platform/upstream/llvm.git
5 years agoGSYM: Add the llvm::gsym::Header header class with tests
Greg Clayton [Tue, 17 Sep 2019 17:46:13 +0000 (17:46 +0000)]
GSYM: Add the llvm::gsym::Header header class with tests

This patch adds the llvm::gsym::Header class which appears at the start of a stand alone GSYM file, or in the first bytes of the GSYM data in a GSYM section within a file. Added encode and decode methods with full error handling and full tests.

Differential Revision: https://reviews.llvm.org/D67666

llvm-svn: 372149

5 years ago[OPENMP] Rework the test, NFC.
Alexey Bataev [Tue, 17 Sep 2019 17:44:27 +0000 (17:44 +0000)]
[OPENMP] Rework the test, NFC.

llvm-svn: 372148

5 years ago[OPENMP5.0]Introduce attribute for declare variant directive.
Alexey Bataev [Tue, 17 Sep 2019 17:36:49 +0000 (17:36 +0000)]
[OPENMP5.0]Introduce attribute for declare variant directive.

Added attribute for declare variant directive. It will allow to handle
declare variant directive at the codegen and will allow to add extra
checks.

llvm-svn: 372147

5 years ago[TableGen] CodeGenMapTable - Don't dereference a dyn_cast result. NFCI.
Simon Pilgrim [Tue, 17 Sep 2019 17:32:15 +0000 (17:32 +0000)]
[TableGen] CodeGenMapTable - Don't dereference a dyn_cast result. NFCI.

The static analyzer is warning about potential null dereferences of dyn_cast<> results - in these cases we can safely use cast<> directly as we know that these cases should all be the correct type, which is why its working atm and anyway cast<> will assert if they aren't.

llvm-svn: 372146

5 years ago[ARM][AsmParser] Don't dereference a dyn_cast result. NFCI.
Simon Pilgrim [Tue, 17 Sep 2019 17:26:14 +0000 (17:26 +0000)]
[ARM][AsmParser] Don't dereference a dyn_cast result. NFCI.

The static analyzer is warning about potential null dereferences of dyn_cast<> results - in these cases we can safely use cast<> directly as we know that these cases should all be the correct type, which is why its working atm and anyway cast<> will assert if they aren't.

llvm-svn: 372145

5 years agoFix MSVC lambda capture warnings. NFCI.
Simon Pilgrim [Tue, 17 Sep 2019 17:24:55 +0000 (17:24 +0000)]
Fix MSVC lambda capture warnings. NFCI.

llvm-svn: 372144

5 years agoRemove asan test for strncat(x, y, 0)
David Bolvansky [Tue, 17 Sep 2019 17:17:30 +0000 (17:17 +0000)]
Remove asan test for strncat(x, y, 0)

llvm-svn: 372143

5 years agoReland "[SLC] Preserve attrs for strncpy(x, "", y) -> memset(align 1 x, '\0', y)"
David Bolvansky [Tue, 17 Sep 2019 17:12:24 +0000 (17:12 +0000)]
Reland "[SLC] Preserve attrs for strncpy(x, "", y) -> memset(align 1 x, '\0', y)"

llvm-svn: 372142

5 years ago[ASAN] Adjust asan tests due to new optimizations
David Bolvansky [Tue, 17 Sep 2019 17:07:31 +0000 (17:07 +0000)]
[ASAN] Adjust asan tests due to new optimizations

llvm-svn: 372141

5 years ago[PowerPC] Exploit single instruction load-and-splat for word and doubleword
Nemanja Ivanovic [Tue, 17 Sep 2019 16:45:20 +0000 (16:45 +0000)]
[PowerPC] Exploit single instruction load-and-splat for word and doubleword

We currently produce a load, followed by (possibly a move for integers and) a
splat as separate instructions. VSX has always had a splatting load for
doublewords, but as of Power9, we have it for words as well. This patch just
exploits these instructions.

Differential revision: https://reviews.llvm.org/D63624

llvm-svn: 372139

5 years ago[MemorySSA] Fix phi insertion when inserting a def.
Alina Sbirlea [Tue, 17 Sep 2019 16:33:35 +0000 (16:33 +0000)]
[MemorySSA] Fix phi insertion when inserting a def.

Summary:
When inserting a Def, the current algorithm is walking edges backward
and inserting new Phis where needed. There may be additional Phis needed
in the IDF of the newly inserted Def and Phis.
Adding Phis in the IDF of the Def was added ina  previous patch, but we
may also need other Phis in the IDF of the newly added Phis.

Reviewers: george.burgess.iv

Subscribers: Prazek, sanjoy.google, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D67637

llvm-svn: 372138

5 years ago[MemorySSA] Update MSSA for non-conventional AA.
Alina Sbirlea [Tue, 17 Sep 2019 16:31:37 +0000 (16:31 +0000)]
[MemorySSA] Update MSSA for non-conventional AA.

Summary:
Regularly when moving an instruction that may not read or write memory,
the instruction is not modelled in MSSA, so not action is necessary.
For a non-conventional AA pipeline, MSSA needs to explicitly check when
creating accesses, so as to not model instructions that may not read and
write memory.

Reviewers: george.burgess.iv

Subscribers: Prazek, sanjoy.google, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D67562

llvm-svn: 372137

5 years agoMove DK_Misexpect for compatability with getNextAvailablePluginDiagnosticKind
Petr Hosek [Tue, 17 Sep 2019 16:27:36 +0000 (16:27 +0000)]
Move DK_Misexpect for compatability with getNextAvailablePluginDiagnosticKind

First identified after D66324 landed.

