]> granicus.if.org Git - llvm/log
llvm
7 years ago[LICM] Report failing to hoist conditionally-executed loads
Adam Nemet [Wed, 11 Jan 2017 04:39:49 +0000 (04:39 +0000)]
[LICM] Report failing to hoist conditionally-executed loads

These are interesting again because the user may not be aware that this
is a common reason preventing LICM.

A const is removed from an instruction pointer declaration in order to
pass it to ORE.

Differential Revision: https://reviews.llvm.org/D27940

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291649 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[LICM] Report failing to hoist a load with an invariant address
Adam Nemet [Wed, 11 Jan 2017 04:39:45 +0000 (04:39 +0000)]
[LICM] Report failing to hoist a load with an invariant address

These are interesting because lack of precision in alias information
could be standing in the way of this optimization.

An example is the case in the test suite that I showed in the DevMeeting
talk:

http://lab.llvm.org:8080/artifacts/opt-view_test-suite/build/MultiSource/Benchmarks/FreeBench/distray/CMakeFiles/distray.dir/html/_org_test-suite_MultiSource_Benchmarks_FreeBench_distray_distray.c.html#L236

canSinkOrHoistInst is also used from LoopSink, which does not use
opt-remarks so we need to take ORE as an optional argument.

Differential Revision: https://reviews.llvm.org/D27939

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291648 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix typo in comment
Adam Nemet [Wed, 11 Jan 2017 04:39:41 +0000 (04:39 +0000)]
Fix typo in comment

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291647 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[LICM] Report successful hoist/sink/promotion
Adam Nemet [Wed, 11 Jan 2017 04:39:35 +0000 (04:39 +0000)]
[LICM] Report successful hoist/sink/promotion

Differential Revision: https://reviews.llvm.org/D27938

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291646 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[DAGCombiner] Teach DAG combiner to fold (vselect (N0 xor AllOnes), N1, N2) -> (vsele...
Craig Topper [Wed, 11 Jan 2017 04:02:23 +0000 (04:02 +0000)]
[DAGCombiner] Teach DAG combiner to fold (vselect (N0 xor AllOnes), N1, N2) -> (vselect N0, N2, N1). Only do this if the target indicates its vector boolean type is ZeroOrNegativeOneBooleanContent.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291645 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDAGCombiner: Add hasOneUse checks to fadd/fma combine
Matt Arsenault [Wed, 11 Jan 2017 02:02:12 +0000 (02:02 +0000)]
DAGCombiner: Add hasOneUse checks to fadd/fma combine

Even with aggressive fusion enabled, this requires duplicating
the fmul, or increases an fadd to another fma which is not an
improvement.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291642 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Target] Fix some Clang-tidy modernize and Include What You Use warnings; other minor...
Eugene Zelenko [Wed, 11 Jan 2017 01:45:03 +0000 (01:45 +0000)]
[Target] 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@291641 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRe-commit r289955: [X86] Fold (setcc (cmp (atomic_load_add x, -C) C), COND) to (setcc...
Hans Wennborg [Wed, 11 Jan 2017 01:36:57 +0000 (01:36 +0000)]
Re-commit r289955: [X86] Fold (setcc (cmp (atomic_load_add x, -C) C), COND) to (setcc (LADD x, -C), COND) (PR31367)

This was reverted because it would miscompile code where the cmp had
multiple uses. That was due to a deficiency in the existing code, which
was fixed in r291630 (see the PR for details).

This re-commit includes an extra test for the kind of code that got
miscompiled: @test_sub_1_setcc_jcc.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291640 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agotools/llvm-xray: Avoid std::errc::protocol_* to appease mingw, like r285261.
NAKAMURA Takumi [Wed, 11 Jan 2017 01:06:57 +0000 (01:06 +0000)]
tools/llvm-xray: Avoid std::errc::protocol_* to appease mingw, like r285261.

They are oriented from winsock and mingw doesn't import them.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291636 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoInstSimplify: Refactor function to use more switches
Matt Arsenault [Wed, 11 Jan 2017 00:57:54 +0000 (00:57 +0000)]
InstSimplify: Refactor function to use more switches

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291634 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove unused field.
Zachary Turner [Wed, 11 Jan 2017 00:56:53 +0000 (00:56 +0000)]
Remove unused field.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291633 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Dont run combineSetCCAtomicArith() when the cmp has multiple uses
Hans Wennborg [Wed, 11 Jan 2017 00:49:54 +0000 (00:49 +0000)]
[X86] Dont run combineSetCCAtomicArith() when the cmp has multiple uses

We would miscompile the following:

  void g(int);
  int f(volatile long long *p) {
    bool b = __atomic_fetch_add(p, 1, __ATOMIC_SEQ_CST) < 0;
    g(b ? 12 : 34);
    return b ? 56 : 78;
  }

into

  pushq   %rax
  lock            incq    (%rdi)
  movl    $12, %eax
  movl    $34, %edi
  cmovlel %eax, %edi
  callq   g(int)
  testq   %rax, %rax   <---- Bad.
  movl    $56, %ecx
  movl    $78, %eax
  cmovsl  %ecx, %eax
  popq    %rcx
  retq

because the code failed to take into account that the cmp has multiple
uses, replaced one of them, and left the other one comparing garbage.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291630 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[RegBankSelect] Improve the output of the debug messages.
Quentin Colombet [Wed, 11 Jan 2017 00:48:41 +0000 (00:48 +0000)]
[RegBankSelect] Improve the output of the debug messages.

Add more information about mapping cost and chosen solution.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291629 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeView/PDB] Rename a bunch of files.
Zachary Turner [Wed, 11 Jan 2017 00:35:43 +0000 (00:35 +0000)]
[CodeView/PDB] Rename a bunch of files.

