platform/upstream/llvm.git
6 years ago[sanitizers] Add init function to set alignment of low level allocator
Walter Lee [Tue, 21 Nov 2017 01:01:32 +0000 (01:01 +0000)]
[sanitizers] Add init function to set alignment of low level allocator

ASan requires that the min alignment be at least the shadow
granularity, so add an init function to do that.

Differential Revision: https://reviews.llvm.org/D39473

llvm-svn: 318717

6 years ago[X86] Remove 'mm3now' from isValidFeatureName.
Craig Topper [Tue, 21 Nov 2017 00:33:26 +0000 (00:33 +0000)]
[X86] Remove 'mm3now' from isValidFeatureName.

The correct spelling is '3dnow' which is already in the list.

llvm-svn: 318716

6 years agoxray-record-yaml.h: Remove unused file
David Blaikie [Tue, 21 Nov 2017 00:33:17 +0000 (00:33 +0000)]
xray-record-yaml.h: Remove unused file

llvm-svn: 318715

6 years agollvm-rc/ResourceScriptTokenList.h: Turns this into a .def file to imply that it's...
David Blaikie [Tue, 21 Nov 2017 00:23:19 +0000 (00:23 +0000)]
llvm-rc/ResourceScriptTokenList.h: Turns this into a .def file to imply that it's non-modular

Also undef the macros at the end of the file to make it easier to use.

llvm-svn: 318714

6 years agoXRayRecord.h: Add missing #include
David Blaikie [Tue, 21 Nov 2017 00:23:19 +0000 (00:23 +0000)]
XRayRecord.h: Add missing #include

llvm-svn: 318713

6 years agoYAML/XRay/std::vector: Fix ODR violation by removing local specialization
David Blaikie [Tue, 21 Nov 2017 00:23:17 +0000 (00:23 +0000)]
YAML/XRay/std::vector: Fix ODR violation by removing local specialization

There's a generic partial specialization for all std::vector<T> that
does what's desired, so no need for this full specialization that's
causing an ODR violation anyway.

llvm-svn: 318712

6 years ago[SCCP] If we replace with a constant, we can't replace with a range.
Davide Italiano [Tue, 21 Nov 2017 00:21:52 +0000 (00:21 +0000)]
[SCCP] If we replace with a constant, we can't replace with a range.

This microoptimization is NFC.

llvm-svn: 318711

6 years agoRevert r318678 to fix Clang test
Richard Trieu [Tue, 21 Nov 2017 00:12:18 +0000 (00:12 +0000)]
Revert r318678 to fix Clang test

r318678 caused the Clang test CodeGen/ms-inline-asm.c to start failing.

llvm-svn: 318710

6 years ago[msan] Don't sanitize "nosanitize" instructions
Vitaly Buka [Mon, 20 Nov 2017 23:37:56 +0000 (23:37 +0000)]
[msan] Don't sanitize "nosanitize" instructions

Reviewers: eugenis

Subscribers: hiraditya, llvm-commits

Differential Revision: https://reviews.llvm.org/D40205

llvm-svn: 318708

6 years ago[tsan] Fix sigaction implementation when it's called only to get handler
Vitaly Buka [Mon, 20 Nov 2017 23:37:46 +0000 (23:37 +0000)]
[tsan] Fix sigaction implementation when it's called only to get handler

Reviewers: eugenis

Subscribers: kubamracek, llvm-commits, krytarowski

Differential Revision: https://reviews.llvm.org/D40272

llvm-svn: 318707

6 years ago[SelectionDAG] When promoting the result of a VSELECT, make sure we promote the condi...
Craig Topper [Mon, 20 Nov 2017 23:08:50 +0000 (23:08 +0000)]
[SelectionDAG] When promoting the result of a VSELECT, make sure we promote the condition to the SetCC type for the final result type not the original type.

Normally this would be cleaned up by promoting the condition operand next. But in the attached case we promoted the result from v2i48 to v2i64 and the condition from v2i1 to v2i48. Then we tried to "promote" the v2i48 condition back to v2i1 because that's what the SetCC result type for v2i64 is on X86 with VLX. But promote is either a NOP or SIGN_EXTEND and this would need a truncation.

With the change here we now get the SetCC type of v2i1 when we're handling the result promotion and the operand no longer needs to be promoted itself.

Fixes PR35272.

llvm-svn: 318706

6 years ago[analyzer] Diagnose stack leaks via block captures
Alexander Shaposhnikov [Mon, 20 Nov 2017 22:53:30 +0000 (22:53 +0000)]
[analyzer] Diagnose stack leaks via block captures

This diff extends StackAddrEscapeChecker
to catch stack addresses leaks via block captures
if the block is executed asynchronously or
returned from a function.

Differential revision: https://reviews.llvm.org/D39438

llvm-svn: 318705

6 years ago[Sparc] efficient pattern for UINT_TO_FP conversion
Fedor Sergeev [Mon, 20 Nov 2017 22:33:58 +0000 (22:33 +0000)]
[Sparc] efficient pattern for UINT_TO_FP conversion

Summary:
        while investigating performance degradation of imagick benchmark
        there were found inefficient pattern for UINT_TO_FP conversion.
        That pattern causes RAW hazard in assembly code. Specifically,
        uitofp IR operator results in poor assembler :

        st          %i0, [%fp - 952]
        ldd         [%fp - 952], %f0

        it stores 32-bit integer register into memory location and then
        loads 64-bit floating point data from that location.
        That is exactly RAW hazard case. To optimize that case it is
        possible to use SPISD::ITOF and SPISD::XTOF for conversion from
        integer to floating point data type and to use ISD::BITCAST to
        copy from integer register into floating point register.
        The fix is to write custom UINT_TO_FP pattern using SPISD::ITOF,
        SPISD::XTOF, ISD::BITCAST.

Patch by Alexey Lapshin

Reviewers: fedor.sergeev, jyknight, dcederman, lero_chris

Reviewed By: jyknight

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D36875

llvm-svn: 318704

6 years agoAdd ADL support to range based <algorithm> extensions
David Blaikie [Mon, 20 Nov 2017 22:12:55 +0000 (22:12 +0000)]
Add ADL support to range based <algorithm> extensions

This adds support for ADL in the range based <algorithm> extensions
(llvm::for_each etc.).

Also adds the helper functions llvm::adl::begin and llvm::adl::end which wrap
std::begin and std::end with ADL support.

Saw this was missing from a recent llvm weekly post about adding llvm::for_each
and thought I might add it.

Patch by Stephen Dollberg!

Differential Revision: https://reviews.llvm.org/D40006

llvm-svn: 318703

6 years agoRevert 318668, which is associated with a broken patch
Erich Keane [Mon, 20 Nov 2017 22:10:28 +0000 (22:10 +0000)]
Revert 318668, which is associated with a broken patch

llvm-svn: 318702

6 years agoRelax pdb.test checks for debug chunk sizes
Reid Kleckner [Mon, 20 Nov 2017 21:53:24 +0000 (21:53 +0000)]
Relax pdb.test checks for debug chunk sizes

llvm-svn: 318700

6 years agoMerge .xdata into .rdata by default
Reid Kleckner [Mon, 20 Nov 2017 21:49:35 +0000 (21:49 +0000)]
Merge .xdata into .rdata by default

Summary: MSVC does this. The user can override it with their own /merge: flag.

