platform/upstream/llvm.git
6 years ago[WebAssembly] Initial wasm linker implementation
Sam Clegg [Fri, 17 Nov 2017 18:14:09 +0000 (18:14 +0000)]
[WebAssembly] Initial wasm linker implementation

This linker backend is still a work in progress but is
enough to link simple programs including linking against
library archives.

Differential Revision: https://reviews.llvm.org/D34851

llvm-svn: 318539

6 years ago[AST] Partially revert r318341 to fix two broken tests on llvm-clang-x86_64-expensive...
Eugene Zelenko [Fri, 17 Nov 2017 18:09:48 +0000 (18:09 +0000)]
[AST] Partially revert r318341 to fix two broken tests on llvm-clang-x86_64-expensive-checks-win (NFC).

llvm-svn: 318538

6 years agoclang-format: remove trailing lines in lamdas and arrow functions.
Martin Probst [Fri, 17 Nov 2017 18:06:33 +0000 (18:06 +0000)]
clang-format: remove trailing lines in lamdas and arrow functions.

Summary:
clang-format already removes empty lines at the beginning & end of
blocks:

    int x() {

      foo();  // lines before and after will be removed.

    }

However because lamdas and arrow functions are parsed as expressions,
the existing logic to remove empty lines in UnwrappedLineFormatter
doesn't handle them.

This change special cases arrow functions in ContinuationIndenter to
remove empty lines:

    x = []() {

      foo();  // lines before and after will now be removed.

    };

Reviewers: djasper

Subscribers: klimek, cfe-commits

Differential Revision: https://reviews.llvm.org/D40178

llvm-svn: 318537

6 years ago[OPENMP] Codegen for `target simd` construct.
Alexey Bataev [Fri, 17 Nov 2017 17:57:25 +0000 (17:57 +0000)]
[OPENMP] Codegen for `target simd` construct.

Added codegen support for `target simd` directive.

llvm-svn: 318536

6 years agoTry to fix the windows build.
Rafael Espindola [Fri, 17 Nov 2017 17:46:39 +0000 (17:46 +0000)]
Try to fix the windows build.

llvm-svn: 318535

6 years agoUse TempFile in dsymutil.
Rafael Espindola [Fri, 17 Nov 2017 17:33:09 +0000 (17:33 +0000)]
Use TempFile in dsymutil.

I don't think there is any functionality change, but the code is
easier to understand IMHO.

llvm-svn: 318534

6 years agoRevert "[MC] Fix regression tests on Windows when git “core.autocrlf” is set to true."
Rafael Espindola [Fri, 17 Nov 2017 17:31:20 +0000 (17:31 +0000)]
Revert "[MC] Fix regression tests on Windows when git “core.autocrlf” is set to true."

This reverts commit r318528.

MC/AsmParser/preserve-comments-crlf.s fails on linux.

llvm-svn: 318533

6 years ago[CMake] Allow for Solaris ld -V output on stdout
Michal Gorny [Fri, 17 Nov 2017 17:12:14 +0000 (17:12 +0000)]
[CMake] Allow for Solaris ld -V output on stdout

In recent versions of Solaris 11.4 (previously 12), ld -V output went to
stdout instead of stderr. Since AddLLVM.cmake only expects it on stderr,
Solaris ld wasn't properly detected and options not understood by it are
passed during the build.

The following patch fixes this, allowing for both variants.

Tested on i386-pc-solaris2.11.4 (on top of D35755 which is needed for
proper Solaris support).

Patch by Rainer Orth.

Differential Revision: https://reviews.llvm.org/D39601

llvm-svn: 318532

6 years ago[AArch64] Adjust the cost model for Exynos M1 and M2
Evandro Menezes [Fri, 17 Nov 2017 16:42:15 +0000 (16:42 +0000)]
[AArch64] Adjust the cost model for Exynos M1 and M2

Improve the accuracy of the model by specifying the proper number of uops.

llvm-svn: 318531

6 years agoIndent code blocks so they are actually treated as such
Stephan Bergmann [Fri, 17 Nov 2017 16:34:36 +0000 (16:34 +0000)]
Indent code blocks so they are actually treated as such

llvm-svn: 318530

6 years agoFix skipping of flags in getClangStripDependencyFileAdjuster
Dave Lee [Fri, 17 Nov 2017 16:27:21 +0000 (16:27 +0000)]
Fix skipping of flags in getClangStripDependencyFileAdjuster

Summary:
The ArgumentsAdjuster returned from `getClangStripDependencyFileAdjuster` will
skip dependency flags, and also their associated values for those flags that
take an argument. This change corrects the handling of the `-MD` and `-MMD`
flags, which do not take an argument.

Reviewers: saugustine, klimek, alexshap

Reviewed By: alexshap

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D40024

llvm-svn: 318529

6 years ago[MC] Fix regression tests on Windows when git “core.autocrlf” is set to true.
Zhen Cao [Fri, 17 Nov 2017 16:17:56 +0000 (16:17 +0000)]
[MC] Fix regression tests on Windows when git “core.autocrlf” is set to true.

Differential Revision: https://reviews.llvm.org/D39737

llvm-svn: 318528

6 years ago[OMPT] Fix inaccuracies in worksharing tests
Jonas Hahnfeld [Fri, 17 Nov 2017 15:26:44 +0000 (15:26 +0000)]
[OMPT] Fix inaccuracies in worksharing tests

These tests were failing rarely on my MacBook when there was some
activity in the background. Read: one of a thousand executions?

 * sections.c missed the sorting based on thread ids. This worked
   as long as the master thread finished its section before the
   worker thread started the second one but failed if the master
   thread was put to sleep by the OS.
 * The checks in single.c assumed that the master thread executes
   the single region which works most of the time because it is
   usually faster than the newly spawned worker thread.

