32, using the VS_32 pseudo register
class which accept VGPRs, SGPRs, or immediates in the encoding.
Some specific instructions with the same encoding operand do not want
to allow immediates or SGPRs, but the encoding format is different
in this case than a regular VGPR_32 operand.

This allows specifying the encoding should be treated the same
without introducing yet another dummy register class.

llvm-svn: 275929

8 years agoAMDGPU: Fix test name and broken CHECK-LABEL
Matt Arsenault [Mon, 18 Jul 2016 23:09:51 +0000 (23:09 +0000)]
AMDGPU: Fix test name and broken CHECK-LABEL

llvm-svn: 275928

8 years ago[utils] Generate html reports with the code coverage utility script
Vedant Kumar [Mon, 18 Jul 2016 22:50:10 +0000 (22:50 +0000)]
[utils] Generate html reports with the code coverage utility script

Instead of extracting raw coverage mappings into an artifact directory,
actually generate useful html reports for a given list of binaries with
symbol demangling turned on.

No tests, but this is actively being used to drive the (still nascent)
coverage bot.

llvm-svn: 275927

8 years ago[OpenMP] Fix incorrect diagnostics in map clause
Kelvin Li [Mon, 18 Jul 2016 22:49:16 +0000 (22:49 +0000)]
[OpenMP] Fix incorrect diagnostics in map clause

Having the following code pattern will result in incorrect diagnostic

int main() {
  int arr[10];
#pragma omp target data map(arr[:])
#pragma omp target map(arr)
  {}
}

t.cpp:4:24: error: original storage of expression in data environment is shared
      but data environment do not fully contain mapped expression storage
#pragma omp target map(arr)
                       ^~~
t.cpp:3:29: note: used here
#pragma omp target data map(arr[:])
                            ^~~~~~
1 error generated.

Patch by David S.

Differential Revision: https://reviews.llvm.org/D22075

llvm-svn: 275926

8 years agoFix some minor issues found by Coverity.
Richard Smith [Mon, 18 Jul 2016 22:37:35 +0000 (22:37 +0000)]
Fix some minor issues found by Coverity.

llvm-svn: 275925

8 years ago[Coverage] Remove '..' from filenames *after* getting an absolute path
Vedant Kumar [Mon, 18 Jul 2016 22:32:02 +0000 (22:32 +0000)]
[Coverage] Remove '..' from filenames *after* getting an absolute path

Failure to do this breaks relative paths which begin with '..'.

This issue was caught by the (still nascent) coverage bot.

llvm-svn: 275924

8 years agoFix -Wreturn-type with gcc 4.8 and libc++
Matt Arsenault [Mon, 18 Jul 2016 22:12:46 +0000 (22:12 +0000)]
Fix -Wreturn-type with gcc 4.8 and libc++

llvm-svn: 275922

8 years ago[llvm-profdata] Speed up merging by using a thread pool
Vedant Kumar [Mon, 18 Jul 2016 22:02:39 +0000 (22:02 +0000)]
[llvm-profdata] Speed up merging by using a thread pool

Add a "-j" option to llvm-profdata to control the number of threads
used. Auto-detect NumThreads when it isn't specified, and avoid spawning
threads when they wouldn't be beneficial.

I tested this patch using a raw profile produced by clang (147MB). Here is the
time taken to merge 4 copies together on my laptop:

  No thread pool: 112.87s user 5.92s system 97% cpu 2:01.08 total
  With 2 threads: 134.99s user 26.54s system 164% cpu 1:33.31 total

Differential Revision: https://reviews.llvm.org/D22438

llvm-svn: 275921

8 years ago[NVPTX] Make sure we adjust alignment at all call sites
Artem Belevich [Mon, 18 Jul 2016 21:58:48 +0000 (21:58 +0000)]
[NVPTX] Make sure we adjust alignment at all call sites

.. including calls from kernel functions that were
ignored by mistake before.

llvm-svn: 275920

8 years ago[PM] Convert Loop Strength Reduce pass to new PM
Dehao Chen [Mon, 18 Jul 2016 21:41:50 +0000 (21:41 +0000)]
[PM] Convert Loop Strength Reduce pass to new PM