Patch By: paulkirth
Differential Revision: https://reviews.llvm.org/D67648

llvm-svn: 372136

5 years agoGSYM: add encoding and decoding to FunctionInfo
Greg Clayton [Tue, 17 Sep 2019 16:15:49 +0000 (16:15 +0000)]
GSYM: add encoding and decoding to FunctionInfo

This patch adds encoding and decoding of the FunctionInfo objects along with full error handling and tests. Full details of the FunctionInfo encoding format appear in the FunctionInfo.h header file.

Differential Revision: https://reviews.llvm.org/D67506

llvm-svn: 372135

5 years ago[ARM] Add a SelectTAddrModeImm7 for MVE narrow loads and stores
David Green [Tue, 17 Sep 2019 15:32:28 +0000 (15:32 +0000)]
[ARM] Add a SelectTAddrModeImm7 for MVE narrow loads and stores

We were previously using the SelectT2AddrModeImm7 for both normal and narrowing
MVE loads/stores. As the narrowing instructions do not accept sp as a register,
it makes little sense to optimise a FrameIndex into the load, only to have to
recover that later on. This adds a SelectTAddrModeImm7 which does not do that
folding, and uses it for narrowing load/store patterns.

Differential Revision: https://reviews.llvm.org/D67489

llvm-svn: 372134

5 years ago[ARM] Fixup pipeline test. NFC
David Green [Tue, 17 Sep 2019 15:25:24 +0000 (15:25 +0000)]
[ARM] Fixup pipeline test. NFC

llvm-svn: 372133

5 years ago[ARM] Reserve an emergency spill slot for fp16 addressing modes that need it
David Green [Tue, 17 Sep 2019 15:23:09 +0000 (15:23 +0000)]
[ARM] Reserve an emergency spill slot for fp16 addressing modes that need it

Similar to D67327, but this time for the FP16 VLDR and VSTR instructions that
use the AddrMode5FP16 addressing mode. We need to reserve an emergency spill
slot for instructions that will be out of range to use sp directly.
AddrMode5FP16 is 8 bits with a scale of 2.

Differential Revision: https://reviews.llvm.org/D67483

llvm-svn: 372132

5 years ago[OPENMP]Try to rework the test to pacify the buildbots, NFC.
Alexey Bataev [Tue, 17 Sep 2019 15:11:52 +0000 (15:11 +0000)]
[OPENMP]Try to rework the test to pacify the buildbots, NFC.

llvm-svn: 372130

5 years ago[clang-format] Fix cleanup of `AnnotatedLine` to include children nodes.
Yitzhak Mandelbaum [Tue, 17 Sep 2019 15:10:39 +0000 (15:10 +0000)]
[clang-format] Fix cleanup of `AnnotatedLine` to include children nodes.

Summary:
AnnotatedLine has a tree structure, and things like the body of a lambda will be
a child of the lambda expression. For example,

    [&]() { foo(a); };

will have an AnnotatedLine with a child:

    [&]() {};
     '- foo(a);

Currently, when the `Cleaner` class analyzes the affected lines, it does not
cleanup the lines' children nodes, which results in missed cleanup
opportunities, like the lambda body in the example above.

This revision extends the algorithm to visit children, thereby fixing the above problem.

Patch by Eric Li.

Reviewers: krasimir

Subscribers: cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D67659

llvm-svn: 372129

5 years ago[clangd] Fix another TSAN issue
Benjamin Kramer [Tue, 17 Sep 2019 14:56:11 +0000 (14:56 +0000)]
[clangd] Fix another TSAN issue

llvm-svn: 372128

5 years ago[RISCV] Unbreak the build
Benjamin Kramer [Tue, 17 Sep 2019 14:27:31 +0000 (14:27 +0000)]
[RISCV] Unbreak the build

llvm-svn: 372127

5 years ago[ARM] Fix for buildbots
Sam Parker [Tue, 17 Sep 2019 14:21:36 +0000 (14:21 +0000)]
[ARM] Fix for buildbots

Remove setPreservesCFG from ARMConstantIslandPass and add a couple
of -verify-machine-dom-info instances into the existing codegen
tests.

llvm-svn: 372126

5 years agoRevert "[SLC] Preserve attrs for strncpy(x, "", y) -> memset(align 1 x, '\0', y)"
Krasimir Georgiev [Tue, 17 Sep 2019 14:15:23 +0000 (14:15 +0000)]
Revert "[SLC] Preserve attrs for strncpy(x, "", y) -> memset(align 1 x, '\0', y)"

Summary:
This reverts commit r372101.

Causes ASAN build bot failures:

http://lab.llvm.org:8011/builders/sanitizer-ppc64be-linux/builds/14176
From http://lab.llvm.org:8011/builders/sanitizer-ppc64be-linux/builds/14176/steps/64-bit%20check-asan/logs/stdio:

```
[ RUN      ] AddressSanitizer.StrNCatOOBTest
/home/buildbots/ppc64be-sanitizer/sanitizer-ppc64be/build/llvm-project/compiler-rt/lib/asan/tests/asan_str_test.cpp:462: Failure
Death test: strncat(to - 1, from, 0)
    Result: failed to die.
```

Subscribers: llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D67658

llvm-svn: 372125

5 years agoAdd SpellingNotCalculated to Attribute Enums to suppress UBSan warnings
Erich Keane [Tue, 17 Sep 2019 14:11:51 +0000 (14:11 +0000)]
Add SpellingNotCalculated to Attribute Enums to suppress UBSan warnings

UBSan downstreams noticed that the assignment of SpellingNotCalculated
to the spellings caused warnings.

llvm-svn: 372124

5 years agogn build: (manually) merge r372076
Nico Weber [Tue, 17 Sep 2019 14:06:05 +0000 (14:06 +0000)]
gn build: (manually) merge r372076

llvm-svn: 372123

5 years ago[llvm-readobj/llvm-objdump] - Improve how tool locate the dynamic table and report...
George Rimar [Tue, 17 Sep 2019 13:58:46 +0000 (13:58 +0000)]
[llvm-readobj/llvm-objdump] - Improve how tool locate the dynamic table and report warnings about that.