Differential Revision: https://reviews.llvm.org/D39853

llvm-svn: 318527

6 years ago[AMDGPU][MC][GFX9][disassembler] Corrected decoding of op_sel_hi for v_mad_mix*
Dmitry Preobrazhensky [Fri, 17 Nov 2017 15:15:40 +0000 (15:15 +0000)]
[AMDGPU][MC][GFX9][disassembler] Corrected decoding of op_sel_hi for v_mad_mix*

See bug 35148: https://bugs.llvm.org//show_bug.cgi?id=35148

Reviewers: tamazov, SamWot, arsenm

Differential Revision: https://reviews.llvm.org/D39492

llvm-svn: 318526

6 years ago[clang-format] Add text proto filename detection
Krasimir Georgiev [Fri, 17 Nov 2017 15:10:49 +0000 (15:10 +0000)]
[clang-format] Add text proto filename detection

Summary: Adds text proto filename detection.

Reviewers: klimek

Reviewed By: klimek

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D40120

llvm-svn: 318525

6 years ago[Support][CachePruning] Fix regression in pruning interval
Ben Dunbobbin [Fri, 17 Nov 2017 14:42:18 +0000 (14:42 +0000)]
[Support][CachePruning] Fix regression in pruning interval

Fixed broken comparison.
borked by: rL284966 (see: https://reviews.llvm.org/D25730).

Differential Revision: https://reviews.llvm.org/D40119

This is a second attempt to commit this.
The first attempt broke lld and gold tests that had been written against
the incorrect behaivour.

llvm-svn: 318524

6 years ago[clang-tidy] Fix an oversight after renaming a check
Gabor Horvath [Fri, 17 Nov 2017 12:28:58 +0000 (12:28 +0000)]
[clang-tidy] Fix an oversight after renaming a check

llvm-svn: 318523

6 years ago[clang-tidy] Add a check for undelegated copy of base classes
Gabor Horvath [Fri, 17 Nov 2017 12:23:30 +0000 (12:23 +0000)]
[clang-tidy] Add a check for undelegated copy of base classes

Finds copy constructors where the constructor don't call
the copy constructor of the base class.

```
class X : public Copyable {
    X(const X &other) {} // Copyable(other) is missing
};
```

Differential Revision: https://reviews.llvm.org/D33722

llvm-svn: 318522

6 years ago[gold] Fix broken thin LTO cache test NFC
Andrew Ng [Fri, 17 Nov 2017 12:00:57 +0000 (12:00 +0000)]
[gold] Fix broken thin LTO cache test NFC

Fix test as it is assuming that the cache pruning is always being
performed by default. Explicitly set prune interval to 0s to ensure
pruning is always performed.

llvm-svn: 318520

6 years ago[ELF] - Teach LLD to use information from .debug_str for error reporting.
George Rimar [Fri, 17 Nov 2017 11:57:47 +0000 (11:57 +0000)]
[ELF] - Teach LLD to use information from .debug_str for error reporting.

Recently we teached LLD to report line numbers for duplicate variables
definitions, though currently LLD is unable to do that for case when
strings are not built in .debug_info, but stored in .debug_str instead.
That is because out LLDDwarfObj does not handle .debug_str yet.
Patch fixes that.

Differential revision: https://reviews.llvm.org/D39542

llvm-svn: 318519

6 years ago[ELF] - Fix BB after r318516
George Rimar [Fri, 17 Nov 2017 11:39:26 +0000 (11:39 +0000)]
[ELF] - Fix BB after r318516

Test did not run under windows because uses zlib.

Content of section changed because now strings
merging is applied.

llvm-svn: 318518

6 years ago[nfc] Iwyu: forward-declare/include raw_ostream in zone algo
Philip Pfaffe [Fri, 17 Nov 2017 11:34:29 +0000 (11:34 +0000)]
[nfc] Iwyu: forward-declare/include  raw_ostream in zone algo

llvm-svn: 318517

6 years ago[ELF] - Allow applying SHF_MERGE optimization for relocatable output.
George Rimar [Fri, 17 Nov 2017 11:27:57 +0000 (11:27 +0000)]
[ELF] - Allow applying SHF_MERGE optimization for relocatable output.

This fixes PR35223.

Here I enabled SHF_MERGE section content merging for -r like
we do for regular linking.

Differential revision: https://reviews.llvm.org/D40026

llvm-svn: 318516

6 years agoImplement more accurate penalty & trade-offs while breaking protruding tokens.
Manuel Klimek [Fri, 17 Nov 2017 11:17:15 +0000 (11:17 +0000)]
Implement more accurate penalty & trade-offs while breaking protruding tokens.

For each line that we break in a protruding token, compute whether the
penalty of breaking is actually larger than the penalty of the excess
characters. Only break if that is the case.

llvm-svn: 318515

6 years agoUse llvm-config.h instead of config.h
Ilya Biryukov [Fri, 17 Nov 2017 10:09:02 +0000 (10:09 +0000)]
Use llvm-config.h instead of config.h

To fix standalone builds broken by r318411 (config.h is private to llvm).

llvm-svn: 318514

6 years agoIgnore -no-ctors-in-init-array.
Rui Ueyama [Fri, 17 Nov 2017 08:17:36 +0000 (08:17 +0000)]
Ignore -no-ctors-in-init-array.

Patch by dxf@google.com.

llvm-svn: 318513

6 years agoAdd a missing #include.
Rui Ueyama [Fri, 17 Nov 2017 08:17:21 +0000 (08:17 +0000)]
Add a missing #include.

llvm-svn: 318512

6 years ago[MinGW] Define __ARM_DWARF_EH__ for MinGW/ARM
Martin Storsjo [Fri, 17 Nov 2017 08:06:49 +0000 (08:06 +0000)]
[MinGW] Define __ARM_DWARF_EH__ for MinGW/ARM

Since SVN r318510, the MinGW/ARM configuration defaults to
dwarf exception handling.

