oota-llvm.git
10 years agoSupport/Unix: use ScopedLock wherever possible
Dylan Noblesmith [Sat, 23 Aug 2014 22:49:17 +0000 (22:49 +0000)]
Support/Unix: use ScopedLock wherever possible

Only one function remains a bit too complicated
for a simple mutex guard. No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216335 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agocmake: actually test -Wcomment
Dylan Noblesmith [Sat, 23 Aug 2014 21:10:58 +0000 (21:10 +0000)]
cmake: actually test -Wcomment

This test was testing nothing, as only -Werror was ever
being added to the compiler flags.

You can see the final nitty-gritty compiler invocation in
CMakeFiles/CMakeOutput.log (for successful tests) and
CMakeFiles/CMakeError.log (for failed tests).

Before:
Building C object CMakeFiles/cmTryCompileExec3385359576.dir/src.c.o
/usr/bin/clang   -fPIC -Wall -W -Wno-unused-parameter -Wwrite-strings -Wmissing-field-initializers -pedantic -Wno-long-long -Wcovered-switch-default  -DC_WCOMMENT_ALLOWS_LINE_WRAP  -Werror   -o CMakeFiles/cmTryCompileExec3385359576.dir/src.c.o   -c /home/nobled/code/llvm-b9/CMakeFiles/CMakeTmp/src.c

After:
Building C object CMakeFiles/cmTryCompileExec3385359576.dir/src.c.o
/usr/bin/clang   -fPIC -Wall -W -Wno-unused-parameter -Wwrite-strings -Wmissing-field-initializers -pedantic -Wno-long-long -Wcovered-switch-default  -DC_WCOMMENT_ALLOWS_LINE_WRAP  -Werror -Wcomment   -o CMakeFiles/cmTryCompileExec3385359576.dir/src.c.o   -c /home/nobled/code/llvm-b9/CMakeFiles/CMakeTmp/src.c

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216328 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agocmake: disable -Wnon-virtual-dtor when it gives false positives
Dylan Noblesmith [Sat, 23 Aug 2014 21:10:56 +0000 (21:10 +0000)]
cmake: disable -Wnon-virtual-dtor when it gives false positives

clang has only been smart enough not to trigger -Wnon-virtual-dtor
warnings on final classes since r208449 (in clang 3.5). Building
with older versions is extremely noisy, so disable the warning
on those compilers.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216327 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert "ARM: improve RTABI 4.2 conformance on Linux"
Chad Rosier [Sat, 23 Aug 2014 18:29:43 +0000 (18:29 +0000)]
Revert "ARM: improve RTABI 4.2 conformance on Linux"

This reverts commit r215862 due to nightly failures.  Will work on getting a
reduced test case, but I wanted to get our bots green in the meantime.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216325 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert "ARM: mark missing functions from RTABI"
Chad Rosier [Sat, 23 Aug 2014 18:29:40 +0000 (18:29 +0000)]
Revert "ARM: mark missing functions from RTABI"

This reverts commit r215863.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216324 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Start fixing a really subtle and terrible form of miscompile in
Chandler Carruth [Sat, 23 Aug 2014 10:25:15 +0000 (10:25 +0000)]
[x86] Start fixing a really subtle and terrible form of miscompile in
these DAG combines.

The DAG auto-CSE thing is truly terrible. Due to it, when RAUW-ing
a node with its operand, you can cause its uses to CSE to itself, which
then causes their uses to become your uses which causes them to be
picked up by the RAUW. For nodes that are determined to be "no-ops",
this is "fine". But if the RAUW is one of several steps to enact
a transformation, this causes the DAG to really silently eat an discard
nodes that you would never expect. It took days for me to actually
pinpoint a test case triggering this and a really frustrating amount of
time to even comprehend the bug because I never even thought about the
ability of RAUW to iteratively consume nodes due to CSE-ing them into
itself.

To fix this, we have to build up a brand-new chain of operations any
time we are combining across (potentially) intervening nodes. But once
the logic is added to do this, another issue surfaces: CombineTo eagerly
deletes the one node combined, *but no others*. This is... really
frustrating. If deleting it makes its operands become dead, those
operand nodes often won't go onto the worklist in the
order you would want -- they're already on it and not near the top. That
means things higher on the worklist will get combined prior to these
dead nodes being GCed out of the worklist, and if the chain is long, the
immediate users won't be enough to re-detect where the root of the chain
is that became single-use again after deleting the dead nodes. The
better way to do this is to never immediately delete nodes, and instead
to just enqueue them so we can recursively delete them. The
combined-from node is typically not on the worklist anyways by virtue of
having been popped off.... But that in turn breaks other tests that
*require* CombineTo to delete unused nodes. :: sigh ::

Fortunately, there is a better way. This whole routine should have been
returning the replacement rather than using CombineTo which is quite
hacky. Switch to that, and all the pieces fall together.

