oota-llvm.git
7 years agoScalarEvolution.cpp: Appease g++-4.7. He missed implicit "this" in lambda.
NAKAMURA Takumi [Thu, 5 Mar 2015 01:02:45 +0000 (01:02 +0000)]
ScalarEvolution.cpp: Appease g++-4.7. He missed implicit "this" in lambda.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231331 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove the conditional addition of the execution dependency fixing
Eric Christopher [Thu, 5 Mar 2015 00:28:55 +0000 (00:28 +0000)]
Remove the conditional addition of the execution dependency fixing
pass from the ARM backend as the pass itself will detect any use
of the appropriate register class.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231324 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoCleanup and remove a chunk of getARMSubtarget calls in the
Eric Christopher [Thu, 5 Mar 2015 00:23:40 +0000 (00:23 +0000)]
Cleanup and remove a chunk of getARMSubtarget calls in the
ARM TargetMachine pass pipeline construction by pushing them down
into the appropriate pass.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231323 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTurn off .debug_pubnames/pubtypes for PS4.
Paul Robinson [Thu, 5 Mar 2015 00:08:27 +0000 (00:08 +0000)]
Turn off .debug_pubnames/pubtypes for PS4.

Differential Revision: http://reviews.llvm.org/D8067

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231322 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoInitializer lists are supported in MSVC 2013. Since that's our minimum required versi...
Aaron Ballman [Wed, 4 Mar 2015 23:17:31 +0000 (23:17 +0000)]
Initializer lists are supported in MSVC 2013. Since that's our minimum required version, we can move that to the list of acceptable C++11 features.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231313 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Support] Increase timeout for the LockFileManager back to 5 mins.
Argyrios Kyrtzidis [Wed, 4 Mar 2015 22:54:38 +0000 (22:54 +0000)]
[Support] Increase timeout for the LockFileManager back to 5 mins.

Waiting for just 1 min may not be enough for some contexts.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231309 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoImprove test robustness
Matthias Braun [Wed, 4 Mar 2015 22:31:18 +0000 (22:31 +0000)]
Improve test robustness

Improve test robustness in preparation of coming commits:
- Avoid undefs which may get propagated too much.
- Remove several pointless add 0, instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231307 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[IndVarSimplify] use the "canonical" way to infer no-wrap.
Sanjoy Das [Wed, 4 Mar 2015 22:24:23 +0000 (22:24 +0000)]
[IndVarSimplify] use the "canonical" way to infer no-wrap.

Summary:
rL225282 introduced an ad-hoc way to promote some additions to nuw or
nsw.  Since then SCEV has become smarter in directly proving no-wrap;
and using the canonical "ext(A op B) == ext(A) op ext(B)" method of
proving no-wrap is just as powerful now.  Rip out the existing
complexity in favor of getting SCEV to do all the heaving lifting
internally.

This change does not add any unit tests because it is supposed to be a
non-functional change.  Tests added in rL225282 and rL226075 are valid
tests for this change.

Reviewers: atrick, majnemer

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D7981

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231306 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[SCEV] make SCEV smarter about proving no-wrap.
Sanjoy Das [Wed, 4 Mar 2015 22:24:17 +0000 (22:24 +0000)]
[SCEV] make SCEV smarter about proving no-wrap.

Summary:
Teach SCEV to prove no overflow for an add recurrence by proving
something about the range of another add recurrence a loop-invariant
distance away from it.

Reviewers: atrick, hfinkel

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D7980

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231305 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoProvide an explicit move ctor because MSVC can't synthesize one
David Blaikie [Wed, 4 Mar 2015 22:20:52 +0000 (22:20 +0000)]
Provide an explicit move ctor because MSVC can't synthesize one

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231303 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[dsymutil] Add minimal code to emit DIE trees.
Frederic Riss [Wed, 4 Mar 2015 22:07:44 +0000 (22:07 +0000)]
[dsymutil] Add minimal code to emit DIE trees.

This commit adds code to emit DIE trees that have been pruned from the
parts that haven't been marked as kept in the previous pass.

It works by 'cloning' the input DIE tree (as read by libDebugInfoDwarf)
into a tree of DIE objects. Cloning the DIEs means essentially cloning
their attributes. The code in this commit does only handle scalar and
block attributes (scalar because they are trivial, blocks because they
can't be easily replaced by a scalr placeholder), all the other ones
are replaced by placeholder zero values and will be handled in
further commits.

The added tests mostly check that the DIE tree has the correct layout and
also verify that a few chosen scalar and block attributes correctly make
their way into the output.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231300 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDWARFFormValue: Add getAsSignedConstant method.
Frederic Riss [Wed, 4 Mar 2015 22:07:41 +0000 (22:07 +0000)]
DWARFFormValue: Add getAsSignedConstant method.

The implementation accepts explicitely signed forms (DW_FORM_sdata),
but also unsigned forms as long as they fit in an int64_t.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231299 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTeach DIEInteger to emit FORM_strp and FORM_ref_addr attributes.
Frederic Riss [Wed, 4 Mar 2015 22:07:36 +0000 (22:07 +0000)]
Teach DIEInteger to emit FORM_strp and FORM_ref_addr attributes.

To be used/tested by llvm-dsymutil. (llvm-dsymutil does a 'static' link,
no need for relocations for most things, so it'll just emit raw integers
for most attributes)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231298 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMake the DWARFAbbreviationDeclaration::AttributeSpec type public.
Frederic Riss [Wed, 4 Mar 2015 22:07:30 +0000 (22:07 +0000)]
Make the DWARFAbbreviationDeclaration::AttributeSpec type public.

