Commit Graph

133178 Commits

Author SHA1 Message Date
James Molloy 53298a1808 [ARM] Shrink post-indexed LDR and STR to LDM/STM
A Thumb-2 post-indexed LDR instruction such as:

  ldr.w r0, [r1], #4

Can be rewritten as:

  ldm.n r1!, {r0}

LDMs can be more expensive than LDRs on some cores, so this has been enabled only in minsize mode.

llvm-svn: 272002
2016-06-07 12:13:34 +00:00
James Molloy 75afc95112 [ARM] Transform LDMs into writeback form to save code size
If we have an LDM that uses only low registers and doesn't write to its base register:

  ldm.w r0, {r1, r2, r3}

And that base register is dead after the LDM, then we can convert it to writeback form and use a narrow encoding:

  ldm.n r0!, {r1, r2, r3}

Obviously, this introduces a new register write and so can cause WAW hazards, so I've enabled it only in minsize mode. This is a code size trick that ARM Compiler 5 ("armcc") does that we don't.

llvm-svn: 272000
2016-06-07 11:47:24 +00:00
George Rimar cd36e182d2 [llvm-readobj] - Teach llvm-readobj to dump .gnu.version_r sections
SHT_GNU_verneed (.gnu.version_r) is a version dependency section.
It was the last symbol versioning relative section that was not dumped,
now it is.

Differential revision: http://reviews.llvm.org/D21024

llvm-svn: 271998
2016-06-07 11:04:49 +00:00
Peter Smith 353a2286e2 [ARM] Incorrect relocation type for Thumb2 B<cond>.w
The Thumb2 conditional branch B<cond>.W has a different encoding (T3) 
to the unconditional branch B.W (T4) as it needs to record <cond>. 
As the encoding is different the B<cond>.W is given a different 
relocation type. 

ELF for the ARM Architecture 4.6.1.6 (Table-13) states that 
R_ARM_THM_JUMP19 should be used for B<cond>.W. At present the 
MC layer is using the R_ARM_THM_JUMP24 from B.W.

This change makes B<cond>.W use R_ARM_THM_JUMP19 and alters the 
existing test that checks for R_ARM_THM_JUMP24 to expect 
R_ARM_THM_JUMP19.

llvm-svn: 271997
2016-06-07 10:34:33 +00:00
Simon Pilgrim db9893fb90 [InstCombine][AVX2] Add support for simplifying AVX2 per-element shifts to native shifts
Unlike native shifts, the AVX2 per-element shift instructions VPSRAV/VPSRLV/VPSLLV handle out of range shift values (logical shifts set the result to zero, arithmetic shifts splat the sign bit).

If the shift amount is constant we can sometimes convert these instructions to native shifts:

1 - if all shift amounts are in range then the conversion is trivial.
2 - out of range arithmetic shifts can be clamped to the (bitwidth - 1) (a legal shift amount) before conversion.
3 - logical shifts just return zero if all elements have out of range shift amounts.

In addition, UNDEF shift amounts are handled - either as an UNDEF shift amount in a native shift or as an UNDEF in the logical 'all out of range' zero constant special case for logical shifts.

Differential Revision: http://reviews.llvm.org/D19675

llvm-svn: 271996
2016-06-07 10:27:15 +00:00
Simon Pilgrim 91e3ac8293 [InstCombine][SSE] Add MOVMSK constant folding (PR27982)
This patch adds support for folding undef/zero/constant inputs to MOVMSK instructions.

The SSE/AVX versions can be fully folded, but the MMX version can only handle undef inputs.

Differential Revision: http://reviews.llvm.org/D20998

llvm-svn: 271990
2016-06-07 08:18:35 +00:00
Craig Topper 2f90c1fedf [AVX512] Allow avx2 and sse41 nontemporal load intrinsics to select EVEX encoded instructions when VLX is enabled.
llvm-svn: 271988
2016-06-07 07:27:57 +00:00
Craig Topper e1cac15feb [AVX512] Remove unnecessary mayLoad, mayStore, hasSidEffects flags from instructions that have patterns that imply them. Add the same set of flags to instructions that don't have patterns to imply them.
llvm-svn: 271987
2016-06-07 07:27:54 +00:00
Craig Topper 0fcf925699 [AVX512] Add NoVLX to a couple patterns that have VLX equivalents. Ordering of the patterns in the .td file protects this, but its better to be explicit.
llvm-svn: 271986
2016-06-07 07:27:51 +00:00
Lang Hames 9dc125b004 [Kaleidoscope] Update Chapter 3 of the "Implementing a Language" tutorial to
take into account modernizations in r246002 and r270381.

Patch based on http://reviews.llvm.org/D20954 by Miroslav Hrncir.
Thanks Miroslav!

llvm-svn: 271985
2016-06-07 05:40:08 +00:00
Zachary Turner 92d9e978eb [pdb] Fix broken unit tests after r271982.
llvm-svn: 271983
2016-06-07 05:32:48 +00:00
Zachary Turner d8447990b0 [pdb] Use MappedBlockStream to parse the PDB directory.
In order to efficiently write PDBs, we need to be able to make a
StreamWriter class similar to a StreamReader, which can transparently deal
with writing to discontiguous streams, and we need to use this for all
writing, similar to how we use StreamReader for all reading.

Most discontiguous streams are the typical numbered streams that appear in
a PDB file and are described by the directory, but the exception to this,
that until now has been parsed by hand, is the directory itself.
MappedBlockStream works by querying the directory to find out which blocks
a stream occupies and various other things, so naturally the same logic
could not possibly work to describe the blocks that the directory itself
resided on.

To solve this, I've introduced an abstraction IPDBStreamData, which allows
the client to query for the list of blocks occupied by the stream, as well
as the stream length. I provide two implementations of this: one which
queries the directory (for indexed streams), and one which queries the
super block (for the directory stream).

This has the side benefit of vastly simplifying the code to parse the
directory. Whereas before a mini state machine was rolled by hand, now we
simply use FixedStreamArray to read out the stream sizes, then build a
vector of FixedStreamArrays for the stream map, all in just a few lines of
code.

Reviewed By: ruiu
Differential Revision: http://reviews.llvm.org/D21046

llvm-svn: 271982
2016-06-07 05:28:55 +00:00
Dan Liew d15c106c9b [LibFuzzer] s/dataflow sanitizer/DataflowSanitizer/
llvm-svn: 271980
2016-06-07 04:44:49 +00:00
Dan Liew 0d09f14554 [LibFuzzer] Disable building and running LSan tests on Apple platforms because LSan is not currently supported.
Differential Revision: http://reviews.llvm.org/D20947

llvm-svn: 271979
2016-06-07 04:44:39 +00:00
Saleem Abdulrasool 532dcbc2c5 ARM: correct TLS access on WoA
TLS access requires an offset from the TLS index.  The index itself is the
section-relative distance of the symbol.  For ARM, the relevant relocation
(IMAGE_REL_ARM_SECREL) is applied as a constant.  This means that the value may
not be an immediate and must be lowered into a constant pool.  This offset will
not be base relocated.  We were previously emitting the actual address of the
symbol which would be base relocated and would therefore be the vaue offset by
the ImageBase + TLS Offset.

llvm-svn: 271974
2016-06-07 03:15:07 +00:00
Saleem Abdulrasool ce4eee4951 ARM: clang-format a couple of switches, add comments
clang-format a couple of switches in preparation for a future change.  Add some
enumeration comments

llvm-svn: 271973
2016-06-07 03:15:01 +00:00
Saleem Abdulrasool 0cd0cb992a ARM: normalise space in the patterns
Just adjust the whitespace for the selection patterns.  NFC.

llvm-svn: 271972
2016-06-07 03:14:57 +00:00
Kostya Serebryany 712fc9803a [sanitizer] Initial implementation of a Hardened Allocator
Summary:
This is an initial implementation of a Hardened Allocator based on Sanitizer Common's CombinedAllocator.
It aims at mitigating heap based vulnerabilities by adding several features to the base allocator, while staying relatively fast.
The following were implemented:
- additional consistency checks on the allocation function parameters and on the heap chunks;
- use of checksum protected chunk header, to detect corruption;
- randomness to the allocator base;
- delayed freelist (quarantine), to mitigate use after free and overall determinism.
Additional mitigations are in the works.