Differential Revision: https://reviews.llvm.org/D39533

llvm-svn: 318511

6 years ago[ARM] Use dwarf exception handling on MinGW
Martin Storsjo [Fri, 17 Nov 2017 08:04:40 +0000 (08:04 +0000)]
[ARM] Use dwarf exception handling on MinGW

Enabling and using dwarf exceptions seems like an easier path
to take, than to make the COFF/ARM backend output EHABI directives.
Previously, no EH model was enabled at all on this target.

There's no point in setting UseIntegratedAssembler to false since
GNU binutils doesn't support Windows on ARM, and since we don't
need to support external assembler, we don't need to use register
numbers in cfi directives.

Differential Revision: https://reviews.llvm.org/D39532

llvm-svn: 318510

6 years ago[SelectionDAG] Allow custom vector widening through ReplaceNodeResults to handle...
Craig Topper [Fri, 17 Nov 2017 07:03:57 +0000 (07:03 +0000)]
[SelectionDAG] Allow custom vector widening through ReplaceNodeResults to handle nodes with chain outputs.

Previously we were assuming all results were vectors and calling SetWidenedVector, but if its a chain result we should just replace uses instead.

This fixes an error found by expensive checks after r318368.

llvm-svn: 318509

6 years ago[IRCE] Remove folding of two range checks into RANGE_CHECK_BOTH
Max Kazantsev [Fri, 17 Nov 2017 06:49:26 +0000 (06:49 +0000)]
[IRCE] Remove folding of two range checks into RANGE_CHECK_BOTH

The logic of replacing of a couple `RANGE_CHECK_LOWER + RANGE_CHECK_UPPER`
into `RANGE_CHECK_BOTH` in fact duplicates the logic of range intersection which
happens when we calculate safe iteration space. Effectively, the result of intersection of
these ranges doesn't differ from the range of merged range check.

We chose to remove duplicating logic in favor of code simplicity.

Differential Revision: https://reviews.llvm.org/D39589

llvm-svn: 318508

6 years agoChange path used in a test from r318503 to work on windows
Bruno Cardoso Lopes [Fri, 17 Nov 2017 04:26:07 +0000 (04:26 +0000)]
Change path used in a test from r318503 to work on windows

http://lab.llvm.org:8011/builders/llvm-clang-lld-x86_64-scei-ps4-windows10pro-fast/builds/13565

llvm-svn: 318507

6 years agoAMDGPU: Replace list of SMEM buffer opcodes
Matt Arsenault [Fri, 17 Nov 2017 04:18:26 +0000 (04:18 +0000)]
AMDGPU: Replace list of SMEM buffer opcodes

llvm-svn: 318506

6 years agoAMDGPU: Fix breaking SMEM clauses
Matt Arsenault [Fri, 17 Nov 2017 04:18:24 +0000 (04:18 +0000)]
AMDGPU: Fix breaking SMEM clauses

This was completely ignoring subregisters,
so was not very useful. Also only break them
if xnack is actually enabled.

llvm-svn: 318505

6 years agoFix LLDB build.
Eugene Zemtsov [Fri, 17 Nov 2017 03:28:58 +0000 (03:28 +0000)]
Fix LLDB build.

It was broken by r318489.

llvm-svn: 318504

6 years ago[PCH+Modules] Improve diagnosticts to help out users pass an extra header search...
Bruno Cardoso Lopes [Fri, 17 Nov 2017 03:24:11 +0000 (03:24 +0000)]
[PCH+Modules] Improve diagnosticts to help out users pass an extra header search path

When mixing PCH and Implicit Modules, missing a header search path
can lead to the implicit built PCM to complaint about not finding its
matching module map.

Instead of adding more magic to implicit modules engine, add a note to
help the user add the appropriate path.

rdar://problem/33388847

llvm-svn: 318503

6 years ago[llvm-profdata] Fix a dangling reference to an error string
Vedant Kumar [Fri, 17 Nov 2017 02:58:23 +0000 (02:58 +0000)]
[llvm-profdata] Fix a dangling reference to an error string

llvm-svn: 318502

6 years ago[GISel]: DCE copy instructions during legalization
Aditya Nandakumar [Fri, 17 Nov 2017 02:44:55 +0000 (02:44 +0000)]
[GISel]: DCE copy instructions during legalization

We might have instructions such as ext(copy(trunc)), and while cleaning
up legalization artifacts, we can also dce the copies that are in
between legalization artifacts.

llvm-svn: 318501

6 years agoUpdate xcode project file to build new ppc64le files.
Jason Molenda [Fri, 17 Nov 2017 01:55:50 +0000 (01:55 +0000)]
Update xcode project file to build new ppc64le files.

llvm-svn: 318500

6 years ago[SelectionDAG] Consolidate (t|T)ransferDbgValues methods, NFC (reapply)
Vedant Kumar [Fri, 17 Nov 2017 01:48:33 +0000 (01:48 +0000)]
[SelectionDAG] Consolidate (t|T)ransferDbgValues methods, NFC (reapply)

TransferDbgValues (capital 'T') is wired into ReplaceAllUsesWith, and
transferDbgValues (lowercase 't') is used elsewhere (e.g in Legalize).

Both functions should be doing the exact same thing. This patch
consolidates the logic into one place.

This was reverted in r318455 because some newly introduced asserts,
which I thought were NFC, were firing. I filed PR35338. For now I've
weakened the asserts.

Testing: check-llvm, check-clang, and a stage2 Rel+Deb build of clang

Differential Revision: https://reviews.llvm.org/D40104

llvm-svn: 318498

6 years agoRemove IsLocal.
Rafael Espindola [Fri, 17 Nov 2017 01:37:50 +0000 (01:37 +0000)]
Remove IsLocal.