Before this patch we gave a priority to a dynamic table found
from the section header.

It was discussed (here: https://reviews.llvm.org/D67078?id=218356#inline-602082)
that probably preferring the table from PT_DYNAMIC is better,
because it is what runtime loader sees.

This patch makes the table from PT_DYNAMIC be chosen at first place if it is available.
But also it adds logic to fall back to SHT_DYNAMIC if the table from the dynamic segment is
broken or fall back to use no table if both are broken.

It adds a few more diagnostic warnings for the logic above.

Differential revision: https://reviews.llvm.org/D67547

llvm-svn: 372122

5 years ago[ARM] Fix for buildbots
Sam Parker [Tue, 17 Sep 2019 13:46:26 +0000 (13:46 +0000)]
[ARM] Fix for buildbots

Add --verifymachineinstrs and update the remaining low overhead loop
tests.

llvm-svn: 372121

5 years ago[RISCV][NFC] Use NoRegister instead of 0 literal
Luis Marques [Tue, 17 Sep 2019 13:34:17 +0000 (13:34 +0000)]
[RISCV][NFC] Use NoRegister instead of 0 literal

Summary: Trivial cleanup.

Reviewers: asb, lenary

Reviewed By: lenary

Subscribers: hiraditya, rbar, johnrusso, simoncook, apazos, sabuasal, niosHD, kito-cheng, shiva0217, jrtc27, MaskRay, zzheng, edward-jones, rogfer01, MartinMosbeck, brucehoult, the_o, rkruppe, PkmX, jocewei, psnobl, benna, Jim, s.egerton, pzheng, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D67526

llvm-svn: 372120

5 years ago[OpenCL] Tidy up some comments; NFC
Sven van Haastregt [Tue, 17 Sep 2019 13:32:56 +0000 (13:32 +0000)]
[OpenCL] Tidy up some comments; NFC

llvm-svn: 372119

5 years ago[X86] X86DAGToDAGISel::tryFoldLoad - assert root/parent pointers are non-null. NFCI.
Simon Pilgrim [Tue, 17 Sep 2019 13:27:02 +0000 (13:27 +0000)]
[X86] X86DAGToDAGISel::tryFoldLoad - assert root/parent pointers are non-null. NFCI.

Silences a static analyzer warning.

llvm-svn: 372118

5 years agoInterleavedAccessInfo - Don't dereference a dyn_cast result. NFCI.
Simon Pilgrim [Tue, 17 Sep 2019 13:25:56 +0000 (13:25 +0000)]
InterleavedAccessInfo - Don't dereference a dyn_cast result. NFCI.

llvm-svn: 372117

5 years ago[LoopVectorize] Don't dereference a dyn_cast result. NFCI.
Simon Pilgrim [Tue, 17 Sep 2019 13:24:54 +0000 (13:24 +0000)]
[LoopVectorize] Don't dereference a dyn_cast result. NFCI.

The static analyzer is warning about potential null dereferences of dyn_cast<> results, we can use cast<> directly as we know that these cases should all be CastInst, which is why its working atm and anyway cast<> will assert if they aren't.

llvm-svn: 372116

5 years ago[ARM] Fix for MVE load/store stack accesses
David Green [Tue, 17 Sep 2019 12:58:51 +0000 (12:58 +0000)]
[ARM] Fix for MVE load/store stack accesses

MVE loads and stores have a 7 bit immediate range, scaled by the length of the type. This needs to be taught to the stack estimation code to ensure that an emergency spill slot is reserved in case we run out of registers when materialising stack indices.

Also the narrowing loads/stores can be created with frame indices even though they do not accept SP as a register. We need in those cases to make sure we have an emergency register to use as the frame base, as SP can never be used.

Differential Revision: https://reviews.llvm.org/D67327

llvm-svn: 372114

5 years agoHide implementation details in namespaces.
Benjamin Kramer [Tue, 17 Sep 2019 12:56:29 +0000 (12:56 +0000)]
Hide implementation details in namespaces.

llvm-svn: 372113

5 years agolldb: move a test input to the test Inputs dir
Krasimir Georgiev [Tue, 17 Sep 2019 12:23:03 +0000 (12:23 +0000)]
lldb: move a test input to the test Inputs dir

Summary:
This makes the input file for a new test added in r372060 directly
available in the Inputs subdirectory of the test dir.

Differential Revision: https://reviews.llvm.org/D67655

llvm-svn: 372112

5 years ago[ARM][LowOverheadLoops] Add LR def safety check
Sam Parker [Tue, 17 Sep 2019 12:19:32 +0000 (12:19 +0000)]
[ARM][LowOverheadLoops] Add LR def safety check

Converting the *LoopStart pseudo instructions into DLS/WLS results in
LR being defined. These instructions were inserted on the assumption
that LR would already contain the loop counter because a mov is
introduced during ISel as the the consumers in the loop can only use
LR. That assumption proved wrong!

So perform a safety check, finding an appropriate place to insert the
DLS/WLS instructions or revert if this isn't possible.

Differential Revision: https://reviews.llvm.org/D67539

llvm-svn: 372111

5 years ago[llvm-readobj] - Test PPC64 relocations properly.
George Rimar [Tue, 17 Sep 2019 12:05:39 +0000 (12:05 +0000)]
[llvm-readobj] - Test PPC64 relocations properly.

We had a precompiled binary committed and not all of the relocations
supported were tested. This patch fixes this.

Differential revision: https://reviews.llvm.org/D67617

llvm-svn: 372110

5 years ago[obj2yaml] - Support PPC64 relocation types.
George Rimar [Tue, 17 Sep 2019 12:00:55 +0000 (12:00 +0000)]
[obj2yaml] - Support PPC64 relocation types.