Reviewers: eugenis, aizatsky, pcc, krasin, vitalybuka, glider, dvyukov, kcc

Subscribers: kubabrecka, filcab, llvm-commits

Differential Revision: http://reviews.llvm.org/D20084

llvm-svn: 271968
2016-06-07 01:20:26 +00:00
Rui Ueyama 4a1ebae537 Add comments.
llvm-svn: 271967
2016-06-07 00:59:04 +00:00
Reid Kleckner 6f3406df67 Re-land "[codeview] Emit information about global variables"
This reverts commit r271962 and reinstantes r271957.

MSVC's linker doesn't appear to like it if you have an empty symbol
substream, so only open a symbol substream if we're going to emit
something about globals into it.

Makes check-asan pass.

llvm-svn: 271965
2016-06-07 00:02:03 +00:00
Reid Kleckner 4ece163c92 Try one more time to pacify -Wpessimizing-move, MSVC, libstdc++4.7, and the world without a named variable
llvm-svn: 271964
2016-06-06 23:46:14 +00:00
Vedant Kumar 8d0e861e9b Revert "Retry^2 "[llvm-profdata] Add option to ingest filepaths from a file""
This reverts commit r271953. It's still breaking on Windows, though the
list initialization issue is fixed:

http://bb.pgr.jp/builders/ninja-clang-i686-msc19-R/builds/3751

llvm-svn: 271963
2016-06-06 23:43:56 +00:00
Reid Kleckner e8a236fc2e Revert "[codeview] Emit information about global variables"
This reverts commit r271957, it broke check-asan on Windows.

llvm-svn: 271962
2016-06-06 23:41:38 +00:00
Michael Kuperstein a0c6ae02a5 [InstCombine] scalarizePHI should not assume the code it sees has been CSE'd
scalarizePHI only looked for phis that have exactly two uses - the "latch"
use, and an extract. Unfortunately, we can not assume all equivalent extracts
are CSE'd, since InstCombine itself may create an extract which is a duplicate
of an existing one. This extends it to handle several distinct extracts from
the same index.

This should fix at least some of the  performance regressions from PR27988.

Differential Revision: http://reviews.llvm.org/D20983

llvm-svn: 271961
2016-06-06 23:38:33 +00:00
Rui Ueyama 4bc848047b Fix CRLF -> LF.
llvm-svn: 271960
2016-06-06 23:35:52 +00:00
Reid Kleckner 52a155fca3 Attempt to work around lack of std::map::emplace in libstdc++4.7
llvm-svn: 271958
2016-06-06 23:28:03 +00:00
Reid Kleckner 87eddf723d [codeview] Emit information about global variables
This currently emits everything as S_GDATA32, which isn't right for
things like thread locals, but it's a start.

llvm-svn: 271957
2016-06-06 23:23:47 +00:00
Peter Collingbourne bb738170a9 Verifier: Simplify and fix issue where we were not verifying unmaterialized functions.
Arrange to call verify(Function &) on each function, followed by
verify(Module &), whether the verifier is being used from the pass or
from verifyModule(). As a side effect, this fixes an issue that caused
us not to call verify(Function &) on unmaterialized functions from
verifyModule().

Differential Revision: http://reviews.llvm.org/D21042

llvm-svn: 271956
2016-06-06 23:21:27 +00:00
Rui Ueyama ba0aab94cc [pdbdump] Verify the size of TPI hash records.
llvm-svn: 271954
2016-06-06 23:19:23 +00:00
Vedant Kumar f051269a7f Retry^2 "[llvm-profdata] Add option to ingest filepaths from a file"
Changes since the initial commit:
- Normalize file paths read from the file to prevent Windows path
  separators from escaping parts of the path.
- Since we need to store the normalized file paths in WeightedFile,
  don't do tricky things to keep the source MemoryBuffer alive.
- Don't use list-initialization for a std::string in WeightedFile.

Differential Revision: http://reviews.llvm.org/D20980

llvm-svn: 271953
2016-06-06 23:17:22 +00:00
Vedant Kumar 87886425bd Revert "Retry "[llvm-profdata] Add option to ingest filepaths from a file"
This reverts commit r271949. It breaks the Windows build:

http://lab.llvm.org:8011/builders/clang-x64-ninja-win7/builds/12796

llvm-svn: 271952
2016-06-06 23:01:42 +00:00
Vedant Kumar d8ee75b8f5 Retry "[llvm-profdata] Add option to ingest filepaths from a file"
Changes since the initial commit:
- Normalize file paths read from the file to prevent Windows path
  separators from escaping parts of the path.
- Since we need to store the normalized file paths in WeightedFile,
  don't do tricky things to keep the source MemoryBuffer alive.

Differential Revision: http://reviews.llvm.org/D20980

llvm-svn: 271949
2016-06-06 22:39:22 +00:00
Peter Collingbourne 71bb79430f Verifier: Remove dead code.
Remove previously unreachable code that verifies that a function definition has
an entry block. By definition, a function definition has at least one block.

llvm-svn: 271948
2016-06-06 22:32:52 +00:00
Chris Bieneman 71ef6178e1 Updating release notes for CMake version bump
CMake 3.4.3 is now required for building LLVM-based projects.

llvm-svn: 271945
2016-06-06 22:02:16 +00:00
Rui Ueyama 2c5384ae4c [pdbdump] Print section header flags.
llvm-svn: 271943
2016-06-06 21:34:55 +00:00
Chris Bieneman 468208a1de [yaml2obj] Get rid of MachO header union
This is based on post-commit feedback from Sean Silva.

llvm-svn: 271942
2016-06-06 21:18:43 +00:00
Zachary Turner 25e8b05d03 [llvm-pdbdump] Dump stream sizes and stream blocks to yaml.
llvm-svn: 271940
2016-06-06 20:37:17 +00:00
Zachary Turner 7120a478fa [llvm-pdbdump] Dump MSF headers to YAML.
This is the simplest possible patch to get some kind of YAML
output.  All it dumps is the MSF header fields so that in
theory an empty MSF file could be reconstructed.

Reviewed By: ruiu, majnemer
Differential Revision: http://reviews.llvm.org/D20971

llvm-svn: 271939
2016-06-06 20:37:05 +00:00
Dan Liew 5914407185 [LibFuzzer] Provide stub implementation of __sanitizer_cov_trace_pc_indir
Calls to this function are currently injected by the
``SanitizerCoverageModule`` pass when the both the ``indirect-calls``
and ``trace-pc`` sanitizer coverage options are enabled and the code
being instrumented has indirect calls. Previously because LibFuzzer did
not define this function this would lead to link errors when building
some of the tests on OSX.

Differential Revision: http://reviews.llvm.org/D20946

llvm-svn: 271938
2016-06-06 20:27:09 +00:00
Matt Arsenault 02458c2d27 AMDGPU: Add function for getting instruction size
llvm-svn: 271936
2016-06-06 20:10:33 +00:00
Matt Arsenault 3b2e2a59e8 AMDGPU: Fix constantexpr addrspacecasts
If we had a constant group address space cast the queue pointer
wasn't enabled for the function, resulting in a crash on noreg
later.

llvm-svn: 271935
2016-06-06 20:03:31 +00:00
Davide Italiano fea0a4c5b2 [PM] Preserve the correct set of analyses for GVN.
llvm-svn: 271934
2016-06-06 20:01:50 +00:00
Davide Italiano 82c447823b [GVN] Switch dump() definition over to LLVM_DUMP_METHOD.
llvm-svn: 271932
2016-06-06 19:24:27 +00:00
Michael Zolotukhin 19edbadfc5 [LoopUnrollAnalyzer] Fix a crash in analyzeLoopUnrollCost.
In some cases, when simplifying with SCEV, we might consider pointer values as
just usual integer values.  Thus, we might get a different type from what we
had originally in the map of simplified values, and hence we need to check
types before operating on the values.

This fixes PR28015.

llvm-svn: 271931
2016-06-06 19:21:40 +00:00
Haicheng Wu 9ed77af89d Fix a test case. NFC.
llvm-svn: 271930
2016-06-06 19:11:53 +00:00
Geoff Berry 43e5160d0e Reapply [LSR] Create fewer redundant instructions.
Summary:
Fix LSRInstance::HoistInsertPosition() to check the original insert
position block first for a canonical insertion point that is dominated
by all inputs.  This leads to SCEV being able to reuse more instructions
since it currently tracks the instructions it creates for reuse by
keeping a table of <Value, insert point> pairs.