It was already exposed through the iterators anyway.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231297 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUpdate LangRef for explicit type changes to 'load' instruction
David Blaikie [Wed, 4 Mar 2015 22:06:14 +0000 (22:06 +0000)]
Update LangRef for explicit type changes to 'load' instruction

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231296 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoExpand variables when evaluating absolute expressions.
Rafael Espindola [Wed, 4 Mar 2015 22:03:21 +0000 (22:03 +0000)]
Expand variables when evaluating absolute expressions.

This allows for variables to be used in .size.
This matches gnu AS functionality.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231295 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUpdate LangRef for getelementptr explicit type changes
David Blaikie [Wed, 4 Mar 2015 22:02:58 +0000 (22:02 +0000)]
Update LangRef for getelementptr explicit type changes

Here's a rough/first draft - it at least hits the actual textual IR
examples and some of the phrasing. It's probably worth a full pass over,
but I'm not sure how much these docs should reflect the strange
intermediate state we're in anyway.

Totally open to lots of review/feedback/suggestions.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231294 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agodon't repeat class / function / variable names in comments; NFC
Sanjay Patel [Wed, 4 Mar 2015 21:49:03 +0000 (21:49 +0000)]
don't repeat class / function / variable names in comments; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231292 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoSupport standard DWARF TLS opcode; Darwin and PS4 use it.
Paul Robinson [Wed, 4 Mar 2015 20:55:11 +0000 (20:55 +0000)]
Support standard DWARF TLS opcode; Darwin and PS4 use it.

Differential Revision: http://reviews.llvm.org/D8018

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231286 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd LLVM support for PPC cryptography builtins
Nemanja Ivanovic [Wed, 4 Mar 2015 20:44:33 +0000 (20:44 +0000)]
Add LLVM support for PPC cryptography builtins
Review: http://reviews.llvm.org/D7955

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231285 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTry to satisfy sanitizer lint check
Reid Kleckner [Wed, 4 Mar 2015 20:38:59 +0000 (20:38 +0000)]
Try to satisfy sanitizer lint check

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231284 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd a FIXME for PR22796, broken ordering of ClassInfo in TableGen
David Blaikie [Wed, 4 Mar 2015 19:56:44 +0000 (19:56 +0000)]
Add a FIXME for PR22796, broken ordering of ClassInfo in TableGen

As discussed (at length) in code review of r222935, with Duncan.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231282 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix the build of the gold-plugin and examples.
Rafael Espindola [Wed, 4 Mar 2015 19:15:29 +0000 (19:15 +0000)]
Fix the build of the gold-plugin and examples.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231279 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd missing <atomic> include to PassRegistry.h
Reid Kleckner [Wed, 4 Mar 2015 19:06:17 +0000 (19:06 +0000)]
Add missing <atomic> include to PassRegistry.h

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231277 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd a lock() function in PassRegistry to speed up multi-thread synchronization.
Erik Eckstein [Wed, 4 Mar 2015 18:57:11 +0000 (18:57 +0000)]
Add a lock() function in PassRegistry to speed up multi-thread synchronization.

When calling lock() after all passes are registered, the PassRegistry doesn't need a mutex anymore to look up passes.
This speeds up multithreaded llvm execution by ~5% (tested with 4 threads).
In an asserts build of llvm this has an even bigger impact.

Note that it's not required to use the lock function.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231276 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRecommit r231221: "Devirtualize ~parser<T> by making it protected in base classes...
David Blaikie [Wed, 4 Mar 2015 18:52:32 +0000 (18:52 +0000)]
Recommit r231221: "Devirtualize ~parser<T> by making it protected in base classes and making derived classes final"

Reverted in r231254 due to a self-hosting crash of Clang (see Clang
PR22793). Workaround the crash by using {} instead of = default to
define a dtor.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231274 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoBring r231132 back with a fix.
Rafael Espindola [Wed, 4 Mar 2015 18:51:45 +0000 (18:51 +0000)]
Bring r231132 back with a fix.

The issue was that we were always printing the remarks. Fix that and add a test
showing that it prints nothing if -pass-remarks is not given.

Original message:
Correctly handle -pass-remarks in the gold plugin.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231273 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMake DataLayout Non-Optional in the Module
Mehdi Amini [Wed, 4 Mar 2015 18:43:29 +0000 (18:43 +0000)]
Make DataLayout Non-Optional in the Module

Summary:
DataLayout keeps the string used for its creation.

As a side effect it is no longer needed in the Module.
This is "almost" NFC, the string is no longer
canonicalized, you can't rely on two "equals" DataLayout
having the same string returned by getStringRepresentation().

Get rid of DataLayoutPass: the DataLayout is in the Module

The DataLayout is "per-module", let's enforce this by not
duplicating it more than necessary.
One more step toward non-optionality of the DataLayout in the
module.

Make DataLayout Non-Optional in the Module

Module->getDataLayout() will never returns nullptr anymore.

Reviewers: echristo

Subscribers: resistor, llvm-commits, jholewinski

Differential Revision: http://reviews.llvm.org/D7992

From: Mehdi Amini <mehdi.amini@apple.com>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231270 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "unique_ptrify ValID::ConstantStructElts"
Reid Kleckner [Wed, 4 Mar 2015 18:31:10 +0000 (18:31 +0000)]
Revert "unique_ptrify ValID::ConstantStructElts"

This reverts r231200 and r231204. The second one added an explicit move
ctor for MSVC.