We do not support them and fail with llvm_unreachable currently.
This is not the only target we do not support and also seems we are missing
the tests for those we have already. But I needed this one for another patch,
so posted it separatelly.

Relocation names are taken from llvm\include\llvm\BinaryFormat\ELFRelocs\PowerPC64.def

Differential revision: https://reviews.llvm.org/D67615

llvm-svn: 372109

5 years ago[yaml2obj/obj2yaml] - Allow setting an arbitrary values for e_machine.
George Rimar [Tue, 17 Sep 2019 11:51:26 +0000 (11:51 +0000)]
[yaml2obj/obj2yaml] - Allow setting an arbitrary values for e_machine.

Currently we only allow using a known named constants
for `Machine` field in YAML documents.

This patch allows using any numbers (valid or "unknown")
and adds test cases for current and new functionality.

With this it is possible to write a test cases for really unknown
EM_* targets.

Differential revision: https://reviews.llvm.org/D67652

llvm-svn: 372108

5 years ago[docs] Make --version text more correct
James Henderson [Tue, 17 Sep 2019 11:43:42 +0000 (11:43 +0000)]
[docs] Make --version text more correct

Follow-up to r371983. Referring to "this program" in the description of
the --version option in the documentation isn't exactly correct, because
the docs are not part of the program, and so "this program" doesn't
really refer to anything. This patch brings the other users of this
terminology into line with the new updates to llvm-size and
llvm-strings.

Reviewed by: alexshap, MaskRay

Differential Revision: https://reviews.llvm.org/D67618

llvm-svn: 372107

5 years ago[RISCV] Switch to the Machine Scheduler
Luis Marques [Tue, 17 Sep 2019 11:15:35 +0000 (11:15 +0000)]
[RISCV] Switch to the Machine Scheduler

Most of the test changes are trivial instruction reorderings and differing
register allocations, without any obvious performance impact.

Differential Revision: https://reviews.llvm.org/D66973

llvm-svn: 372106

5 years ago[Attributor][Fix] Initialize the cache prior to using it
Johannes Doerfert [Tue, 17 Sep 2019 10:52:41 +0000 (10:52 +0000)]
[Attributor][Fix] Initialize the cache prior to using it

Summary:
There were segfaults as we modified and iterated the instruction maps in
the cache at the same time. This was happening because we created new
instructions while we populated the cache. This fix changes the order
in which we perform these actions. First, the caches for the whole
module are created, then we start to create abstract attributes.

I don't have a unit test but the LLVM test suite exposes this problem.

Reviewers: uenoku, sstefan1

Subscribers: hiraditya, bollu, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D67232

llvm-svn: 372105

5 years agoRevert Patch from Phabricator
Luis Marques [Tue, 17 Sep 2019 10:52:09 +0000 (10:52 +0000)]
Revert Patch from Phabricator

This reverts r372092 (git commit e38695a0255c9e7b53639f349f8101bae1ce5c04)

llvm-svn: 372104

5 years ago[X86] Use APInt::getLowBitsSet helper. NFCI.
Simon Pilgrim [Tue, 17 Sep 2019 10:51:30 +0000 (10:51 +0000)]
[X86] Use APInt::getLowBitsSet helper. NFCI.

Also avoids a static analyzer warning about out of range shifts.

llvm-svn: 372103

5 years agoAdd SemanticRanges to Clangd server.
Utkarsh Saxena [Tue, 17 Sep 2019 10:28:05 +0000 (10:28 +0000)]
Add SemanticRanges to Clangd server.

Summary:
Adds Semantic Ranges capabilities to Clangd server.
Also adds tests for running it via clangd server.

This differs from the LSP spec as the spec needs this to be evaluated on multiple 'pos' and the expected output is an list of list of semantic ranges.
This is majorly for multi cursor and assuming this is a rare thing, we don't want to optimize make things complicated just for this.
This should be done in the LSP level by queueing one request per 'pos' in the input.

LSP Spec:
https://github.com/microsoft/language-server-protocol/blob/dbaeumer/3.15/specification.md#textDocument_selectionRange

Reviewers: hokein

Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, kadircet, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D67650

llvm-svn: 372102

5 years ago[SLC] Preserve attrs for strncpy(x, "", y) -> memset(align 1 x, '\0', y)
David Bolvansky [Tue, 17 Sep 2019 10:25:38 +0000 (10:25 +0000)]
[SLC] Preserve attrs for strncpy(x, "", y) -> memset(align 1 x, '\0', y)

llvm-svn: 372101

5 years ago[SVE][MVT] Fixed-length vector MVT ranges
Graham Hunter [Tue, 17 Sep 2019 10:19:23 +0000 (10:19 +0000)]
[SVE][MVT] Fixed-length vector MVT ranges

  * Reordered MVT simple types to group scalable vector types
    together.
  * New range functions in MachineValueType.h to only iterate over
    the fixed-length int/fp vector types.
  * Stopped backends which don't support scalable vector types from
    iterating over scalable types.

Reviewers: sdesmalen, greened

Reviewed By: greened

Differential Revision: https://reviews.llvm.org/D66339

llvm-svn: 372099

5 years ago[InstCombine] Annotate strdup with deref_or_null
David Bolvansky [Tue, 17 Sep 2019 10:12:48 +0000 (10:12 +0000)]
[InstCombine] Annotate strdup with deref_or_null

llvm-svn: 372098

5 years ago[NFCI] Fixed buildbots
David Bolvansky [Tue, 17 Sep 2019 10:03:45 +0000 (10:03 +0000)]
[NFCI] Fixed buildbots

llvm-svn: 372097

5 years ago[SimplifyLibCalls] Fix -Wunused-result after D53342/r372091
Fangrui Song [Tue, 17 Sep 2019 09:56:55 +0000 (09:56 +0000)]
[SimplifyLibCalls] Fix -Wunused-result after D53342/r372091

llvm-svn: 372096

5 years ago[NFC] Updated test
David Bolvansky [Tue, 17 Sep 2019 09:53:14 +0000 (09:53 +0000)]
[NFC] Updated test

llvm-svn: 372095

5 years ago[ELF][AARCH64] Refactor AArchErrataFix to match changes in ARMErrataFix NFC.
Peter Smith [Tue, 17 Sep 2019 09:49:30 +0000 (09:49 +0000)]
[ELF][AARCH64] Refactor AArchErrataFix to match changes in ARMErrataFix NFC.

