Commit Graph

224132 Commits

Author SHA1 Message Date
Dehao Chen 1012be120a Perform InstructioinCombiningPass before SampleProfile pass.
Summary: SampleProfile pass needs to be performed after InstructionCombiningPass, which helps eliminate un-inlinable function calls.

Reviewers: davidxl, dnovillo

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D17742

llvm-svn: 262419
2016-03-01 22:53:02 +00:00
Simon Pilgrim 2e6a9a290b Updated SSE41 builtin tests to more closely match the llvm fast-isel equivalent tests
llvm-svn: 262418
2016-03-01 22:38:33 +00:00
Kostya Serebryany 3d95dd9149 [libFuzzer] deprecate exit_on_first flag
llvm-svn: 262417
2016-03-01 22:33:14 +00:00
David Blaikie f72dbc101c llvm-dwp: Add missing copyright notice to llvm-dwp.cpp
Addressing feedback on IRC by Sean Silva.

llvm-svn: 262416
2016-03-01 22:29:00 +00:00
Kostya Serebryany 228d5b1ce4 [libFuzzer] add generic signal handlers so that libFuzzer can report at least something if ASan is not handlig the signals for us. Remove abort_on_timeout flag.
llvm-svn: 262415
2016-03-01 22:19:21 +00:00
John McCall 07daf721fb Mangle extended qualifiers in the proper order and mangle the
ARC ownership-convention function type modifications.

According to the Itanium ABI, vendor extended qualifiers are
supposed to be mangled in reverse-alphabetical order before
any CVR qualifiers.  The ARC function type conventions are
plausibly order-significant (they are associated with the
function type), which permits us to ignore the need to correctly
inter-order them with any other vendor qualifiers on the parameter
and return types.

Implementing these rules correctly is technically an ABI break.
Apple is comfortable with the risk of incompatibility here for
the ARC features, and I believe that address-space qualification
is still uncommon enough to allow us to adopt the conforming
rule without serious risk.  Still, targets which make heavy
use of address space qualification may want to revert to the
non-conforming order.

llvm-svn: 262414
2016-03-01 22:18:03 +00:00
Rui Ueyama 148945eb17 Fix --help message.
llvm-svn: 262413
2016-03-01 22:17:09 +00:00
Simon Pilgrim a3ad9cd793 [X86][SSE41] Added missing fast-isel intrinsics tests
Match IR generated in clang/test/CodeGen/sse41-builtins.c

llvm-svn: 262412
2016-03-01 22:05:05 +00:00
Colin LeMahieu 2d497a0078 [NFC] Convert tabs to spaces.
llvm-svn: 262411
2016-03-01 22:05:03 +00:00
Simon Pilgrim 9c03d1313c [X86][XOP] Regenerated intrinsics tests
llvm-svn: 262410
2016-03-01 21:58:50 +00:00
Daniel Sanders c1eb1e0419 Explicitly select IAS on one more RUN line in new embed-bitcode.c test.
This should fix clang-cmake-mips builder since MIPS does not have IAS enabled
by default (yet).

llvm-svn: 262409
2016-03-01 21:57:22 +00:00
Matthias Braun 37d884fdf4 AArch64: Reenable CompleteModel for A53, A57 and Kryo models
The fixes in r262393 completed them as well.

llvm-svn: 262408
2016-03-01 21:55:35 +00:00
Ted Woodward 7071c5fd64 Fix bug with function resolution when using IR Interpreter
Summary: Recent changes to the expression parser broke function name resolution when using the IR interpreter instead of JIT. This patch changes the IRMemoryMap ivar in InterpreterStackFrame to an IRExecutionUnitSP (which is a subclass), allowing InterpreterStackFrame::ResolveConstantValue() to call FindSymbol() on the name of the Value when it's a FunctionVal. It also changes IRExecutionUnit::FindInSymbols() to call GetFileAddress() on the symball if ResolveCallableAddress() fails and there is no valid Process.