Reviewers: ruiu, pcc

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D40197

llvm-svn: 318699

6 years agoRevert r318669/318694
Erich Keane [Mon, 20 Nov 2017 21:46:29 +0000 (21:46 +0000)]
Revert r318669/318694

Broke some libclang tests, so reverting for now.

llvm-svn: 318698

6 years agoResubmit "Refactor debuginfo-tests" again.
Zachary Turner [Mon, 20 Nov 2017 21:41:36 +0000 (21:41 +0000)]
Resubmit "Refactor debuginfo-tests" again.

This was reverted due to the tests being run twice on some
build bots.  Each run had a slightly different configuration
due to the way in which it was being invoked.  This fixes
the problem (albeit in a somewhat hacky way).  Hopefully in
the future we can get rid of the workflow of running
debuginfo-tests as part of clang, and then this hack can
go away.

llvm-svn: 318697

6 years agoFix a lld-x86_64-darwin13 build error.
Hiroshi Yamauchi [Mon, 20 Nov 2017 21:38:43 +0000 (21:38 +0000)]
Fix a lld-x86_64-darwin13 build error.

Summary:
Fix this build error

http://lab.llvm.org:8011/builders/lld-x86_64-darwin13/builds/15112/steps/build_Lld/logs/stdio

after https://reviews.llvm.org/rL318693

Reviewers: davidxl

Reviewed By: davidxl

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D40266

llvm-svn: 318696

6 years agobpf: add a test case for trunc-op optimization
Yonghong Song [Mon, 20 Nov 2017 21:37:58 +0000 (21:37 +0000)]
bpf: add a test case for trunc-op optimization

Commit b5cbc7760ab8 ("[bpf] allow direct and indirect calls")
allowed more than one function in the bpf program, and
commit 114353884415 ("bpf: fix a bug in trunc-op optimization")
fixed a bug in trunc-op optimization which only showed up
with more than one function in the bpf program.

This patch added a test case for trunc-op optimization
for bpf programs with two functions. Reverting commit
"bpf: fix a bug in trunc-op optimization" will cause
failure for this test case.

Signed-off-by: Yonghong Song <yhs@fb.com>
llvm-svn: 318695

6 years agoInclude test files for rL318668
Erich Keane [Mon, 20 Nov 2017 21:15:01 +0000 (21:15 +0000)]
Include test files for  rL318668

Forgotten when doing my SVN commit.

llvm-svn: 318694

6 years agoAdd heuristics for irreducible loop metadata under PGO
Hiroshi Yamauchi [Mon, 20 Nov 2017 21:03:38 +0000 (21:03 +0000)]
Add heuristics for irreducible loop metadata under PGO

Summary:
Add the following heuristics for irreducible loop metadata:

- When an irreducible loop header is missing the loop header weight metadata,
  give it the minimum weight seen among other headers.
- Annotate indirectbr targets with the loop header weight metadata (as they are
  likely to become irreducible loop headers after indirectbr tail duplication.)

These greatly improve the accuracy of the block frequency info of the Python
interpreter loop (eg. from ~3-16x off down to ~40-55% off) and the Python
performance (eg. unpack_sequence from ~50% slower to ~8% faster than GCC) due to
better register allocation under PGO.

Reviewers: davidxl

Reviewed By: davidxl

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D39980

llvm-svn: 318693

6 years ago[OpenMP] Initial implementation of code generation for pragma 'teams distribute paral...
Carlo Bertolli [Mon, 20 Nov 2017 20:46:39 +0000 (20:46 +0000)]
[OpenMP] Initial implementation of code generation for pragma 'teams distribute parallel for' on host

https://reviews.llvm.org/D40187

This patch implements code gen for 'teams distribute parallel for' on the host, including all its clauses and related regression tests.

llvm-svn: 318692

6 years ago[MIPS] Add test case to check calculation of GOT pages in case of using a linker...
Simon Atanasyan [Mon, 20 Nov 2017 20:40:45 +0000 (20:40 +0000)]
[MIPS] Add test case to check calculation of GOT pages in case of using a linker script. NFC

llvm-svn: 318691

6 years agoFix std::string::data() symbol during library build.
Eric Fiselier [Mon, 20 Nov 2017 20:23:27 +0000 (20:23 +0000)]
Fix std::string::data() symbol during library build.

The non-const data() member of std::string is only exposed
in C++17 and beyond. However std::string is externally instantiated
and so the member function needs to be exposed to be externally instantiated.

On Linux and OS X this shouldn't cause a problem, because
_LIBCPP_INLINE_VISIBILITY ensures the symbol is always inlined.

However on Windows, the symbol gets marked dllimport, but
there is no definition to import, causing link errors.

llvm-svn: 318690

6 years agoCOFF: Correctly handle relocations against early discarded sections.
Peter Collingbourne [Mon, 20 Nov 2017 19:37:07 +0000 (19:37 +0000)]
COFF: Correctly handle relocations against early discarded sections.

Don't crash if we encounter a reference to an early discarded section
(such as .drectve). Instead, handle them the same way as sections
discarded by comdat merging, i.e. either print an error message or
(for debug sections) silently ignore the relocation.

Differential Revision: https://reviews.llvm.org/D40235

llvm-svn: 318689

6 years agoAdd missing test.
Rafael Espindola [Mon, 20 Nov 2017 19:23:36 +0000 (19:23 +0000)]
Add missing test.

Without this we could remove

SS->getFile<ELFT>()->IsUsed = true;

from addUndefined and no test would fail.

llvm-svn: 318688

6 years agoFix spelling in comment. NFCI.
Simon Pilgrim [Mon, 20 Nov 2017 19:18:33 +0000 (19:18 +0000)]
Fix spelling in comment. NFCI.

llvm-svn: 318687

6 years agoRevert "[SelectionDAG] Make sorting predicate stronger to remove non-deterministic...
Mandeep Singh Grang [Mon, 20 Nov 2017 19:17:11 +0000 (19:17 +0000)]
Revert "[SelectionDAG] Make sorting predicate stronger to remove non-deterministic ordering"

This broke the bots. Reverting this until I can fix the failures.

This reverts commit 5a3db2856d12a3c4b400f487d39f8f05989e79f0.

llvm-svn: 318686

6 years ago[AArch64] Adjust the cost model for Exynos M1 and M2
Evandro Menezes [Mon, 20 Nov 2017 19:11:56 +0000 (19:11 +0000)]
[AArch64] Adjust the cost model for Exynos M1 and M2

Fix the modeling of test and branch.

llvm-svn: 318685

6 years agoRevert "Fix out-of-order stepping behavior in programs with sunk instructions."
Paul Robinson [Mon, 20 Nov 2017 19:07:52 +0000 (19:07 +0000)]
Revert "Fix out-of-order stepping behavior in programs with sunk instructions."