D67284 introduced ARMErrataFix.cpp which was derived from
AArch64ErrataFix.cpp. There were some useful refactoring changes made to
ARMErrataFix.cpp made as part of the review. This change applies the
relevant changes back to AArch64ErrataFix.cpp.

Main changes are:
- Old style variable names in comments like IS, are now new style isec.
- Simplify init() collection of mappingSymbols to always start with a code
mapping symbol.
- Simplify logic in mergeCmp().
- Fix one 80 column overflow caused by IS -> isec transformation.

Differential Revision: https://reviews.llvm.org/D67622

llvm-svn: 372094

5 years ago[NFC} Updated test
David Bolvansky [Tue, 17 Sep 2019 09:45:52 +0000 (09:45 +0000)]
[NFC} Updated test

llvm-svn: 372093

5 years agoPatch from Phabricator
Luis Marques [Tue, 17 Sep 2019 09:43:08 +0000 (09:43 +0000)]
Patch from Phabricator

llvm-svn: 372092

5 years ago[SimplifyLibCalls] Mark known arguments with nonnull
David Bolvansky [Tue, 17 Sep 2019 09:32:52 +0000 (09:32 +0000)]
[SimplifyLibCalls] Mark known arguments with nonnull

Reviewers: efriedma, jdoerfert

Reviewed By: jdoerfert

Subscribers: ychen, rsmith, joerg, aaron.ballman, lebedev.ri, uenoku, jdoerfert, hfinkel, javed.absar, spatel, dmgreen, llvm-commits

Differential Revision: https://reviews.llvm.org/D53342

llvm-svn: 372091

5 years ago[lldb] [Process/gdb-remote] Fix defaulting signal to invalid in action list
Michal Gorny [Tue, 17 Sep 2019 09:31:00 +0000 (09:31 +0000)]
[lldb] [Process/gdb-remote] Fix defaulting signal to invalid in action list

Fix processing of "C" packet with signal for the whole process to
default signal value for action list to LLDB_INVALID_SIGNAL_NUMBER
rather than 0.

Differential Revision: https://reviews.llvm.org/D67625

llvm-svn: 372090

5 years ago[llvm-readobj] - Fix BB after r372087.
George Rimar [Tue, 17 Sep 2019 09:26:49 +0000 (09:26 +0000)]
[llvm-readobj] - Fix BB after r372087.

Seems I forgot to update the number of bytes checked.

llvm-svn: 372089

5 years ago[llvm-ar] Parse 'h' and '-h': display help and exit
Fangrui Song [Tue, 17 Sep 2019 09:25:52 +0000 (09:25 +0000)]
[llvm-ar] Parse 'h' and '-h': display help and exit

Support `llvm-ar h` and `llvm-ar -h` because they may be what users try
at first. Note, operation 'h' is undocumented in GNU ar.

Reviewed By: jhenderson

Differential Revision: https://reviews.llvm.org/D67560

llvm-svn: 372088

5 years ago[llvm-readobj] - Fix a TODO in elf-reloc-zero-name-or-value.test.
George Rimar [Tue, 17 Sep 2019 09:12:10 +0000 (09:12 +0000)]
[llvm-readobj] - Fix a TODO in elf-reloc-zero-name-or-value.test.

The "TODO" mentioned was:

"Add test for symbol with no name but with a value once yaml2obj allows
referencing symbols with no name from relocations."

We can do it now.

Differential revision: https://reviews.llvm.org/D67609

llvm-svn: 372087

5 years ago[AMDGPU]: PHI Elimination hooks added for custom COPY insertion. Fixed
Alexander Timofeev [Tue, 17 Sep 2019 09:08:58 +0000 (09:08 +0000)]
[AMDGPU]: PHI Elimination hooks added for custom COPY insertion. Fixed

Defferential Revision: https://reviews.llvm.org/D67101

Reviewers: rampitec, vpykhtin
llvm-svn: 372086

5 years ago[ARM] LE support in ConstantIslands
Sam Parker [Tue, 17 Sep 2019 09:08:05 +0000 (09:08 +0000)]
[ARM] LE support in ConstantIslands

The low-overhead branch extension provides a loop-end 'LE' instruction
that performs no decrement nor compare, it just jumps backwards. This
patch modifies the constant islands pass to try to insert LE
instructions in place of a Thumb2 conditional branch, instead of
shrinking it. This only happens if a cmp can be converted to a cbn/z
and used to exit the loop.

Differential Revision: https://reviews.llvm.org/D67404

llvm-svn: 372085

5 years ago[LoopUnroll] Use LoopSize+1 as threshold, to allow unrolling loops matching LoopSize.
Florian Hahn [Tue, 17 Sep 2019 09:02:48 +0000 (09:02 +0000)]
[LoopUnroll] Use LoopSize+1 as threshold, to allow unrolling loops matching LoopSize.

We use `< UP.Threshold` later on, so we should use LoopSize + 1, to
allow unrolling if the result won't exceed to loop size.

Fixes PR43305.

Reviewers: efriedma, dmgreen, paquette

Reviewed By: dmgreen

Differential Revision: https://reviews.llvm.org/D67594

llvm-svn: 372084

5 years ago[llvm-readobj] - Refactor the code.
George Rimar [Tue, 17 Sep 2019 08:53:18 +0000 (08:53 +0000)]
[llvm-readobj] - Refactor the code.