This change broke the clang-cl self-host due to weirdness in MSVC's
implementation of std::map::insert. Somehow we lost our rvalue ref-ness
when going through variadic placement new:

  template <class _Objty, class... _Types>
  void construct(_Objty *_Ptr,
                 _Types &&... _Args) { // construct _Objty(_Types...) at _Ptr
    ::new ((void *)_Ptr) _Objty(_STD forward<_Types>(_Args)...);
  }

For some reason, Clang decided to call the deleted std::pair copy
constructor at this point. Needs further investigation, once I can
build.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231269 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert the test commit.
Wei Mi [Wed, 4 Mar 2015 17:44:22 +0000 (17:44 +0000)]
Revert the test commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231264 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTest commit. It will be reverted in the next commit.
Wei Mi [Wed, 4 Mar 2015 17:41:17 +0000 (17:41 +0000)]
Test commit. It will be reverted in the next commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231262 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUpdate the out-of-date dwarf expressions in these testcases.
Adrian Prantl [Wed, 4 Mar 2015 17:39:59 +0000 (17:39 +0000)]
Update the out-of-date dwarf expressions in these testcases.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231261 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix DwarfExpression::AddMachineRegExpression so it doesn't read past the
Adrian Prantl [Wed, 4 Mar 2015 17:39:33 +0000 (17:39 +0000)]
Fix DwarfExpression::AddMachineRegExpression so it doesn't read past the
end of an expression that ends with DW_OP_plus.
Caught by the ASAN build bots.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231260 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoR600/SI: Add an intrinsic for S_FLBIT_I32 / V_FFBH_I32
Marek Olsak [Wed, 4 Mar 2015 17:33:45 +0000 (17:33 +0000)]
R600/SI: Add an intrinsic for S_FLBIT_I32 / V_FFBH_I32

Required by OpenGL (ARB_gpu_shader5).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231259 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTest commit. Removed an unnecessary space
Nemanja Ivanovic [Wed, 4 Mar 2015 17:09:12 +0000 (17:09 +0000)]
Test commit. Removed an unnecessary space

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231257 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoExplicitly default ilistTest::Node's copy constructor
David Blaikie [Wed, 4 Mar 2015 17:01:18 +0000 (17:01 +0000)]
Explicitly default ilistTest::Node's copy constructor

In the presence of a user-declared dtor, calling an implicit copy ctor
is deprecated in C++11.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231256 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert r231221, "Devirtualize ~parser<T> by making it protected in base classes and...
NAKAMURA Takumi [Wed, 4 Mar 2015 16:24:40 +0000 (16:24 +0000)]
Revert r231221, "Devirtualize ~parser<T> by making it protected in base classes and making derived classes final"

It broke seflhosting.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231254 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert r231132, "Correctly handle -pass-remarks in the gold plugin.", for now, to...
NAKAMURA Takumi [Wed, 4 Mar 2015 16:24:28 +0000 (16:24 +0000)]
Revert r231132, "Correctly handle -pass-remarks in the gold plugin.", for now, to suppress log floodng in LTO.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231253 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMutate TargetLowering::shouldExpandAtomicRMWInIR to specifically dictate how AtomicRM...
JF Bastien [Wed, 4 Mar 2015 15:47:57 +0000 (15:47 +0000)]
Mutate TargetLowering::shouldExpandAtomicRMWInIR to specifically dictate how AtomicRMWInsts are expanded.

Summary:
In PNaCl, most atomic instructions have their own @llvm.nacl.atomic.* function, each one, with a few exceptions, represents a consistent behaviour across all NaCl-supported targets. Unfortunately, the atomic RMW operations nand, [u]min, and [u]max aren't directly represented by any such @llvm.nacl.atomic.* function. This patch refines shouldExpandAtomicRMWInIR in TargetLowering so that a future `Le32TargetLowering` class can selectively inform the caller how the target desires the atomic RMW instruction to be expanded (ie via load-linked/store-conditional for ARM/AArch64, via cmpxchg for X86/others?, or not at all for Mips) if at all.

This does not represent a behavioural change and as such no tests were added.

Patch by: Richard Diamond.

Reviewers: jfb

Reviewed By: jfb

Subscribers: jfb, aemerson, t.p.northover, llvm-commits

Differential Revision: http://reviews.llvm.org/D7713

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231250 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[mips][microMIPS] Make usage of ADDU16 and SUBU16 by code generator
Jozef Kolek [Wed, 4 Mar 2015 15:47:42 +0000 (15:47 +0000)]
[mips][microMIPS] Make usage of ADDU16 and SUBU16 by code generator

Differential Revision: http://reviews.llvm.org/D7609

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231249 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[PowerPC] Remove unnecessary and incomplete commentary
Bill Schmidt [Wed, 4 Mar 2015 14:56:05 +0000 (14:56 +0000)]
[PowerPC] Remove unnecessary and incomplete commentary

This "itinerary class map" in PPCSchedule.td is incomplete and
redundant with the actual code.  As it provides no value, we've
decided to remove it.

No functional change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231246 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][FastISel] Simplify the logic in method X86SelectSIToFP.
Andrea Di Biagio [Wed, 4 Mar 2015 14:23:25 +0000 (14:23 +0000)]
[X86][FastISel] Simplify the logic in method X86SelectSIToFP.

The target-independent selection algorithm in FastISel already knows how
to select a SINT_TO_FP if the target is SSE but not AVX.

On targets that have SSE but not AVX, the tablegen'd 'fastEmit' functions
for ISD::SINT_TO_FP know how to select instruction X86::CVTSI2SSrr
(for an i32 to f32 conversion) and X86::CVTSI2SDrr (for an i32 to f64
conversion).

This patch simplifies the logic in method X86SelectSIToFP knowing that
the code would not be reachable if the subtarget doesn't have AVX.
No functional change intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231243 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoasan: do not instrument direct inbounds accesses to stack variables
Dmitry Vyukov [Wed, 4 Mar 2015 13:27:53 +0000 (13:27 +0000)]
asan: do not instrument direct inbounds accesses to stack variables

Do not instrument direct accesses to stack variables that can be
proven to be inbounds, e.g. accesses to fields of structs on stack.

But it eliminates 33% of instrumentation on webrtc/modules_unittests
(number of memory accesses goes down from 290152 to 193998) and
reduces binary size by 15% (from 74M to 64M) and improved compilation time by 6-12%.