This reverts commit 30419e150cd940893a13b345e85f96053850208f.
aka r318679.  It caused "sanitizer-windows" bot to fail.

llvm-svn: 318684

6 years agoCOFF: Remove the SymbolBodies vector, and rename SparseSymbolBodies to Symbols.
Peter Collingbourne [Mon, 20 Nov 2017 18:52:53 +0000 (18:52 +0000)]
COFF: Remove the SymbolBodies vector, and rename SparseSymbolBodies to Symbols.

Differential Revision: https://reviews.llvm.org/D40237

llvm-svn: 318683

6 years agoCOFF: Stop requiring comdat sections to have an external leader to participate in...
Peter Collingbourne [Mon, 20 Nov 2017 18:51:29 +0000 (18:51 +0000)]
COFF: Stop requiring comdat sections to have an external leader to participate in ICF.

This requirement was added in r254578 to fix pr25686. However, it
appears to have originated from a misdiagnosis of the problem: link.exe
refused to merge the two sections because they are non-executable,
not because they have internal leaders. If I set up a similar scenario
with functions instead of globals I see that link.exe merges them.

Differential Revision: https://reviews.llvm.org/D40236

llvm-svn: 318682

6 years ago[AutoComplete] Use stronger sort predicate for autocomplete candidates to remove...
Mandeep Singh Grang [Mon, 20 Nov 2017 18:49:14 +0000 (18:49 +0000)]
[AutoComplete] Use stronger sort predicate for autocomplete candidates to remove non-deterministic ordering

Summary: This fixes the failure in test/Driver/autocomplete.c uncovered by D39245.

Reviewers: yamaguchi, teemperor, ruiu

Reviewed By: yamaguchi, ruiu

Subscribers: cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D40234

llvm-svn: 318681

6 years ago[SelectionDAG] Make sorting predicate stronger to remove non-deterministic ordering
Mandeep Singh Grang [Mon, 20 Nov 2017 18:46:11 +0000 (18:46 +0000)]
[SelectionDAG] Make sorting predicate stronger to remove non-deterministic ordering

Summary:
This fixes failures in the following tests uncovered by D39245:
        LLVM :: CodeGen/ARM/ifcvt3.ll
        LLVM :: CodeGen/ARM/switch-minsize.ll
        LLVM :: CodeGen/X86/switch.ll

Reviewers: hans, efriedma

Reviewed By: hans

Subscribers: fhahn, aemerson, kristof.beyls, llvm-commits

Differential Revision: https://reviews.llvm.org/D39995

llvm-svn: 318680

6 years agoFix out-of-order stepping behavior in programs with sunk instructions.
Paul Robinson [Mon, 20 Nov 2017 18:42:17 +0000 (18:42 +0000)]
Fix out-of-order stepping behavior in programs with sunk instructions.

MachineSink attempts to place instructions near the basic blocks where
they are needed.  Once an instruction has been sunk, its location
relative to other instructions is no longer consistent with the
original source code. In order to ensure correct single-stepping and
profiling, the debug location for sunk instructions is either merged
with the insertion point or erased if the target successor block is
empty.

Patch by Matthew Voss!

Differential Revision: https://reviews.llvm.org/D39933

llvm-svn: 318679

6 years ago[X86] Avoid unecessary opsize byte in segment move to memory
Nirav Dave [Mon, 20 Nov 2017 18:38:55 +0000 (18:38 +0000)]
[X86] Avoid unecessary opsize byte in segment move to memory

Summary:

Segment moves to memory are always 16-bit. Remove invalid 32 and 64
bit variants.

Fixes PR34478.

Reviewers: rnk, craig.topper

Subscribers: llvm-commits, hiraditya

Differential Revision: https://reviews.llvm.org/D39847

llvm-svn: 318678

6 years ago[SROA] Correctly invalidate analyses when dead instructions deleted
Teresa Johnson [Mon, 20 Nov 2017 18:33:38 +0000 (18:33 +0000)]
[SROA] Correctly invalidate analyses when dead instructions deleted

Summary:
SROA can fail in rewriting alloca but still rewrite a phi resulting
in dead instruction elimination. The Changed flag was not being set
correctly, resulting in downstream passes using stale analyses.
The included test case will assert during the second BDCE pass as a
result.

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D39921

llvm-svn: 318677

6 years agoTry to fix syntax for !NetBSD in TSAN_INTERCEPTORs
Kamil Rytarowski [Mon, 20 Nov 2017 18:27:39 +0000 (18:27 +0000)]
Try to fix syntax for !NetBSD in TSAN_INTERCEPTORs

llvm-svn: 318676

6 years ago[AMDGPU][MC][GFX8][GFX9] Corrected names of integer v_{add/addc/sub/subrev/subb/subbrev}
Dmitry Preobrazhensky [Mon, 20 Nov 2017 18:24:21 +0000 (18:24 +0000)]
[AMDGPU][MC][GFX8][GFX9] Corrected names of integer v_{add/addc/sub/subrev/subb/subbrev}

See bug 34765: https://bugs.llvm.org//show_bug.cgi?id=34765

Reviewers: tamazov, SamWot, arsenm, vpykhtin

Differential Revision: https://reviews.llvm.org/D40088

llvm-svn: 318675

6 years agoHandle NetBSD specific indirection of libpthread functions
Kamil Rytarowski [Mon, 20 Nov 2017 18:07:52 +0000 (18:07 +0000)]
Handle NetBSD specific indirection of libpthread functions

Summary:
Correct handling of libpthread(3) functions in TSan/NetBSD:

 - pthread_cond_init(3),
 - pthread_cond_signal(3),
 - pthread_cond_broadcast(3),
 - pthread_cond_wait(3),
 - pthread_cond_destroy(3),
 - pthread_mutex_init(3),
 - pthread_mutex_destroy(3),
 - pthread_mutex_trylock(3),
 - pthread_rwlock_init(3),
 - pthread_rwlock_destroy(3),
 - pthread_rwlock_rdlock(3),
 - pthread_rwlock_tryrdlock(3),
 - pthread_rwlock_wrlock(3),
 - pthread_rwlock_trywrlock(3),
 - pthread_rwlock_unlock(3),
 - pthread_once(3).

Code out of the libpthread(3) context uses the libc symbols
that are prefixed with __libc_, for example: __libc_cond_init.

This caused that these functions were invisible to sanitizers on NetBSD.
Intercept the libc-specific ones and add them as NetBSD-specific aliases
for the common pthread(3) ones.

NetBSD needs to intercept both functions, as the regularly named ones
are used internally in libpthread(3).