Originally reviewed in http://reviews.llvm.org/D18001

Reviewers: atrick

Subscribers: llvm-commits, mzolotukhin, mcrosier

Differential Revision: http://reviews.llvm.org/D18480

llvm-svn: 271929
2016-06-06 19:10:46 +00:00
Rui Ueyama ef2b488482 [pdbdump] Print out New FPO stream contents.
The data strucutre in the new FPO stream is described in the
PE/COFF spec. There is one record per function if frame pointer
is omitted.

Differential Revision: http://reviews.llvm.org/D20999

llvm-svn: 271926
2016-06-06 18:39:21 +00:00
Haicheng Wu 77ea344786 [MBP] Reduce code size by running tail merging in MBP.
The code layout that TailMerging (inside BranchFolding) works on is not the
final layout optimized based on the branch probability. Generally, after
BlockPlacement, many new merging opportunities emerge.

This patch calls Tail Merging after MBP and calls MBP again if Tail Merging
merges anything.

Differential Revision: http://reviews.llvm.org/D20276

llvm-svn: 271925
2016-06-06 18:36:07 +00:00
Lang Hames e038aae25a [Kaleidoscope][BuildingAJIT] Fix hyphenation in chapter 2 title.
llvm-svn: 271924
2016-06-06 18:35:44 +00:00
Haicheng Wu 664b561f01 [BranchFolding] Replace MachineBlockFrequencyInfo with MBFIWrapper. NFC.
Differential Revision: http://reviews.llvm.org/D20184

llvm-svn: 271923
2016-06-06 18:35:41 +00:00
Alina Sbirlea 76c4a855bc [cpu-detection] Substantial refactor of Host CPU detection code (x86)
Summary:
Following D20970 (committed as r271726).
This is a substantial refactoring of the host CPU detection code.

There is no functionality change intended, but the changes are extensive.

Definitions of architecture types and subtypes are by no means exhaustive or
perfectly defined, but a fair starting point.
Suggestions for futher improvements are welcome.

Reviewers: llvm-commits

Differential Revision: http://reviews.llvm.org/D20988

llvm-svn: 271921
2016-06-06 18:29:59 +00:00
Lang Hames d29ee53628 [Kaleidoscope][BuildingAJIT] More cleanup of Chapter 2.
Streamline some wording, fix a bug in the markup for the layer interface table.

llvm-svn: 271917
2016-06-06 18:22:47 +00:00
Lang Hames 706db2e609 [Kaleidoscope][BuildingAJIT] Clean up sentence, remove comments from code block.
llvm-svn: 271913
2016-06-06 18:07:23 +00:00
Sanjay Patel 6a333c3ed9 [InstCombine] limit icmp transform to ConstantInt (PR28011)
In r271810 ( http://reviews.llvm.org/rL271810 ), I loosened the check
above this to work for any Constant rather than ConstantInt. AFAICT, 
that part makes sense if we can determine that the shrunken/extended 
constant remained equal. But it doesn't make sense for this later 
transform where we assume that the constant DID change. 

This could assert for a ConstantExpr:
https://llvm.org/bugs/show_bug.cgi?id=28011

And it could be wrong for a vector as shown in the added regression test.

llvm-svn: 271908
2016-06-06 16:56:57 +00:00
Sanjay Patel 027c469158 regenerate checks
llvm-svn: 271904
2016-06-06 16:03:06 +00:00
Sanjay Patel 70aa568c4e regenerate checks
llvm-svn: 271903
2016-06-06 15:55:00 +00:00
Artem Tamazov 135487767b [AMDGPU][llvm-mc] v_cndmask_b32: src2 is mandatory; do not enforce VOP2 when src2 == VCC.
Another step for unification llvm assembler/disassembler with sp3.
Besides, CodeGen output is a bit improved, thus changes in CodeGen tests.
Assembler/Disassembler tests updated/added.

Differential Revision: http://reviews.llvm.org/D20796

llvm-svn: 271900
2016-06-06 15:23:43 +00:00
Matthew Simpson e3e3b994ae [LAA] Use load and store vectors (NFC)
Contributed-by: Aditya Kumar <hiraditya@msn.com>
Differential Revision: http://reviews.llvm.org/D20953

llvm-svn: 271895
2016-06-06 14:15:41 +00:00
Igor Breger edafb0595e [KNL] Fix UMULO lowering.
Differential Revision: http://reviews.llvm.org/D21013

llvm-svn: 271891
2016-06-06 12:24:52 +00:00
Benjamin Kramer f21beb2c74 Remove dead function with incredibly broken assert.
Found by clang-tidy's misc-assert-side-effect.

llvm-svn: 271887
2016-06-06 12:10:42 +00:00
Filipe Cabecinhas 6e7d5467c0 [NFC] Silence gcc warning (-Wsign-compare)
llvm-svn: 271882
2016-06-06 10:49:56 +00:00
Craig Topper 33350cc406 [AVX512] Remove masked palignr intrinsics and auto-upgrade them to native IR of vector shuffle and select.
llvm-svn: 271872
2016-06-06 06:12:54 +00:00
NAKAMURA Takumi 500b0a49d2 LLVM_BUILD_32_BITS: Add -m32 with CMAKE_C*_FLAGS. [CMP0056]
With CMP0056, try_compile() uses also CMAKE_EXE_LINKER_FLAG.
It caused mismatch between CMAKE_CXX_FLAGS and CMAKE_EXE_LINKER_FLAGS, to fail to examine CXX_SUPPORTS_CXX11 with -m32.

FYI, before this, try_compile() tries without -m32 regardless of LLVM_BUILD_32_BITS.

llvm-svn: 271871
2016-06-06 05:54:55 +00:00
Craig Topper 143446d5c1 [AVX512] Add PALIGNR shuffle lowering for v32i16 and v16i32.
llvm-svn: 271870
2016-06-06 05:39:10 +00:00
Craig Topper ccad6d57c1 [AVX512] Update tests to show shuffle decoding for vpshuflw/vpshufhw.
llvm-svn: 271869
2016-06-06 05:39:07 +00:00
Lang Hames 3242f65ea9 [Kaleidoscope][BuildingAJIT] Split up the code-block describing the substitution
of OptimizeLayer for CompileLayer in Chapter 2.

Hopefully this will read a little more clearly.

llvm-svn: 271868
2016-06-06 05:07:52 +00:00
Lang Hames 38eb0312ba [Kaleidoscope][BuildingAJIT] Fix code-blocks in Chapter 2.
llvm-svn: 271867
2016-06-06 04:53:59 +00:00
Lang Hames c499d2a7c2 [Kaleidoscope][BuildingAJIT] Add tutorial text for Chapter 2.
This chapter discusses IR optimizations, the ORC IRTransformLayer, and the ORC
layer concept itself.

The text is still pretty rough, but I think the main ideas are there. Feedback
is very welcome, as always.

llvm-svn: 271865
2016-06-06 03:28:12 +00:00
Nick Lewycky bfad015559 Fix spelling and capitalization in comments. NFC
llvm-svn: 271862
2016-06-06 01:51:23 +00:00
NAKAMURA Takumi f5fe079e03 Trailing whitespace.
llvm-svn: 271861
2016-06-06 00:31:45 +00:00
NAKAMURA Takumi 9ff93f9421 Untabify.
llvm-svn: 271860
2016-06-06 00:31:28 +00:00
Eli Friedman ee89505799 LICM: Don't sink stores out of loops that may throw.
Summary:
This hasn't been caught before because it requires noalias or similarly
strong alias analysis to actually reproduce.

Fixes http://llvm.org/PR27952 .

Reviewers: hfinkel, sanjoy

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D20944

llvm-svn: 271858
2016-06-05 22:13:52 +00:00
Sanjoy Das b7e861a488 Add safety check to InstCombiner::commonIRemTransforms
Since FoldOpIntoPhi speculates the binary operation to potentially each
of the predecessors of the PHI node (pulling it out of arbitrary control
dependence in the process), we can FoldOpIntoPhi only if we know the
operation doesn't have UB.

This also brings up an interesting profitability question -- the way it
is written today, commonIRemTransforms will hoist out work from
dynamically dead code into code that will execute at runtime.  Perhaps
that isn't the best canonicalization?