It's a straightforward refactoring that allows to simplify and encapsulate the code.

Differential revision: https://reviews.llvm.org/D67624

llvm-svn: 372083

5 years ago[Clang] Pragma vectorize_width() implies vectorize(enable)
Sjoerd Meijer [Tue, 17 Sep 2019 08:43:11 +0000 (08:43 +0000)]
[Clang] Pragma vectorize_width() implies vectorize(enable)

Specifying the vectorization width was supposed to implicitly enable
vectorization, except that it wasn't really doing this. It was only
setting the vectorize.width metadata, but not vectorize.enable.

This should fix PR27643.

Differential Revision: https://reviews.llvm.org/D66290

llvm-svn: 372082

5 years ago[llvm-objcopy] - Remove python invocations from 2 test cases.
George Rimar [Tue, 17 Sep 2019 08:38:53 +0000 (08:38 +0000)]
[llvm-objcopy] - Remove python invocations from 2 test cases.

It is possible to use yaml2obj to create sections with overlapping sh_offset now.
This patch does that.

Differential revision: https://reviews.llvm.org/D67610

llvm-svn: 372081

5 years ago[RISCV] Add option aliases: -mcmodel=medany and -mcmodel=medlow
Kito Cheng [Tue, 17 Sep 2019 08:19:17 +0000 (08:19 +0000)]
[RISCV] Add option aliases: -mcmodel=medany and -mcmodel=medlow

RISC-V GCC use -mcmodel=medany and -mcmodel=medlow, but LLVM use
-mcmodel=small and -mcmodel=medium.

Add those two option aliases for provide same user interface between
GCC and LLVM.

Reviewed By: lenary

Differential Revision: https://reviews.llvm.org/D67066

llvm-svn: 372080

5 years ago[bugpoint] Add support for -Oz and properly enable -Os.
Florian Hahn [Tue, 17 Sep 2019 08:14:09 +0000 (08:14 +0000)]
[bugpoint] Add support for -Oz and properly enable -Os.

This patch adds -Oz as option and also properly enables support for -Os.
Currently, the existing check for -Os is dead, because the enclosing if
only checks of O1, O2 and O3.

There is still a difference between the -Oz pipeline compared to opt,
but I have not been able to track that down yet.

Reviewers: bogner, sebpop, efriedma

Reviewed By: efriedma

Differential Revision: https://reviews.llvm.org/D67593

llvm-svn: 372079

5 years ago[RISCV] Define __riscv_cmodel_medlow and __riscv_cmodel_medany correctly
Kito Cheng [Tue, 17 Sep 2019 08:09:56 +0000 (08:09 +0000)]
[RISCV] Define __riscv_cmodel_medlow and __riscv_cmodel_medany correctly

RISC-V LLVM was only implement small/medlow code model, so it defined
__riscv_cmodel_medlow directly without check.

Now, we have medium/medany code model in RISC-V back-end, it should
define according the actually code model.

Reviewed By: lewis-revill

Differential Revision: https://reviews.llvm.org/D67065

llvm-svn: 372078

5 years agoReland "[lldb][NFC] Make ApplyObjcCastHack less scary"
Raphael Isemann [Tue, 17 Sep 2019 07:58:01 +0000 (07:58 +0000)]
Reland "[lldb][NFC] Make ApplyObjcCastHack less scary"

First version had a typo.

llvm-svn: 372077

5 years ago[ARM][MVE] Add invalidForTailPredication to TSFlags
Sam Parker [Tue, 17 Sep 2019 07:43:04 +0000 (07:43 +0000)]
[ARM][MVE] Add invalidForTailPredication to TSFlags

Set this bit for the MVE reduction instructions to prevent a loop from
becoming tail predicated in their presence.

Differential Revision: https://reviews.llvm.org/D67444

llvm-svn: 372076

5 years ago[Attributor] Use Alias Analysis in noalias callsite argument deduction
Hideto Ueno [Tue, 17 Sep 2019 06:53:27 +0000 (06:53 +0000)]
[Attributor] Use Alias Analysis in noalias callsite argument deduction

Summary: This patch adds a check of alias analysis in `noalias` callsite argument deduction.

Reviewers: jdoerfert, sstefan1

Reviewed By: jdoerfert

Subscribers: hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D67604

llvm-svn: 372075

5 years ago[clangd] Remove stale comment. NFC
Ilya Biryukov [Tue, 17 Sep 2019 06:49:50 +0000 (06:49 +0000)]
[clangd] Remove stale comment. NFC

llvm-svn: 372074

5 years ago[Attributor] Create helper struct for handling analysis getters
Hideto Ueno [Tue, 17 Sep 2019 05:45:18 +0000 (05:45 +0000)]
[Attributor] Create helper struct for handling analysis getters

Summary: This patch introduces a helper struct `AnalysisGetter` to put together analysis getters. In this patch, a getter for `AAResult` is also added for  `noalias`.

Reviewers: jdoerfert, sstefan1

Reviewed By: jdoerfert

Subscribers: hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D67603

llvm-svn: 372072

5 years ago[git-llvm] Do not reinvent `@{upstream}` (take 2)
David Zarzycki [Tue, 17 Sep 2019 04:44:13 +0000 (04:44 +0000)]
[git-llvm] Do not reinvent `@{upstream}` (take 2)

This makes git-llvm more of a thin wrapper around git while temporarily
maintaining backwards compatibility with past git-llvm behavior.

Using @{upstream} makes git-llvm more robust when used with a nontrivial
local repository.

https://reviews.llvm.org/D67389

llvm-svn: 372070

5 years ago[X86] Split oversized vXi1 vector arguments and return values into scalars on avx512...
Craig Topper [Tue, 17 Sep 2019 04:41:14 +0000 (04:41 +0000)]
[X86] Split oversized vXi1 vector arguments and return values into scalars on avx512 targets.