Summary: Convert Loop String Reduce pass to new PM

Reviewers: davidxl, silvas

Subscribers: junbuml, sanjoy, mzolotukhin, llvm-commits

Differential Revision: https://reviews.llvm.org/D22468

llvm-svn: 275919

8 years agoUpdate doxygen description for `WriteBitcodeToFile()` API (NFC)
Mehdi Amini [Mon, 18 Jul 2016 21:29:24 +0000 (21:29 +0000)]
Update doxygen description for `WriteBitcodeToFile()` API (NFC)

llvm-svn: 275917

8 years ago[PM] Port FunctionImport Pass to new PM
Teresa Johnson [Mon, 18 Jul 2016 21:22:24 +0000 (21:22 +0000)]
[PM] Port FunctionImport Pass to new PM

Summary: Port FunctionImport Pass to new PM.

Reviewers: mehdi_amini, davide

Subscribers: davidxl, llvm-commits

Differential Revision: https://reviews.llvm.org/D22475

llvm-svn: 275916

8 years agoRevert rL275912.
Wei Mi [Mon, 18 Jul 2016 21:14:43 +0000 (21:14 +0000)]
Revert rL275912.

llvm-svn: 275915

8 years agoAdd missing headers after header cleanup in r275882.
Chaoren Lin [Mon, 18 Jul 2016 21:11:43 +0000 (21:11 +0000)]
Add missing headers after header cleanup in r275882.

llvm-svn: 275914

8 years ago[Coverage] Normalize '..' out of filename strings
Vedant Kumar [Mon, 18 Jul 2016 21:01:27 +0000 (21:01 +0000)]
[Coverage] Normalize '..' out of filename strings

This fixes the issue of having duplicate entries for the same file in a
coverage report s.t none of the entries actually displayed the correct
coverage information.

llvm-svn: 275913

8 years agoUse uniforms set to populate VecValuesToIgnore.
Wei Mi [Mon, 18 Jul 2016 20:59:53 +0000 (20:59 +0000)]
Use uniforms set to populate VecValuesToIgnore.

For instructions in uniform set, they will not have vector versions so
add them to VecValuesToIgnore.
For induction vars, those only used in uniform instructions or consecutive
ptrs instructions have already been added to VecValuesToIgnore above. For
those induction vars which are only used in uniform instructions or
non-consecutive/non-gather scatter ptr instructions, the related phi and
update will also be added into VecValuesToIgnore set.

The change will make the vector RegUsages estimation less conservative.

Differential Revision: https://reviews.llvm.org/D20474

llvm-svn: 275912

8 years agorefactor SimplifySelectInst; NFCI
Sanjay Patel [Mon, 18 Jul 2016 20:56:53 +0000 (20:56 +0000)]
refactor SimplifySelectInst; NFCI

llvm-svn: 275911

8 years agoWrite isUInt using template specializations to work around an incorrect MSVC warning.
Justin Lebar [Mon, 18 Jul 2016 20:40:35 +0000 (20:40 +0000)]
Write isUInt using template specializations to work around an incorrect MSVC warning.

Summary:
Per D22441, MSVC warns on our old implementation of isUInt<64>.  It sees
uint64_t(1) << 64 and doesn't realize that it's not going to be
executed.  Writing as a template specialization is ugly, but prevents
the warning.

Reviewers: RKSimon

Subscribers: majnemer, llvm-commits

Differential Revision: https://reviews.llvm.org/D22472

llvm-svn: 275909

8 years agoadd tests for missed sext transform
Sanjay Patel [Mon, 18 Jul 2016 20:37:51 +0000 (20:37 +0000)]
add tests for missed sext transform

llvm-svn: 275908

8 years ago[Sema] Create a separate group for incompatible function pointer warning
Bruno Cardoso Lopes [Mon, 18 Jul 2016 20:37:06 +0000 (20:37 +0000)]
[Sema] Create a separate group for incompatible function pointer warning

Give incompatible function pointer warning its own diagnostic group
but still leave it as a subgroup of incompatible-pointer-types. This is in
preparation to promote -Wincompatible-function-pointer-types to error on
darwin.