Fixes PR27968.

llvm-svn: 271857
2016-06-05 21:17:04 +00:00
Sanjoy Das 0dcd1d859c Add test case for InstCombiner::commonIRemTransforms; NFC
The PHI case in commonIRemTransforms was untested; add a trivial test
case.

llvm-svn: 271856
2016-06-05 21:17:00 +00:00
Eli Friedman aa77fa0036 Fix deadlock in ThreadPool unittest.
(Yes, this only deadlocks on a computer with a single core; I'm using
a virtual machine.)

llvm-svn: 271855
2016-06-05 21:15:46 +00:00
Davide Italiano a1cbc3f8cc [Internalize] Test that __stack_chk_{guard, fail} are not internalized.
r154645 introduced this feature without test. This should have better
coverage now.

llvm-svn: 271853
2016-06-05 19:08:54 +00:00
Filipe Cabecinhas 6328f8e9e6 [BitCode] Make sure atomicrmw's argument is an actual PointerType
llvm-svn: 271851
2016-06-05 18:43:40 +00:00
Filipe Cabecinhas 036e73c8bf [BitCode] Make sure storeatomic's argument is an actual PointerType
llvm-svn: 271850
2016-06-05 18:43:33 +00:00
Filipe Cabecinhas fc2a3c98e9 [BitCode] Diagnose GEPs with no indices
llvm-svn: 271849
2016-06-05 18:43:26 +00:00
Filipe Cabecinhas 2849b48fea [BitCode] Don't allow constants of void type.
llvm-svn: 271848
2016-06-05 18:43:17 +00:00
Mike Spertus 3e45d77835 Add Visual Studio Visualizer for ArrayRef
Modeled after visualization of llvm::SmallVec

llvm-svn: 271847
2016-06-05 18:34:02 +00:00
Sanjoy Das 4d4339d1e8 [PM] Port IndVarSimplify to the new pass manager
Summary:
There are some rough corners, since the new pass manager doesn't have
(as far as I can tell) LoopSimplify and LCSSA, so I've updated the
tests to run them separately in the old pass manager in the lit tests.
We also don't have an equivalent for AU.setPreservesCFG() in the new
pass manager, so I've left a FIXME.

Reviewers: bogner, chandlerc, davide

Subscribers: sanjoy, mcrosier, llvm-commits

Differential Revision: http://reviews.llvm.org/D20783

llvm-svn: 271846
2016-06-05 18:01:19 +00:00
Sanjoy Das f90e28d6fd [IndVars] Remove -liv-reduce
It is an off-by-default option that no one seems to use[0], and given
that SCEV directly understands the overflow instrinsics there is no real
need for it anymore.

[0]: http://lists.llvm.org/pipermail/llvm-dev/2016-April/098181.html

llvm-svn: 271845
2016-06-05 18:01:12 +00:00
Sanjay Patel 0fab306eb5 fix checks
update_test_checks.py got confused matching the variable names. 

llvm-svn: 271844
2016-06-05 17:54:56 +00:00
Sanjay Patel a6fbc82392 [InstCombine] allow vector icmp bool transforms
llvm-svn: 271843
2016-06-05 17:49:45 +00:00
Sanjay Patel 54d7010627 add tests to show missing vector transforms
llvm-svn: 271842
2016-06-05 17:32:58 +00:00
Sanjay Patel 51dc83c052 regenerate checks
llvm-svn: 271841
2016-06-05 17:29:45 +00:00
Sanjay Patel 009c3da65f update test to use FileCheck
llvm-svn: 271840
2016-06-05 17:13:09 +00:00
Sanjay Patel 5f0217f42e fix documentation comments and other clean-ups; NFC
llvm-svn: 271839
2016-06-05 16:46:18 +00:00
Sanjay Patel f48b909f28 update test to use FileCheck
llvm-svn: 271838
2016-06-05 16:41:20 +00:00
Sanjay Patel 8a3b6d0d8b update test to FileCheck
llvm-svn: 271837
2016-06-05 16:29:15 +00:00
Simon Pilgrim 64c6de4525 [X86][XOP] Added VPERMIL2PD/VPERMIL2PS raw mask decoding for target shuffle combines
llvm-svn: 271834
2016-06-05 15:21:30 +00:00
Simon Pilgrim 478295dadd [X86][XOP] Added VPERMIL2PD/VPERMIL2PS as a target shuffle type
llvm-svn: 271831
2016-06-05 15:01:45 +00:00
Simon Pilgrim 163987a235 [X86][XOP] Tidied up DecodeVPERMIL2PMask to more closely match DecodeVPERMILPMask.
llvm-svn: 271830
2016-06-05 14:33:43 +00:00
Craig Topper 8eeda57a40 [AVX512] Add support for lowering PALIGNR for v64i8.
Could do this for other types to, but this is what's needed to replace the instrinsic with native IR in clang.

llvm-svn: 271828
2016-06-05 06:29:12 +00:00
Craig Topper 5a315d4613 [AVX512] Split command lines and regenerate a test to prepare for a future commit.
llvm-svn: 271827
2016-06-05 06:29:08 +00:00
Craig Topper 9f51c9ef15 [AVX512] Fix PANDN combining for v4i32/v8i32 when VLX is enabled.
v4i32/v8i32 ANDs aren't promoted to v2i64/v4i64 when VLX is enabled.

llvm-svn: 271826
2016-06-05 05:35:11 +00:00
Xinliang David Li 41f4a05245 Fix wrong comment in header /NFC
llvm-svn: 271825
2016-06-05 05:18:20 +00:00
Xinliang David Li e537cbf1a3 Fix pass declaration order /NFC
llvm-svn: 271824
2016-06-05 05:15:45 +00:00
Xinliang David Li 64dbb295b6 [PM] Port GCOVProfiler pass to the new pass manager
llvm-svn: 271823
2016-06-05 05:12:23 +00:00
Xinliang David Li fb3137c3b3 [PM] code refactoring /NFC
llvm-svn: 271822
2016-06-05 03:40:03 +00:00
Chandler Carruth 806faaef8b Disable the use of std::call_once on PowerPC due to an apparent bug in
libstdc++ (or in compilers, or somewhere, I can't track it down) that
causes unittests that use INITIALIZE_PASS to crash.

The analysis I've been able to do is that inside libstdc++'s
implementation of std::call_once, it uses pthread_once, and when that
returns an error code it throws std::system_error which then eventually
calls std::terminate.

Hopefully some of the folks who work on PPC can try to sort out what's
going on here. Until then, they'll have to use the fallback
implementation.

llvm-svn: 271821
2016-06-05 02:46:01 +00:00
NAKAMURA Takumi 379e7efd94 Threading.h: Update \param. [-Wdocumentation]
llvm-svn: 271819
2016-06-05 00:15:44 +00:00
Sanjay Patel 6f8f47b358 [InstCombine] less 'CI' confusion; NFC
Change the name of the ICmpInst to 'ICmp' and the Constant (was a ConstantInt) to 'C',
so that it's hopefully clearer that 'CI' refers to CastInst in this context.

While we're scrubbing, fix the documentation comment and use 'auto' with 'dyn_cast'.

llvm-svn: 271817
2016-06-05 00:12:32 +00:00
David Majnemer 2482e1c017 [SimplifyCFG] Don't kill empty cleanuppads with multiple uses
A basic block could contain:
  %cp = cleanuppad []
  cleanupret from %cp unwind to caller

This basic block is empty and is thus a candidate for removal.  However,
there can be other uses of %cp outside of this basic block.  This is
only possible in unreachable blocks.

Make our transform more correct by checking that the pad has a single
user before removing the BB.