I suspect the same kind of miscompile is possible in the half-shuffle
folding code, and potentially the recursive folding code. I'll be
switching those over to a pattern more like this one for safety's sake
even though I don't immediately have any test cases for them. Note that
the only way I got a test case for this instance was with *heavily* DAG
combined 256-bit shuffle sequences generated by my fuzzer. ;]

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216319 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoProgrammersManual: the flag is called -debug-only
Hans Wennborg [Sat, 23 Aug 2014 04:34:58 +0000 (04:34 +0000)]
ProgrammersManual: the flag is called -debug-only

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216316 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agollvm-cov: test: add xfail for the big-endian buildbots
Alex Lorenz [Sat, 23 Aug 2014 00:47:24 +0000 (00:47 +0000)]
llvm-cov: test: add xfail for the big-endian buildbots

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216310 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert r215611 because it caused the infinite loop in bug 20736. There is a reduced...
Nick Lewycky [Sat, 23 Aug 2014 00:45:03 +0000 (00:45 +0000)]
Revert r215611 because it caused the infinite loop in bug 20736. There is a reduced testcase in that bug.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216307 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd a test case for SROA where the store size is bigger than slice size. The
Yunzhong Gao [Fri, 22 Aug 2014 23:27:04 +0000 (23:27 +0000)]
Add a test case for SROA where the store size is bigger than slice size. The
test case was fixed in r216248.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216303 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd support for comdats to the gold plugin.
Rafael Espindola [Fri, 22 Aug 2014 23:26:10 +0000 (23:26 +0000)]
Add support for comdats to the gold plugin.

There are two parts to this. First, the plugin needs to tell gold the comdat by
setting comdat_key.

What gets things a bit more complicated is that gold only seems
symbols. In particular, if A is an alias to B, it only sees the symbols
A and B. It can then ask us to keep symbol A but drop symbol B. What
we have to do instead is to create an internal version of B and make A
an alias to that.

At some point some of this logic should be moved to lib/Linker so that
we don't map a Constant to an internal version just to have lib/Linker
map that again to the destination module.

The reason for implementing this in tools/gold for now is simplicity.
With it in place it should be possible to update clang to use comdats
for constructors and destructors on ELF without breaking the LTO
bootstrap. Once that is done I intend to come back and improve the
interface lib/Linker exposes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216302 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agollvm-cov: add code coverage tool that's based on coverage mapping format and clang...
Alex Lorenz [Fri, 22 Aug 2014 22:56:03 +0000 (22:56 +0000)]
llvm-cov: add code coverage tool that's based on coverage mapping format and clang's pgo.

This commit expands llvm-cov's functionality by adding support for a new code coverage
tool that uses LLVM's coverage mapping format and clang's instrumentation based profiling.
The gcov compatible tool can be invoked by supplying the 'gcov' command as the first argument,
or by modifying the tool's name to end with 'gcov'.

Differential Revision: http://reviews.llvm.org/D4445

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216300 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[SROA] Fold a PHI node if all its incoming values are the same
Jingyue Wu [Fri, 22 Aug 2014 22:45:57 +0000 (22:45 +0000)]
[SROA] Fold a PHI node if all its incoming values are the same

Summary:
Fixes PR20425.

During slice building, if all of the incoming values of a PHI node are the same, replace the PHI node with the common value. This simplification makes alloca's used by PHI nodes easier to promote.

Test Plan: Added three more tests in phi-and-select.ll

Reviewers: nlewycky, eliben, meheff, chandlerc

Reviewed By: chandlerc

Subscribers: zinovy.nis, hfinkel, baldrick, llvm-commits

Differential Revision: http://reviews.llvm.org/D4659

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216299 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoARM / x86_64 varargs: Don't save regparms in prologue without va_start
Reid Kleckner [Fri, 22 Aug 2014 21:59:26 +0000 (21:59 +0000)]
ARM / x86_64 varargs: Don't save regparms in prologue without va_start

There's no need to do this if the user doesn't call va_start. In the
future, we're going to have thunks that forward these register
parameters with musttail calls, and they won't need these spills for
handling va_start.

Most of the test suite changes are adding va_start calls to existing
tests to keep things working.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216294 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoClear the llvm release notes to make room for 3.6.
Rafael Espindola [Fri, 22 Aug 2014 21:57:38 +0000 (21:57 +0000)]
Clear the llvm release notes to make room for 3.6.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216292 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd the start of the support for llvm-objdump’s -private-headers for Mach-O files.
Kevin Enderby [Fri, 22 Aug 2014 20:35:18 +0000 (20:35 +0000)]
Add the start of the support for llvm-objdump’s -private-headers for Mach-O files.