Since we always have Binding in the current symbol design IsLocal is
redundant.

llvm-svn: 318497

6 years agoRecognize another magic token sent in the LaunchInfo's EventData.
Jim Ingham [Fri, 17 Nov 2017 01:22:34 +0000 (01:22 +0000)]
Recognize another magic token sent in the LaunchInfo's EventData.

<rdar://problem/32505028>

llvm-svn: 318496

6 years ago"source list -a" was calling DisplaySourceLinesWithNumbers incorrectly.
Jim Ingham [Fri, 17 Nov 2017 01:19:59 +0000 (01:19 +0000)]
"source list -a" was calling DisplaySourceLinesWithNumbers incorrectly.

The parameters were just passed in the wrong order.

<rdar://problem/35150446>

llvm-svn: 318495

6 years ago[CUDA] Remove implementations of nexttoward.
Justin Lebar [Fri, 17 Nov 2017 01:15:43 +0000 (01:15 +0000)]
[CUDA] Remove implementations of nexttoward.

Summary:
__builtin_nexttoward lowers to a libcall, e.g. nexttowardf(), that CUDA
does not have.

Rather than try to implement it, we simply remove these functions --
nvcc doesn't support them either, and nextafter, which does work, does
essentially the same thing on GPUs, because GPUs don't have long double.

Reviewers: tra

Subscribers: cfe-commits, sanjoy

Differential Revision: https://reviews.llvm.org/D40152

llvm-svn: 318494

6 years ago[asan] Test ASan instrumentation for shadow scale value of 5
Walter Lee [Fri, 17 Nov 2017 01:15:31 +0000 (01:15 +0000)]
[asan] Test ASan instrumentation for shadow scale value of 5

Add additional RUN clauses to test for -asan-mapping-scale=5 in
selective tests, with special CHECK statements where needed.

Differential Revision: https://reviews.llvm.org/D39775

llvm-svn: 318493

6 years agoUpdate for layering fix in LLVM CodeGen<>Target
David Blaikie [Fri, 17 Nov 2017 01:07:20 +0000 (01:07 +0000)]
Update for layering fix in LLVM CodeGen<>Target

llvm-svn: 318491

6 years agoFix a bunch more layering of CodeGen headers that are in Target
David Blaikie [Fri, 17 Nov 2017 01:07:10 +0000 (01:07 +0000)]
Fix a bunch more layering of CodeGen headers that are in Target

All these headers already depend on CodeGen headers so moving them into
CodeGen fixes the layering (since CodeGen depends on Target, not the
other way around).

llvm-svn: 318490

6 years agoDon't #include MemoryBuffer.h from Host.h.
Zachary Turner [Fri, 17 Nov 2017 01:00:35 +0000 (01:00 +0000)]
Don't #include MemoryBuffer.h from Host.h.

It turns out this #include isn't used from Host.h anyway,
but by having it it causes circular include dependencies.
This issues only surfaced while I was working on a separate
patch, so I'm submitting this first so that it's independent
of the other, unrelated patch.

llvm-svn: 318489

6 years ago[fuzzer] Initialize PcDescr buffer before calling __sanitizer_symbolize_pc
Vitaly Buka [Fri, 17 Nov 2017 00:46:59 +0000 (00:46 +0000)]
[fuzzer] Initialize PcDescr buffer before calling __sanitizer_symbolize_pc

__sanitizer_symbolize_pc is not instrumented so msan assumes that PcDescr
was not initialized.

llvm-svn: 318488

6 years ago[PredicateInfo] Add comment about why we require stable sort
Mandeep Singh Grang [Fri, 17 Nov 2017 00:43:24 +0000 (00:43 +0000)]
[PredicateInfo] Add comment about why we require stable sort

llvm-svn: 318487

6 years agoRe-revert "Refactor debuginfo-tests"
Zachary Turner [Fri, 17 Nov 2017 00:41:18 +0000 (00:41 +0000)]
Re-revert "Refactor debuginfo-tests"

This is still broken because it causes certain tests to be
run twice with slightly different configurations, which is
wrong in some cases.

You can observe this by running:

  ninja -nv check-all | grep debuginfo-tests

And seeing that it passes clang/test and clang/test/debuginfo-tests
to lit, which causes it to run debuginfo-tests twice.  The fix is
going to involve either:

  a) figuring out that we're running in this "deprecated" configuration,
     and then deleting the clang/test/debuginfo-tests path, which should
     cause it to behave identically to before, or:
  b) make lit smart enough that it doesn't descend into a sub-suite if
     that sub-suite already has a lit.cfg file.

llvm-svn: 318486

6 years agoCurrent implementation of Value::replaceUsesExceptBlockAddr() uses UseList
Dmitry Mikulin [Fri, 17 Nov 2017 00:30:24 +0000 (00:30 +0000)]
Current implementation of Value::replaceUsesExceptBlockAddr() uses UseList
iterator to walk the list which keeps changing inside the loop. When the
UseList contains several uses with the same user, we end processing the same
user more than once, which leads to an assert.

With this fix, unique users are saved and processed later to avoid
processing duplicates.

Differential Revision: https://reviews.llvm.org/D39864

llvm-svn: 318477

6 years ago[libFuzzer] Fix flaky merge.test.
Matt Morehouse [Fri, 17 Nov 2017 00:11:12 +0000 (00:11 +0000)]
[libFuzzer] Fix flaky merge.test.

llvm-svn: 318475

6 years agoPR22763: if a defaulted (non-user-provided) special member function is
Richard Smith [Thu, 16 Nov 2017 23:54:56 +0000 (23:54 +0000)]
PR22763: if a defaulted (non-user-provided) special member function is
explicitly instantiated, still emit it with each use.