This fixes PR28005.

llvm-svn: 271816
2016-06-04 23:50:03 +00:00
David Majnemer 36b7b08d4f [DebugInfo, PDB] Use sparse bitfields for the name map
The name map might not be densely packed on disk.  Using a sparse map
will save memory in such situations.

llvm-svn: 271811
2016-06-04 22:47:39 +00:00
Sanjay Patel ea8a211169 [InstCombine] allow vector constants for cast+icmp fold
This is step 1 of unknown towards fixing PR28001:
https://llvm.org/bugs/show_bug.cgi?id=28001

llvm-svn: 271810
2016-06-04 22:04:05 +00:00
Simon Pilgrim 2ead861d07 [X86][XOP] Added VPERMIL2PD/VPERMIL2PS shuffle mask comment decoding
llvm-svn: 271809
2016-06-04 21:44:28 +00:00
Sanjay Patel 8e63999bee [InstCombine] add test for missing vector optimization
llvm-svn: 271808
2016-06-04 21:41:25 +00:00
Sanjay Patel c774f8c265 clean-up; NFC
llvm-svn: 271807
2016-06-04 21:20:44 +00:00
Sanjay Patel 4c42211c6f [InstCombine] add test for missing vector optimization
llvm-svn: 271806
2016-06-04 21:20:03 +00:00
Sanjay Patel 58a92a327d [InstCombine] minimize test case and use FileCheck
llvm-svn: 271805
2016-06-04 21:04:59 +00:00
Sanjay Patel 4c204230fc fix formatting, punctuation; NFC
llvm-svn: 271804
2016-06-04 20:39:22 +00:00
Simon Pilgrim ba319ded5e [Analysis] Enabled BITREVERSE as a vectorizable intrinsic
Allows XOP to vectorize BITREVERSE - other targets will follow as their costmodels improve.

llvm-svn: 271803
2016-06-04 20:21:07 +00:00
Craig Topper e609bd6600 [X86] Add the VR128L/H and VR256L/H to the list of vector register classes for inline asm constraints. Also fix the comment on the function.
llvm-svn: 271802
2016-06-04 20:15:08 +00:00
Chandler Carruth fe1ffb912d [LPM] Reinstate r271781 which reinstated r271652 to replace the
CALL_ONCE_... macro in the legacy pass manager with the new
llvm::call_once facility.

Nothing changed sicne the last attempt in r271781 which I reverted in
r271788. At least one of the failures I saw was spurious, and I want to
make sure the other failures are real before I work around them -- they
appeared to only effect ppc64le and ppc64be.

Original commit message of r271781:
----
[LPM] Reinstate r271652 to replace the CALL_ONCE_... macro in the legacy
pass manager with the new llvm::call_once facility.

This reverts commit r271657 and re-applies r271652 with a fix to
actually work with arguments. In the original version, we just ended up
directly calling std::call_once via ADL because of the std::once_flag
argument. The llvm::call_once never worked with arguments. Now,
llvm::call_once is a variadic template that perfectly forwards
everything. As a part of this it had to move to the header and we use
a generic functor rather than an explict function pointer. It would be
nice to use std::invoke here but we don't have it yet. That means
pointer to members won't work here, but that seems a tolerable
compromise.

I've also tested this by forcing the fallback path, so hopefully it
sticks this time.
----

Original commit message of r271652:
----
[LPM] Replace the CALL_ONCE_... macro in the legacy pass manager with
the new llvm::call_once facility.

This facility matches the standard APIs and when the platform supports
it actually directly uses the standard provided functionality. This is
both more efficient on some platforms and much more TSan friendly.

The only remaining user of the cas_flag and home-rolled atomics is the
fallback implementation of call_once. I have a patch that removes them
entirely, but it needs a Windows patch to land first.

This alone substantially cleans up the macros for the legacy pass
manager, and should subsume some of the work Mehdi was doing to clear
the path for TSan testing of ThinLTO, a really important step to have
reliable upstream testing of ThinLTO in all forms.
----

llvm-svn: 271800
2016-06-04 19:57:55 +00:00
Saleem Abdulrasool 1fcdc23a6e X86: enable TLS on Windows itanium
Windows itanium is nearly identical to windows-msvc (MS ABI for C, itanium for
C++).  Enable the TLS support for the target similar to the MSVC model.

llvm-svn: 271797
2016-06-04 18:27:22 +00:00
Simon Pilgrim fd2eda4f64 [X86][AVX2] Fix v16i16 SHL lowering (PR27730)
The AVX2 v16i16 shift lowering works by unpacking to 2 x v8i32, performing the shift and then truncating the result.

The unpacking is used to place the values in the upper 16-bits so that we can correctly sign-extend for SRA shifts. Unfortunately we weren't ensuring that the lower 16-bits were zero to ensure that SHL correctly shifts in zero bits.

llvm-svn: 271796
2016-06-04 16:45:33 +00:00
David Majnemer 8c46a4ceea [AsmPrinter, CodeView] There are some more ways of getting wchar_t
C++ has a builtin type called wchar_t.  Clang also provides a type
called __wchar_t in C mode.

In C mode, wchar_t can be a typedef to unsigned short.

llvm-svn: 271793
2016-06-04 15:40:33 +00:00
David Majnemer 862a8ae812 [CodeView] Fix a busted assert in TypeTableBuilder::writeClass
It was checking for Union when it should have checked for Interface.

llvm-svn: 271792
2016-06-04 15:40:31 +00:00
David Majnemer 9a59584211 [CodeView] Validate the vftable offset
llvm-svn: 271791
2016-06-04 15:40:29 +00:00
David Majnemer 067e3d0cc5 [TypeStreamMerger] visitUnknownMember was supposed to be visitUnknownType
llvm-svn: 271790
2016-06-04 15:40:27 +00:00
Simon Pilgrim fda22d66fc [InstCombine][MMX] Extend SimplifyDemandedUseBits MOVMSK support to MMX
Add the MMX implementation to the SimplifyDemandedUseBits SSE/AVX MOVMSK support added in D19614

Requires a minor tweak as llvm.x86.mmx.pmovmskb takes a x86_mmx argument - so we have to be explicit about the implied v8i8 vector type.

llvm-svn: 271789
2016-06-04 13:42:46 +00:00
Chandler Carruth fe9466fe2c [LPM] Revert r271781 which was a re-commit of r271652.
There appears to be a strange exception thrown and crash using call_once
on a PPC build bot, and a *really* weird windows link error for
GCMetadata.obj. Still need to investigate the cause of both problems.

Original change summary:
[LPM] Reinstate r271652 to replace the CALL_ONCE_... macro in the legacy
pass manager with the new llvm::call_once facility.

llvm-svn: 271788
2016-06-04 09:36:40 +00:00
Chandler Carruth fa4890e068 [LPM] Reinstate r271652 to replace the CALL_ONCE_... macro in the legacy
pass manager with the new llvm::call_once facility.

This reverts commit r271657 and re-applies r271652 with a fix to
actually work with arguments. In the original version, we just ended up
directly calling std::call_once via ADL because of the std::once_flag
argument. The llvm::call_once never worked with arguments. Now,
llvm::call_once is a variadic template that perfectly forwards
everything. As a part of this it had to move to the header and we use
a generic functor rather than an explict function pointer. It would be
nice to use std::invoke here but we don't have it yet. That means
pointer to members won't work here, but that seems a tolerable
compromise.

I've also tested this by forcing the fallback path, so hopefully it
sticks this time.

Original commit message:
----
[LPM] Replace the CALL_ONCE_... macro in the legacy pass manager with
the new llvm::call_once facility.

This facility matches the standard APIs and when the platform supports
it actually directly uses the standard provided functionality. This is
both more efficient on some platforms and much more TSan friendly.

The only remaining user of the cas_flag and home-rolled atomics is the
fallback implementation of call_once. I have a patch that removes them
entirely, but it needs a Windows patch to land first.

This alone substantially cleans up the macros for the legacy pass
manager, and should subsume some of the work Mehdi was doing to clear
the path for TSan testing of ThinLTO, a really important step to have
reliable upstream testing of ThinLTO in all forms.

llvm-svn: 271781
2016-06-04 07:25:44 +00:00
Craig Topper 6ae375c9ba [X86] Use smaller types to shrink the intrinsic lowering tables by about 12K.
llvm-svn: 271776
2016-06-04 04:32:17 +00:00
Craig Topper 5250634334 [X86] Use X86ISD::ABS for lowering pabs SSSE3/AVX intrinsics to match AVX512. Should allow those intrinsics to use the EVEX encoded instructions and get the extra registers when available.
llvm-svn: 271775
2016-06-04 04:32:15 +00:00
Petr Hosek 17f7a24807 [MC] Check the upper bound in truncate assertion
The truncateToSize function already has assertion to check the
lower boundary for the number bytes, but it does not check the
upper boundary which could still lead to usage errors.