This adds the printing of the mach header. Load command printing will be next.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216285 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd a few missing mach header flags.
Kevin Enderby [Fri, 22 Aug 2014 20:34:31 +0000 (20:34 +0000)]
Add a few missing mach header flags.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216284 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix PR17239 by changing the semantics of the RemainingArgsClass Option kind
Reid Kleckner [Fri, 22 Aug 2014 19:29:17 +0000 (19:29 +0000)]
Fix PR17239 by changing the semantics of the RemainingArgsClass Option kind

This patch contains the LLVM side of the fix of PR17239.

This bug that happens because the /link (clang-cl.exe argument) is
marked as "consume all remaining arguments". However, when inside a
response file, /link should only consume all remaining arguments inside
the response file where it is located, not the entire command line after
expansion.

My patch will change the semantics of the RemainingArgsClass kind to
always consume only until the end of the response file when the option
originally came from a response file. There are only two options in this
class: dash dash (--) and /link.

Reviewed By: rnk

Differential Revision: http://reviews.llvm.org/D4899

Patch by Rafael Auler!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216280 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Use READ2/WRITE2 instructions for 64-bit mem ops with 32-bit alignment
Tom Stellard [Fri, 22 Aug 2014 18:49:35 +0000 (18:49 +0000)]
R600/SI: Use READ2/WRITE2 instructions for 64-bit mem ops with 32-bit alignment

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216279 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Use a ComplexPattern for DS loads and stores
Tom Stellard [Fri, 22 Aug 2014 18:49:33 +0000 (18:49 +0000)]
R600/SI: Use a ComplexPattern for DS loads and stores

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216278 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Wrap local memory pointer in AssertZExt on SI
Tom Stellard [Fri, 22 Aug 2014 18:49:31 +0000 (18:49 +0000)]
R600/SI: Wrap local memory pointer in AssertZExt on SI

These pointers are really just offsets and they will always be
less than 16-bits.  Using AssertZExt allows us to use computeKnownBits
to prove that these values are positive.  We will use this information
in a later commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216277 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Use correct helper class for DS_WRITE2 instructions
Tom Stellard [Fri, 22 Aug 2014 18:49:28 +0000 (18:49 +0000)]
R600/SI: Use correct helper class for DS_WRITE2 instructions

DS_1A uses a single offset encoding, so offset1 wasn't being
encoded.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216276 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[ARM] Move the implementation of the target hooks related to copy-related
Quentin Colombet [Fri, 22 Aug 2014 18:05:22 +0000 (18:05 +0000)]
[ARM] Move the implementation of the target hooks related to copy-related
instruction from ARMInstrInfo to ARMBaseInstrInfo.
That way, thumb mode can also benefit from the advanced copy optimization.

<rdar://problem/12702965>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216274 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstCombine: Don't unconditionally preserve 'nuw' when shrinking constants
David Majnemer [Fri, 22 Aug 2014 17:11:04 +0000 (17:11 +0000)]
InstCombine: Don't unconditionally preserve 'nuw' when shrinking constants

Consider:
  %add = add nuw i32 %a, -16777216
  %and = and i32 %add, 255

Regardless of whether or not we demand the sign bit of %add, we cannot
replace -16777216 with 2130706432 without also removing 'nuw' from the
instruction.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216273 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstCombine: sub nsw %x, C -> add nsw %x, -C if C isn't INT_MIN
David Majnemer [Fri, 22 Aug 2014 16:41:23 +0000 (16:41 +0000)]
InstCombine: sub nsw %x, C -> add nsw %x, -C if C isn't INT_MIN

We can preserve nsw during this transform if -C won't overflow.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216269 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[Support] Fix the overflow bug in ULEB128 decoding.
Alex Lorenz [Fri, 22 Aug 2014 16:29:45 +0000 (16:29 +0000)]
[Support] Fix the overflow bug in ULEB128 decoding.

Differential Revision: http://reviews.llvm.org/D5029

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216268 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[mips] Don't use odd-numbered float registers for double arguments for fastcc
Sasa Stankovic [Fri, 22 Aug 2014 09:23:22 +0000 (09:23 +0000)]
[mips] Don't use odd-numbered float registers for double arguments for fastcc
calling convention if FP is 64-bit and +nooddspreg is used.

Differential Revision: http://reviews.llvm.org/D4981.diff

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216262 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstCombine: Don't unconditionally preserve 'nsw' when shrinking constants
David Majnemer [Fri, 22 Aug 2014 07:56:32 +0000 (07:56 +0000)]
InstCombine: Don't unconditionally preserve 'nsw' when shrinking constants

Consider:
  %add = add nsw i32 %a, -16777216
  %and = and i32 %add, 255

Regardless of whether or not we demand the sign bit of %add, we cannot
replace -16777216 with 2130706432 without also removing 'nsw' from the
instruction.

This fixes PR20377.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216261 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agofix: SLPVectorizer crashes for unreachable blocks containing not schedulable instruct...
Erik Eckstein [Fri, 22 Aug 2014 01:18:39 +0000 (01:18 +0000)]
fix: SLPVectorizer crashes for unreachable blocks containing not schedulable instructions.