We don't emit a definition of the member with an explicit instantiation
definition (and indeed it appears that we're not allowed to, since an explicit
instantiation definition does not constitute an odr-use and only odr-use
permits definition for defaulted special members). So we still need to emit a
weak definition with each use.

This also makes defaulted-in-class declarations behave more like
implicitly-declared special members, which matches their design intent.
And it matches the way this problem was solved in GCC.

llvm-svn: 318474

6 years ago[ARM] 't' asm constraint should accept i32
Yi Kong [Thu, 16 Nov 2017 23:38:17 +0000 (23:38 +0000)]
[ARM] 't' asm constraint should accept i32

't' constraint normally only accepts f32 operands, but for VCVT the
operands can be i32. LLVM is overly restrictive and rejects asm like:

  float foo() {
    float result;
    __asm__ __volatile__(
      "vcvt.f32.s32 %[result], %[arg1]\n"
      : [result]"=t"(result)
      : [arg1]"t"(0x01020304) );
    return result;
  }

Relax the value type for 't' constraint to either f32 or i32.

Differential Revision: https://reviews.llvm.org/D40137

llvm-svn: 318472

6 years ago[asan] Properly mark or disable tests that only work with shadow scale of 3
Walter Lee [Thu, 16 Nov 2017 23:29:19 +0000 (23:29 +0000)]
[asan] Properly mark or disable tests that only work with shadow scale of 3

Differential Revision: https://reviews.llvm.org/D39774

llvm-svn: 318471

6 years ago[asan] Port tests to shadow scale of 5
Walter Lee [Thu, 16 Nov 2017 23:28:50 +0000 (23:28 +0000)]
[asan] Port tests to shadow scale of 5

The tests are ported as follows:

contiguous_container_crash.cc
use-after-delete.cc
use-after-free.cc
  Replace hardwired shadow granularity in CHECK statements with regex.

max_redzone.cc
  Bump max_redzone parameter to 32.

memset_test.cc
  Bump size parameter of __asan_poison_memory_region to 32.

scariness_score_test.cc
  For "far-from-bounds" heap overflow, make sure overflow is more than
  one shadow granularity away.

  At large shadow granularity, there is not enough redzone between
  stack elements to detect far-from-bounds, so fake out that test.

Differential Revision: https://reviews.llvm.org/D39773

llvm-svn: 318470

6 years ago[asan] Add lit feature to indicate compiler_rt's shadow scale value
Walter Lee [Thu, 16 Nov 2017 23:28:25 +0000 (23:28 +0000)]
[asan] Add lit feature to indicate compiler_rt's shadow scale value

This will be used to mark tests that require a specific shadow scale.

Differential Revision: https://reviews.llvm.org/D39772

llvm-svn: 318469

6 years ago[X86] Add DAG combine to remove sext i32->i64 from gather/scatter instructions.
Craig Topper [Thu, 16 Nov 2017 23:09:06 +0000 (23:09 +0000)]
[X86] Add DAG combine to remove sext i32->i64 from gather/scatter instructions.

Only do this pre-legalize in case we're using the sign extend to legalize for KNL.

This recovers all of the tests that changed when I stopped SelectionDAGBuilder from deleting sign extends.

There's more work that could be done here particularly to fix the i8->i64 test case that experienced split.

llvm-svn: 318468

6 years ago[X86] Add gather test with index sign extended from i8 type.
Craig Topper [Thu, 16 Nov 2017 23:09:03 +0000 (23:09 +0000)]
[X86] Add gather test with index sign extended from i8 type.

Previously SelectionDAGBuilder would remove this sign extend leading to a failure during isel.

The codegen here isn't very nice as we ended up triggering a split.

llvm-svn: 318467

6 years ago[X86] Don't remove sign extend of gather/scatter indices during SelectionDAGBuilder.
Craig Topper [Thu, 16 Nov 2017 23:08:57 +0000 (23:08 +0000)]
[X86] Don't remove sign extend of gather/scatter indices during SelectionDAGBuilder.

The sign extend might be from an i16 or i8 type and was inserted by InstCombine to match the pointer width. X86 gather legalization isn't currently detecting this to reinsert a sign extend to make things legal.

It's a bit weird for the SelectionDAGBuilder to do this kind of optimization in the first place. With this removed we can at least lean on InstCombine somewhat to ensure the index is i32 or i64.

I'll work on trying to recover some of the test cases by removing sign extends in the backend when its safe to do so with an understanding of the current legalizer capabilities.

This should fix PR30690.

llvm-svn: 318466

6 years ago[libFuzzer] Make merge-sigusr.test work on Mac.
Matt Morehouse [Thu, 16 Nov 2017 23:06:17 +0000 (23:06 +0000)]
[libFuzzer] Make merge-sigusr.test work on Mac.

llvm-svn: 318465

6 years ago[Support] Support NetBSD PaX MPROTECT in sys::Memory.
Lang Hames [Thu, 16 Nov 2017 23:04:44 +0000 (23:04 +0000)]
[Support] Support NetBSD PaX MPROTECT in sys::Memory.

Removes AllocateRWX, setWritable and setExecutable from sys::Memory and
standardizes on allocateMappedMemory / protectMappedMemory. The
allocateMappedMemory method is updated to request full permissions for memory
blocks so that they can be marked executable later.

llvm-svn: 318464

6 years agoAttempt to fix inscrutible build break...
David Blaikie [Thu, 16 Nov 2017 22:40:02 +0000 (22:40 +0000)]
Attempt to fix inscrutible build break...

llvm-svn: 318463

6 years agoFix some undefined beahvior in FileMapping.
Zachary Turner [Thu, 16 Nov 2017 22:39:55 +0000 (22:39 +0000)]
Fix some undefined beahvior in FileMapping.