Differential Revision: http://reviews.llvm.org/D20755

llvm-svn: 271773
2016-06-04 04:02:18 +00:00
Taewook Oh 99497fdebd Revert commit r271704, a patch that enables warnings for non-portable #include and #import paths (Corresponding clang patch has been reverted by r271761). Patches are reverted because they generate lots of unadressable warnings for windows and fail tests under ASAN.
llvm-svn: 271764
2016-06-04 03:36:12 +00:00
Chandler Carruth 57cd7ef1c9 [sancov] Revert r271695 which broke all of the PPC bots.
Original commit message:
[sancov] Run sancov tests on more platforms

The only tests that need to be run on Linux are the ones that use C++
demangling. I'm assuming they will fail on Mac, since __cxa_demangle
there won't handle the non-double-underscore prefixed mangled names.

llvm-svn: 271763
2016-06-04 03:28:27 +00:00
Chandler Carruth 0c30f89cca [llvm-profdata] Revert r271709 and the 3 subsequent commits - the code
and/or tests aren't working on Windows currently.

There seems to be some problem with quoting the file paths. I don't
understand the test structure here or the code well enough to try to
come up with a way to correctly handle paths with back slashes in them,
and this has caused the Windows builds to be failing for 7 hours now, so
I'm reverting the whole thing to bring them back to life. Sorry for the
disruption, but a couple of these were bug fixes anyways that can be
folded into a fresh commit.

Reverts the following patches:

r271756: Clean up the way we create the input filenames buffer (NFC)
r271748: Fix use-after-free from discarded MemoryBuffer (NFC)
r271710: Fix option description (NFC)
r271709: Add option to ingest filepaths from a file
llvm-svn: 271760
2016-06-04 03:08:01 +00:00
Devin Coughlin 7cb374882d STLExtras: Add convenience is_contained() function.
This commit adds a convenience is_contained() function
which checks if an element exists in a container. It is part of a larger
series of patches adding an MPI checker to the clang static analyzer.

Reviewers: dblaikie,bkramer

A patch by Alexander Droste!

Differential Revision:http://reviews.llvm.org/D16053

llvm-svn: 271757
2016-06-04 00:49:46 +00:00
Vedant Kumar f771a050db [llvm-profdata] Clean up the way we create the input filenames buffer (NFC)
Create the buffer before calling parseInputFilenamesFile(), and add a
comment explaining why this is done.

Thanks to David Li for the suggestion!

llvm-svn: 271756
2016-06-04 00:36:28 +00:00
Bruno Cardoso Lopes 42b1f65f03 [LockFileManager] Improve error output by using better error messages
This is currently used by clang to lock access to modules; improve the
error message so that clang can use better output messages from locking
error issues.

rdar://problem/26529101

Differential Review: http://reviews.llvm.org/D20942

llvm-svn: 271755
2016-06-04 00:34:00 +00:00
Adrian Prantl 2905d54af3 Testcase cleanup: Remove a redundant test input.
llvm-svn: 271753
2016-06-04 00:10:17 +00:00
Matthias Braun c25c9ccbcb MIR: Support MachineMemOperands without associated value
This is allowed (though used rarely) and useful to keep your tests
short.

llvm-svn: 271752
2016-06-04 00:06:31 +00:00
Xinliang David Li ff2873742e Replace hard coded probability threshold with parameter /NFC
llvm-svn: 271751
2016-06-03 23:48:36 +00:00
Vedant Kumar 8a3d717868 [llvm-profdata] Fix use-after-free from discarded MemoryBuffer (NFC)
Thanks to Justin Bogner for pointing this out!

Caught by ASAN.

llvm-svn: 271748
2016-06-03 23:12:38 +00:00
Xinliang David Li 6c44e9e33d [pgo] extend r271532 to darwin platform
llvm-svn: 271746
2016-06-03 23:02:28 +00:00
Easwaran Raman 019e0bf592 Reapply r271728 after adding move cobstructor for ProfileSummaryInfo
llvm-svn: 271745
2016-06-03 22:54:26 +00:00
Derek Bruening 9ef5772154 [esan|wset] Optionally assume intra-cache-line accesses
Summary:
Adds an option -esan-assume-intra-cache-line which causes esan to assume
that a single memory access touches just one cache line, even if it is not
aligned, for better performance at a potential accuracy cost.  Experiments
show that the performance difference can be 2x or more, and accuracy loss
is typically negligible, so we turn this on by default.  This currently
applies just to the working set tool.

Reviewers: aizatsky

Subscribers: vitalybuka, zhaoqin, kcc, eugenis, llvm-commits

Differential Revision: http://reviews.llvm.org/D20978

llvm-svn: 271743
2016-06-03 22:29:52 +00:00
Mehdi Amini 448dd8c5c6 Add a FIXME note in the release notes about documenting ThinLTO
llvm-svn: 271742
2016-06-03 21:45:34 +00:00
Mike Aizatsky 70fd3e412a [libfuzzer] hiding custom mutator handling in MutationDispatcher.
Summary: Refactoring, no functional changes.

Differential Revision: http://reviews.llvm.org/D20975

llvm-svn: 271740
2016-06-03 21:34:29 +00:00
Easwaran Raman 94edaaaefb Revert r271728 as it breaks Windows build
llvm-svn: 271738
2016-06-03 21:14:26 +00:00
Rui Ueyama fd97bf1f76 pdbdump: print out TPI hashes.
Differential Revision: http://reviews.llvm.org/D20945

llvm-svn: 271736
2016-06-03 20:48:51 +00:00
Justin Bogner 07bf5349ee Re-apply "SDAG: Update ChainNodesMatched as nodes are deleted"
My first attempt at this had an overly aggressive assert - chain nodes
will only be removed, but we could hit the assert if a non-chain node
was CSE'd (NodeToMatch, for instance).

This reapplies r271706 by reverting r271713 and fixing an assert.

Original message:

Avoid relying on UB by looking into deleted nodes for a marker value.
Instead, update the list of chain nodes as we go.

llvm-svn: 271733
2016-06-03 20:47:40 +00:00
Teresa Johnson 4cafbdd4bd [ThinLTO/gold] Pass import lists by reference (NFC)
In my earlier patch r271690 I missed passing a map by reference in one
place.

llvm-svn: 271732
2016-06-03 20:46:32 +00:00
Easwaran Raman d142050f3a Analysis pass to access profile summary info
Differential Revision: http://reviews.llvm.org/D20648

llvm-svn: 271728
2016-06-03 20:37:19 +00:00
Reid Kleckner ab1dfaae06 Fix non-Windows build when inserting a move only type into a map
llvm-svn: 271727
2016-06-03 20:29:51 +00:00
Alina Sbirlea 400eb02775 [cpu-detection] Naming convention
Summary:
    Follow-up to D20926 (committed as r271595, r271596).
    This patch is in preparation for a substantial refactoring of the code.

    No functionality changed.

Differential Revision: http://reviews.llvm.org/D20970

llvm-svn: 271726
2016-06-03 20:27:50 +00:00
Reid Kleckner f27f3f8491 [Symbolize] Check if the PE file has a PDB and emit an error if we can't load it
Summary:
Previously we would try to load PDBs for every PE executable we tried to
symbolize. If that failed, we would fall back to DWARF. If there wasn't
any DWARF, we'd print mostly useless symbol information using the export
table.

With this change, we only try to load PDBs for executables that claim to
have them. If that fails, we can now print an error rather than falling
back silently. This should make it a lot easier to diagnose and fix
common symbolization issues, such as not having DIA or not having a PDB.

Reviewers: zturner, eugenis

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D20982

llvm-svn: 271725
2016-06-03 20:25:09 +00:00
Chad Rosier 9faa5bcf13 [AArch64] Move tests from r271677 to a more appropriately named file. NFC.
llvm-svn: 271718
2016-06-03 20:11:09 +00:00
Chad Rosier be879ea751 [AArch64] Spot SBFX-compatible code expressed with sign_extend.
This is very similar to r271677, but for extracts from i32 with the SIGN_EXTEND
acting on a arithmetic shift.

llvm-svn: 271717
2016-06-03 20:05:49 +00:00
Derek Bruening 4252a16c35 [esan] Specify which tool via a global variable
Summary:
Adds a global variable to specify the tool, to support handling early
interceptors that invoke instrumented code and require shadow memory to be
initialized prior to __esan_init() being invoked.

Reviewers: aizatsky

Subscribers: vitalybuka, zhaoqin, kcc, eugenis, llvm-commits

Differential Revision: http://reviews.llvm.org/D20973

llvm-svn: 271715
2016-06-03 19:40:37 +00:00
Justin Bogner 737c136176 Revert "SDAG: Update ChainNodesMatched as nodes are deleted"
Seeing failures in CodeGen/Generic/icmp-illegal.ll on quite a few
bots.