In unreachable blocks it's legal to have instructions like "%x = op %x".
Such instuctions are not schedulable. Therefore the SLPVectorizer has to check for
unreachable blocks and ignore them.

Fixes bug 20646.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216256 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[dfsan] Fix non-determinism bug in non-zero label check annotator.
Peter Collingbourne [Fri, 22 Aug 2014 01:18:18 +0000 (01:18 +0000)]
[dfsan] Fix non-determinism bug in non-zero label check annotator.

We now use a std::vector instead of a DenseSet to store the list of
label checks so that we can iterate over it deterministically.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216255 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoValueTracking: Figure out more bits when looking at add/sub
David Majnemer [Fri, 22 Aug 2014 00:40:43 +0000 (00:40 +0000)]
ValueTracking: Figure out more bits when looking at add/sub

Given something like X01XX + X01XX, we know that the result must look
like X1XXX.

Adapted from a patch by Richard Smith, test-case written by me.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216250 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSROA: Handle a case of store size being smaller than allocation size
Reid Kleckner [Fri, 22 Aug 2014 00:09:56 +0000 (00:09 +0000)]
SROA: Handle a case of store size being smaller than allocation size

In this case, we are creating an x86_fp80 slice for a union from C where
the padding bytes may contain real data. An x86_fp80 alloca is 16 bytes,
and that's just fine. We can't, however, use regular loads and stores to
access the slice, because the store size is only 10 bytes / 80 bits.
Instead, use memcpy and memset.

Fixes PR18726.

Reviewed By: chandlerc

Differential Revision: http://reviews.llvm.org/D5012

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216248 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert "X86: Align the stack on word boundaries in LowerFormalArguments()"
Duncan P. N. Exon Smith [Thu, 21 Aug 2014 23:36:08 +0000 (23:36 +0000)]
Revert "X86: Align the stack on word boundaries in LowerFormalArguments()"

This (mostly) reverts commit r216119.

Somewhere during the review Reid committed r214980 which fixed this
another way, and I neglected to check that the testcase still failed
before committing.

I've left test/CodeGen/X86/aligned-variadic.ll around in case it adds
extra coverage.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216246 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd an explicit move constructor to SrcBuffer
Reid Kleckner [Thu, 21 Aug 2014 23:24:08 +0000 (23:24 +0000)]
Add an explicit move constructor to SrcBuffer

MSVC can't synthesize the explicit one.  Instead it tries to emit a copy
ctor which would call the deleted copy ctor of unique_ptr.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216244 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[FastISel][AArch64] Add support for variable shift.
Juergen Ributzka [Thu, 21 Aug 2014 23:06:07 +0000 (23:06 +0000)]
[FastISel][AArch64] Add support for variable shift.

This adds the missing variable shift support for value type i8, i16, and i32.

This fixes <rdar://problem/18095685>.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216242 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMinor refactor to make applying patches from 'Add a "probe-stack" attribute' review...
Philip Reames [Thu, 21 Aug 2014 22:53:49 +0000 (22:53 +0000)]
Minor refactor to make applying patches from 'Add a "probe-stack" attribute' review thread out of order easier.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216241 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoUse DILexicalBlockFile, rather than DILexicalBlock, to track discriminator changes...
David Blaikie [Thu, 21 Aug 2014 22:45:21 +0000 (22:45 +0000)]
Use DILexicalBlockFile, rather than DILexicalBlock, to track discriminator changes to ensure discriminator changes don't introduce new DWARF DW_TAG_lexical_blocks.

Somewhat unnoticed in the original implementation of discriminators, but
it could cause instructions to end up in new, small,
DW_TAG_lexical_blocks due to the use of DILexicalBlock to track
discriminator changes.

Instead, use DILexicalBlockFile which we already use to track file
changes without introducing new scopes, so it works well to track
discriminator changes in the same way.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216239 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoname change: isPow2DivCheap -> isPow2SDivCheap
Sanjay Patel [Thu, 21 Aug 2014 22:31:48 +0000 (22:31 +0000)]
name change: isPow2DivCheap -> isPow2SDivCheap

isPow2DivCheap

That name doesn't specify signed or unsigned.

Lazy as I am, I eventually read the function and variable comments. It turns out that this is strictly about signed div. But I discovered that the comments are wrong:

   srl/add/sra

is not the general sequence for signed integer division by power-of-2. We need one more 'sra':

   sra/srl/add/sra

That's the sequence produced in DAGCombiner. The first 'sra' may be removed when dividing by exactly '2', but that's a special case.

This patch corrects the comments, changes the name of the flag bit, and changes the name of the accessor methods.

No functional change intended.