The optimization is guarded by asan-opt-stack flag that is off by default.

http://reviews.llvm.org/D7583

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231241 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[mips] Rename the LA/LI/DLI TableGen definitions and classes. NFC.
Toma Tabacu [Wed, 4 Mar 2015 13:01:14 +0000 (13:01 +0000)]
[mips] Rename the LA/LI/DLI TableGen definitions and classes. NFC.

Summary:
Use more reasonable names for these pseudo-instructions.
As there's only one definition tied to any one of these classes, I named them with abbreviated versions of their respective class' name.

Reviewers: dsanders

Reviewed By: dsanders

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D7831

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231240 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[mips] Keep the parameter list of Filler::searchRange() consistent. NFC.
Vasileios Kalintiris [Wed, 4 Mar 2015 12:37:58 +0000 (12:37 +0000)]
[mips] Keep the parameter list of Filler::searchRange() consistent. NFC.

Summary:
Move the "Filler" parameter to the end of the parameter list as it is,
conceptually, the only output parameter of that function.

Reviewers: dsanders

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D7726

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231239 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[MBP] Fix a really horrible bug in MachineBlockPlacement, but behind
Chandler Carruth [Wed, 4 Mar 2015 12:18:08 +0000 (12:18 +0000)]
[MBP] Fix a really horrible bug in MachineBlockPlacement, but behind
a flag for now.

First off, thanks to Daniel Jasper for really pointing out the issue
here. It's been here forever (at least, I think it was there when
I first wrote this code) without getting really noticed or fixed.

The key problem is what happens when two reasonably common patterns
happen at the same time: we outline multiple cold regions of code, and
those regions in turn have diamonds or other CFGs for which we can't
just topologically lay them out. Consider some C code that looks like:

  if (a1()) { if (b1()) c1(); else d1(); f1(); }
  if (a2()) { if (b2()) c2(); else d2(); f2(); }
  done();

Now consider the case where a1() and a2() are unlikely to be true. In
that case, we might lay out the first part of the function like:

  a1, a2, done;

And then we will be out of successors in which to build the chain. We go
to find the best block to continue the chain with, which is perfectly
reasonable here, and find "b1" let's say. Laying out successors gets us
to:

  a1, a2, done; b1, c1;

At this point, we will refuse to lay out the successor to c1 (f1)
because there are still un-placed predecessors of f1 and we want to try
to preserve the CFG structure. So we go get the next best block, d1.

... wait for it ...

Except that the next best block *isn't* d1. It is b2! d1 is waaay down
inside these conditionals. It is much less important than b2. Except
that this is exactly what we didn't want. If we keep going we get the
entire set of the rest of the CFG *interleaved*!!!

  a1, a2, done; b1, c1; b2, c2; d1, f1; d2, f2;

So we clearly need a better strategy here. =] My current favorite
strategy is to actually try to place the block whose predecessor is
closest. This very simply ensures that we unwind these kinds of CFGs the
way that is natural and fitting, and should minimize the number of cache
lines instructions are spread across.

It also happens to be *dead simple*. It's like the datastructure was
specifically set up for this use case or something. We only push blocks
onto the work list when the last predecessor for them is placed into the
chain. So the back of the worklist *is* the nearest next block.

Unfortunately, a change like this is going to cause *soooo* many
benchmarks to swing wildly. So for now I'm adding this under a flag so
that we and others can validate that this is fixing the problems
described, that it seems possible to enable, and hopefully that it fixes
more of our problems long term.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231238 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[mips] Specify the correct value type when combining a CMovFP node.
Vasileios Kalintiris [Wed, 4 Mar 2015 12:10:18 +0000 (12:10 +0000)]
[mips] Specify the correct value type when combining a CMovFP node.

This commit fixes a bug introduced in r230956 where we were creating
CMovFP_{T,F} nodes with multiple return value types (one for each operand).
With this change the return value type of the new node is the same as the
value type of the True/False operands of the original node.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231237 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd a flag to experiment with outlining optional branches.
Daniel Jasper [Wed, 4 Mar 2015 11:05:34 +0000 (11:05 +0000)]
Add a flag to experiment with outlining optional branches.

In a CFG with the edges A->B->C and A->C, B is an optional branch.

LLVM's default behavior is to lay the blocks out naturally, i.e. A, B,
C, in order to improve code locality and fallthroughs. However, if a
function contains many of those optional branches only a few of which
are taken, this leads to a lot of unnecessary icache misses. Moving B
out of line can work around this.

Review: http://reviews.llvm.org/D7719

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231230 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix PR22408 - LLVM producing AArch64 TLS relocations that GNU linkers cannot handle...
Kristof Beyls [Wed, 4 Mar 2015 09:12:08 +0000 (09:12 +0000)]
Fix PR22408 - LLVM producing AArch64 TLS relocations that GNU linkers cannot handle yet.

As is described at http://llvm.org/bugs/show_bug.cgi?id=22408, the GNU linkers
ld.bfd and ld.gold currently only support a subset of the whole range of AArch64
ELF TLS relocations. Furthermore, they assume that some of the code sequences to
access thread-local variables are produced in a very specific sequence.
When the sequence is not as the linker expects, it can silently mis-relaxe/mis-optimize
the instructions.
Even if that wouldn't be the case, it's good to produce the exact sequence,
as that ensures that linkers can perform optimizing relaxations.