We were starting to get some name clashes between llvm-pdbdump
and the common CodeView framework, so I took this opportunity
to rename a bunch of files to more accurately describe their
usage.  This also helps in llvm-pdbdump to distinguish
between different files and whether they are used for pretty
dump mode or raw dump mode.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291627 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeView] Add TypeDatabase class.
Zachary Turner [Wed, 11 Jan 2017 00:35:08 +0000 (00:35 +0000)]
[CodeView] Add TypeDatabase class.

This creates a centralized class in which to store type records.
It stores types as an array of entries, which matches the
notion of a type stream being a topologically sorted DAG.
Logic to build up such a database was already being used in
CVTypeDumper, so CVTypeDumper is now updated to to read from
a TypeDatabase which is filled out by an earlier visitor in
the pipeline.

Differential Revision: https://reviews.llvm.org/D28486

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291626 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd better documentation for iterator facade subclasses.
Zachary Turner [Wed, 11 Jan 2017 00:34:41 +0000 (00:34 +0000)]
Add better documentation for iterator facade subclasses.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291625 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoInstSimplify: Eliminate fabs on known positive
Matt Arsenault [Wed, 11 Jan 2017 00:33:24 +0000 (00:33 +0000)]
InstSimplify: Eliminate fabs on known positive

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291624 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[gmock] Teach gmock ElementsAre and BeginEndDistanceIs matchers to
Chandler Carruth [Wed, 11 Jan 2017 00:16:03 +0000 (00:16 +0000)]
[gmock] Teach gmock ElementsAre and BeginEndDistanceIs matchers to
handle generic ranges by using std::begin and std::end rather than
requiring things to look exactly like an STL container.

Much of the credit for this goes to Dave Blaikie who helped me figure
out the right incantations.

This will probably be re-designed when I send this to the maintainers of
gmock, so I've instead structured it to change is little as possible
while it is a local patch. That makes it somewhat ugly, but I think a focused
change is better for getting this to work for LLVM today and letting the
upstream maintainers figure out the correct long-term pattern.

Differential Revision: https://reviews.llvm.org/D28288

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291623 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU/EG,CM: Add fp16 conversion instructions
Jan Vesely [Wed, 11 Jan 2017 00:12:39 +0000 (00:12 +0000)]
AMDGPU/EG,CM: Add fp16 conversion instructions

Differential Revision: https://reviews.llvm.org/D28164

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291622 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "[PGO] Turn off comdat renaming in IR PGO by default"
Rong Xu [Tue, 10 Jan 2017 23:54:31 +0000 (23:54 +0000)]
Revert "[PGO] Turn off comdat renaming in IR PGO by default"

This patch reverts r291588: [PGO] Turn off comdat renaming in IR PGO by default,
as we are seeing some hash mismatches in our internal tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291621 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[InstCombine] add a wrapper for a common pair of transforms; NFCI
Sanjay Patel [Tue, 10 Jan 2017 23:49:07 +0000 (23:49 +0000)]
[InstCombine] add a wrapper for a common pair of transforms; NFCI

Some of the callers are artificially limiting this transform to integer types;
this should make it easier to incrementally remove that restriction.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291620 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[loop-unroll] Properly populate LoopInfo for loops cloned in LoopUnrollRuntime.
Florian Hahn [Tue, 10 Jan 2017 23:43:35 +0000 (23:43 +0000)]
[loop-unroll] Properly populate LoopInfo for loops cloned in LoopUnrollRuntime.

Summary:
This fixes Transforms/LoopUnroll/runtime-loop3.ll which failed with
EXTENSIVE_DEBUG, because the cloned basic blocks were not added to the
correct sub-loops in LoopUnrollRuntime.cpp.

Reviewers: dexonsmith, mzolotukhin

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D28482

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291619 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[TM] Restore default TargetOptions in TargetMachine::resetTargetOptions.
Justin Lebar [Tue, 10 Jan 2017 23:43:04 +0000 (23:43 +0000)]
[TM] Restore default TargetOptions in TargetMachine::resetTargetOptions.

Summary:
Previously if you had

 * a function with the fast-math-enabled attr, followed by
 * a function without the fast-math attr,

the second function would inherit the first function's fast-math-ness.

This means that mixing fast-math and non-fast-math functions in a module
was completely broken unless you explicitly annotated every
non-fast-math function with "unsafe-fp-math"="false".  This appears to
have been broken since r176986 (March 2013), when the resetTargetOptions
function was introduced.

This patch tests the correct behavior as best we can.  I don't think I
can test FPDenormalMode and NoTrappingFPMath, because they aren't used
in any backends during function lowering.  Surprisingly, I also can't
find any uses at all of LessPreciseFPMAD affecting generated code.

The NVPTX/fast-math.ll test changes are an expected result of fixing
this bug.  When FMA is disabled, we emit add as "add.rn.f32", which
prevents fma combining.  Before this patch, fast-math was enabled in all
functions following the one which explicitly enabled it on itself, so we
were emitting plain "add.f32" where we should have generated
"add.rn.f32".

Reviewers: mkuper

Subscribers: hfinkel, majnemer, jholewinski, nemanjai, llvm-commits

Differential Revision: https://reviews.llvm.org/D28507

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291618 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[NVPTX] Add CHECK-LABEL where appropriate to fast-math.ll test.
Justin Lebar [Tue, 10 Jan 2017 23:42:46 +0000 (23:42 +0000)]
[NVPTX] Add CHECK-LABEL where appropriate to fast-math.ll test.

Also fix up whitespace.

Test-only change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291617 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AArch64] Consider all vector types for FeatureSlowMisaligned128Store
Evandro Menezes [Tue, 10 Jan 2017 23:42:21 +0000 (23:42 +0000)]
[AArch64] Consider all vector types for FeatureSlowMisaligned128Store

The original code considered only v2i64 as slow for this feature. This patch
consider all 128-bit long vector types as slow candidates.