Differential Revision: http://reviews.llvm.org/D5010

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216237 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PeepholeOptimizer] Enable the advanced copy optimization by default.
Quentin Colombet [Thu, 21 Aug 2014 22:23:52 +0000 (22:23 +0000)]
[PeepholeOptimizer] Enable the advanced copy optimization by default.

The advanced copy optimization does not yield any difference on the whole llvm
test-suite + SPECs, either in compile time or runtime (binaries are identical),
but has a big potential when data go back and forth between register files as
demonstrated with test/CodeGen/ARM/adv-copy-opt.ll.

Note: This was measured for both Os and O3 for armv7s, arm64, and x86_64.

<rdar://problem/12702965>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216236 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoWhitespace change to reduce diff in future patch.
Philip Reames [Thu, 21 Aug 2014 22:19:16 +0000 (22:19 +0000)]
Whitespace change to reduce diff in future patch.

Patch 2 of 11 in 'Add a "probe-stack" attribute' review thread

Patch by: john.kare.alsaker@gmail.com

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216235 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[X86] Split out the logic to select the stack probe function (NFC)
Philip Reames [Thu, 21 Aug 2014 22:15:20 +0000 (22:15 +0000)]
[X86] Split out the logic to select the stack probe function (NFC)

Patch 1 of 11 in 'Add a "probe-stack" attribute' review thread.

Patch by: <john.kare.alsaker@gmail.com>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216233 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd hooks for emitLeading/TrailingFence
Robin Morisset [Thu, 21 Aug 2014 22:09:25 +0000 (22:09 +0000)]
Add hooks for emitLeading/TrailingFence

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216232 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRename AtomicExpandLoadLinked into AtomicExpand
Robin Morisset [Thu, 21 Aug 2014 21:50:01 +0000 (21:50 +0000)]
Rename AtomicExpandLoadLinked into AtomicExpand

AtomicExpandLoadLinked is currently rather ARM-specific. This patch is the first of
a group that aim at making it more target-independent. See
http://lists.cs.uiuc.edu/pipermail/llvmdev/2014-August/075873.html
for details

The command line option is "atomic-expand"

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216231 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PeepholeOptimizer] Update the kill flags when extending the live-range of the
Quentin Colombet [Thu, 21 Aug 2014 21:34:06 +0000 (21:34 +0000)]
[PeepholeOptimizer] Update the kill flags when extending the live-range of the
source of a copy.

<rdar://problem/12702965>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216229 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix a URL (NFC)
Justin Bogner [Thu, 21 Aug 2014 21:09:24 +0000 (21:09 +0000)]
Fix a URL (NFC)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216228 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[FastISel][AArch64] Use the correct register class to make the MI verifier happy.
Juergen Ributzka [Thu, 21 Aug 2014 20:57:57 +0000 (20:57 +0000)]
[FastISel][AArch64] Use the correct register class to make the MI verifier happy.

This is mostly achieved by providing the correct register class manually,
because getRegClassFor always returns the GPR*AllRegClass for MVT::i32 and
MVT::i64.

Also cleanup the code to use the FastEmitInst_* method whenever possible. This
makes sure that the operands' register class is properly constrained. For all
the remaining cases this adds the missing constrainOperandRegClass calls for
each operand.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216225 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoExplicitly pass ownership of the MemoryBuffer to AddNewSourceBuffer using std::unique_ptr
David Blaikie [Thu, 21 Aug 2014 20:44:56 +0000 (20:44 +0000)]
Explicitly pass ownership of the MemoryBuffer to AddNewSourceBuffer using std::unique_ptr

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216223 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Teach moveToVALU how to handle more S_LOAD_* instructions
Tom Stellard [Thu, 21 Aug 2014 20:41:00 +0000 (20:41 +0000)]
R600/SI: Teach moveToVALU how to handle more S_LOAD_* instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216220 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Make sure SCRATCH_WAVE_OFFSET is added as Live-In to the function
Tom Stellard [Thu, 21 Aug 2014 20:40:58 +0000 (20:40 +0000)]
R600/SI: Make sure SCRATCH_WAVE_OFFSET is added as Live-In to the function

This fixes a crash in an ocl conformance test.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216219 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Remove unused SGPR spilling code
Tom Stellard [Thu, 21 Aug 2014 20:40:56 +0000 (20:40 +0000)]
R600/SI: Remove unused SGPR spilling code

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216218 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Use eliminateFrameIndex() to expand SGPR spill pseudos
Tom Stellard [Thu, 21 Aug 2014 20:40:54 +0000 (20:40 +0000)]
R600/SI: Use eliminateFrameIndex() to expand SGPR spill pseudos

This will simplify the SGPR spilling and also allow us to use
MachineFrameInfo for calculating offsets, which should be more
reliable than our custom code.

This fixes a crash in some cases where a register would be spilled
in a branch such that the VGPR defined for spilling did not dominate
all the uses when restoring.