Reviewers: spyffe

Subscribers: lldb-commits

Differential Revision: http://reviews.llvm.org/D17745

llvm-svn: 262407
2016-03-01 21:53:26 +00:00
Simon Pilgrim b1f5c62d5f [X86][AVX2] Regenerated 256-bit vector / 64-bit element permute tests
llvm-svn: 262406
2016-03-01 21:53:12 +00:00
Tim Northover d59b23a5ae Fix typo. NFC.
llvm-svn: 262405
2016-03-01 21:45:22 +00:00
Michael Kruse c7e0d9c216 Fix non-synthesizable loop exit values.
Polly recognizes affine loops that ScalarEvolution does not, in
particular those with loop conditions that depend on hoisted invariant
loads. Check for SCEVAddRec dependencies on such loops and do not
consider their exit values as synthesizable because SCEVExpander would
generate them as expressions that depend on the original induction
variables. These are not available in generated code.

llvm-svn: 262404
2016-03-01 21:44:06 +00:00
Simon Pilgrim 89244ba84a [X86][AVX2] Regenerated horizontal add/sub tests
llvm-svn: 262403
2016-03-01 21:43:55 +00:00
Daniel Jasper 56ef6aca2b [clang-format] Detect constructor initializers preceded by `noexcept`.
Patch by Erik Kessler, thank you.

llvm-svn: 262402
2016-03-01 21:41:58 +00:00
Simon Pilgrim 46e57fd073 [X86][AVX2] Regenerated intrinsics tests
llvm-svn: 262401
2016-03-01 21:38:41 +00:00
Colin LeMahieu 5cb6eea664 [Hexagon] Modifying r262258 to only be in effect in the hand assembler path, not the integrated assembler.
llvm-svn: 262400
2016-03-01 21:37:41 +00:00
Matthias Braun a939bd07d1 TableGen: Display helpfull message for incomplete models.
llvm-svn: 262399
2016-03-01 21:36:12 +00:00
Simon Pilgrim f2f5626b84 [X86][AVX] Fixed triple/arch clash in test case
We were specifying a x64 triple and then overriding with a x86 arch. 

llvm-svn: 262398
2016-03-01 21:33:08 +00:00
Matt Arsenault b36d462fac DAGCombiner: Turn truncate of a bitcasted vector to an extract
On AMDGPU where operations i64 operations are often bitcasted to v2i32
and back, this pattern shows up regularly where it breaks some
expected combines on i64, such as load width reducing.

This fixes some test failures in a future commit when i64 loads
are changed to promote.

llvm-svn: 262397
2016-03-01 21:31:53 +00:00
Rafael Espindola 889e45601e Add LLVMBuild for ObjectYAML.
Should fix the DBUILD_SHARED_LIBS bots.

llvm-svn: 262396
2016-03-01 21:29:33 +00:00
David Blaikie ddb27369a0 Revert "llvm-dwp: Keep ObjectFiles alive until object emission their contents can be referenced directly rather than copied"
Accidentally committed.

This reverts commit r262389.

llvm-svn: 262395
2016-03-01 21:24:04 +00:00
Jacques Pienaar ea9f25a740 [lanai] Add ELF enum value and relocations.
Add ELF enum value and relocations for Lanai backed.

General Lanai backend discussion on llvm-dev thread "[RFC] Lanai backend" (http://lists.llvm.org/pipermail/llvm-dev/2016-February/095118.html).

Differential Revision: http://reviews.llvm.org/D17008

llvm-svn: 262394
2016-03-01 21:21:42 +00:00
Matthias Braun a6cfb6f682 AArch64: Add missing schedinfo, check completeness for cyclone
This adds some missing generic schedule info definitions, enables
completeness checking for cyclone and fixes a typo uncovered by that.

Differential Revision: http://reviews.llvm.org/D17748

llvm-svn: 262393
2016-03-01 21:20:31 +00:00
Kit Barton e725669483 [Power9] Implement new vector compare, extract, insert instructions
This change implements the following vector operations:

  - Vector Compare Not Equal
    - vcmpneb(.) vcmpneh(.) vcmpnew(.)
    - vcmpnezb(.) vcmpnezh(.) vcmpnezw(.)
  - Vector Extract Unsigned
    - vextractub vextractuh vextractuw vextractd
    - vextublx vextubrx vextuhlx vextuhrx vextuwlx vextuwrx
  - Vector Insert
    - vinsertb vinserth vinsertw vinsertd

26 instructions.