Sponsored by <The NetBSD Foundation>

Reviewers: joerg, dvyukov, vitalybuka

Reviewed By: dvyukov

Subscribers: kubamracek, llvm-commits, #sanitizers

Tags: #sanitizers

Differential Revision: https://reviews.llvm.org/D40243

llvm-svn: 318673

6 years ago[Docs] Regenerate the command line option reference.
Craig Topper [Mon, 20 Nov 2017 18:07:43 +0000 (18:07 +0000)]
[Docs] Regenerate the command line option reference.

llvm-svn: 318672

6 years agoAdd DemangleFunctionName for backtracing on NetBSD
Kamil Rytarowski [Mon, 20 Nov 2017 18:06:26 +0000 (18:06 +0000)]
Add DemangleFunctionName for backtracing on NetBSD

Summary:
NetBSD uses indirection for old threading functions for historical reasons
The mangled names are internal implementation detail and should not be
exposed even in backtraces.

Sponsored by <The NetBSD Foundation>

Reviewers: joerg, vitalybuka, dvyukov

Reviewed By: dvyukov

Subscribers: kubamracek, llvm-commits, #sanitizers

Tags: #sanitizers

Differential Revision: https://reviews.llvm.org/D40251

llvm-svn: 318671

6 years ago[clang-tidy] revert hicpp-multiway-paths-covered
Jonas Toth [Mon, 20 Nov 2017 18:01:35 +0000 (18:01 +0000)]
[clang-tidy] revert hicpp-multiway-paths-covered

The address sanitizer found a stackoverflow with this patch.
There is no obvious fix. This patch will be reapplied when the problem
is found.

llvm-svn: 318670

6 years agoFor Linux/gnu compatibility, preinclude <stdc-predef.h> if the file is available
Erich Keane [Mon, 20 Nov 2017 17:57:42 +0000 (17:57 +0000)]
For Linux/gnu compatibility, preinclude <stdc-predef.h> if the file is available

As reported in llvm bugzilla 32377.
Here’s a patch to add preinclude of stdc-predef.h.

The gcc documentation says “On GNU/Linux, <stdc-predef.h> is pre-included.”
See https://gcc.gnu.org/gcc-4.8/porting_to.html;

The preinclude is inhibited with –ffreestanding.

Basically I fixed the failing test cases by adding –ffreestanding which inhibits
this behavior.

I fixed all the failing tests, including some in extra/test, there's a separate
patch for that which is linked here

Patch By: mibintc

Differential Revision: https://reviews.llvm.org/D34158

llvm-svn: 318669

6 years agoextra test modifications for D34158
Erich Keane [Mon, 20 Nov 2017 17:57:20 +0000 (17:57 +0000)]
extra test modifications for D34158

When adding support for D34158 which changes preprocessed output, I needed to
make tiny test corrections for these. Adding the option -ffreestanding
suppresses the new behavior, and that's the change I made to fix the tests.

Patch By: mibintc

Differential Revision: https://reviews.llvm.org/D34624

llvm-svn: 318668

6 years agoRevert r318456 "Issue -Wempty-body warnings for else blocks"
Hans Wennborg [Mon, 20 Nov 2017 17:48:54 +0000 (17:48 +0000)]
Revert r318456 "Issue -Wempty-body warnings for else blocks"

This caused warnings also when the if or else comes from macros. There was an
attempt to fix this in r318556, but that introduced new problems and was
reverted. Reverting this too until the whole issue is sorted.

> This looks like it was just an oversight.
>
> Fixes http://llvm.org/pr35319
>
> git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@318456 91177308-0d34-0410-b5e6-96231b3b80d8

llvm-svn: 318667

6 years ago[asan] Use dynamic shadow on 32-bit Android, try 2.
Evgeniy Stepanov [Mon, 20 Nov 2017 17:41:57 +0000 (17:41 +0000)]
[asan] Use dynamic shadow on 32-bit Android, try 2.

Summary:
This change reverts r318575 and changes FindDynamicShadowStart() to
keep the memory range it found mapped PROT_NONE to make sure it is
not reused. We also skip MemoryRangeIsAvailable() check, because it
is (a) unnecessary, and (b) would fail anyway.

Reviewers: pcc, vitalybuka, kcc

Subscribers: srhines, kubamracek, mgorny, llvm-commits, hiraditya

Differential Revision: https://reviews.llvm.org/D40203

llvm-svn: 318666

6 years agoRevert r318556 "Loosen -Wempty-body warning"
Hans Wennborg [Mon, 20 Nov 2017 17:38:16 +0000 (17:38 +0000)]
Revert r318556 "Loosen -Wempty-body warning"

It seems this somehow made -Wempty-body fire in some macro cases where
it didn't before, e.g.

  ../../third_party/ffmpeg/libavcodec/bitstream.c(169,5):  error: if statement has empty body [-Werror,-Wempty-body]
      ff_dlog(NULL, "new table index=%d size=%d\n", table_index, table_size);
      ^
  ../../third_party/ffmpeg\libavutil/internal.h(276,80):  note: expanded from macro 'ff_dlog'
  #   define ff_dlog(ctx, ...) do { if (0) av_log(ctx, AV_LOG_DEBUG, __VA_ARGS__); } while (0)
                                                                                 ^
  ../../third_party/ffmpeg/libavcodec/bitstream.c(169,5):  note: put the
  semicolon on a separate line to silence this warning

Reverting until this can be figured out.

> Do not show it when `if` or `else` come from macros.
> E.g.,
>
>     #define USED(A) if (A); else
>     #define SOME_IF(A) if (A)
>
>     void test() {
>       // No warnings are shown in those cases now.
>       USED(0);
>       SOME_IF(0);
>     }
>
> Patch by Ilya Biryukov!
>
> Differential Revision: https://reviews.llvm.org/D40185

llvm-svn: 318665

6 years ago[Expression parser] Remove logging of field that is no longer in clang's TargetOption...
Craig Topper [Mon, 20 Nov 2017 17:24:47 +0000 (17:24 +0000)]
[Expression parser] Remove logging of field that is no longer in clang's TargetOptions struct.

The Reciprocals field was moved to CodeGenOptions. I don't think lldb was really using this field so I'm just removing the mention.

llvm-svn: 318664

6 years ago[CodeGen] Move Reciprocals option from TargetOptions to CodeGenOptions
Craig Topper [Mon, 20 Nov 2017 17:09:22 +0000 (17:09 +0000)]
[CodeGen] Move Reciprocals option from TargetOptions to CodeGenOptions

Diffrential Revision: https://reviews.llvm.org/D40226

llvm-svn: 318662

6 years ago[MachineCSE] Add new callback for is caller preserved or constant physregs
Tony Jiang [Mon, 20 Nov 2017 16:55:07 +0000 (16:55 +0000)]
[MachineCSE] Add new callback for is caller preserved or constant physregs

The instructions addis,addi, bl are used to calculate the address of TLS thread
local variables. These TLS access code sequences are generated repeatedly every
time the thread local variable is accessed. By communicating to Machine CSE that
X2 is guaranteed to have the same value within the same function call (so called
Caller Preserved Physical Register), the redundant TLS access code sequences are
cleaned up.