This fixes a crash in an ocl conformance test.  The test requries
register spilling and is too big to include.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216217 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoR600/SI: Handle VCC in SIRegisterInfo::getPhysRegSubReg()
Tom Stellard [Thu, 21 Aug 2014 20:40:50 +0000 (20:40 +0000)]
R600/SI: Handle VCC in SIRegisterInfo::getPhysRegSubReg()

This fixes a crash in an ocl conformance test.  The test requries
register spilling and is too big to include.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216216 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRewrite the gold plugin to fix pr19901.
Rafael Espindola [Thu, 21 Aug 2014 20:28:55 +0000 (20:28 +0000)]
Rewrite the gold plugin to fix pr19901.

There is a fundamental difference between how the gold API and lib/LTO view
the LTO process.

The gold API talks about a particular symbol in a particular file. The lib/LTO
API talks about a symbol in the merged module.

The merged module is then defined in terms of the IR semantics. In particular,
a linkonce_odr GV is only copied if it is used, since it is valid to drop
unused linkonce_odr GVs.

In the testcase in pr19901 both properties collide. What happens is that gold
asks us to keep a particular linkonce_odr symbol, but the IR linker doesn't
copy it to the merged module and we never have a chance to ask lib/LTO to keep
it.

This patch fixes it by having a more direct implementation of the gold API. If
it asks us to keep a symbol, we change the linkage so it is not linkonce. If it
says we can drop a symbol, we do so. All of this before we even send the module
to lib/Linker.

Since now we don't have to produce LTO_SYMBOL_SCOPE_DEFAULT_CAN_BE_HIDDEN,
during symbol resolution we can use a temporary LLVMContext and do lazy
module loading. This allows us to keep the minimum possible amount of
allocated memory around. This should also allow as much parallelism as
we want, since there is no shared context.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216215 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSatiate the sanitizer build bot
Jonathan Roelofs [Thu, 21 Aug 2014 20:09:15 +0000 (20:09 +0000)]
Satiate the sanitizer build bot

This fixes a missing initializer from r216182

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216212 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMove some logic to populateLTOPassManager.
Rafael Espindola [Thu, 21 Aug 2014 20:03:44 +0000 (20:03 +0000)]
Move some logic to populateLTOPassManager.

This will avoid code duplication in the next commit which calls it directly
from the gold plugin.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216211 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[AVX512] Add class to group common template arguments related to vector type
Adam Nemet [Thu, 21 Aug 2014 19:50:07 +0000 (19:50 +0000)]
[AVX512] Add class to group common template arguments related to vector type

We discussed the issue of generality vs. readability of the AVX512 classes
recently.  I proposed this approach to try to hide and centralize the mappings
we commonly perform based on the vector type.  A new class X86VectorVTInfo
captures these.

The idea is to pass an instance of this class to classes/multiclasses instead
of the corresponding ValueType.  Then the class/multiclass can use its field
for things that derive from the type rather than passing all those as separate
arguments.

I modified avx512_valign to demonstrate this new approach.  As you can see
instead of 7 related template parameters we now have one.  The downside is
that we have to refer to fields for the derived values.  I named the argument
'_' in order to make this as invisible as possible.  Please let me know if you
absolutely hate this.  (Also once we allow local initializations in
multiclasses we can recover the original version by assigning the fields to
local variables.)

Another possible use-case for this class is to directly map things, e.g.:

  RegisterClass KRC = X86VectorVTInfo<32, i16>.KRC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216209 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoCoverage Mapping: add function's hash to coverage function records.
Alex Lorenz [Thu, 21 Aug 2014 19:23:25 +0000 (19:23 +0000)]
Coverage Mapping: add function's hash to coverage function records.

The profile data format was recently updated and the new indexing api
requires the code coverage tool to know the function's hash as well
as the function's name to get the execution counts for a function.

Differential Revision: http://reviews.llvm.org/D4994

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216207 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agollvm-gcc is dead.
Rafael Espindola [Thu, 21 Aug 2014 19:22:24 +0000 (19:22 +0000)]
llvm-gcc is dead.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216206 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[LIT] Remove documentation for method since it does not exist
Eric Fiselier [Thu, 21 Aug 2014 18:52:58 +0000 (18:52 +0000)]
[LIT] Remove documentation for  method since it does not exist

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216204 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRespect LibraryInfo in populateLTOPassManager and use it. NFC.
Rafael Espindola [Thu, 21 Aug 2014 18:49:52 +0000 (18:49 +0000)]
Respect LibraryInfo in populateLTOPassManager and use it. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216203 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove dead code. NFC.
Rafael Espindola [Thu, 21 Aug 2014 18:11:21 +0000 (18:11 +0000)]
Remove dead code. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216201 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[AArch64] Run a peephole pass right after AdvSIMD pass.
Quentin Colombet [Thu, 21 Aug 2014 18:10:07 +0000 (18:10 +0000)]
[AArch64] Run a peephole pass right after AdvSIMD pass.
The AdvSIMD pass may produce copies that are not coalescer-friendly. The
peephole optimizer knows how to fix that as demonstrated in the test case.