Previously we tried to split them into narrower v64i1 or v16i1
pieces that each got promoted to vXi8 and then passed in a zmm
or xmm register. But this crashes when you need to pass more
pieces than available registers reserved for argument passing.

The scalarizing done here generates much longer and slower code,
but is consistent with the behavior of avx2 and earlier targets
for these types.

Fixes PR43323.

llvm-svn: 372069

5 years ago[X86] Allow masked VBROADCAST instructions to be turned into BLENDM with a broadcast...
Craig Topper [Tue, 17 Sep 2019 04:41:10 +0000 (04:41 +0000)]
[X86] Allow masked VBROADCAST instructions to be turned into BLENDM with a broadcast load to avoid a copy.

The BLENDM instructions allow an 2 sources and an independent
destination while masked VBROADCAST has the destination tied
to the source.

llvm-svn: 372068

5 years ago[X86] Add support for commuting EVEX VCMP instructons with any immediate value.
Craig Topper [Tue, 17 Sep 2019 04:41:05 +0000 (04:41 +0000)]
[X86] Add support for commuting EVEX VCMP instructons with any immediate value.

Previously we limited to the EQ/NE/TRUE/FALSE/ORD/UNORD immediates.

llvm-svn: 372067

5 years ago[X86] Add test case for missed opportunity to commute a VCMP instruction after unfold...
Craig Topper [Tue, 17 Sep 2019 04:41:01 +0000 (04:41 +0000)]
[X86] Add test case for missed opportunity to commute a VCMP instruction after unfolding one load in order to fold another load.

llvm-svn: 372066

5 years ago[X86] Enable commuting of EVEX VCMP for all immediate values during isel.
Craig Topper [Tue, 17 Sep 2019 04:40:58 +0000 (04:40 +0000)]
[X86] Enable commuting of EVEX VCMP for all immediate values during isel.

llvm-svn: 372065

5 years ago[test] Disable reproducer dump test on Windows
Jonas Devlieghere [Tue, 17 Sep 2019 03:58:32 +0000 (03:58 +0000)]
[test] Disable reproducer dump test on Windows

llvm-svn: 372064

5 years agoFix reliance on -flax-vector-conversions in AVX intrinsics headers and
Richard Smith [Tue, 17 Sep 2019 03:56:30 +0000 (03:56 +0000)]
Fix reliance on -flax-vector-conversions in AVX intrinsics headers and
corresponding tests.

llvm-svn: 372063

5 years agoFix reliance on lax vector conversions in tests for x86 intrinsics.
Richard Smith [Tue, 17 Sep 2019 03:56:28 +0000 (03:56 +0000)]
Fix reliance on lax vector conversions in tests for x86 intrinsics.

llvm-svn: 372062

5 years agoRemove reliance on lax vector conversions from altivec.h in VSX mode.
Richard Smith [Tue, 17 Sep 2019 03:56:26 +0000 (03:56 +0000)]
Remove reliance on lax vector conversions from altivec.h in VSX mode.

llvm-svn: 372061

5 years ago[ScriptInterpreter] Initialize globals when loading a scripting module.
Jonas Devlieghere [Tue, 17 Sep 2019 03:55:58 +0000 (03:55 +0000)]
[ScriptInterpreter] Initialize globals when loading a scripting module.

The LoadScriptingModule used by command script import wasn't
initializing the LLDB global variables (things like `lldb.frame` and
`lldb.debugger`). They would get initialized however when running the
interactive script interpreter or running a single script line (e.g.
`script print(lldb.frame)`). This patch fixes that by properly
initializing the globals when loading a Python module.

Differential revision: https://reviews.llvm.org/D67644

llvm-svn: 372060

5 years ago[ELF][Hexagon] Allow PT_LOAD to have overlapping p_offset ranges on EM_HEXAGON
Fangrui Song [Tue, 17 Sep 2019 02:45:38 +0000 (02:45 +0000)]
[ELF][Hexagon] Allow PT_LOAD to have overlapping p_offset ranges on EM_HEXAGON

Port the D64906 technique to EM_HEXAGON. This concludes the patch series.

Differential Revision: https://reviews.llvm.org/D67605

llvm-svn: 372059

5 years agoPush lambda scope earlier when transforming lambda expression
Nicholas Allegra [Tue, 17 Sep 2019 01:43:33 +0000 (01:43 +0000)]
Push lambda scope earlier when transforming lambda expression

Differential Revision: https://reviews.llvm.org/D66067

llvm-svn: 372058

5 years agoRevert "[lldb][NFC] Make ApplyObjcCastHack less scary"
Jim Ingham [Tue, 17 Sep 2019 00:44:48 +0000 (00:44 +0000)]
Revert "[lldb][NFC] Make ApplyObjcCastHack less scary"

This reverts commit 21641a2f6dbac22653befd03496e0850537882ff.

It was causing the following test failures:

lldb-Suite.lang/objc/objc-class-method.TestObjCClassMethod.py
lldb-Suite.lang/objc/foundation.TestObjCMethodsString.py
lldb-Suite.lang/objc/foundation.TestConstStrings.py
lldb-Suite.lang/objc/radar-9691614.TestObjCMethodReturningBOOL.py
lldb-Suite.lang/objc/foundation.TestObjCMethodsNSArray.py

llvm-svn: 372057

5 years ago[libFuzzer] Always print DSO map on Fuchsia libFuzzer launch
Jake Ehrlich [Tue, 17 Sep 2019 00:34:41 +0000 (00:34 +0000)]
[libFuzzer] Always print DSO map on Fuchsia libFuzzer launch

Fuchsia doesn't have /proc/id/maps, so it relies on the kernel logging system
to provide the DSO map to be able to symbolize in the context of ASLR. The DSO
map is logged automatically on Fuchsia when encountering a crash or writing to
the sanitizer log for the first time in a process. There are several cases
where libFuzzer doesn't encounter a crash, e.g. on timeouts, OOMs, and when
configured to print new PCs as they become covered, to name a few. Therefore,
this change always writes to the sanitizer log on startup to ensure the DSO map
is available in the log.