Differential Revision: https://reviews.llvm.org/D39173

llvm-svn: 318661

6 years ago[AMDGPU] Update test r600.amdgpu-alias-analysis.ll
Yaxun Liu [Mon, 20 Nov 2017 16:53:13 +0000 (16:53 +0000)]
[AMDGPU] Update test r600.amdgpu-alias-analysis.ll

Manually update test r600.amdgpu-alias-analysis.ll for amdgiz environment
since it cannot be done by script.

The two pointers are swapped in the output because PrintResults in
AliasAnalysisEvaluator.cpp sorts the strings obtained from printAsOperand
before printing them.

Differential Revision: https://reviews.llvm.org/D40131

llvm-svn: 318660

6 years ago[sanitizer] New attempt at using runtime checks for Android logging
Kostya Kortchinsky [Mon, 20 Nov 2017 16:39:49 +0000 (16:39 +0000)]
[sanitizer] New attempt at using runtime checks for Android logging

Summary:
This is a second attempt after D40100 induced racey crashes with ASan
(due to `__android_log_write` and the `strncpy` interceptor on API >= 21).

This new version checks the runtime API level to be <= `ANDROID_KITKAT` for
the use  of `__android_log_write`, otherwise we use `syslog`, which should
conform with the previous behavior.

Unfortunately despite numerous efforts I couldn't reproduce the original
crashes in my environments so I couldn't test that the fix was actually
preventing crashes.

Reviewers: eugenis

Reviewed By: eugenis

Subscribers: llvm-commits, srhines, kubamracek

Differential Revision: https://reviews.llvm.org/D40149

llvm-svn: 318659

6 years agoFixed OMP doacross implementation on 32-bit platforms.
Andrey Churbanov [Mon, 20 Nov 2017 16:00:42 +0000 (16:00 +0000)]
Fixed OMP doacross implementation on 32-bit platforms.

Differential Revision: https://reviews.llvm.org/D40171

llvm-svn: 318658

6 years ago[mips] Reorder target specific passes
Simon Dardis [Mon, 20 Nov 2017 15:59:18 +0000 (15:59 +0000)]
[mips] Reorder target specific passes

Move the hazard scheduling pass to after the long branch pass, as the
long branch pass can create forbiddden slot hazards. Rather than complicating
the implementation of the long branch pass to handle forbidden slot hazards,
just reorder the passes.

llvm-svn: 318657

6 years ago[ELF] Fix rewriting of --sysroot in reproducer tars
Alexander Richardson [Mon, 20 Nov 2017 15:49:48 +0000 (15:49 +0000)]
[ELF] Fix rewriting of --sysroot in reproducer tars

Summary:
I noticed that the reproducers files I was getting from building CheriBSD
didn't work because the --sysroot option was not being rewritten. I've
updated the test to also verify that the rewritten path matches uses a
FileCheck capature instead of a {{.+}} regex

Reviewers: ruiu

Reviewed By: ruiu

Subscribers: llvm-commits, emaste

Differential Revision: https://reviews.llvm.org/D40125

llvm-svn: 318656

6 years ago[ELF] Fall back to search dirs for linker scripts specified with -T
Alexander Richardson [Mon, 20 Nov 2017 15:43:20 +0000 (15:43 +0000)]
[ELF] Fall back to search dirs for linker scripts specified with -T

Summary:
This matches the behaviour of ld.bfd:
https://sourceware.org/binutils/docs/ld/Options.html#Options

If scriptfile does not exist in the current directory, ld looks for it in
the directories specified by any preceding '-L' options. Multiple '-T'
options accumulate.

Reviewers: ruiu, grimar

Reviewed By: ruiu, grimar

Subscribers: emaste, llvm-commits

Differential Revision: https://reviews.llvm.org/D40129

llvm-svn: 318655

6 years agoCorrect return types of NetBSD specific functions
Kamil Rytarowski [Mon, 20 Nov 2017 15:39:30 +0000 (15:39 +0000)]
Correct return types of NetBSD specific functions

The __libc_mutex_lock, __libc_mutex_unlock and __libc_thr_setcancelstate
functions return int, not void.

This does not seem to introduce a functional change, however it looks
better with fixed the function prototype.

Sponsored by <The NetBSD Foundation>

llvm-svn: 318654

6 years ago[ELF][MIPS] Fix crash in LLD when linking code that needs PIC thunks
Alexander Richardson [Mon, 20 Nov 2017 15:37:19 +0000 (15:37 +0000)]
[ELF][MIPS] Fix crash in LLD when linking code that needs PIC thunks

Summary:
The bug triggers when the following conditions are met:
    - A thunk is created in a given input section S
    - A linker script is specified
    - There is at least one matcher in the linker script .text section output
      that does not match any of the sections in the input files, before the matcher
      that matches section S.

The issue was found when linking the FreeBSD kernel for MIPS when built
with -fPIC. Patch by Alfredo Mazzinghi.

Reviewers: ruiu, psmith, atanasyan

Reviewed By: ruiu

Subscribers: peter.smith, emaste, sdardis, krytarowski, llvm-commits

Differential Revision: https://reviews.llvm.org/D40174

llvm-svn: 318653

6 years ago[SystemZ] Bugfix for handling of subregisters in getRegAllocationHints().
Jonas Paulsson [Mon, 20 Nov 2017 14:54:03 +0000 (14:54 +0000)]
[SystemZ]  Bugfix for handling of subregisters in getRegAllocationHints().

The 32 bit subreg indices of GR128 registers must also be checked for in
getRC32().

Review: Ulrich Weigand.
llvm-svn: 318652

6 years ago[PPC] Heuristic to choose between a X-Form VSX ld/st vs a X-Form FP ld/st.
Tony Jiang [Mon, 20 Nov 2017 14:38:30 +0000 (14:38 +0000)]
[PPC] Heuristic to choose between a X-Form VSX ld/st vs a X-Form FP ld/st.

The VSX versions have the advantage of a full 64-register target whereas the FP
ones have the advantage of lower latency and higher throughput. So what we’re
after is using the faster instructions in low register pressure situations and
using the larger register file in high register pressure situations.

The heuristic chooses between the following 7 pairs of instructions.
PPC::LXSSPX vs PPC::LFSX
PPC::LXSDX vs PPC::LFDX
PPC::STXSSPX vs PPC::STFSX
PPC::STXSDX vs PPC::STFDX
PPC::LXSIWAX vs PPC::LFIWAX
PPC::LXSIWZX vs PPC::LFIWZX
PPC::STXSIWX vs PPC::STFIWX

Differential Revision: https://reviews.llvm.org/D38486

llvm-svn: 318651

6 years ago[AArch64][TableGen] Skip tied result operands for InstAlias
Sander de Smalen [Mon, 20 Nov 2017 14:36:40 +0000 (14:36 +0000)]
[AArch64][TableGen] Skip tied result operands for InstAlias

Summary:
This patch fixes an issue so that the right alias is printed when the instruction has tied operands. It checks the number of operands in the resulting instruction as opposed to the alias, and then skips over tied operands that should not be printed in the alias.