<rdar://problem/12702965>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216200 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[FastISel][AArch64] Factor out ANDWri instruction generation into a helper function...
Juergen Ributzka [Thu, 21 Aug 2014 18:02:25 +0000 (18:02 +0000)]
[FastISel][AArch64] Factor out ANDWri instruction generation into a helper function. NFCI.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216199 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoThumb1 load/store optimizer: Improve code to materialize new base register.
Moritz Roth [Thu, 21 Aug 2014 17:11:03 +0000 (17:11 +0000)]
Thumb1 load/store optimizer: Improve code to materialize new base register.

There are two add-immediate instructions in Thumb1: tADDi8 and tADDi3. Only
the latter supports using different source and destination registers, so
whenever we materialize a new base register (at a certain offset) we'd do
so by moving the base register value to the new register and then adding in
place. This patch changes the code to use a single tADDi3 if the offset is
small enough to fit in 3 bits.

Differential Revision: http://reviews.llvm.org/D5006

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216193 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoUse returns_nonnull in BumpPtrAllocator and MallocAllocator to avoid null-check in...
Hans Wennborg [Thu, 21 Aug 2014 17:10:00 +0000 (17:10 +0000)]
Use returns_nonnull in BumpPtrAllocator and MallocAllocator to avoid null-check in placement new

In both Clang and LLVM, this is a common pattern:

  Size = sizeof(DeclRefExpr) + SomeExtraStuff;
  void *Mem = Context.Allocate(Size, llvm::alignOf<DeclRefExpr>());
  return new (Mem) DeclRefExpr(...);

The annoying thing is that because the default placement-new operator has a
nothrow specification, the compiler will insert a null check of Mem before
calling the DeclRefExpr constructor. This null check is redundant for us,
because we expect the allocation functions to never return null.

By annotating the allocator functions with returns_nonnull, we can optimize
away these checks. Compiling clang with a recent version of Clang and measuring
with:

  $ perf stat -r20 bin/clang.patch -fsyntax-only -w gcc.c && perf stat -r20 bin/clang.orig -fsyntax-only -w gcc.c

Shows a 2.4% speed-up (+- 0.8%).

The pattern occurs in LLVM too. Measuring with -O3 (and now using bzip2.c
instead, because it's smaller):

  $ perf stat -r20 bin/clang.patch -O3 -w bzip2.c  &&  perf stat -r20 bin/clang.orig -O3 -w bzip2.c

Shows 4.4 % speed-up (+- 1%).

If anyone knows of a similar attribute we can use for MSVC, or some other
technique to get rid off the null check there, please let me know.

Differential Revision: http://reviews.llvm.org/D4989

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216192 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[FastISel][AArch64] Remove redundant test.
Juergen Ributzka [Thu, 21 Aug 2014 16:40:05 +0000 (16:40 +0000)]
[FastISel][AArch64] Remove redundant test.

These tests and many more are already covered by fast-isel-addressing-modes.ll.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216186 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd a thread-model knob for lowering atomics on baremetal & single threaded systems
Jonathan Roelofs [Thu, 21 Aug 2014 14:35:47 +0000 (14:35 +0000)]
Add a thread-model knob for lowering atomics on baremetal & single threaded systems

http://reviews.llvm.org/D4984

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216182 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoHandle inlining in populateLTOPassManager like in populateModulePassManager.
Rafael Espindola [Thu, 21 Aug 2014 13:35:30 +0000 (13:35 +0000)]
Handle inlining in populateLTOPassManager like in populateModulePassManager.

No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216178 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[CLNUP] Remove return after llvm_unreachable. Thanks to Hal Finkel for pointing.
Zinovy Nis [Thu, 21 Aug 2014 13:30:05 +0000 (13:30 +0000)]
[CLNUP] Remove return after llvm_unreachable. Thanks to Hal Finkel for pointing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216176 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDAGCombiner: Make concat_vector combine safe for EVTs and concat_vectors with many...
Benjamin Kramer [Thu, 21 Aug 2014 13:28:02 +0000 (13:28 +0000)]
DAGCombiner: Make concat_vector combine safe for EVTs and concat_vectors with many arguments.

PR20677

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216175 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMove DisableGVNLoadPRE from populateLTOPassManager to PassManagerBuilder.
Rafael Espindola [Thu, 21 Aug 2014 13:13:17 +0000 (13:13 +0000)]
Move DisableGVNLoadPRE from populateLTOPassManager to PassManagerBuilder.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216174 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoX86AsmPrinter MCJIT MSVC bug fix.
Josh Klontz [Thu, 21 Aug 2014 12:55:27 +0000 (12:55 +0000)]
X86AsmPrinter MCJIT MSVC bug fix.