In internal tests, extending this feature to all 128-bit vector types
resulted in an overall improvement of 1% on Exynos M1.

Differential revision: https://reviews.llvm.org/D27998

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291616 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Constant fold when immediate is materialized
Matt Arsenault [Tue, 10 Jan 2017 23:32:04 +0000 (23:32 +0000)]
AMDGPU: Constant fold when immediate is materialized

In future commits these patterns will appear after moveToVALU changes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291615 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[loop-unroll] Factor out code to update LoopInfo (NFC).
Florian Hahn [Tue, 10 Jan 2017 23:24:54 +0000 (23:24 +0000)]
[loop-unroll] Factor out code to update LoopInfo (NFC).

Move the code to update LoopInfo for cloned basic blocks to
addClonedBlockToLoopInfo, as suggested in
https://reviews.llvm.org/D28482.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291614 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMove the section name from GlobalObject to the LLVMContext
Reid Kleckner [Tue, 10 Jan 2017 23:23:58 +0000 (23:23 +0000)]
Move the section name from GlobalObject to the LLVMContext

Summary:
Convention wisdom says that bytes in Function are precious, and the
vast, vast majority of globals do not live in special sections. Even
when they do, they tend to live in the same section. Store the section
name on the LLVMContext in a StringSet, and maintain a map from
GlobalObject* to section name like we do for metadata, prefix data, etc.

The fact that we've survived this long wasting at least three pointers
of space in Function suggests that Function bytes are perhaps not as
precious as we once thought. Given that most functions have metadata
attachments when debug info is enabled, we might consider adding a
pointer here to make that access more efficient.

Reviewers: jlebar, dexonsmith, mehdi_amini

Subscribers: mehdi_amini, aprantl, llvm-commits

Differential Revision: https://reviews.llvm.org/D28150

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291613 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoInstCombine: Set operands instead of creating new call
Matt Arsenault [Tue, 10 Jan 2017 23:17:52 +0000 (23:17 +0000)]
InstCombine: Set operands instead of creating new call

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291612 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoInstCombine: fdiv -x, -y -> fdiv x, y
Matt Arsenault [Tue, 10 Jan 2017 23:08:54 +0000 (23:08 +0000)]
InstCombine: fdiv -x, -y -> fdiv x, y

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291611 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoCodeGen: Allow small copyable blocks to "break" the CFG.
Kyle Butt [Tue, 10 Jan 2017 23:04:30 +0000 (23:04 +0000)]
CodeGen: Allow small copyable blocks to "break" the CFG.

When choosing the best successor for a block, ordinarily we would have preferred
a block that preserves the CFG unless there is a strong probability the other
direction. For small blocks that can be duplicated we now skip that requirement
as well.

Differential revision: https://reviews.llvm.org/D27742

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291609 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove unused CONVERT_RNDSAT intrinsics
Matt Arsenault [Tue, 10 Jan 2017 22:38:02 +0000 (22:38 +0000)]
Remove unused CONVERT_RNDSAT intrinsics

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291607 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd the 'googlemock' component of Google Test to LLVM's unittest libraries.
Chandler Carruth [Tue, 10 Jan 2017 22:32:26 +0000 (22:32 +0000)]
Add the 'googlemock' component of Google Test to LLVM's unittest libraries.

I have two immediate motivations for adding this:
1) It makes writing expectations in tests *dramatically* easier. A
   quick example that is a taste of what is possible:

     std::vector<int> v = ...;
     EXPECT_THAT(v, UnorderedElementsAre(1, 2, 3));

   This checks that v contains '1', '2', and '3' in some order. There
   are a wealth of other helpful matchers like this. They tend to be
   highly generic and STL-friendly so they will in almost all cases work
   out of the box even on custom LLVM data structures.

   I actually find the matcher syntax substantially easier to read even
   for simple assertions:

     EXPECT_THAT(a, Eq(b));
     EXPECT_THAT(b, Ne(c));

   Both of these make it clear what is being *tested* and what is being
   *expected*. With `EXPECT_EQ` this is implicit (the LHS is expected,
   the RHS is tested) and often confusing. With `EXPECT_NE` it is just
   not clear. Even the failure error messages are superior with the
   matcher based expectations.

2) When testing any kind of generic code, you are continually defining
   dummy types with interfaces and then trying to check that the
   interfaces are manipulated in a particular way. This is actually what
   mocks are *good* for -- testing *interface interactions*. With
   generic code, there is often no "fake" or other object that can be
   used.

   For a concrete example of where this is currently causing significant
   pain, look at the pass manager unittests which are riddled with
   counters incremented when methods are called. All of these could be
   replaced with mocks. The result would be more effective at testing
   the code by having tighter constraints. It would be substantially
   more readable and maintainable when updating the code. And the error
   messages on failure would have substantially more information as
   mocks automatically record stack traces and other information *when
   the API is misused* instead of trying to diagnose it after the fact.

I expect that #1 will be the overwhelming majority of the uses of gmock,
but I think that is sufficient to justify having it. I would actually
like to update the coding standards to encourage the use of matchers
rather than any other form of `EXPECT_...` macros as they are IMO
a strict superset in terms of functionality and readability.

I think that #2 is relatively rarely useful, but there *are* cases where
it is useful. Historically, I think misuse of actual mocking as
described in #2 has led to resistance towards this framework. I am
actually sympathetic to this -- mocking can easily be overused. However
I think this is not a significant concern in LLVM. First and foremost,
LLVM has very careful and rare exposure of abstract interfaces or
dependency injection, which are the most prone to abuse with mocks. So
there are few opportunities to abuse them. Second, a large fraction of
LLVM's unittests are testing *generic code* where mocks actually make
tremendous sense. And gmock is well suited to building interfaces that
exercise generic libraries. Finally, I still think we should be willing
to have testing utilities in tree even if they should be used rarely. We
can use code review to help guide the usage here.