Differential Revision: https://reviews.llvm.org/D22248

rdar://problem/12907612

llvm-svn: 275907

8 years agoAdd missing header in ClangFuzzer (after r275882 cleanup)
Mehdi Amini [Mon, 18 Jul 2016 20:33:09 +0000 (20:33 +0000)]
Add missing header in ClangFuzzer (after r275882 cleanup)

llvm-svn: 275906

8 years agoAllow iOS and tvOS version numbers with 2-digit major version numbers.
Bob Wilson [Mon, 18 Jul 2016 20:29:14 +0000 (20:29 +0000)]
Allow iOS and tvOS version numbers with 2-digit major version numbers.

rdar://problem/26921601

llvm-svn: 275905

8 years agoBump version # to 4.0.0
Marshall Clow [Mon, 18 Jul 2016 20:27:19 +0000 (20:27 +0000)]
Bump version # to 4.0.0

llvm-svn: 275904

8 years agobuild_llvm_package.bat: update version to 4.0.0
Hans Wennborg [Mon, 18 Jul 2016 20:26:46 +0000 (20:26 +0000)]
build_llvm_package.bat: update version to 4.0.0

llvm-svn: 275903

8 years ago[interception] Remove extra whitespace to appease linters (NFC)
Vedant Kumar [Mon, 18 Jul 2016 20:07:27 +0000 (20:07 +0000)]
[interception] Remove extra whitespace to appease linters (NFC)

Attempt to fix:

  http://lab.llvm.org:8011/builders/clang-s390x-linux/builds/7774

llvm-svn: 275901

8 years agoauto-generate checks
Sanjay Patel [Mon, 18 Jul 2016 20:06:51 +0000 (20:06 +0000)]
auto-generate checks

llvm-svn: 275899

8 years agoRevert r273099 "If the revision number starts with r, drop it. It will get added...
Hans Wennborg [Mon, 18 Jul 2016 20:06:27 +0000 (20:06 +0000)]
Revert r273099 "If the revision number starts with r, drop it. It will get added back"

This doesn't seem to work with Bash:

$ /work/llvm/utils/release/merge.sh --proj llvm --rev r275870
/work/llvm/utils/release/merge.sh: line 34: ${$1#r}: bad substitution

I get the same error with and without a leading 'r'.

llvm-svn: 275898

8 years ago[Driver] Compute effective target triples once per job (NFCI)
Vedant Kumar [Mon, 18 Jul 2016 19:56:38 +0000 (19:56 +0000)]
[Driver] Compute effective target triples once per job (NFCI)

Compute an effective target triple exactly once in ConstructJob(), and
then simply pass around references to it. This eliminates wasteful
re-computation of effective triples (e.g in getARMFloatABI()).

Differential Revision: https://reviews.llvm.org/D22290

llvm-svn: 275895

8 years ago[Driver] Make Driver::DefaultTargetTriple private (NFCI)
Vedant Kumar [Mon, 18 Jul 2016 19:56:33 +0000 (19:56 +0000)]
[Driver] Make Driver::DefaultTargetTriple private (NFCI)

No in-tree targets access this `DefaultTargetTriple` directly, and usage
of default triples is generally discouraged. Make the field private.

This is part of en effort to make the clang driver use effective triples
more pervasively.

Differential Revision: https://reviews.llvm.org/D22289

llvm-svn: 275894

8 years ago[NVPTX] Force minimum alignment of 4 for byval arguments of device-side functions.
Artem Belevich [Mon, 18 Jul 2016 19:54:56 +0000 (19:54 +0000)]
[NVPTX] Force minimum alignment of 4 for byval arguments of device-side functions.

Taking address of a byval variable in PTX is legal, but currently runs
into miscompilation by ptxas on sm_50+ (NVIDIA issue 1789042).
Work around the issue by enforcing minimum alignment on byval arguments
of device functions.

The change is a no-op on SASS level for sm_3x where ptxas already aligns
local copy by at least 4.

Differential Revision: https://reviews.llvm.org/D22428

llvm-svn: 275893

8 years ago[compiler-rt] Fix incorrect handling of indirect load.
Etienne Bergeron [Mon, 18 Jul 2016 19:50:55 +0000 (19:50 +0000)]
[compiler-rt] Fix incorrect handling of indirect load.

Summary:
Indirect load are relative offset from RIP.

The current trampoline implementation is incorrectly
copying these instructions which make some unittests
crashing.

This patch is not fixing the unittests but it's fixing
the crashes. The functions are no longer hooked.

Patches will come soon to fix these unittests.