Phabricator: http://reviews.llvm.org/D15916
llvm-svn: 262392
2016-03-01 20:51:57 +00:00
Sanjay Patel 18988ae66c [x86] use getBitcast()
This isn't quite NFC because some of the SDLocs may change which could
cause scheduling differences. But no regression tests are affected and
there is no functional change intended.

llvm-svn: 262391
2016-03-01 20:47:02 +00:00
David Blaikie bcb1c7e369 Fix some warnings a bit harder/different
This is an alternate fix to 262378 and a fix to a pessimizing-move
warning.

llvm-svn: 262390
2016-03-01 20:41:17 +00:00
David Blaikie 578f2cdd2d llvm-dwp: Keep ObjectFiles alive until object emission their contents can be referenced directly rather than copied
llvm-svn: 262389
2016-03-01 20:40:02 +00:00
Geoff Berry a0df341082 Revert "[AArch64] Fix isLegalAddImmediate() to return true for valid negative values."
Revert r262248 in an attempt to fix the clang-native-aarch64-full
    bot and to investigate a performance regression in
    SingleSource/Benchmarks/CoyoteBench/huffbench

llvm-svn: 262388
2016-03-01 20:28:52 +00:00
Vasileios Kalintiris 36901dd1c3 Revert "[mips] Promote the result of SETCC nodes to GPR width."
This reverts commit r262316.

It seems that my change breaks an out-of-tree chromium buildbot, so
I'm reverting this in order to investigate the situation further.

llvm-svn: 262387
2016-03-01 20:25:43 +00:00
Kit Barton 01cd2e7a4b New file to track implementation status of new POWER9 instructions
llvm-svn: 262386
2016-03-01 20:19:43 +00:00
Ekaterina Romanova c207006bbb This patch adds doxygen comments for the intrinsincs in the header file popcntintrin.h.
The doxygen comments are automatically generated based on Sony's intrinsics documentation.

Differential Revision: http://reviews.llvm.org/D17550 

llvm-svn: 262385
2016-03-01 20:04:57 +00:00
Matthias Braun 17cb57995e TableGen: Check scheduling models for completeness
TableGen checks at compiletime that for scheduling models with
"CompleteModel = 1" one of the following holds:

- Is marked with the hasNoSchedulingInfo flag
- The instruction is a subclass of Sched
- There are InstRW definitions in the scheduling model

Typical steps necessary to complete a model:

- Ensure all pseudo instructions that are expanded before machine
  scheduling (usually everything handled with EmitYYY() functions in
  XXXTargetLowering).
- If a CPU does not support some instructions mark the corresponding
  resource unsupported: "WriteRes<WriteXXX, []> { let Unsupported = 1; }".
- Add missing scheduling information.

Differential Revision: http://reviews.llvm.org/D17747

llvm-svn: 262384
2016-03-01 20:03:21 +00:00
Matthias Braun 8e0a734fc5 TableGen: Add hasNoSchedulingInfo to instructions
This introduces a new flag that indicates that a specific instruction
will never be present when the MachineScheduler runs and therefore needs
no scheduling information.

This is in preparation for an upcoming commit which checks completeness
of a scheduling model when tablegen runs.

Differential Revision: http://reviews.llvm.org/D17728

llvm-svn: 262383
2016-03-01 20:03:11 +00:00
Reid Kleckner 8f1b1f5921 Reword a misleading comment discussing landingpads and SEH
SEH doesn't use landingpads anymore.

llvm-svn: 262382
2016-03-01 19:51:48 +00:00
Justin Lebar 2daaa1ceca [NVPTX] Annotate param loads/stores as mayLoad/mayStore.
Summary:
Tablegen was unable to determine that param loads/stores were actually
reading or writing from memory.  I think this isn't a problem in
practice for param stores, because those occur in a block right before
we make our call.  But param loads don't have to at the very beginning
of a function, so should be annotated as mayLoad so we don't incorrectly
optimize them.