For a longer and more complete discussion of this, see the llvm-dev
thread here:
http://lists.llvm.org/pipermail/llvm-dev/2017-January/108672.html

The general consensus seems that this is a reasonable direction to start
down, but that doesn't mean we should race ahead and use this
everywhere. I have one test that is blocked on this to land and that was
specifically used as an example. Before widespread adoption, I'm going
to work up some (brief) guidelines as some of these facilities should be
used sparingly and carefully.

Differential Revision: https://reviews.llvm.org/D28156

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291606 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMake the test accept different OpCode values since it doesn't really care about the...
Douglas Yung [Tue, 10 Jan 2017 22:10:22 +0000 (22:10 +0000)]
Make the test accept different OpCode values since it doesn't really care about the value.

Differential Revision: https://reviews.llvm.org/D28487

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291605 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDAG: Avoid OOB when legalizing vector indexing
Matt Arsenault [Tue, 10 Jan 2017 22:02:30 +0000 (22:02 +0000)]
DAG: Avoid OOB when legalizing vector indexing

If a vector index is out of bounds, the result is supposed to be
undefined but is not undefined behavior. Change the legalization
for indexing the vector on the stack so that an out of bounds
index does not create an out of bounds memory access.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291604 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[WebAssembly] Only RAUW a constant once in FixFunctionBitcasts
Derek Schuff [Tue, 10 Jan 2017 21:59:53 +0000 (21:59 +0000)]
[WebAssembly] Only RAUW a constant once in FixFunctionBitcasts

When we collect 2 uses of a function in FindUses and then RAUW when we
visit the first, we end up visiting the wrapper (because the second was
RAUW'd).  We still want to use RAUW instead of just Use->set() because
it has special handling for Constants, so this patch just ensures that
only one use of each constant is added to the work list.

Differential Revision: https://reviews.llvm.org/D28504

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291603 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoCorrect object file for implicit const test
Victor Leschuk [Tue, 10 Jan 2017 21:30:42 +0000 (21:30 +0000)]
Correct object file for implicit const test

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291601 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDebugInfo: support for DW_FORM_implicit_const
Victor Leschuk [Tue, 10 Jan 2017 21:18:26 +0000 (21:18 +0000)]
DebugInfo: support for DW_FORM_implicit_const

Support for DW_FORM_implicit_const DWARFv5 feature.
When this form is used attribute value goes to .debug_abbrev section (as SLEB).
As this form would break any debug tool which doesn't support DWARFv5
it is guarded by dwarf version check. Attempt to use this form with
dwarf version <= 4 is considered a fatal error.

Differential Revision: https://reviews.llvm.org/D28456

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291599 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix memory leak in a unit test.
Rui Ueyama [Tue, 10 Jan 2017 20:07:58 +0000 (20:07 +0000)]
Fix memory leak in a unit test.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291595 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[llvm-config] Canonicalize CMake booleans to 0/1
Michal Gorny [Tue, 10 Jan 2017 19:55:51 +0000 (19:55 +0000)]
[llvm-config] Canonicalize CMake booleans to 0/1

Following the similar change to lit configuration, ensure that all CMake
booleans are canonicalized to 0/1 when being passed to llvm-config. This
fixes the incorrect interpretation of values when user passes another
value than the ON/OFF, and simplifies the code by removing unnecessary
string matching.

Furthermore, the code for --has-rtti and --has-global-isel has been
modified to print consistent values indepdently of the boolean used by
passed by the user to CMake. Sadly, the code already implicitly used
different values for the two (YES/NO for --has-rtti, ON/OFF for
--has-global-isel).

Include tests for all booleans and multi-value options in llvm-config.

Differential Revision: https://reviews.llvm.org/D28366

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291593 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CMake] Handle common options for runtimes build
Petr Hosek [Tue, 10 Jan 2017 19:47:05 +0000 (19:47 +0000)]
[CMake] Handle common options for runtimes build

All the existing runtimes relies on flags which are set by AddLLVM
and HandleLLVMOptions. In the standalone case, they would include
these themselves, but when being built using LLVM runtimes we should
include these in the top-level runtimes CMake files.

Differential Revision: https://reviews.llvm.org/D28389

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291590 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[LV] Don't panic when encountering the IV of an outer loop.
Michael Kuperstein [Tue, 10 Jan 2017 19:32:30 +0000 (19:32 +0000)]
[LV] Don't panic when encountering the IV of an outer loop.

Bail out instead of asserting when we encounter this situation,
which can actually happen.

The reason the test uses the new PM is that the "bad" phi, incidentally, gets
cleaned up by LoopSimplify. But LICM can create this kind of phi and preserve
loop simplify form, so the cleanup has no chance to run.

This fixes PR31190.
We may want to solve this in a less conservative manner, since this phi is
actually uniform within the inner loop (or we may want LICM to output a cleaner
promotion to begin with).

Differential Revision: https://reviews.llvm.org/D28490

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291589 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[PGO] Turn off comdat renaming in IR PGO by default
Rong Xu [Tue, 10 Jan 2017 19:30:20 +0000 (19:30 +0000)]
[PGO] Turn off comdat renaming in IR PGO by default

Summary:
In IR PGO we append the function hash to comdat functions to avoid the
potential hash mismatch. This turns out not legal in some cases: if the comdat
function is address-taken and used in comparison. Renaming changes the semantic.

This patch turns off comdat renaming by default.

To alleviate the hash mismatch issue, we now rename the profile variable
for comdat functions. Profile allows co-existing multiple versions of profiles
with different hash value. The inlined copy will always has the correct profile
counter. The out-of-line copy might not have the correct count. But we will
not have the bogus mismatch warning.