This was broken when building a 32-bit native toolchain, as
shifting a size_t right by 32 is UB when sizeof(size_t) == 8.

llvm-svn: 318462

6 years agollvm-readobj/ARMEHABIPrinter.h: Make this a real/modular header
David Blaikie [Thu, 16 Nov 2017 22:30:36 +0000 (22:30 +0000)]
llvm-readobj/ARMEHABIPrinter.h: Make this a real/modular header

Had several non-inline/strong function definitions that needed to be
marked inline, etc.

llvm-svn: 318461

6 years ago[asan] Avoid assert failure for non-default shadow scale
Walter Lee [Thu, 16 Nov 2017 22:02:58 +0000 (22:02 +0000)]
[asan] Avoid assert failure for non-default shadow scale

Rather than assertion failing, we can fall back to the
non-optimized version which works for any shadow scale.

Differential Revision: https://reviews.llvm.org/D39474

llvm-svn: 318460

6 years agoConvert the last use of sys::fs::createUniqueFile in bugpoint.
Rafael Espindola [Thu, 16 Nov 2017 21:53:51 +0000 (21:53 +0000)]
Convert the last use of sys::fs::createUniqueFile in bugpoint.

llvm-svn: 318459

6 years agoConvert another use of createUniqueFile to TempFile::create.
Rafael Espindola [Thu, 16 Nov 2017 21:40:10 +0000 (21:40 +0000)]
Convert another use of createUniqueFile to TempFile::create.

This one requires a new small feature in TempFile: the ability to keep
the temporary file with the temporary name.

llvm-svn: 318458

6 years ago[CMake][runtimes] Use cmake_parse_arguments in runtimes functions
Petr Hosek [Thu, 16 Nov 2017 21:28:54 +0000 (21:28 +0000)]
[CMake][runtimes] Use cmake_parse_arguments in runtimes functions

Passing lists to functions in CMake is tricky, any list argument
has to be quoted otherwise it'll be expanded. To avoid this issue,
use cmake_parse_arguments in runtime functions and pass lists using
a keyword argument which eliminates any ambiguity when dealing with
lists.

Differential Revision: https://reviews.llvm.org/D40087

llvm-svn: 318457

6 years agoIssue -Wempty-body warnings for else blocks
Reid Kleckner [Thu, 16 Nov 2017 21:26:18 +0000 (21:26 +0000)]
Issue -Wempty-body warnings for else blocks

This looks like it was just an oversight.

Fixes http://llvm.org/pr35319

llvm-svn: 318456

6 years agoRevert "[SelectionDAG] Consolidate (t|T)ransferDbgValues methods, NFC."
Vedant Kumar [Thu, 16 Nov 2017 21:08:51 +0000 (21:08 +0000)]
Revert "[SelectionDAG] Consolidate (t|T)ransferDbgValues methods, NFC."

This reverts commit r318448. It looks like some of the asserts need to
be weakened.

http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-autoconf/builds/16296

llvm-svn: 318455

6 years ago[RISCV] Fix 64-bit data layout mismatch between backend and target description
Mandeep Singh Grang [Thu, 16 Nov 2017 20:30:49 +0000 (20:30 +0000)]
[RISCV] Fix 64-bit data layout mismatch between backend and target description

Reviewers: asb

Reviewed By: asb

Subscribers: rbar, johnrusso, simoncook, jordy.potman.lists, llvm-commits

Differential Revision: https://reviews.llvm.org/D40145

llvm-svn: 318454

6 years ago[libFuzzer] add more tests for merge
Kostya Serebryany [Thu, 16 Nov 2017 20:25:45 +0000 (20:25 +0000)]
[libFuzzer] add more tests for merge

llvm-svn: 318453

6 years ago[X86] Pre-truncate gather/scatter indices that have element sizes larger than 64...
Craig Topper [Thu, 16 Nov 2017 20:23:22 +0000 (20:23 +0000)]
[X86] Pre-truncate gather/scatter indices that have element sizes larger than 64-bits before Legalize.

The wider element type will normally cause legalize to try to split and scalarize the gather/scatter, but we can't handle that. Instead, truncate the index early so the gather/scatter node is insulated from the legalization.

This really shouldn't happen in practice since InstCombine will normalize index types to the same size as pointers.

llvm-svn: 318452

6 years ago[X86] DAGCombinerInfo is in TargetLowering not X86TargetLowering.
Craig Topper [Thu, 16 Nov 2017 20:23:17 +0000 (20:23 +0000)]
[X86] DAGCombinerInfo is in TargetLowering not X86TargetLowering.

llvm-svn: 318451

6 years ago[DAGCombiner] Use cast instead of an unchecked dyn_cast.
Craig Topper [Thu, 16 Nov 2017 20:23:12 +0000 (20:23 +0000)]
[DAGCombiner] Use cast instead of an unchecked dyn_cast.

llvm-svn: 318450

6 years agollvm-objdump: Add missing requires for the test in r313921
Justin Bogner [Thu, 16 Nov 2017 20:20:56 +0000 (20:20 +0000)]
llvm-objdump: Add missing requires for the test in r313921

This test needs an arm backend in addition to the arm64 one that's
already checked in the test directory.

llvm-svn: 318449

6 years ago[SelectionDAG] Consolidate (t|T)ransferDbgValues methods, NFC.
Vedant Kumar [Thu, 16 Nov 2017 19:50:24 +0000 (19:50 +0000)]
[SelectionDAG] Consolidate (t|T)ransferDbgValues methods, NFC.

TransferDbgValues (capital 'T') is wired into ReplaceAllUsesWith, and
transferDbgValues (lowercase 't') is used elsewhere (e.g in Legalize).

Both functions should be doing the exact same thing. This patch
consolidates the logic into one place.