Author: aarongreen
Differential Revision: https://reviews.llvm.org/D66233

llvm-svn: 372056

5 years ago[OPENMP] Fix the test, NFC
Alexey Bataev [Tue, 17 Sep 2019 00:08:50 +0000 (00:08 +0000)]
[OPENMP] Fix the test, NFC

llvm-svn: 372055

5 years agollvm-reduce: Clean out previous test temp/output dir, since it was a dir and now...
David Blaikie [Mon, 16 Sep 2019 23:56:26 +0000 (23:56 +0000)]
llvm-reduce: Clean out previous test temp/output dir, since it was a dir and now it's used as just a single file

llvm-svn: 372054

5 years agollvm-reduce: Remove some string copies
David Blaikie [Mon, 16 Sep 2019 23:54:57 +0000 (23:54 +0000)]
llvm-reduce: Remove some string copies

llvm-svn: 372053

5 years ago[test] Fail gracefully if the regex doesn't match
Jonas Devlieghere [Mon, 16 Sep 2019 23:49:42 +0000 (23:49 +0000)]
[test] Fail gracefully if the regex doesn't match

This test is failing on the Fedora bot (staging). Rather than failing
with an IndexError, we should trigger an assert and dump the log when
the regex doesn't match.

llvm-svn: 372052

5 years agoRevert r372035: "[lit] Make internal diff work in pipelines"
Joel E. Denny [Mon, 16 Sep 2019 23:47:46 +0000 (23:47 +0000)]
Revert r372035: "[lit] Make internal diff work in pipelines"

This breaks a Windows bot.

llvm-svn: 372051

5 years ago[GlobalISel] Partially revert r371901.
Amara Emerson [Mon, 16 Sep 2019 23:46:03 +0000 (23:46 +0000)]
[GlobalISel] Partially revert r371901.

r371901 was overeager and widenScalarDst() and the like in the legalizer
attempt to increment the insert point given in order to add new instructions
after the currently legalizing inst. In cases where the insertion point is not
exactly the current instruction, then callers need to de-compensate for the
behaviour by decrementing the insertion iterator before calling them. It's not
a nice state of affairs, for now just undo the problematic parts of the change.

llvm-svn: 372050

5 years agollvm-reduce: Make tests shell-independent by passing the interpreter on the command...
David Blaikie [Mon, 16 Sep 2019 23:41:19 +0000 (23:41 +0000)]
llvm-reduce: Make tests shell-independent by passing the interpreter on the command line rather than using #! in the test file

llvm-svn: 372049

5 years agoAdd libc to path mappings in git-llvm.
David L. Jones [Mon, 16 Sep 2019 23:36:35 +0000 (23:36 +0000)]
Add libc to path mappings in git-llvm.

llvm-svn: 372048

5 years agoFix swig python package path
Haibo Huang [Mon, 16 Sep 2019 23:31:16 +0000 (23:31 +0000)]
Fix swig python package path

Summary:
The path defined in CMakeLists.txt doesn't match the path generated in
our python script. This change fixes that.

LLVM_LIBRARY_OUTPUT_INTDIR is defined as:

${CMAKE_BINARY_DIR}/${CMAKE_CFG_INTDIR}/lib${LLVM_LIBDIR_SUFFIX})

On the other hand, the path of site-package is generaged in
get_framework_python_dir_windows() in finishSwigPythonLLDB.py as:
(Dispite its name, the function is used for everything other than xcode)

prefix/cmakeBuildConfiguration/distutils.sysconfig.get_python_lib()

From lldb/CMakeLists.txt, we can see that:
prefix=${CMAKE_BINARY_DIR},
cmakeBuildConfiguration=${CMAKE_CFG_INTDIR}

And from python source code, we can see get_python_lib() always returns
lib/pythonx.y/site-packages for posix, or Lib/site-packages for windows:
https://github.com/python/cpython/blob/3.8/Lib/distutils/sysconfig.py#L128

We should make them match each other.

Subscribers: mgorny, lldb-commits

Tags: #lldb

Differential Revision: https://reviews.llvm.org/D67583

llvm-svn: 372047

5 years ago[Reproducer] Implement dumping packets.
Jonas Devlieghere [Mon, 16 Sep 2019 23:31:06 +0000 (23:31 +0000)]
[Reproducer] Implement dumping packets.

This patch completes the dump functionality by adding support for
dumping a reproducer's GDB remote packets.

Differential revision: https://reviews.llvm.org/D67636

llvm-svn: 372046

5 years agoFix warning: lambda capture 'temp_file_path' is not used
Jonas Devlieghere [Mon, 16 Sep 2019 22:55:49 +0000 (22:55 +0000)]
Fix warning: lambda capture 'temp_file_path' is not used

llvm-svn: 372044

5 years ago[PowerPC] Cust lower fpext v2f32 to v2f64 from extract_subvector v4f32
Nemanja Ivanovic [Mon, 16 Sep 2019 22:54:52 +0000 (22:54 +0000)]
[PowerPC] Cust lower fpext v2f32 to v2f64 from extract_subvector v4f32

Add the missing piece of r372029.
Somehow when the patch for review D61961 was committed, only the test case
went in and the code didn't. This of course caused all kinds of build bot
breaks.
This patch just adds the code for that patch.

Author: Lei Huang
Differential revision: https://reviews.llvm.org/D61961

llvm-svn: 372043