Reviewers: davidxl

Subscribers: llvm-commits, xur

Differential Revision: https://reviews.llvm.org/D28416

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291588 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM] Remove rbit intrinsics and autoupgrade to generic bitreverse.
Chad Rosier [Tue, 10 Jan 2017 19:23:51 +0000 (19:23 +0000)]
[ARM] Remove rbit intrinsics and autoupgrade to generic bitreverse.

Testing already covered by CodeGen/ARM/rbit.ll

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291587 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Add tests for HasMultipleConditionRegisters
Matt Arsenault [Tue, 10 Jan 2017 19:08:15 +0000 (19:08 +0000)]
AMDGPU: Add tests for HasMultipleConditionRegisters

This was enabled without many specific tests or the comment.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291586 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CostModel][X86] Add AVX512VL vector shift cost tests.
Simon Pilgrim [Tue, 10 Jan 2017 19:04:12 +0000 (19:04 +0000)]
[CostModel][X86] Add AVX512VL vector shift cost tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291585 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX512]Improving shuffle lowering by using AVX-512 EXPAND* instructions
Michael Zuckerman [Tue, 10 Jan 2017 18:57:17 +0000 (18:57 +0000)]
[X86][AVX512]Improving shuffle lowering by using AVX-512 EXPAND* instructions

This patch fix PR31351: https://llvm.org/bugs/show_bug.cgi?id=31351

1.  This patch adds new type of shuffle lowering
2.  We can use the expand instruction, When the shuffle pattern is as following:
    { 0*a[0]0*a[1]...0*a[n] , n >=0 where a[] elements in a ascending order}.

Reviewers: 1. igorb
           2. guyblank
           3. craig.topper
           4. RKSimon

Differential Revision: https://reviews.llvm.org/D28352

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291584 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoLangRef: Note that calls also support fast math flags
Matt Arsenault [Tue, 10 Jan 2017 18:06:38 +0000 (18:06 +0000)]
LangRef: Note that calls also support fast math flags

This is already documented on the call instruction, but
not in the list of supported instructions in the fast math
flag section.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291578 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[SimplifyLibCalls] Propagate fast math flags while optimizing pow().
Davide Italiano [Tue, 10 Jan 2017 18:02:05 +0000 (18:02 +0000)]
[SimplifyLibCalls] Propagate fast math flags while optimizing pow().

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291577 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AArch64] Add support for lowering bitreverse to the rbit instruction.
Chad Rosier [Tue, 10 Jan 2017 17:20:33 +0000 (17:20 +0000)]
[AArch64] Add support for lowering bitreverse to the rbit instruction.

Differential Revision: https://reviews.llvm.org/D28379

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291575 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[mips] Fix Mips MSA instrinsics
Simon Dardis [Tue, 10 Jan 2017 16:40:57 +0000 (16:40 +0000)]
[mips] Fix Mips MSA instrinsics

The usage of some MIPS MSA instrinsics that took immediates could crash LLVM
during lowering. This patch addresses that behaviour. Crucially this patch
also makes the use of intrinsics with out of range immediates as producing an
internal error.

The ld,st instrinsics would trigger an assertion failure for MIPS64 as their
lowering would attempt to add an i32 offset to a i64 pointer.

Reviewers: vkalintiris, slthakur

Differential Revision: https://reviews.llvm.org/D25438

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291571 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[mips] Honour -mno-odd-spreg for vector splat (again)
Simon Dardis [Tue, 10 Jan 2017 15:53:10 +0000 (15:53 +0000)]
[mips] Honour -mno-odd-spreg for vector splat (again)

Previous the lowering of FILL_FW would use the MSA128W register class when
performing a vector splat. Instead it should be honouring -mno-odd-spreg and
only use the even registers when performing a splat from word to vector
register.

Logical follow-on from r230235.

This fixes PR/31369.

A previous commit was missing the test case and had another differential
in it.

Reviewers: slthakur

Differential Revision: https://reviews.llvm.org/D28373

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291566 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "[mips] Honour -mno-odd-spreg for vector splat"
Simon Dardis [Tue, 10 Jan 2017 13:57:44 +0000 (13:57 +0000)]
Revert "[mips] Honour -mno-odd-spreg for vector splat"

This reverts commit r291556. It was a mixture of two differentials and
was missing a test.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291562 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRuntimeDyldELF: implement R_AARCH64_PREL64 reloc
Eugene Leviant [Tue, 10 Jan 2017 11:05:30 +0000 (11:05 +0000)]
RuntimeDyldELF: implement R_AARCH64_PREL64 reloc

Differential revision: https://reviews.llvm.org/D28122

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291558 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[mips] Honour -mno-odd-spreg for vector splat
Simon Dardis [Tue, 10 Jan 2017 10:28:37 +0000 (10:28 +0000)]
[mips] Honour -mno-odd-spreg for vector splat

Previous the lowering of FILL_FW would use the MSA128W register class when
performing a vector splat. Instead it should be honouring -mno-odd-spreg and
only use the even registers when performing a splat from word to vector
register.

Logical follow-on from r230235.

This fixes PR/31369.

Reviewers: slthakur

Differential Revision: https://reviews.llvm.org/D28373

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291556 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[DAGCombiner] Merge together duplicate checks for folding fold (select C, 1, X) ...
Craig Topper [Tue, 10 Jan 2017 07:42:57 +0000 (07:42 +0000)]
[DAGCombiner] Merge together duplicate checks for folding fold (select C, 1, X) -> (or C, X)  and  folding (select C, X, 0) -> (and C, X). Also be consistent about checking that both the condition and the result type are i1. NFC

I guess previously we just assumed if the result type was i1, then the condition type must also be i1?