Reviewers: jholewinski

Subscribers: jholewinski, llvm-commits

Differential Revision: http://reviews.llvm.org/D17471

llvm-svn: 262381
2016-03-01 19:44:22 +00:00
Justin Lebar 536c8b7446 [NVPTX] Remove workaround for tablegen crash in NVPTXInstrInfo.td.
Summary: Looks like this was caused by a typo.

Reviewers: jholewinski

Subscribers: jholewinski, llvm-commits, tra

Differential Revision: http://reviews.llvm.org/D17357

llvm-svn: 262380
2016-03-01 19:44:20 +00:00
David Majnemer 25eb165f18 [MSVC Compat] Correctly handle finallys nested within finallys
We'd lose track of the parent CodeGenFunction, leading us to get
confused with regard to which function a nested finally belonged to.

Differential Revision: http://reviews.llvm.org/D17752

llvm-svn: 262379
2016-03-01 19:42:53 +00:00
Reid Kleckner d2da0f0cac Fix -Wnon-virtual-dtor warnings
llvm-svn: 262378
2016-03-01 19:39:54 +00:00
George Rimar 262b927d36 [ELF] - handle of command line options incompatible with -r
Patch handles the command line options
incompatible with relocatable output.

Differential revision: http://reviews.llvm.org/D17758

llvm-svn: 262377
2016-03-01 19:38:51 +00:00
Owen Anderson 7ea02fc787 Fix an issue where fast math flags were dropped during scalarization.
Most portions of InstCombine properly propagate fast math flags, but
apparently the vector scalarization section was overlooked.

llvm-svn: 262376
2016-03-01 19:35:52 +00:00
Michael Kruse b3a7935d54 [SCEVValidator] Remove redundant visit.
SCEVAddRecExpr::getStart() is synonymous to SCEVAddRecExpr::getOperand(0)
which will be visited in the following loop anyway.

llvm-svn: 262375
2016-03-01 19:30:54 +00:00
Sanjoy Das f1e9cae00e [SCEV] Minor cleanup: rename method, C++11'ify; NFC
llvm-svn: 262374
2016-03-01 19:28:01 +00:00
Justin Lebar b5ca00a58d [NVPTX] Use different, convergent MIs for convergent calls.
Summary:
Calls sometimes need to be convergent.  This is already handled at the
LLVM IR level, but it also needs to be handled at the MI level.

Ideally we'd propagate convergence from instructions, down through the
selection DAG, and into MIs.  But this is Hard, and would affect
optimizations in the SDNs -- right now only SDNs with two operands have
any flags at all.

Instead, here's a much simpler hack: Add new opcodes for NVPTX for
convergent calls, and generate these when lowering convergent LLVM
calls.

Reviewers: jholewinski

Subscribers: jholewinski, chandlerc, joker.eph, jhen, tra, llvm-commits

Differential Revision: http://reviews.llvm.org/D17423

llvm-svn: 262373
2016-03-01 19:24:03 +00:00
Justin Lebar 93e7a9b91c [NVPTX] Nix hack used to emit '{' and '}' for NVPTX calls.
Summary: Tablegen understands backslash as an escape char; that's sufficient.

Reviewers: jholewinski

Subscribers: llvm-commits, tra, jholewinski

Differential Revision: http://reviews.llvm.org/D17432

llvm-svn: 262372
2016-03-01 19:24:00 +00:00
Justin Lebar 877f5acc60 [NVPTX] Reformat NVPTXInstrInfo.td, and add additional comments.
Summary:
Also simplify some of the embedded C++ logic.

No functional changes.

Reviewers: jholewinski

Subscribers: llvm-commits, tra, jholewinski

Differential Revision: http://reviews.llvm.org/D17354

llvm-svn: 262371
2016-03-01 19:23:30 +00:00
David Majnemer 791b88b6da [X86] Elide references to _chkstk for dynamic allocas
The _chkstk function is called by the compiler to probe the stack in an
order consistent with Windows' expectations.  However, it is possible to
elide the call to _chkstk and manually adjust the stack pointer if we
can prove that the allocation is fixed size and smaller than the probe
size.

This shrinks chrome.dll, chrome_child.dll and chrome.exe by a
cummulative ~133 KB.

Differential Revision: http://reviews.llvm.org/D17679

llvm-svn: 262370
2016-03-01 19:20:23 +00:00