This allows to generate the preferred assembly syntax for the AArch64 'ins' instruction, which should always be displayed as 'mov' according to the ARM Architecture Reference Manual. Several unit tests have changed as a result, but only to reflect the preferred disassembly. Some other InstAlias patterns (movk/bic/orr) needed a slight adjustment to stop them becoming the default and breaking other unit tests.

Please note that the patch is mostly the same as https://reviews.llvm.org/D29219 which was reverted because of an issue found when running TableGen with the Address Sanitizer. That issue has been addressed in this iteration of the patch.

Reviewers: rengolin, stoklund, huntergr, SjoerdMeijer, rovka

Reviewed By: rengolin, SjoerdMeijer

Subscribers: fhahn, aemerson, javed.absar, kristof.beyls, llvm-commits

Differential Revision: https://reviews.llvm.org/D40030

llvm-svn: 318650

6 years agoAMDGPU: Partial ILP scheduler port from SelectionDAG to SchedulingDAG (experimental)
Valery Pykhtin [Mon, 20 Nov 2017 14:35:53 +0000 (14:35 +0000)]
AMDGPU: Partial ILP scheduler port from SelectionDAG to SchedulingDAG (experimental)

Differential revision: https://reviews.llvm.org/D39897

llvm-svn: 318649

6 years ago[ARM] For assembler files recognize -Xassembler or -Wa, -mthumb
Peter Smith [Mon, 20 Nov 2017 13:53:55 +0000 (13:53 +0000)]
[ARM] For assembler files recognize -Xassembler or -Wa, -mthumb

Attempt to fix warning picked up by buildbot.

llvm-svn: 318648

6 years ago[ARM] For assembler files recognize -Xassembler or -Wa, -mthumb
Peter Smith [Mon, 20 Nov 2017 13:43:55 +0000 (13:43 +0000)]
[ARM] For assembler files recognize -Xassembler or -Wa, -mthumb

The Unified Arm Assembler Language is designed so that the majority of
assembler files can be assembled for both Arm and Thumb with the choice
made as a compilation option.

The way this is done in gcc is to pass -mthumb to the assembler with either
-Wa,-mthumb or -Xassembler -mthumb. This change adds support for these
options to clang. There is no assembler equivalent of -mno-thumb, -marm or
-mno-arm so we don't need to recognize these.

Ideally we would do all of the processing in
CollectArgsForIntegratedAssembler(). Unfortunately we need to change the
triple and at that point it is too late. Instead we look for the option
earlier in ComputeLLVMTriple().

Fixes PR34519

Differential Revision: https://reviews.llvm.org/D40127

llvm-svn: 318647

6 years agoHandle NetBSD specific indirection of libpthread functions
Kamil Rytarowski [Mon, 20 Nov 2017 12:21:11 +0000 (12:21 +0000)]
Handle NetBSD specific indirection of libpthread functions

Summary:
Correct handling of three libpthread(3) functions on NetBSD:

 - pthread_mutex_lock(3),
 - pthread_mutex_unlock(3),
 - pthread_setcancelstate(3).

Code out of the libpthread(3) context uses the libc symbols:

 - __libc_mutex_lock,
 - __libc_mutex_unlock,
 - __libc_thr_setcancelstate.

The threading library (libpthread(3)) defines strong aliases:

 - __strong_alias(__libc_mutex_lock,pthread_mutex_lock)
 - __strong_alias(__libc_mutex_unlock,pthread_mutex_unlock)
 - __strong_alias(__libc_thr_setcancelstate,pthread_setcancelstate)

This caused that these functions were invisible to sanitizers on NetBSD.
Intercept the libc-specific ones and add them as NetBSD-specific aliases
for the common pthread(3) ones.

NetBSD needs to intercept both functions, as the regularly named ones
are used internally in libpthread(3).

Sponsored by <The NetBSD Foundation>

Reviewers: joerg, dvyukov, vitalybuka

Reviewed By: dvyukov

Subscribers: llvm-commits, kubamracek, #sanitizers

Tags: #sanitizers

Differential Revision: https://reviews.llvm.org/D40241

llvm-svn: 318646

6 years ago[LV] Model masking in VPlan, introducing VPInstructions
Gil Rapaport [Mon, 20 Nov 2017 12:01:47 +0000 (12:01 +0000)]
[LV] Model masking in VPlan, introducing VPInstructions

This patch adds a new abstraction layer to VPlan and leverages it to model the planned
instructions that manipulate masks (AND, OR, NOT), introduced during predication.

The new VPValue and VPUser classes model how data flows into, through and out
of a VPlan, forming the vertices of a planned Def-Use graph. The new
VPInstruction class is a generic single-instruction Recipe that models a
planned instruction along with its opcode, operands and users. See
VectorizationPlan.rst for more details.

Differential Revision: https://reviews.llvm.org/D38676

llvm-svn: 318645

6 years ago[Driver] Add a cc1 flag for the new TBAA metadata format
Ivan A. Kosarev [Mon, 20 Nov 2017 11:16:16 +0000 (11:16 +0000)]
[Driver] Add a cc1 flag for the new TBAA metadata format

This patch starts a series of changes to add support for the new
TBAA metadata format proposed in this llvm-dev thread:

http://lists.llvm.org/pipermail/llvm-dev/2017-November/118748.html

Differential Revision: https://reviews.llvm.org/D39955

llvm-svn: 318644

6 years ago[ARM GlobalISel] Add test for RSBri. NFC
Diana Picus [Mon, 20 Nov 2017 11:05:31 +0000 (11:05 +0000)]
[ARM GlobalISel] Add test for RSBri. NFC

Add instruction selector test for RSBri, which is derived from
AsI1_rbin_irs, and make sure it doesn't get mistaken for SUBri, which is
derived from the very similar AsI1_bin_irs pattern.

llvm-svn: 318643

6 years ago[ARM GlobalISel] Clean up binary operator tests. NFC
Diana Picus [Mon, 20 Nov 2017 10:35:35 +0000 (10:35 +0000)]
[ARM GlobalISel] Clean up binary operator tests. NFC

Remove some of the instruction selector tests for binary operators (and,
or, xor). These are all derived from the same kind of TableGen pattern,
AsI1_bin_irs, so there's no point in testing all of them.

llvm-svn: 318642

6 years ago[LV][X86] Support of AVX2 Gathers code generation and update the LV with this
Mohammed Agabaria [Mon, 20 Nov 2017 08:18:12 +0000 (08:18 +0000)]
[LV][X86] Support of AVX2 Gathers code generation and update the LV with this

This patch depends on: https://reviews.llvm.org/D35348

Support of pattern selection of masked gathers of AVX2 (X86\AVX2 code gen)
Update LoopVectorize to generate gathers for AVX2 processors.