This reverts r271706.

llvm-svn: 271713
2016-06-03 19:40:06 +00:00
Zachary Turner d311739a5f [llvm-pdbdump] Introduce an abstraction for the output style.
This opens the door to introducing a YAML outputter which can be
used for machine consumption.  Currently the yaml output style
is unimplemented and returns an error if you try to use it.

Reviewed By: rnk, ruiu
Differential Revision: http://reviews.llvm.org/D20967

llvm-svn: 271712
2016-06-03 19:28:33 +00:00
Alina Sbirlea 8c9ad10a7f Test commit. Removes some spaces. No functionality changed.
Summary:
Test commit. Removes some spaces.
No functionality changed.

Reviewers: llvm-commits

Differential Revision: http://reviews.llvm.org/D20972

llvm-svn: 271711
2016-06-03 19:20:37 +00:00
Vedant Kumar d45a277fea [llvm-profdata] Fix option description (NFC)
llvm-svn: 271710
2016-06-03 19:10:25 +00:00
Vedant Kumar 5c276d0e5d [llvm-profdata] Add option to ingest filepaths from a file
Differential Revision: http://reviews.llvm.org/D20980

llvm-svn: 271709
2016-06-03 19:05:20 +00:00
Justin Bogner 6f6d012e32 SDAG: Update ChainNodesMatched as nodes are deleted
Avoid relying on UB by looking into deleted nodes for a marker value.
Instead, update the list of chain nodes as we go.

llvm-svn: 271706
2016-06-03 18:50:11 +00:00
Taewook Oh dfec58e80c In openFileForRead, attempt to fetch the actual name of the file on disk -- including case -- so that clang can later warn about non-portable #include and #import directives.
Differential Revision: http://reviews.llvm.org/D19842

Patch by Eric Niebler

llvm-svn: 271704
2016-06-03 18:38:39 +00:00
Derek Schuff 5859a9ed80 [WebAssembly] Emit type signatures for declared functions
Under emscripten, C code can take the address of a function implemented
in Javascript (which is exposed via an import in wasm). Because imports
do not have linear memory address in wasm, we need to generate a thunk
to be the target of the indirect call; it call the import directly.

To make this possible, LLVM needs to emit the type signatures for these
functions, because they may not be called directly or referred to other
than where the address is taken.

This uses s new .s directive (.functype) which specifies the signature.

Differential Revision: http://reviews.llvm.org/D20891

Re-apply r271599 but instead of bailing with an error when a declared
function has multiple returns, replace it with a pointer argument. Also
add the test case I forgot to 'git add' last time around.

llvm-svn: 271703
2016-06-03 18:34:36 +00:00
Justin Bogner 1785503dd3 SDAG: Replace some unreachable code with an assert. NFC
The current node shouldn't be (and isn't) removed partway through
selection.

llvm-svn: 271699
2016-06-03 18:09:53 +00:00
Reid Kleckner 98df480c8e [sancov] Disable these tests if there is no X86 backend
Copied from test/CodeGen/X86

llvm-svn: 271698
2016-06-03 18:07:32 +00:00
Mike Aizatsky 3c1782f23b [libfuzzer] splitting fuzzer.test
llvm-svn: 271697
2016-06-03 18:05:22 +00:00
Reid Kleckner e2bef1f143 [sancov] Run sancov tests on more platforms
The only tests that need to be run on Linux are the ones that use C++
demangling. I'm assuming they will fail on Mac, since __cxa_demangle
there won't handle the non-double-underscore prefixed mangled names.

llvm-svn: 271695
2016-06-03 17:51:42 +00:00
Ben Craig eff9278a7e Doxygen for FoldingSet::reserve and FoldingSet::capacity
llvm-svn: 271694
2016-06-03 17:50:14 +00:00
Teresa Johnson 56c68353fa [ThinLTO/gold] Pass down the imports lists from the thin link (NFC)
No longer need to compute the imports in each backend thread.

llvm-svn: 271690
2016-06-03 17:15:54 +00:00
Chris Bieneman 4c423773d8 [yaml2obj] Sort MachO LinkEdit write operations based on offset
This re-applies r271611, and hopefully the bots won't break this time.

Although ld64 always outputs linkedit data in the same order, it isn't actually required to. This change makes yaml2obj resilient if the offsets are in arbitrary order.

llvm-svn: 271687
2016-06-03 16:58:05 +00:00
Vedant Kumar 7f5200162b Trim some spaces (NFC)
llvm-svn: 271681
2016-06-03 15:58:59 +00:00
Reid Kleckner a8d5740757 [codeview] Add basic record type translation
This only translates data members for now. Translating overloaded
methods is complicated, so I stopped short of doing that.

Reviewers: aaboud

Differential Revision: http://reviews.llvm.org/D20924

llvm-svn: 271680
2016-06-03 15:58:20 +00:00
Sjoerd Meijer 9bc93f6298 Code size optimisation: do not inline memcpy if this expansion results
in more instructions than the libary call.

Differential Revision: http://reviews.llvm.org/D20958

llvm-svn: 271678
2016-06-03 15:38:55 +00:00
Chad Rosier 2d658703e1 [AArch64] Spot SBFX-compatbile code expressed with sign_extend_inreg.
We were assuming all SBFX-like operations would have the shl/asr form, but often
when the field being extracted is an i8 or i16, we end up with a
SIGN_EXTEND_INREG acting on a shift instead.

This is a port of r213754 from ARM to AArch64.

llvm-svn: 271677
2016-06-03 15:00:09 +00:00
Sanjay Patel 6cf18af1c5 [InstCombine] look through bitcasts to find selects
There was concern that creating bitcasts for the simpler potential select pattern:

define <2 x i64> @vecBitcastOp1(<4 x i1> %cmp, <2 x i64> %a) {
  %a2 = add <2 x i64> %a, %a
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %bc = bitcast <4 x i32> %sext to <2 x i64>
  %and = and <2 x i64> %a2, %bc
  ret <2 x i64> %and
}

might lead to worse code for some targets, so this patch is matching the larger
patterns seen in the test cases.

The motivating example for this patch is this IR produced via SSE intrinsics in C:

define <2 x i64> @gibson(<2 x i64> %a, <2 x i64> %b) {
  %t0 = bitcast <2 x i64> %a to <4 x i32>
  %t1 = bitcast <2 x i64> %b to <4 x i32>
  %cmp = icmp sgt <4 x i32> %t0, %t1
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %t2 = bitcast <4 x i32> %sext to <2 x i64>
  %and = and <2 x i64> %t2, %a
  %neg = xor <4 x i32> %sext, <i32 -1, i32 -1, i32 -1, i32 -1>
  %neg2 = bitcast <4 x i32> %neg to <2 x i64>
  %and2 = and <2 x i64> %neg2, %b
  %or = or <2 x i64> %and, %and2
  ret <2 x i64> %or
}

For an AVX target, this is currently:

vpcmpgtd  %xmm1, %xmm0, %xmm2
vpand     %xmm0, %xmm2, %xmm0
vpandn    %xmm1, %xmm2, %xmm1
vpor      %xmm1, %xmm0, %xmm0
retq

With this patch, it becomes:

vpmaxsd   %xmm1, %xmm0, %xmm0

Differential Revision: http://reviews.llvm.org/D20774

llvm-svn: 271676
2016-06-03 14:42:07 +00:00
Artem Tamazov f88397c84c [test/AMDGPU] Square-braced-syntax for registers: add macro test/example.
Test added as per discussion in http://reviews.llvm.org/D20588.
The macro is just a demonstration, useless in practice.
Coding style fixes.

Differential Revision: http://reviews.llvm.org/D20797

llvm-svn: 271675
2016-06-03 14:41:17 +00:00
Zachary Turner 71f725735c [pdb] Add string table offsets to check output.
llvm-svn: 271674
2016-06-03 14:22:46 +00:00
Simon Pilgrim ff35eecd90 [X86][AVX512] Fixed 512-bit vector nontemporal load alignment
llvm-svn: 271673
2016-06-03 14:12:43 +00:00
Sjoerd Meijer d906bf1369 RAS extensions are part of ARMv8.2-A. This change enables them by introducing a
new instruction to ARM and AArch64 targets and several system registers.