This patch:

* implements support for 16MiB TLS area size instead of 4GiB TLS area size. Ideally clang
  would grow an -mtls-size option to allow support for both, but that's not part of this patch.
* by default doesn't produce local dynamic access patterns, as even modern ld.bfd and ld.gold
  linkers do not support the associated relocations. An option (-aarch64-elf-ldtls-generation)
  is added to enable generation of local dynamic code sequence, but is off by default.
* makes sure that the exact expected code sequence for local dynamic and general dynamic
  accesses is produced, by making use of a new pseudo instruction. The patch also removes
  two (AArch64ISD::TLSDESC_BLR, AArch64ISD::TLSDESC_CALL) pre-existing AArch64-specific pseudo
  SDNode instructions that are superseded by the new one (TLSDESC_CALLSEQ).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231227 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Tablegen] Use correct result number variables with the pattern nodes they go with...
Craig Topper [Wed, 4 Mar 2015 09:04:54 +0000 (09:04 +0000)]
[Tablegen] Use correct result number variables with the pattern nodes they go with when handling SDTCisSameAs. No functional change as they are always both 0 unless you try to define a multi result type profile that uses SDTCisSame on one of the other results.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231226 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoExplicitly default DenseMapTest::CtorTest::operator=
David Blaikie [Wed, 4 Mar 2015 07:57:45 +0000 (07:57 +0000)]
Explicitly default DenseMapTest::CtorTest::operator=

Using the implicit default copy assignment operator in the presence of a
user-declared copy ctor is deprecated in C++11.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231225 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove explicit RNSuccIterator copy assignment in favor of implicit default
David Blaikie [Wed, 4 Mar 2015 07:51:50 +0000 (07:51 +0000)]
Remove explicit RNSuccIterator copy assignment in favor of implicit default

Asserting that the source and destination iterators are from the same
region is unnecessary - there's no reason to disallow reassignment from
any regions, so long as they aren't compared.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231224 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agouse = default instead of {}
David Blaikie [Wed, 4 Mar 2015 07:35:04 +0000 (07:35 +0000)]
use = default instead of {}

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231223 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMake format_object_base explicitly copyable, so format_objects can be copied without...
David Blaikie [Wed, 4 Mar 2015 07:35:02 +0000 (07:35 +0000)]
Make format_object_base explicitly copyable, so format_objects can be copied without relying on the implicit copy ctor

Use of the implicit copy ctor is deprecated in C++11 in the presence of
a user declared dtor.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231222 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDevirtualize ~parser<T> by making it protected in base classes and making derived...
David Blaikie [Wed, 4 Mar 2015 07:29:01 +0000 (07:29 +0000)]
Devirtualize ~parser<T> by making it protected in base classes and making derived classes final

These objects are never owned/destroyed polymorphically, so there's no
need for a virtual dtor.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231221 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAvoid copying parser objects
David Blaikie [Wed, 4 Mar 2015 07:28:59 +0000 (07:28 +0000)]
Avoid copying parser objects

Use of their copy members is deprecated since they have a user-declared
dtor.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231220 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[DAGCombine] Fix a bug in a BUILD_VECTOR combine
Michael Kuperstein [Wed, 4 Mar 2015 07:27:39 +0000 (07:27 +0000)]
[DAGCombine] Fix a bug in a BUILD_VECTOR combine

When trying to convert a BUILD_VECTOR into a shuffle, we try to split a single source vector that is twice as wide as the destination vector.
We can not do this when we also need the zero vector to create a blend.
This fixes PR22774.

Differential Revision: http://reviews.llvm.org/D8040

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231219 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMake OptionValue explicitly copyable
David Blaikie [Wed, 4 Mar 2015 07:09:53 +0000 (07:09 +0000)]
Make OptionValue explicitly copyable

Since OptionValue (& its base classes) have user-declared dtors, use of
the implicit copy ctor/assignment operator is deprecated in C++11.
Provide them explicitly (defaulted) to avoid depending on this
deprecated feature.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231218 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDevirtualize OptionValue::~OptionValue in favor of protected in the base, with final...
David Blaikie [Wed, 4 Mar 2015 06:57:14 +0000 (06:57 +0000)]
Devirtualize OptionValue::~OptionValue in favor of protected in the base, with final derived classes

These objects are never polymorphically owned, so there's no need for
virtual dtors - just make the dtor protected in the base classes, and
make the derived classes final.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231217 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[MC][Target] Implement support for R_X86_64_SIZE{32,64}.
Davide Italiano [Wed, 4 Mar 2015 06:49:39 +0000 (06:49 +0000)]
[MC][Target] Implement support for R_X86_64_SIZE{32,64}.

Differential Revision: D7990
Reviewed by: rafael, majnemer

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231216 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[llvm-pdbdump] Display full enum definitions.
Zachary Turner [Wed, 4 Mar 2015 06:09:53 +0000 (06:09 +0000)]
[llvm-pdbdump] Display full enum definitions.

This will now display enum definitions both at the global
scope as well as nested inside of classes.  Additionally,
it will no longer display enums at the global scope if the
enum is nested.  Instead, it will omit the definition of
the enum globally and instead emit it in the corresponding
class definition.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231215 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "[ADT] fail-fast iterators for DenseMap"
Chaoren Lin [Wed, 4 Mar 2015 06:05:37 +0000 (06:05 +0000)]
Revert "[ADT] fail-fast iterators for DenseMap"

This reverts commit 4b7263d855006988854036b4a4891fcf19aebe65.

r231125 http://reviews.llvm.org/D7931

This was causing many LLDB tests to fail on OS X, Linux, and FreeBSD.

https://bpaste.net/show/6a23e1f53623

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231214 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMove emitDIE and emitAbbrevs to AsmPrinter. NFC.
Frederic Riss [Wed, 4 Mar 2015 02:30:17 +0000 (02:30 +0000)]
Move emitDIE and emitAbbrevs to AsmPrinter. NFC.

(They are called emitDwarfDIE and emitDwarfAbbrevs in their new home)

llvm-dsymutil wants to reuse that code, but it doesn't have a DwarfUnit or
a DwarfDebug object to call those. It has access to an AsmPrinter though.

Having emitDIE in the AsmPrinter also removes the DwarfFile dependency
on DwarfDebug, and thus the patch drops that field.