Reviewers: delena, zvi, RKSimon, craig.topper, aaboud, igorb

Reviewed By: delena, RKSimon

Differential Revision: https://reviews.llvm.org/D35772

llvm-svn: 318641

6 years ago[X86] Regenerate a test with broadcast comments. NFC
Craig Topper [Mon, 20 Nov 2017 08:15:04 +0000 (08:15 +0000)]
[X86] Regenerate a test with broadcast comments. NFC

llvm-svn: 318640

6 years ago[IRCE] Smart range intersection
Max Kazantsev [Mon, 20 Nov 2017 06:07:57 +0000 (06:07 +0000)]
[IRCE] Smart range intersection

In rL316552, we ban intersection of unsigned latch range with signed range check and vice
versa, unless the entire range check iteration space is known positive. It was a correct
functional fix that saved us from dealing with ambiguous values, but it also appeared
to be a very restrictive limitation. In particular, in the following case:

  loop:
    %iv = phi i32 [ 0, %preheader ], [ %iv.next, %latch]
    %iv.offset = add i32 %iv, 10
    %rc = icmp slt i32 %iv.offset, %len
    br i1 %rc, label %latch, label %deopt

  latch:
    %iv.next = add i32 %iv, 11
    %cond = icmp i32 ult %iv.next, 100
    br it %cond, label %loop, label %exit

Here, the unsigned iteration range is `[0, 100)`, and the safe range for range
check is `[-10, %len - 10)`. For unsigned iteration spaces, we use unsigned
min/max functions for range intersection. Given this, we wanted to avoid dealing
with `-10` because it is interpreted as a very big unsigned value. Semantically, range
check's safe range goes through unsigned border, so in fact it is two disjoint
ranges in IV's iteration space. Intersection of such ranges is not trivial, so we prohibited
this case saying that we are not allowed to intersect such ranges.

What semantics of this safe range actually means is that we can start from `-10` and go
up increasing the `%iv` by one until we reach `%len - 10` (for simplicity let's assume that
`%len - 10`  is a reasonably big positive value).

In particular, this safe iteration space includes `0, 1, 2, ..., %len - 11`. So if we were able to return
safe iteration space `[0, %len - 10)`, we could safely intersect it with IV's iteration space. All
values in this range are non-negative, so using signed/unsigned min/max for them is unambiguous.

In this patch, we alter the algorithm of safe range calculation so that it returnes a subset of the
original safe space which is represented by one continuous range that does not go through wrap.
In order to reach this, we use modified SCEV substraction function. It can be imagined as a function
that substracts by `1` (or `-1`) as long as the further substraction does not cause a wrap in IV iteration
space. This allows us to perform IRCE in many situations when we deal with IV space and range check
of different types (in terms of signed/unsigned).

We apply this approach for both matching and not matching types of IV iteration space and the
range check. One implication of this is that now IRCE became smarter in detection of empty safe
ranges. For example, in this case:
  loop:
    %iv = phi i32 [ %begin, %preheader ], [ %iv.next, %latch]
    %iv.offset = sub i32 %iv, 10
    %rc = icmp ult i32 %iv.offset, %len
    br i1 %rc, label %latch, label %deopt

  latch:
    %iv.next = add i32 %iv, 11
    %cond = icmp i32 ult %iv.next, 100
    br it %cond, label %loop, label %exit

If `%len` was less than 10 but SCEV failed to trivially prove that `%begin - 10 >u %len- 10`,
we could end up executing entire loop in safe preloop while the main loop was still generated,
but never executed. Now, cutting the ranges so that if both `begin - 10` and `%len - 10` overflow,
we have a trivially empty range of `[0, 0)`. This in some cases prevents us from meaningless optimization.

Differential Revision: https://reviews.llvm.org/D39954

llvm-svn: 318639

6 years ago[CGP] Fix the crash caused by enable of complex addr mode
Serguei Katkov [Mon, 20 Nov 2017 05:42:36 +0000 (05:42 +0000)]
[CGP] Fix the crash caused by enable of complex addr mode

We must collect all AddModes even if they are the same.
This is due to Original value is different but we need all original
values collected as they are used as anchors in common phi finding.

Reviewers: john.brawn, reames
Reviewed By: john.brawn
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D40166

llvm-svn: 318638

6 years agoCOFF: Fix invalid comdat sections in test cases.
Peter Collingbourne [Mon, 20 Nov 2017 05:31:37 +0000 (05:31 +0000)]
COFF: Fix invalid comdat sections in test cases.

The comdat sections in these test cases do not comply with the COFF
specification, and link.exe rejects them. I plan to make a change to
how we handle comdat sections which would also cause us to reject them.

llvm-svn: 318637

6 years agoCOFF: Change SparseChunks to be of type std::vector<SectionChunk *>. NFC.
Peter Collingbourne [Mon, 20 Nov 2017 05:31:23 +0000 (05:31 +0000)]
COFF: Change SparseChunks to be of type std::vector<SectionChunk *>. NFC.

llvm-svn: 318636

6 years agoCOFF: Remove unused fields. NFC.
Peter Collingbourne [Mon, 20 Nov 2017 05:31:09 +0000 (05:31 +0000)]
COFF: Remove unused fields. NFC.

llvm-svn: 318635

6 years ago[X86] clzero check in getHostCPUFeatures should use getX86CpuIDAndInfo not getX86CpuI...
Craig Topper [Sun, 19 Nov 2017 23:49:19 +0000 (23:49 +0000)]
[X86] clzero check in getHostCPUFeatures should use getX86CpuIDAndInfo not getX86CpuIDAndInfoEx.

This leaf doesn't take an additional argument.

llvm-svn: 318634

6 years ago[X86] Reorder and reformat the feature bit checks in getHostCPUFeatues to keep the...
Craig Topper [Sun, 19 Nov 2017 23:30:22 +0000 (23:30 +0000)]
[X86] Reorder and reformat the feature bit checks in getHostCPUFeatues to keep the bits in order per register and encourage future additions to be in order too.

llvm-svn: 318633

6 years agoPort ScopInfo to the isl cpp bindings
Philip Pfaffe [Sun, 19 Nov 2017 22:13:34 +0000 (22:13 +0000)]
Port ScopInfo to the isl cpp bindings

Summary:
Most changes are mechanical, but in one place I changed the program semantics
by fixing a likely bug:

In `Scop::hasFeasibleRuntimeContext()`, I'm now explicitely handling the
error-case. Before, when the call to `addNonEmptyDomainConstraints()`
returned a null set, this (probably) accidentally worked because
isl_bool_error converts to true. I'm checking for nullptr now.