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291548 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ObjectYAML] Missed one mixup in the debug_line test
Chris Bieneman [Tue, 10 Jan 2017 06:24:24 +0000 (06:24 +0000)]
[ObjectYAML] Missed one mixup in the debug_line test

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291547 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ObjectYAML] Support for DWARF line tables
Chris Bieneman [Tue, 10 Jan 2017 06:22:49 +0000 (06:22 +0000)]
[ObjectYAML] Support for DWARF line tables

One more try... relanding r291541 with a fix to properly gate MaxOpsPerInst on DWARF version.

Description from r291541:

This patch re-lands r291470, which failed on Linux bots. The issue (I believe) was undefined behavior because the size of llvm::dwarf::LineNumberOps was not explcitly specified or consistently respected. The updated patch adds an explcit underlying type to the enum and preserves the size more correctly.

Original description:

This patch adds support for the DWARF debug_lines section. The line table state machine opcodes are preserved, so this can be used to test the state machine evaluation directly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291546 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMD family 17h (znver1) enablement
Craig Topper [Tue, 10 Jan 2017 06:01:16 +0000 (06:01 +0000)]
AMD family 17h (znver1) enablement

Summary:
This patch enables the following
1. AMD family 17h architecture using "znver1" tune flag (-march, -mcpu).
2. ISAs that are enabled for "znver1" architecture.
3. Checks ADX isa from cpuid to identify "znver1" flag when -march=native is used.
4. ISAs FMA4, XOP are disabled as they are dropped from amdfam17.
5. For the time being, it uses the btver2 scheduler model.
6. Test file is updated to check this flag.

This item is linked to clang review item https://reviews.llvm.org/D28018

Patch by Ganesh Gopalasubramanian

Reviewers: RKSimon, craig.topper

Subscribers: vprasad, RKSimon, ashutosh.nema, llvm-commits

Differential Revision: https://reviews.llvm.org/D28017

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291543 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "[ObjectYAML] Support for DWARF line tables"
Chris Bieneman [Tue, 10 Jan 2017 05:31:23 +0000 (05:31 +0000)]
Revert "[ObjectYAML] Support for DWARF line tables"

This reverts commit r291541.

Still failing on a bot:

http://bb.pgr.jp/builders/cmake-llvm-x86_64-linux/builds/47224/steps/test_llvm/logs/stdio

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291542 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ObjectYAML] Support for DWARF line tables
Chris Bieneman [Tue, 10 Jan 2017 05:25:24 +0000 (05:25 +0000)]
[ObjectYAML] Support for DWARF line tables

This patch re-lands r291470, which failed on Linux bots. The issue (I believe) was undefined behavior because the size of llvm::dwarf::LineNumberOps was not explcitly specified or consistently respected. The updated patch adds an explcit underlying type to the enum and preserves the size more correctly.

Original description:

This patch adds support for the DWARF debug_lines section. The line table state machine opcodes are preserved, so this can be used to test the state machine evaluation directly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291541 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[XRay] Use regular expression for finding symbols
Dean Michael Berris [Tue, 10 Jan 2017 04:32:34 +0000 (04:32 +0000)]
[XRay] Use regular expression for finding symbols

Un-break the test in Windows.

Follow-up on D24376.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291538 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] When lowering uniform shifts, use X86ISD::VZEXT instead of using a ZERO_EXTEND_...
Craig Topper [Tue, 10 Jan 2017 04:12:24 +0000 (04:12 +0000)]
[X86] When lowering uniform shifts, use X86ISD::VZEXT instead of using a ZERO_EXTEND_VECTOR_INREG. If we emit the ZERO_EXTEND_VECTOR_INREG too late it doesn't get lowered properly and makes it through to isel and fails.

Fixes PR31593.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291535 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[DAGCombiner] Remove code for optimizing select (xor Cond, 0), X, Y -> select Cond...
Craig Topper [Tue, 10 Jan 2017 04:12:19 +0000 (04:12 +0000)]
[DAGCombiner] Remove code for optimizing select (xor Cond, 0), X, Y -> select Cond, X, Y. Just let combine on the xor itself take care of it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291534 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[XRay] Fixup includes for modules build
Dean Michael Berris [Tue, 10 Jan 2017 03:21:54 +0000 (03:21 +0000)]
[XRay] Fixup includes for modules build

Remove unnecessary system header include.

Follow-up to D24376.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291533 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix a typo and also test a new machine for commit. NFC.
Xin Tong [Tue, 10 Jan 2017 03:13:52 +0000 (03:13 +0000)]
Fix a typo and also test a new machine for commit. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291532 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[XRay] Don't include <unistd.h> unnecessarily
Dean Michael Berris [Tue, 10 Jan 2017 02:51:13 +0000 (02:51 +0000)]
[XRay] Don't include <unistd.h> unnecessarily

Follow-up to D24376.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291531 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[StructurizeCfg] Update dominator info.
Serge Pavlov [Tue, 10 Jan 2017 02:50:47 +0000 (02:50 +0000)]
[StructurizeCfg] Update dominator info.

In some cases StructurizeCfg updates root node, but dominator info
remains unchanges, it causes crash when expensive checks are enabled.
To cope with this problem a new method was added to DominatorTreeBase
that allows adding new root nodes, it is called in StructurizeCfg to
put dominator tree in sync.

This change fixes PR27488.

Differential Revision: https://reviews.llvm.org/D28114

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291530 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[XRay] Implement `llvm-xray convert` -- trace file conversion
Dean Michael Berris [Tue, 10 Jan 2017 02:38:11 +0000 (02:38 +0000)]
[XRay] Implement `llvm-xray convert` -- trace file conversion

This is the second part of a multi-part change to define additional
subcommands to the `llvm-xray` tool.

This change defines a conversion subcommand to take XRay log files, and
turns them from one format to another (binary or YAML). This currently
only supports the first version of the log file format, defined in the
compiler-rt runtime.

Depends on D21987.