Differential Revision: http://reviews.llvm.org/D8024

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231210 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoConstify AsmPrinter passed to DIE methods.
Frederic Riss [Wed, 4 Mar 2015 02:30:08 +0000 (02:30 +0000)]
Constify AsmPrinter passed to DIE methods.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231209 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix the test for r231201. We don't crash anymore.
Filipe Cabecinhas [Wed, 4 Mar 2015 02:09:40 +0000 (02:09 +0000)]
Fix the test for r231201. We don't crash anymore.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231207 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoWorkaround MSVC not providing implicit move members
David Blaikie [Wed, 4 Mar 2015 02:07:51 +0000 (02:07 +0000)]
Workaround MSVC not providing implicit move members

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231204 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoObject: Add range iterators to Archive symbols
Rui Ueyama [Wed, 4 Mar 2015 02:05:06 +0000 (02:05 +0000)]
Object: Add range iterators to Archive symbols

Also define operator* for symbol iterator just like Archive children iterator.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231203 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUse report_fatal_error instead of unreachable for -fast-isel-abort
Mehdi Amini [Wed, 4 Mar 2015 01:48:39 +0000 (01:48 +0000)]
Use report_fatal_error instead of unreachable for -fast-isel-abort

Suggestion by Andrea Di Biagio

From: Mehdi Amini <mehdi.amini@apple.com>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231201 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agounique_ptrify ValID::ConstantStructElts
David Blaikie [Wed, 4 Mar 2015 01:41:01 +0000 (01:41 +0000)]
unique_ptrify ValID::ConstantStructElts

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231200 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoLLParser: Avoid copying ValIDs, the copy ctor is deprecated in C++11 due to the prese...
David Blaikie [Wed, 4 Mar 2015 01:40:07 +0000 (01:40 +0000)]
LLParser: Avoid copying ValIDs, the copy ctor is deprecated in C++11 due to the presence of a user-declared dtor

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231199 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUse the vanilla func_end symbol for .size.
Rafael Espindola [Wed, 4 Mar 2015 01:35:23 +0000 (01:35 +0000)]
Use the vanilla func_end symbol for .size.

No need to create yet another temp symbol.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231198 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove MCStreamer include which isn't used here. NFC
Pete Cooper [Wed, 4 Mar 2015 01:24:26 +0000 (01:24 +0000)]
Remove MCStreamer include which isn't used here.  NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231195 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoThis file should always have included MCAssembler and not MCStreamer. NFC
Pete Cooper [Wed, 4 Mar 2015 01:24:24 +0000 (01:24 +0000)]
This file should always have included MCAssembler and not MCStreamer.  NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231194 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove MCStreamer.h include from MCContext.h and explictly include it where necessary...
Pete Cooper [Wed, 4 Mar 2015 01:24:11 +0000 (01:24 +0000)]
Remove MCStreamer.h include from MCContext.h and explictly include it where necessary.  NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231193 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRecommit r231168: unique_ptrify LiveRange::segmentSet
David Blaikie [Wed, 4 Mar 2015 01:20:33 +0000 (01:20 +0000)]
Recommit r231168: unique_ptrify LiveRange::segmentSet

GCC 4.7's libstdc++ doesn't have std::map::emplace, but it does have
std::unordered_map::emplace, and the use case here doesn't appear to
need ordering. The container has been changed in a separate/precursor
patch, and now this patch should hopefully build cleanly even with
GCC 4.7.

& then I realized the order of the container did matter, so extra
handling of ordering was added in r231189.

Original commit message:
This makes LiveRange non-copyable, and LiveInterval is already
non-movable (due to the explicit dtor), so now it's non-copyable and
non-movable.

Fix the one case where we were relying on the (deprecated in C++11)
implicit copy ctor of LiveInterval (which happened to work because the
ctor created an object with a null segmentSet, so double-deleting the
null pointer was fine).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231192 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoWeaken the check for a specific movl on the twoaddr-coalesce-3
Eric Christopher [Wed, 4 Mar 2015 01:19:17 +0000 (01:19 +0000)]
Weaken the check for a specific movl on the twoaddr-coalesce-3
test - we only care that there are two moves in the loop and not
which part is relative to which register anyhow.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231191 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoReverse the order libc++ and libc++abi are added in CMake.
Eric Fiselier [Wed, 4 Mar 2015 01:16:43 +0000 (01:16 +0000)]
Reverse the order libc++ and libc++abi are added in CMake.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231190 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRecommit r231175: Change LiveStackAnalysis::SS2IntervalMap from std::map to std:...
David Blaikie [Wed, 4 Mar 2015 01:15:53 +0000 (01:15 +0000)]
Recommit r231175: Change LiveStackAnalysis::SS2IntervalMap from std::map to std::unordered_map

The order of this container was needed at one point - so, at that point
create a temporary array of pointers, sort those, then iterate them.
This keeps lookup efficient (& the lesser issue, of allowing the use of
emplace... ), object identity preserved, and ordered iteration in the
one place that requires it.

While this has no functional change, I realize it does mean allocating
an extra data structure and performing a sort - so if this looks suspect
to anyone regarding perf characteristics, I'm all ears.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231189 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix the x86-upgrade-avx2-vbroadcast.ll test by commenting the CHECK lines
Filipe Cabecinhas [Wed, 4 Mar 2015 00:49:12 +0000 (00:49 +0000)]
Fix the x86-upgrade-avx2-vbroadcast.ll test by commenting the CHECK lines

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231187 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRegisterCoalescer: Gracefully continue if subrange merging fails.
Matthias Braun [Wed, 4 Mar 2015 00:43:50 +0000 (00:43 +0000)]
RegisterCoalescer: Gracefully continue if subrange merging fails.