Reviewers: grosser, Meinersbur, bollu

Reviewed By: Meinersbur

Subscribers: nemanjai, kbarton, pollydev, llvm-commits

Differential Revision: https://reviews.llvm.org/D39971

llvm-svn: 318632

6 years agoRemove 2 unused methods DWARFDebugInfo::Find and their FindCallbackString
Jan Kratochvil [Sun, 19 Nov 2017 19:04:24 +0000 (19:04 +0000)]
Remove 2 unused methods DWARFDebugInfo::Find and their FindCallbackString

Differential revision: https://reviews.llvm.org/D40216

llvm-svn: 318631

6 years ago[x86] add sqrt tests for partially-inline-libcalls (PR31455)
Sanjay Patel [Sun, 19 Nov 2017 17:31:37 +0000 (17:31 +0000)]
[x86] add sqrt tests for partially-inline-libcalls (PR31455)

llvm-svn: 318630

6 years ago[LibCallSimplifier] allow splat vectors for pow(x, 0.5) -> sqrt() transforms
Sanjay Patel [Sun, 19 Nov 2017 16:42:27 +0000 (16:42 +0000)]
[LibCallSimplifier] allow splat vectors for pow(x, 0.5) -> sqrt() transforms

llvm-svn: 318629

6 years ago[LibCallSimplifier] partly fix pow(x, 0.5) -> sqrt() transforms
Sanjay Patel [Sun, 19 Nov 2017 16:13:14 +0000 (16:13 +0000)]
[LibCallSimplifier] partly fix pow(x, 0.5) -> sqrt() transforms

As the first test shows, we could transform an llvm intrinsic which never sets errno
into a libcall which could set errno (even though it's marked readnone?), so that's
not ideal.

It's possible that we can also transform a libcall which could set errno to an
intrinsic given the fast-math-flags constraint, but that's deferred to determine
exactly which set of FMF are needed.

Differential Revision: https://reviews.llvm.org/D40150

llvm-svn: 318628

6 years ago[InstSimplify] fold and/or of fcmp ord/uno when operand is known nnan
Sanjay Patel [Sun, 19 Nov 2017 15:34:27 +0000 (15:34 +0000)]
[InstSimplify] fold and/or of fcmp ord/uno when operand is known nnan

The 'ord' and 'uno' predicates have a logic operation for NAN built into their definitions:

FCMP_ORD   =  7,  ///< 0 1 1 1    True if ordered (no nans)
FCMP_UNO   =  8,  ///< 1 0 0 0    True if unordered: isnan(X) | isnan(Y)

So we can simplify patterns like this:

(fcmp ord (known NNAN), X) && (fcmp ord X, Y) --> fcmp ord X, Y
(fcmp uno (known NNAN), X) || (fcmp uno X, Y) --> fcmp uno X, Y

It might be better to split this into (X uno 0) | (Y uno 0) as a canonicalization, but that
would be another patch.

Differential Revision: https://reviews.llvm.org/D40130

llvm-svn: 318627

6 years agoAdd comments to DWARFCompileUnit length fields/methods
Jan Kratochvil [Sun, 19 Nov 2017 14:35:07 +0000 (14:35 +0000)]
Add comments to DWARFCompileUnit length fields/methods

Differential revision: https://reviews.llvm.org/D40211

llvm-svn: 318626

6 years agoAttempt to fix template depth test failures on older Clang versions
Eric Fiselier [Sun, 19 Nov 2017 09:46:34 +0000 (09:46 +0000)]
Attempt to fix template depth test failures on older Clang versions

llvm-svn: 318625

6 years ago[X86] Add test cases for rndscaless/sd intrinsics.
Craig Topper [Sun, 19 Nov 2017 06:24:26 +0000 (06:24 +0000)]
[X86] Add test cases for rndscaless/sd intrinsics.

Also fix the memop in the ins for these instructions. Not sure what effect this has.

llvm-svn: 318624

6 years ago[X86] Improve load folding of scalar rcp28 and rsqrt28 instructions using sse_load_f3...
Craig Topper [Sun, 19 Nov 2017 05:42:54 +0000 (05:42 +0000)]
[X86] Improve load folding of scalar rcp28 and rsqrt28 instructions using sse_load_f32/f64.

llvm-svn: 318623

6 years agoFix min/max usage in variant
Eric Fiselier [Sun, 19 Nov 2017 04:57:22 +0000 (04:57 +0000)]
Fix min/max usage in variant

llvm-svn: 318622

6 years ago[libc++] Shrink variant's index type when possible
Eric Fiselier [Sun, 19 Nov 2017 04:19:44 +0000 (04:19 +0000)]
[libc++] Shrink variant's index type when possible

Summary:
Currently `std::variant` always uses an unsigned int to store the variant index. However this isn't nessesary and causes `std::variant` to be larger than it needs to be in most cases.

This patch changes the index type to be `unsigned char` when possible, and `unsigned short` or `unsigned int` otherwise, depending on the size (Although it's questionable if it's even possible to create a variant with 65535 elements.

Unfortunately this change is an ABI break, and as such is only enabled in ABI v2.

Reviewers: mpark

Reviewed By: mpark

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D40210

llvm-svn: 318621

6 years ago[X86] Make sure 'knm' is accepted by -target-cpu
Craig Topper [Sun, 19 Nov 2017 04:12:35 +0000 (04:12 +0000)]
[X86] Make sure 'knm' is accepted by -target-cpu

llvm-svn: 318620

6 years ago[X86] Make sure 'knm' and 'cannonlake' are accepted by builtin_cpu_is
Craig Topper [Sun, 19 Nov 2017 04:12:33 +0000 (04:12 +0000)]
[X86] Make sure 'knm' and 'cannonlake' are accepted by builtin_cpu_is

llvm-svn: 318619

6 years agoFix nodiscard test when modules are enabled
Eric Fiselier [Sun, 19 Nov 2017 03:50:35 +0000 (03:50 +0000)]
Fix nodiscard test when modules are enabled

llvm-svn: 318618

6 years ago[X86] Add icelake CPU support for -march.
Craig Topper [Sun, 19 Nov 2017 02:55:15 +0000 (02:55 +0000)]
[X86] Add icelake CPU support for -march.

llvm-svn: 318617

6 years ago[X86] Set __corei7__ preprocessor defines for skylake server and cannonlake.
Craig Topper [Sun, 19 Nov 2017 02:55:14 +0000 (02:55 +0000)]
[X86] Set __corei7__ preprocessor defines for skylake server and cannonlake.

This is the resolution we came to in D38824.

llvm-svn: 318616

6 years ago[bpf] remove unused variable
Alexei Starovoitov [Sun, 19 Nov 2017 02:41:53 +0000 (02:41 +0000)]
[bpf] remove unused variable

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
llvm-svn: 318615

6 years ago[bpf] allow direct and indirect calls
Alexei Starovoitov [Sun, 19 Nov 2017 01:35:00 +0000 (01:35 +0000)]
[bpf] allow direct and indirect calls

kernel verifier is becoming smarter and soon will support
direct and indirect function calls.
Remove obsolete error from BPF backend.
Make call to use PCRel_4 fixup.
'bpf to bpf' calls are distinguished from 'bpf to kernel' calls
by insn->src_reg == BPF_PSEUDO_CALL == 1 which is used as relocation
indicator similar to ld_imm64->src_reg == BPF_PSEUDO_MAP_FD == 1
The actual 'call' instruction remains the same for both
'bpf to kernel' and 'bpf to bpf' calls.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
llvm-svn: 318614