Reviewers: rnk

Subscribers: llvm-commits, wang0109, chrisha

Differential Revision: https://reviews.llvm.org/D22410

llvm-svn: 275892

8 years ago[LoopSimplify] Update LCSSA after separating nested loops.
Michael Zolotukhin [Mon, 18 Jul 2016 19:44:19 +0000 (19:44 +0000)]
[LoopSimplify] Update LCSSA after separating nested loops.

Summary:
Usually LCSSA survives this transformation, but in some cases (see
attached test) it doesn't: values from the original loop after
separating might be used from the outer loop. Before the transformation
it was the same loop, so LCSSA phis were not required.

This fixes PR28272.

Reviewers: sanjoy, hfinkel, chandlerc

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D21665

llvm-svn: 275891

8 years agoRevert "[ARM] Skip inline asm memory operands in DAGToDAGISel"
Vitaly Buka [Mon, 18 Jul 2016 19:44:01 +0000 (19:44 +0000)]
Revert "[ARM] Skip inline asm memory operands in DAGToDAGISel"

Breaks asan, see https://reviews.llvm.org/D22103

This reverts commit r275776.

llvm-svn: 275890

8 years agoRevert "[ARM] Update test to use CHECK-LABEL. NFCI."
Vitaly Buka [Mon, 18 Jul 2016 19:43:58 +0000 (19:43 +0000)]
Revert "[ARM] Update test to use CHECK-LABEL. NFCI."

Breaks asan, see https://reviews.llvm.org/D22103

This reverts commit r275777.

llvm-svn: 275889

8 years ago[MC] Separate non-parsing operations from conditional chains. NFC.
Nirav Dave [Mon, 18 Jul 2016 19:35:21 +0000 (19:35 +0000)]
[MC] Separate non-parsing operations from conditional chains. NFC.

llvm-svn: 275888

8 years ago[compiler-rt] Allow trampoline allocation further and 1 gig.
Etienne Bergeron [Mon, 18 Jul 2016 19:33:05 +0000 (19:33 +0000)]
[compiler-rt] Allow trampoline allocation further and 1 gig.

Summary:
The trampoline allocation limits the memory scanning to 1 gig.
There is an unittest that is allocating a large object which make
it impossible to the trampoline allocator to find a free spot.

see shadow_mapping_failures:
```
char bigchunk[1 << 30];
```

This patch is not fixing the unittest but it's fixing it's infinite
loop behavior.

Reviewers: rnk

Subscribers: llvm-commits, wang0109, chrisha

Differential Revision: https://reviews.llvm.org/D22471

llvm-svn: 275887

8 years agoUnbreak extra tools build post r275882.
Benjamin Kramer [Mon, 18 Jul 2016 19:21:22 +0000 (19:21 +0000)]
Unbreak extra tools build post r275882.

llvm-svn: 275886

8 years agomake macOS 'launch in terminal' bring terminal to the front during launch
Todd Fiala [Mon, 18 Jul 2016 19:15:38 +0000 (19:15 +0000)]
make macOS 'launch in terminal' bring terminal to the front during launch

rdar://25235812

llvm-svn: 275885

8 years ago[GVNHoist] Remove a home-grown version of replaceUsesOfWith
David Majnemer [Mon, 18 Jul 2016 19:14:14 +0000 (19:14 +0000)]
[GVNHoist] Remove a home-grown version of replaceUsesOfWith

replaceUsesOfWith will, on average, consider fewer values when trying
to do the replacement.

No functional change is intended.

llvm-svn: 275884

8 years ago[LCSSA] Post-process PHI-nodes created by SSAUpdate when constructing LCSSA form.
Michael Zolotukhin [Mon, 18 Jul 2016 19:05:08 +0000 (19:05 +0000)]
[LCSSA] Post-process PHI-nodes created by SSAUpdate when constructing LCSSA form.

Summary:
SSAUpdate might insert PHI-nodes inside loops, which can break LCSSA
form unless we fix it up.

This fixes PR28424.