Reviewers: dblaikie, echristo

Subscribers: mehdi_amini, dberris, beanz, llvm-commits

Differential Revision: https://reviews.llvm.org/D24376

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291529 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeGen] Implement the SUnit::print() method
Evandro Menezes [Tue, 10 Jan 2017 01:08:01 +0000 (01:08 +0000)]
[CodeGen] Implement the SUnit::print() method

This method seems to have had a troubled life.  This patch proposes that it
replaces the recently added helper function dumpSUIdentifier.  This way, the
method can be used in other files using the SUnit class.

Differential revision: https://reviews.llvm.org/D28488

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291520 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTry once again to fix the MSVC build of AlignedCharArrayUnion
Reid Kleckner [Tue, 10 Jan 2017 01:05:33 +0000 (01:05 +0000)]
Try once again to fix the MSVC build of AlignedCharArrayUnion

It was complaining about ambiguity between llvm::detail and
llvm::support::detail:
  error C2872: 'detail': ambiguous symbol
  note: could be 'llvm::detail'
  note: or       'llvm::support::detail'

Standardize on llvm::support::detail to hide these symbols further.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291519 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ThinLTO] Hash more part of the config to build cache entries
Mehdi Amini [Tue, 10 Jan 2017 00:55:47 +0000 (00:55 +0000)]
[ThinLTO] Hash more part of the config to build cache entries

This has been fixed in the "new" LTO API used by Gold/LLD, this is
fixing the same issue in the libLTO API used by ld64 (amongst other)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291518 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd an assert for hasLoopInvariantOperands
Xin Tong [Tue, 10 Jan 2017 00:39:49 +0000 (00:39 +0000)]
Add an assert for hasLoopInvariantOperands

Summary: Add an assert for hasLoopInvariantOperands

Reviewers: danielcdh, sanjoy

Subscribers: mzolotukhin, llvm-commits

Differential Revision: https://reviews.llvm.org/D28501

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291516 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert the attempt to optimize the constexpr functions. MSVC does not handle this yet
Reid Kleckner [Tue, 10 Jan 2017 00:29:05 +0000 (00:29 +0000)]
Revert the attempt to optimize the constexpr functions. MSVC does not handle this yet

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291515 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix MSVC build of AlignedCharArrayUnion
Reid Kleckner [Tue, 10 Jan 2017 00:26:56 +0000 (00:26 +0000)]
Fix MSVC build of AlignedCharArrayUnion

Use constexpr recursion for alignof like we do for sizeof. Seems to work
with Clang and MSVC. Also, don't recurse twice to avoid slowdowns in
compilers that don't memoize constexpr results (Clang).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291514 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoLift the 10-type limit for AlignedCharArrayUnion
Sean Callanan [Mon, 9 Jan 2017 23:23:25 +0000 (23:23 +0000)]
Lift the 10-type limit for AlignedCharArrayUnion

This patch uses C++11 parameter packs and constexpr functions
to allow AlignedCharArrayUnion to hold an arbitrary number of
types.

Differential Revision: https://reviews.llvm.org/D28429

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291503 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoCommit a test for match-full-lines.
James Y Knight [Mon, 9 Jan 2017 23:11:25 +0000 (23:11 +0000)]
Commit a test for match-full-lines.

I unfortunately neglected to add it in r260540, but it has been
sitting in my working dir ever since. D'oh.

Modified to work with r290069, which made the CHECK patterns
themselves whitespace-sensitive as well, and remove the test added
then, as this tests both strict and non-strict modes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291499 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[WebAssembly] Add return type annotations in fast isel.
Dan Gohman [Mon, 9 Jan 2017 23:09:38 +0000 (23:09 +0000)]
[WebAssembly] Add return type annotations in fast isel.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291498 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTarWriter: Fix a bug in Ustar header.
Rui Ueyama [Mon, 9 Jan 2017 22:55:00 +0000 (22:55 +0000)]
TarWriter: Fix a bug in Ustar header.

If we split a filename into `Name` and `Prefix`, `Prefix` is at most
145 bytes. We had a bug that didn't split a path correctly. This bug
was pointed out by Rafael in the post commit review.

This patch adds a unit test for TarWriter to verify the fix.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291494 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[NVPTX] Fix some Clang-tidy modernize and Include What You Use warnings; other minor...
Eugene Zelenko [Mon, 9 Jan 2017 22:16:51 +0000 (22:16 +0000)]
[NVPTX] 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@291490 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX512VL] Added AVX512VL to 128/256 bit vector shift tests
Simon Pilgrim [Mon, 9 Jan 2017 22:13:51 +0000 (22:13 +0000)]
[X86][AVX512VL] Added AVX512VL to 128/256 bit vector shift tests

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291488 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRefactor inline threshold update code.
Easwaran Raman [Mon, 9 Jan 2017 21:56:26 +0000 (21:56 +0000)]
Refactor inline threshold update code.

Functional change: Previously, if a callee is cold, we used ColdThreshold if it minimizes the existing threshold. This was irrespective of whether we were optimizing for minsize (-Oz) or not. But -Oz uses very low threshold to begin with and the inlining with -Oz is expected to be tuned for lowering code size, so there is no good reason to set an even lower threshold for cold callees. We now lower the threshold for cold callees only when -Oz is not used. For default values of -inlinethreshold and -inlinecold-threshold, this change has no effect and this simplifies the code.

NFC changes: Group all threshold updates that are guarded by !Caller->optForMinSize() and within that group threshold updates that require profile summary info.

Differential revision: https://reviews.llvm.org/D28369

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291487 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[SimplifyLibCalls] pow(x, -0.5) -> 1.0 / sqrt(x).
Davide Italiano [Mon, 9 Jan 2017 21:55:23 +0000 (21:55 +0000)]
[SimplifyLibCalls] pow(x, -0.5) -> 1.0 / sqrt(x).