There is a known bug where the register coalescer fails to merge
subranges when multiple ranges end up in the "overflow" bit 32 of the
lanemasks. A proper fix for this is complicated so for now this is a
workaround which lets the register coalescer drop the subregister
liveness information (we just loose some precision by that) and
continue.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231186 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDrop the "eh_" from eh_func_begin and eh_func_end.
Rafael Espindola [Wed, 4 Mar 2015 00:27:43 +0000 (00:27 +0000)]
Drop the "eh_" from eh_func_begin and eh_func_end.

They will be used for more than eh tables.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231185 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "unique_ptrify LiveRange::segmentSet"
David Blaikie [Wed, 4 Mar 2015 00:15:02 +0000 (00:15 +0000)]
Revert "unique_ptrify LiveRange::segmentSet"

Apparently something does care about ordering of LiveIntervals... so
revert all that stuff (r231175, r231176, r231177) & take some time to
re-evaluate.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231184 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[RewriteStatepointsForGC] Fix a relocation bug w.r.t values defined by invoke instruc...
Philip Reames [Wed, 4 Mar 2015 00:13:52 +0000 (00:13 +0000)]
[RewriteStatepointsForGC] Fix a relocation bug w.r.t values defined by invoke instructions

RewriteStatepointsForGC pass emits an alloca for each GC pointer which will be relocated. It then inserts stores after def and all relocations, and inserts loads before each use as well. In the end, mem2reg is used to update IR with relocations in SSA form.

However, there is a problem with inserting stores for values defined by invoke instructions. The code didn't expect a def was a terminator instruction, and inserting instructions after these terminators resulted in malformed IR.

This patch fixes this problem by handling invoke instructions as a special case. If the def is an invoke instruction, the store will be inserted at the beginning of the normal destination block. Since return value from invoke instruction does not dominate the unwind destination block, no action is needed there.

Patch by: Chen Li
Differential Revision: http://reviews.llvm.org/D7923

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231183 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove 'llvm.x86.avx2.vbroadcasti128' intrinsic.
Juergen Ributzka [Wed, 4 Mar 2015 00:13:25 +0000 (00:13 +0000)]
Remove 'llvm.x86.avx2.vbroadcasti128' intrinsic.

The intrinsic is no longer generated by the front-end. Remove the intrinsic and
auto-upgrade it to a vector shuffle.

Reviewed by Nadav

This is related to rdar://problem/18742778.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231182 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDocument the LLVM "thunk" attribute added back in r226708
Reid Kleckner [Wed, 4 Mar 2015 00:08:56 +0000 (00:08 +0000)]
Document the LLVM "thunk" attribute added back in r226708

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231181 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUpdate twoaddr-coalesce-3.ll to run on darwin and linux machines:
Eric Christopher [Tue, 3 Mar 2015 23:56:20 +0000 (23:56 +0000)]
Update twoaddr-coalesce-3.ll to run on darwin and linux machines:

a) Default relocation model differences,
b) Different numbers of # in comments

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231178 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd missing header include
David Blaikie [Tue, 3 Mar 2015 23:54:35 +0000 (23:54 +0000)]
Add missing header include

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231177 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRecommit r231168: unique_ptrify LiveRange::segmentSet
David Blaikie [Tue, 3 Mar 2015 23:53:03 +0000 (23:53 +0000)]
Recommit r231168: unique_ptrify LiveRange::segmentSet

GCC 4.7's libstdc++ doesn't have std::map::emplace, but it does have
std::unordered_map::emplace, and the use case here doesn't appear to
need ordering. The container has been changed in a separate/precursor
patch, and now this patch should hopefully build cleanly even with
GCC 4.7.

Original commit message:
This makes LiveRange non-copyable, and LiveInterval is already
non-movable (due to the explicit dtor), so now it's non-copyable and
non-movable.

Fix the one case where we were relying on the (deprecated in C++11)
implicit copy ctor of LiveInterval (which happened to work because the
ctor created an object with a null segmentSet, so double-deleting the
null pointer was fine).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231176 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoChange LiveStackAnalysis::SS2IntervalMap from std::map to std::unordered_map
David Blaikie [Tue, 3 Mar 2015 23:53:00 +0000 (23:53 +0000)]
Change LiveStackAnalysis::SS2IntervalMap from std::map to std::unordered_map

This use case doesn't appear to benefit from ordering, and
std::unordered_map has the advantage that it supports emplace (the
LiveInterval values really shouldn't be copyable or movable & they won't
be in a near-future patch).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231175 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "unique_ptrify LiveRange::segmentSet"
David Blaikie [Tue, 3 Mar 2015 23:44:07 +0000 (23:44 +0000)]
Revert "unique_ptrify LiveRange::segmentSet"

GCC 4.7 *shakes fist* (doesn't have std::map::emplace... )

This reverts commit r231168.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231173 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMove TargetLibraryInfo data from two files into one common .def file.
Jan Wen Voung [Tue, 3 Mar 2015 23:41:58 +0000 (23:41 +0000)]
Move TargetLibraryInfo data from two files into one common .def file.

Summary:
This makes it more obvious that the enum definition and the
"StandardName" array is in sync. Mechanically refactored w/ a
python script.

Test Plan: still compiles

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D7845

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231172 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agounique_ptrify LiveRange::segmentSet
David Blaikie [Tue, 3 Mar 2015 23:30:40 +0000 (23:30 +0000)]
unique_ptrify LiveRange::segmentSet

This makes LiveRange non-copyable, and LiveInterval is already
non-movable (due to the explicit dtor), so now it's non-copyable and
non-movable.

Fix the one case where we were relying on the (deprecated in C++11)
implicit copy ctor of LiveInterval (which happened to work because the
ctor created an object with a null segmentSet, so double-deleting the
null pointer was fine).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231168 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[sanitizer/coverage] Add AFL-style coverage counters (search heuristic for fuzzing).
Kostya Serebryany [Tue, 3 Mar 2015 23:27:02 +0000 (23:27 +0000)]
[sanitizer/coverage] Add AFL-style coverage counters (search heuristic for fuzzing).