Reviewers: sanjoy, chandlerc, hfinkel

Subscribers: uabelho, llvm-commits

Differential Revision: http://reviews.llvm.org/D21997

llvm-svn: 275883

8 years ago[NFC] Header cleanup
Mehdi Amini [Mon, 18 Jul 2016 19:02:11 +0000 (19:02 +0000)]
[NFC] Header cleanup

Summary: Removed unused headers, replaced some headers with forward class declarations

Patch by: Eugene <claprix@yandex.ru>

Differential Revision: https://reviews.llvm.org/D20100

llvm-svn: 275882

8 years agoAMDGPU: Remove pointless dyn_cast_or_null
Matt Arsenault [Mon, 18 Jul 2016 19:00:07 +0000 (19:00 +0000)]
AMDGPU: Remove pointless dyn_cast_or_null

This is already casted above so non-null

llvm-svn: 275881

8 years agoRevert "[analyzer] Add checker modeling potential C++ self-assignment"
Devin Coughlin [Mon, 18 Jul 2016 18:57:50 +0000 (18:57 +0000)]
Revert "[analyzer] Add checker modeling potential C++ self-assignment"

This reverts commit r275820. It is failing on the bots.

llvm-svn: 275880

8 years agoFix -Wmicrosoft-enum-value in GVNHoist.cpp
Reid Kleckner [Mon, 18 Jul 2016 18:53:50 +0000 (18:53 +0000)]
Fix -Wmicrosoft-enum-value in GVNHoist.cpp

llvm-svn: 275879

8 years ago[X86][SSE] Regenerate extraction from promotion test
Simon Pilgrim [Mon, 18 Jul 2016 18:53:15 +0000 (18:53 +0000)]
[X86][SSE] Regenerate extraction from promotion test

Added tests for SSE2 as well as SSE41

llvm-svn: 275878

8 years agoSupport -masm= flag for x86 assembly targets.
Yunzhong Gao [Mon, 18 Jul 2016 18:44:51 +0000 (18:44 +0000)]
Support -masm= flag for x86 assembly targets.

For assembly files without .intel_syntax or .att_syntax directives, allow the
-masm= flag to supply a default assembly dialect. For example,

C:\TMP> type intel.s
.text
mov al,0

C:\TMP> clang -masm=intel -c intel.s

Without this patch, one would need to pass an "-mllvm -x86-asm-syntax=" flag
directly to the backend.
C:\TMP> clang -mllvm --x86-asm-syntax=intel -c intel.s

Differentials Review: http://reviews.llvm.org/D22285

llvm-svn: 275877

8 years ago[X86][SSE] Regenerate extraction+store memop tests
Simon Pilgrim [Mon, 18 Jul 2016 18:44:01 +0000 (18:44 +0000)]
[X86][SSE] Regenerate extraction+store memop tests

Added tests for SSE2 as well as SSE41+AVX

llvm-svn: 275876

8 years ago[X86][SSE] Regenerate truncate+extension memop tests
Simon Pilgrim [Mon, 18 Jul 2016 18:42:33 +0000 (18:42 +0000)]
[X86][SSE] Regenerate truncate+extension memop tests

Added tests for SSE2 as well as SSE41

llvm-svn: 275875

8 years agoR600: Use new barrier intrinsic
Matt Arsenault [Mon, 18 Jul 2016 18:42:17 +0000 (18:42 +0000)]
R600: Use new barrier intrinsic

llvm-svn: 275874

8 years agoAMDGPU: Fix missing switch case warning
Matt Arsenault [Mon, 18 Jul 2016 18:40:51 +0000 (18:40 +0000)]
AMDGPU: Fix missing switch case warning

llvm-svn: 275873

8 years agoRegenerate test
Simon Pilgrim [Mon, 18 Jul 2016 18:38:51 +0000 (18:38 +0000)]
Regenerate test

llvm-svn: 275872

8 years agoAMDGPU: Add intrinsic for s_flbit_i32/v_ffbh_i32
Matt Arsenault [Mon, 18 Jul 2016 18:35:05 +0000 (18:35 +0000)]
AMDGPU: Add intrinsic for s_flbit_i32/v_ffbh_i32

llvm-svn: 275871

8 years agoAMDGPU/R600: Replace barrier intrinsics
Matt Arsenault [Mon, 18 Jul 2016 18:34:59 +0000 (18:34 +0000)]
AMDGPU/R600: Replace barrier intrinsics

llvm-svn: 275870

8 years agoAMDGPU: Remove dead check in AMDGPUPromoteAlloca
Matt Arsenault [Mon, 18 Jul 2016 18:34:53 +0000 (18:34 +0000)]
AMDGPU: Remove dead check in AMDGPUPromoteAlloca

This is currently only called with GEP users. A direct
alloca would only happen with current typed pointers
for arrays which are a perverse case.

Also fix crashes on 0 x and 1 x arrays.

llvm-svn: 275869