Differential Revision: https://reviews.llvm.org/D40104

llvm-svn: 318448

6 years agoFix my typo of PDB_TableType
Reid Kleckner [Thu, 16 Nov 2017 19:41:12 +0000 (19:41 +0000)]
Fix my typo of PDB_TableType

llvm-svn: 318447

6 years agoRemove a FIXME about truncated section names
Martin Storsjo [Thu, 16 Nov 2017 19:36:48 +0000 (19:36 +0000)]
Remove a FIXME about truncated section names

If the linker chose to store the full section name instead of truncating
it, this field doesn't contain a truncated name, but an offset into
the string table of the binary. The string table isn't loaded/mapped
into memory during runtime though, so it's not possible to read the
full section name, unless we try to locate the DLL/EXE on disk that
the HMODULE corresponds to and load that manually.

Due to this, lld now always prefers writing a truncated section name
for sections that will be mapped at runtime, even when debug info is
enabled.

Differential Revision: https://reviews.llvm.org/D39918

llvm-svn: 318446

6 years ago[VirtualFileSystem] Support creating directories then adding files inside
Ben Hamilton [Thu, 16 Nov 2017 19:34:08 +0000 (19:34 +0000)]
[VirtualFileSystem] Support creating directories then adding files inside

Summary:
In https://reviews.llvm.org/D39572 , I added support for specifying
`Type` when invoking `InMemoryFileSystem::addFile()`.

However, I didn't account for the fact that when `Type` is
`directory_file`, we need to construct an `InMemoryDirectory`, not an
`InMemoryFile`, or else clients cannot create files inside that
directory.

This diff fixes the bug and adds a test.

Test Plan: New test added. Ran test with:

  % make -j12 check-clang-tools

Reviewers: bkramer, hokein

Reviewed By: bkramer

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D40140

llvm-svn: 318445

6 years agoFix -Wreturn-type falling off the end of a function in new DIA code
Reid Kleckner [Thu, 16 Nov 2017 19:32:53 +0000 (19:32 +0000)]
Fix -Wreturn-type falling off the end of a function in new DIA code

llvm-svn: 318444

6 years ago[arc] Fix ambiguous overloaded operator error
Daniel Sanders [Thu, 16 Nov 2017 19:16:56 +0000 (19:16 +0000)]
[arc] Fix ambiguous overloaded operator error

lib/Target/ARC/ARCISelLowering.cpp:490:22: error: use of overloaded operator '<<' is ambiguous (with operand types 'llvm::raw_ostream' and 'llvm::MVT::SimpleValueType')
                     << RegVT.getSimpleVT().SimpleTy << "\n");
                     ^  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~

llvm-svn: 318443

6 years agobpf: print backward branch target properly
Yonghong Song [Thu, 16 Nov 2017 19:15:36 +0000 (19:15 +0000)]
bpf: print backward branch target properly

Currently, it prints the backward branch offset as unsigned value
like below:
       7:       7d 34 0b 00 00 00 00 00         if r4 s>= r3 goto 11 <LBB0_3>
       8:       b7 00 00 00 00 00 00 00         r0 = 0
LBB0_2:
       9:       07 00 00 00 01 00 00 00         r0 += 1
      ......
      17:       bf 31 00 00 00 00 00 00         r1 = r3
      18:       6d 32 f6 ff 00 00 00 00         if r2 s> r3 goto 65526 <LBB0_3+0x7FFB0>

The correct print insn 18 should be:
      18:       6d 32 f6 ff 00 00 00 00         if r2 s> r3 goto -10 <LBB0_2>

To provide better clarity and be consistent with kernel verifier output,
the insn 7 output is changed to the following with "+" added to
non-negative branch offset:
       7:       7d 34 0b 00 00 00 00 00         if r4 s>= r3 goto +11 <LBB0_3>

Signed-off-by: Yonghong Song <yhs@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
llvm-svn: 318442

6 years ago[arc] Update TargetInfo to include the new backend name argument
Daniel Sanders [Thu, 16 Nov 2017 19:10:26 +0000 (19:10 +0000)]
[arc] Update TargetInfo to include the new backend name argument

Also update a comment about the usage of RegisterTarget() that didn't mention
the new argument.

llvm-svn: 318441

6 years ago[MS] Apply adjustments after storing 'this'
Reid Kleckner [Thu, 16 Nov 2017 19:09:36 +0000 (19:09 +0000)]
[MS] Apply adjustments after storing 'this'

Summary:
The MS ABI convention is that the 'this' pointer on entry is the address
of the vfptr that was used to make the virtual method call. In other
words, the pointer on entry always points to the base subobject that
introduced the virtual method. Consider this hierarchy:

  struct A { virtual void f() = 0; };
  struct B { virtual void g() = 0; };
  struct C : A, B {
    void f() override;
    void g() override;
  };

On entry to C::g, [ER]CX will contain the address of C's B subobject,
and C::g will have to subtract sizeof(A) to recover a pointer to C.

Before this change, we applied this adjustment in the prologue and
stored the new value into the "this" local variable alloca used for
debug info. However, MSVC does not do this, presumably because it is
often profitable to fold the adjustment into later field accesses. This
creates a problem, because the debugger expects the variable to be
unadjusted. Unfortunately, CodeView doesn't have anything like DWARF
expressions for computing variables that aren't in the program anymore,
so we have to declare 'this' to be the unadjusted value if we want the
debugger to see the right value.

This has the side benefit that, in optimized builds, the 'this' pointer
will usually be available on function entry because it doesn't require
any adjustment.

Reviewers: hans

Subscribers: aprantl, cfe-commits

Differential Revision: https://reviews.llvm.org/D40109

llvm-svn: 318440

6 years ago[libFuzzer] add docs for -reduce_inputs
Kostya Serebryany [Thu, 16 Nov 2017 18:58:14 +0000 (18:58 +0000)]
[libFuzzer] add docs for -reduce_inputs

llvm-svn: 318439

6 years ago[sanitizer] Revert rL318410
Kostya Kortchinsky [Thu, 16 Nov 2017 18:53:18 +0000 (18:53 +0000)]
[sanitizer] Revert rL318410

Summary:
The patch seems to have turned some Android tests flaky. The reason is unclear.
This reverts D40100 in case we can't figure out what is happening.