Differential Revision:  https://reviews.llvm.org/D28479

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291486 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoSupport outputting to /dev/null.
Rafael Espindola [Mon, 9 Jan 2017 21:52:35 +0000 (21:52 +0000)]
Support outputting to /dev/null.

When writing to a non regular file we cannot rename to it. Since we
have to write, we may as well create a temporary file to avoid trying
to create an unique file in /dev when trying to write to /dev/null.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291485 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoPeepholeOptimizer: Do not replace SubregToReg(bitcast like)
Matthias Braun [Mon, 9 Jan 2017 21:38:17 +0000 (21:38 +0000)]
PeepholeOptimizer: Do not replace SubregToReg(bitcast like)

While we can usually replace bitcast like instructions
(MachineInstr::isBitcast()) with a COPY this is not legal if any of the
users uses SUBREG_TO_REG to assert the upper bits of the result are
zero.

Differential Revision: https://reviews.llvm.org/D28474

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291483 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDrive by typo fix
Matthias Braun [Mon, 9 Jan 2017 21:38:14 +0000 (21:38 +0000)]
Drive by typo fix

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291482 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMachineInstr: Print name for subreg index in SUBREG_TO_REG
Matthias Braun [Mon, 9 Jan 2017 21:38:10 +0000 (21:38 +0000)]
MachineInstr: Print name for subreg index in SUBREG_TO_REG

SUBREG_TO_REG takes a subregister index as 3rd operand, print the name
instead of a number. We already do the same for INSERT_SUBREG and
REG_SEQUENCE.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291481 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTarWriter: Set "00" to Ustar version field.
Rui Ueyama [Mon, 9 Jan 2017 21:20:42 +0000 (21:20 +0000)]
TarWriter: Set "00" to Ustar version field.

Most (maybe all?) tar commands can handle tar archives with blank
version fields, but POSIX requires "00" to be set to the field, so
doing it is good for compliance.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291479 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert r291092 because it introduces a crash.
Michael Kuperstein [Mon, 9 Jan 2017 21:04:46 +0000 (21:04 +0000)]
Revert r291092 because it introduces a crash.

See PR31589 for details.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291478 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoX86-specific path: Implemented the fusing of MUL+ADDSUB to FMADDSUB.
Vyacheslav Klochkov [Mon, 9 Jan 2017 20:26:17 +0000 (20:26 +0000)]
X86-specific path: Implemented the fusing of MUL+ADDSUB to FMADDSUB.
Differential Revision: https://reviews.llvm.org/D28087

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291473 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[InstCombine] add test to show missed fold using llvm.assume; NFC
Sanjay Patel [Mon, 9 Jan 2017 20:18:30 +0000 (20:18 +0000)]
[InstCombine] add test to show missed fold using llvm.assume; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291472 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "[ObjectYAML] Support for DWARF line tables"
Chris Bieneman [Mon, 9 Jan 2017 20:04:55 +0000 (20:04 +0000)]
Revert "[ObjectYAML] Support for DWARF line tables"

This reverts commit r291470 due to failing bots:

http://bb.pgr.jp/builders/cmake-llvm-x86_64-linux/builds/47209/steps/test_llvm/logs/stdio

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291471 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ObjectYAML] Support for DWARF line tables
Chris Bieneman [Mon, 9 Jan 2017 20:01:37 +0000 (20:01 +0000)]
[ObjectYAML] Support for DWARF line tables

This patch adds support for the DWARF debug_lines section. The line table state machine opcodes are preserved, so this can be used to test the state machine evaluation directly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291470 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[InstCombine] regenerate checks; NFC
Sanjay Patel [Mon, 9 Jan 2017 19:43:26 +0000 (19:43 +0000)]
[InstCombine] regenerate checks; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291469 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ValueTracking] regenerate checks; NFC
Sanjay Patel [Mon, 9 Jan 2017 19:31:20 +0000 (19:31 +0000)]
[ValueTracking] regenerate checks; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291468 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix function regex in update_tests so it can handle {}'s in function args
Daniel Berlin [Mon, 9 Jan 2017 19:24:19 +0000 (19:24 +0000)]
Fix function regex in update_tests so it can handle {}'s in function args

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291467 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[InstCombine] regenerate checks; NFC
Sanjay Patel [Mon, 9 Jan 2017 19:18:46 +0000 (19:18 +0000)]
[InstCombine] regenerate checks; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291464 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[InstCombine] remove unnecessary attribute comments from test files; NFC
Sanjay Patel [Mon, 9 Jan 2017 19:13:38 +0000 (19:13 +0000)]
[InstCombine] remove unnecessary attribute comments from test files; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291463 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[LV] Fix-up external IV users after updating dominator tree
Matthew Simpson [Mon, 9 Jan 2017 19:05:29 +0000 (19:05 +0000)]
[LV] Fix-up external IV users after updating dominator tree

This patch delays the fix-up step for external induction variable users until
after the dominator tree has been properly updated. This should fix PR30742.
The SCEVExpander in InductionDescriptor::transform can generate code in the
wrong location if the dominator tree is not up-to-date. We should work towards
keeping the dominator tree up-to-date throughout the transformation.

Reference: https://llvm.org/bugs/show_bug.cgi?id=30742
Differential Revision: https://reviews.llvm.org/D28168

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291462 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAMDGPU: Add Assert[SZ]Ext during argument load creation
Matt Arsenault [Mon, 9 Jan 2017 18:52:39 +0000 (18:52 +0000)]
AMDGPU: Add Assert[SZ]Ext during argument load creation

For i16 zeroext arguments when i16 was a legal type, the
known bits information from the truncate was lost. Insert
a zeroext so the known bits optimizations work with the 32-bit
loads.

Fixes code quality regressions vs. SI in min.ll test.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@291461 91177308-0d34-0410-b5e6-96231b3b80d8