Introduce -mllvm -sanitizer-coverage-8bit-counters=1
which adds imprecise thread-unfriendly 8-bit coverage counters.

The run-time library maps these 8-bit counters to 8-bit bitsets in the same way
AFL (http://lcamtuf.coredump.cx/afl/technical_details.txt) does:
counter values are divided into 8 ranges and based on the counter
value one of the bits in the bitset is set.
The AFL ranges are used here: 1, 2, 3, 4-7, 8-15, 16-31, 32-127, 128+.

These counters provide a search heuristic for single-threaded
coverage-guided fuzzers, we do not expect them to be useful for other purposes.

Depending on the value of -fsanitize-coverage=[123] flag,
these counters will be added to the function entry blocks (=1),
every basic block (=2), or every edge (=3).

Use these counters as an optional search heuristic in the Fuzzer library.
Add a test where this heuristic is critical.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231166 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove subtarget dependence in pass pipeline setup for AArch64.
Eric Christopher [Tue, 3 Mar 2015 23:22:40 +0000 (23:22 +0000)]
Remove subtarget dependence in pass pipeline setup for AArch64.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231165 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoWinEH: Remove vestigial EH object
Reid Kleckner [Tue, 3 Mar 2015 23:20:30 +0000 (23:20 +0000)]
WinEH: Remove vestigial EH object

Ultimately, we'll need to leave something behind to indicate which
alloca will hold the exception, but we can figure that out when it comes
time to emit the __CxxFrameHandler3 catch handler table.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231164 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoLangRef: Clarify select's semantics with vector arguments
David Majnemer [Tue, 3 Mar 2015 22:45:47 +0000 (22:45 +0000)]
LangRef: Clarify select's semantics with vector arguments

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231158 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoInstCombine: Ensure select condition types are identical before merging
David Majnemer [Tue, 3 Mar 2015 22:40:36 +0000 (22:40 +0000)]
InstCombine: Ensure select condition types are identical before merging

Selection conditions may be vectors or scalars.  Make sure InstCombine
doesn't indiscriminately assume that a select which is value dependent
on another select have identical select condition types.

This fixes PR22773.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231156 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMoving WinEH outlining tests to an architecture neutral location
Andrew Kaylor [Tue, 3 Mar 2015 22:33:39 +0000 (22:33 +0000)]
Moving WinEH outlining tests to an architecture neutral location

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231155 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAvoid copying LiveInterval, this could lead to a double-delete
David Blaikie [Tue, 3 Mar 2015 22:25:48 +0000 (22:25 +0000)]
Avoid copying LiveInterval, this could lead to a double-delete

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231154 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix a problem where the TwoAddressInstructionPass which generate redundant register...
Eric Christopher [Tue, 3 Mar 2015 22:03:03 +0000 (22:03 +0000)]
Fix a problem where the TwoAddressInstructionPass which generate redundant register moves in a loop.

From:
int M, total;
void foo() {
int i;
for (i = 0; i < M; i++) {
  total = total + i / 2;
}
}

This is the kernel loop:

.LBB0_2: # %for.body

=>This Inner Loop Header: Depth=1
movl %edx, %esi
movl %ecx, %edx
shrl $31, %edx
addl %ecx, %edx
sarl %edx
addl %esi, %edx
incl %ecx
cmpl %eax, %ecx
jl .LBB0_2
--------------------------
The first mov insn "movl %edx, %esi" could be removed if we change "addl %esi, %edx" to "addl %edx, %esi".

The IR before TwoAddressInstructionPass is:
BB#2: derived from LLVM BB %for.body

Predecessors according to CFG: BB#1 BB#2
    %vreg3<def> = COPY %vreg12<kill>; GR32:%vreg3,%vreg12
    %vreg2<def> = COPY %vreg11<kill>; GR32:%vreg2,%vreg11
    %vreg7<def,tied1> = SHR32ri %vreg3<tied0>, 31, %EFLAGS<imp-def,dead>; GR32:%vreg7,%vreg3
    %vreg8<def,tied1> = ADD32rr %vreg3<tied0>, %vreg7<kill>, %EFLAGS<imp-def,dead>; GR32:%vreg8,%vreg3,%vreg7
    %vreg9<def,tied1> = SAR32r1 %vreg8<kill,tied0>, %EFLAGS<imp-def,dead>; GR32:%vreg9,%vreg8
    %vreg4<def,tied1> = ADD32rr %vreg9<kill,tied0>, %vreg2<kill>, %EFLAGS<imp-def,dead>; GR32:%vreg4,%vreg9,%vreg2
    %vreg5<def,tied1> = INC64_32r %vreg3<kill,tied0>, %EFLAGS<imp-def,dead>; GR32:%vreg5,%vreg3
    CMP32rr %vreg5, %vreg0, %EFLAGS<imp-def>; GR32:%vreg5,%vreg0
    %vreg11<def> = COPY %vreg4; GR32:%vreg11,%vreg4
    %vreg12<def> = COPY %vreg5<kill>; GR32:%vreg12,%vreg5
    JL_4 <BB#2>, %EFLAGS<imp-use,kill>
Now TwoAddressInstructionPass will choose vreg9 to be tied with vreg4. However, it doesn't see that there is copy from vreg4 to vreg11 and another copy from vreg11 to vreg2 inside the loop body. To remove those copies, it is necessary to choose vreg2 to be tied with vreg4 instead of vreg9. This code pattern commonly appears when there is reduction operation in a loop.

So check for a reversed copy chain and if we encounter one then we can commute the add instruction so we can avoid a copy.

Patch by Wei Mi.
http://reviews.llvm.org/D7806

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@231148 91177308-0d34-0410-b5e6-96231b3b80d8