Reviewers: eugenis

Reviewed By: eugenis

Subscribers: srhines, kubamracek

Differential Revision: https://reviews.llvm.org/D40138

llvm-svn: 318438

6 years agoFix RISCV build after r318352
Azharuddin Mohammed [Thu, 16 Nov 2017 18:39:31 +0000 (18:39 +0000)]
Fix RISCV build after r318352

Reviewers: asb, apazos, mgrang

Reviewed By: mgrang

Subscribers: rbar, johnrusso, simoncook, jordy.potman.lists, llvm-commits

Differential Revision: https://reviews.llvm.org/D40139

llvm-svn: 318437

6 years ago[PPC] Change i32 constant in store instruction to i64
Guozhi Wei [Thu, 16 Nov 2017 18:27:34 +0000 (18:27 +0000)]
[PPC] Change i32 constant in store instruction to i64

This patch changes all i32 constant in store instruction to i64 with truncation, to increase the chance that the referenced constant can be shared with other i64 constant.

Differential Revision: https://reviews.llvm.org/D39352

llvm-svn: 318436

6 years agoResubmit "Refactor debuginfo-tests"
Zachary Turner [Thu, 16 Nov 2017 18:26:20 +0000 (18:26 +0000)]
Resubmit "Refactor debuginfo-tests"

This was reverted due to some failures on specific darwin buildbots,
the issue being that the new lit configuration was not setting the
SDKROOT environment variable.  We've tested a fix locally and confirmed
that it works, so this patch resubmits everything with the fix
applied.

llvm-svn: 318435

6 years ago[OPENMP] Add support for cancelling inside target parallel for
Alexey Bataev [Thu, 16 Nov 2017 18:20:21 +0000 (18:20 +0000)]
[OPENMP] Add support for cancelling inside target parallel for
directive.

Added missed support for cancelling of target parallel for construct.

llvm-svn: 318434

6 years agoAdd ELF dynamic symbol support to yaml2obj/obj2yaml
Dave Lee [Thu, 16 Nov 2017 18:10:15 +0000 (18:10 +0000)]
Add ELF dynamic symbol support to yaml2obj/obj2yaml

Summary:
This change introduces a `DynamicSymbols` field to the ELF specific YAML
supported by `yaml2obj` and `obj2yaml`. This grouping of symbols provides a way
to represent ELF dynamic symbols. The `DynamicSymbols` structure is identical to
the existing `Symbols`.

Reviewers: compnerd, jakehehrlich, silvas

Reviewed By: silvas

Subscribers: silvas, jakehehrlich, llvm-commits

Differential Revision: https://reviews.llvm.org/D39582

llvm-svn: 318433

6 years agoMark free functions size/empty/data conditionally noexcept.
Marshall Clow [Thu, 16 Nov 2017 17:55:41 +0000 (17:55 +0000)]
Mark free functions size/empty/data conditionally noexcept.

llvm-svn: 318432

6 years agoImplement GetTls() for NetBSD
Kamil Rytarowski [Thu, 16 Nov 2017 17:54:14 +0000 (17:54 +0000)]
Implement GetTls() for NetBSD

Summary:
Reuse the existing code for FreeBSD that is compatible with NetBSD.

Add NetBSD support in tests: tls_race.cc and tls_race2.cc.

Sponsored by <The NetBSD Foundation>

Reviewers: joerg, dvyukov, vitalybuka, kcc, eugenis

Reviewed By: dvyukov

Subscribers: srhines, emaste, kubamracek, llvm-commits, #sanitizers

Tags: #sanitizers

Differential Revision: https://reviews.llvm.org/D40105

llvm-svn: 318431

6 years ago[InstCombine] add tests for pow(); NFC
Sanjay Patel [Thu, 16 Nov 2017 17:49:54 +0000 (17:49 +0000)]
[InstCombine] add tests for pow(); NFC

Also, increase test diversity (and show another bug) by varying the types.

llvm-svn: 318430

6 years agomake exitDsymutil static.
Rafael Espindola [Thu, 16 Nov 2017 17:46:43 +0000 (17:46 +0000)]
make exitDsymutil static.

The objective is to remove it completelly.

This first patch removes the last use outside dsymutil.cpp and makes
it static.

llvm-svn: 318429

6 years agoAllow empty mappings for optional YAML input
Dave Lee [Thu, 16 Nov 2017 17:46:11 +0000 (17:46 +0000)]
Allow empty mappings for optional YAML input

Summary:
This change fixes a bug where `obj2yaml` can in some cases produce YAML that
causes `yaml2obj` to error.

The ELF YAML document structure has a `Sections` mapping, which contains three
mappings, all of which are optional: `Local`, `Global`, and `Weak.` Any one of
these can be missing, but if all three are missing, then `yaml2obj` errors. This
change allows YAML input for cases like this one.

I have tested this with check-llvm and check-lld, and all tests passed.

This change is the result of test failures while working on D39582, which
introduces a `DynamicSymbols` mapping, which will be empty at times.

Reviewers: compnerd, jakehehrlich, silvas, kledzik, mehdi_amini, pcc

Reviewed By: compnerd

Subscribers: silvas, llvm-commits

Differential Revision: https://reviews.llvm.org/D39908

llvm-svn: 318428

6 years agoConvert another use of createUniqueFile to TempFile::create.
Rafael Espindola [Thu, 16 Nov 2017 17:35:50 +0000 (17:35 +0000)]
Convert another use of createUniqueFile to TempFile::create.

llvm-svn: 318427