Patch by: Roger Ferrer Ibanez and Oliver Stannard

Differential Revision: http://reviews.llvm.org/D20282

llvm-svn: 271670
2016-06-03 14:03:27 +00:00
Ben Craig 60adb9229c Adding reserve and capacity methods to FoldingSet
http://reviews.llvm.org/D20930

llvm-svn: 271669
2016-06-03 13:54:48 +00:00
Simon Pilgrim f92d175a78 [X86][AVX512] Added 512-bit vector nontemporal load tests
llvm-svn: 271668
2016-06-03 13:42:49 +00:00
Sjoerd Meijer 9da258d8e5 ARM target does not use printAliasInstr machinery which
forces having special checks in ArmInstPrinter::printInstruction. This
patch addresses this issue.

Not all special checks could be removed: either they involve elaborated
conditions under which the alias is emitted (e.g. ldm/stm on sp may be
pop/push but only if the number of registers is >= 2) or the number
of registers is multivalued (like happens again with ldm/stm) and they
do not match the InstAlias pattern which assumes single-valued operands
in the pattern.

Patch by: Roger Ferrer Ibanez

Differential Revision: http://reviews.llvm.org/D20237

llvm-svn: 271667
2016-06-03 13:19:43 +00:00
Sjoerd Meijer 3c2f785769 AsmWriterEmitter.cpp assumes that all operands of a printed alias
will appear after a blank. This assumption does not hold in the ARM
target.

Patch by: Roger Ferrer Ibanez

Differential Revision: http://reviews.llvm.org/D20234

llvm-svn: 271666
2016-06-03 13:17:37 +00:00
Sjoerd Meijer 84e2f69245 Currently AsmWriterEmiter.cpp (used by tblgen -gen-asm-writer) does not
consider the Predicates attached to InstAlias when generating printAliasInstr.
This forces users of printAliasInstr to check those predicates beforehand.

This commit adds them in the condition set of the IAPrinter object.

Patch by: Roger Ferrer Ibanez

Differential Revision: http://reviews.llvm.org/D20233

llvm-svn: 271665
2016-06-03 13:14:19 +00:00
Sam Kolton a4a99ad1bc [AMDGPU] Assembler: More tests for SDWA instructions. Fix for SDWA float modifiers.
Summary: Depends on D20625

Reviewers: tstellarAMD, vpykhtin, artem.tamazov

Subscribers: arsenm, kzhuravl

Differential Revision: http://reviews.llvm.org/D20674

llvm-svn: 271662
2016-06-03 11:43:09 +00:00
Simon Pilgrim a6022c9a63 [X86][SSE] Added nontemporal load tests
These currently all lower to regular loads, generic nontemporal load support will be added in a future patch

llvm-svn: 271659
2016-06-03 11:00:55 +00:00
Daniel Sanders 43750eab82 [mips] EABI CodeGen is completely untested and seems to have bitrotted. Remove it.
Summary:
There are no tests*, no EABI buildbots, and simple test cases do not work.

* There is a single MIPS16 test using a mips*-gnueabi triple but this test
  doesn't test EABI and the triple doesn't cause EABI to be used.

Reviewers: sdardis

Subscribers: tberghammer, danalbert, srhines, dsanders, sdardis, llvm-commits

Differential Revision: http://reviews.llvm.org/D20906

llvm-svn: 271658
2016-06-03 10:38:09 +00:00
Benjamin Kramer cdaf62684e Revert "[LPM] Replace the CALL_ONCE_... macro in the legacy pass manager with the new llvm::call_once facility."
Fallback path doesn't compile. This reverts commit r271652.

llvm-svn: 271657
2016-06-03 10:36:10 +00:00
Simon Pilgrim 960ca812ed [X86] Added nontemporal scalar store tests
llvm-svn: 271656
2016-06-03 10:30:54 +00:00
Sam Kolton 05ef1c940f [AMDGPU] Assembler: Custom converters for SDWA instructions. Support for _dpp and _sdwa suffixes in mnemonics.
Summary:
Added custom converters for SDWA instruction to support optional operands and modifiers.
Support for _dpp and _sdwa suffixes that allows to force DPP or SDWA encoding for instructions.

Reviewers: tstellarAMD, vpykhtin, artem.tamazov

Subscribers: arsenm, kzhuravl

Differential Revision: http://reviews.llvm.org/D20625

llvm-svn: 271655
2016-06-03 10:27:37 +00:00
Simon Pilgrim 02284541b2 [X86][SSE] Regenerated nontemporal vector store tests and added extra target types
llvm-svn: 271654
2016-06-03 10:24:24 +00:00
Daniel Sanders 7b09493bff [mips] Remove CPU-only triples from llvm-objdump commands.
Summary: They aren't necessary since llvm-objdump can auto-detect the architecture.

Reviewers: sdardis

Subscribers: jfb, dsanders, llvm-commits, sdardis

Differential Revision: http://reviews.llvm.org/D20904

llvm-svn: 271653
2016-06-03 10:22:22 +00:00
Chandler Carruth 0bc1515a03 [LPM] Replace the CALL_ONCE_... macro in the legacy pass manager with
the new llvm::call_once facility.

This facility matches the standard APIs and when the platform supports
it actually directly uses the standard provided functionality. This is
both more efficient on some platforms and much more TSan friendly.

The only remaining user of the cas_flag and home-rolled atomics is the
fallback implementation of call_once. I have a patch that removes them
entirely, but it needs a Windows patch to land first.

This alone substantially cleans up the macros for the legacy pass
manager, and should subsume some of the work Mehdi was doing to clear
the path for TSan testing of ThinLTO, a really important step to have
reliable upstream testing of ThinLTO in all forms.

llvm-svn: 271652
2016-06-03 10:20:02 +00:00
Simon Pilgrim 38b4661b1b [X86] Regenerated nontemporal store tests and added tests for all 128-bit vector types
llvm-svn: 271651
2016-06-03 10:15:36 +00:00
Chandler Carruth 9ac86efd4d Remove bogus initialization of the PPC and Hexagon SelectionDAGISel
subclasses. These are not passes proper. We don't support registering
them, they can't be constructed with default arguments, and the ID is
actually in a base class.

Only these two targets even had any boiler plate to try to do this, and
it had to be munged out of the INITIALIZE_PASS macros to work. What's
worse, the boiler plate has rotted and the "name" of the pass is
actually the description string now!!! =/ All of this is completely
unnecessary. No other target bothers, and nothing breaks if you don't
initialize them because CodeGen has an entirely separate initialization
path that is somewhat more durable than relying on the implicit
initialization the way the 'opt' tool does for registered passes.

llvm-svn: 271650
2016-06-03 10:13:31 +00:00
Chandler Carruth d474144a7a Use the standard INITIALIZE_PASS macro rather than hand rolling a (not
entirely correct) version of its contents.

llvm-svn: 271649
2016-06-03 10:13:29 +00:00
Chandler Carruth 990111e4d8 Reformat the macros in Passes.h with clang-format before editting them.
No functionality changed here.

llvm-svn: 271648
2016-06-03 10:13:26 +00:00
Simon Pilgrim 205f65f62f [X86][AVX2] Relaxed alignment on nontemporal store tests
llvm-svn: 271646
2016-06-03 10:06:59 +00:00
Simon Pilgrim 8ea8940677 [X86][AVX2] Regenerated nontemporal store tests and added tests for all 256-bit vector types
llvm-svn: 271645
2016-06-03 09:56:24 +00:00
Daniel Sanders 6ba3dd6b71 [mips] Implement 'la' macro in PIC mode for O32.
Summary:
N32 support will follow in a later patch since the symbol version of 'la'
incorrectly believes N32 to have 64-bit pointers and rejects it early.

This fixes the three incorrectly expanded 'la' macros found in bionic.

Reviewers: sdardis

Subscribers: dsanders, llvm-commits, sdardis

Differential Revision: http://reviews.llvm.org/D20820

llvm-svn: 271644
2016-06-03 09:53:06 +00:00
Chandler Carruth 345d225da2 Run clang-format over PassSupport.h, mostly to get the macros all
formatted fancily.

I'm working on rewriting these macros to use the new call_once stuff,
but really want to have clang-format work on the edits, so just
re-baselining the entire file here. No changes other than clang-format.

llvm-svn: 271635
2016-06-03 08:46:16 +00:00