Summary:
This bug was introduced in r213006 which makes an assumption that MCSection is COFF for Windows MSVC. This assumption is broken for MCJIT users where ELF is used instead [1]. The fix is to change the MCSection cast to a dyn_cast.

[1] http://lists.cs.uiuc.edu/pipermail/llvmdev/2013-December/068407.html.

Reviewers: majnemer

Reviewed By: majnemer

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D4872

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216173 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[ARM] Enable DP copy, load and store instructions for FPv4-SP
Oliver Stannard [Thu, 21 Aug 2014 12:50:31 +0000 (12:50 +0000)]
[ARM] Enable DP copy, load and store instructions for FPv4-SP

The FPv4-SP floating-point unit is generally referred to as
single-precision only, but it does have double-precision registers and
load, store and GPR<->DPR move instructions which operate on them.
This patch enables the use of these registers, the main advantage of
which is that we now comply with the AAPCS-VFP calling convention.
This partially reverts r209650, which added some AAPCS-VFP support,
but did not handle return values or alignment of double arguments in
registers.

This patch also adds tests for Thumb2 code generation for
floating-point instructions and intrinsics, which previously only
existed for ARM.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216172 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoSort declarations.
Rafael Espindola [Thu, 21 Aug 2014 12:39:07 +0000 (12:39 +0000)]
Sort declarations.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216171 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMake format_object_base's destructor protected and non-virtual.
Benjamin Kramer [Thu, 21 Aug 2014 11:22:05 +0000 (11:22 +0000)]
Make format_object_base's destructor protected and non-virtual.

It's not meant to be used with operator delete and this avoids emitting virtual
dtors for every derived format object.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216170 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoReassociate x + -0.1234 * y into x - 0.1234 * y
Erik Verbruggen [Thu, 21 Aug 2014 10:45:30 +0000 (10:45 +0000)]
Reassociate x + -0.1234 * y into x - 0.1234 * y

This does not require -ffast-math, and it gives CSE/GVN more options to
eliminate duplicate expressions in, e.g.:

  return ((x + 0.1234 * y) * (x - 0.1234 * y));

Differential Revision: http://reviews.llvm.org/D4904

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216169 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoX86: Turn redundant if into an assertion.
Benjamin Kramer [Thu, 21 Aug 2014 10:31:37 +0000 (10:31 +0000)]
X86: Turn redundant if into an assertion.

While there remove noop casts.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216168 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Added _addcarry_ and _subborrow_ intrinsics
Robert Khasanov [Thu, 21 Aug 2014 09:43:43 +0000 (09:43 +0000)]
[x86] Added _addcarry_ and _subborrow_ intrinsics

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216164 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] SMAP: added HasSMAP attribute for CLAC/STAC, corrected attributes
Robert Khasanov [Thu, 21 Aug 2014 09:34:12 +0000 (09:34 +0000)]
[x86] SMAP: added HasSMAP attribute for CLAC/STAC, corrected attributes

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216163 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Broadwell: ADOX/ADCX. Added _addcarryx_u{32|64} intrinsics to LLVM.
Robert Khasanov [Thu, 21 Aug 2014 09:27:00 +0000 (09:27 +0000)]
[x86] Broadwell: ADOX/ADCX. Added _addcarryx_u{32|64} intrinsics to LLVM.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216162 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[x86] Enable Broadwell target.
Robert Khasanov [Thu, 21 Aug 2014 09:16:12 +0000 (09:16 +0000)]
[x86] Enable Broadwell target.
Added FeatureSMAP.

Broadwell ISA includes Haswell ISA + ADX + RDSEED + SMAP

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216161 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[INDVARS] Extend using of widening of induction variables for the cases of "sub nsw...
Zinovy Nis [Thu, 21 Aug 2014 08:25:45 +0000 (08:25 +0000)]
[INDVARS] Extend using of widening of induction variables for the cases of "sub nsw" and "mul nsw" instructions.

Currently only "add nsw" are widened. This patch eliminates tons of "sext" instructions for 64 bit code (and the corresponding target code) in cases like:

int N = 100;
float **A;

void foo(int x0, int x1)
{
        float * A_cur = &A[0][0];
        float * A_next = &A[1][0];
        for(int x = x0; x < x1; ++x).
        {
          // Currently only [x+N] case is widened. Others 2 cases lead to sext.
          // This patch fixes it, so all 3 cases do not need sext.
          const float div = A_cur[x + N] + A_cur[x - N] + A_cur[x * N];
          A_next[x] = div;
        }
}
...
> clang++ test.cpp -march=core-avx2 -Ofast  -fno-unroll-loops -fno-tree-vectorize -S -o -

Differential Revision: http://reviews.llvm.org/D4695

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216160 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoIntelJITEventListener updates to fix breaks by recent changes to EngineBuilder and...
Elena Demikhovsky [Thu, 21 Aug 2014 07:01:55 +0000 (07:01 +0000)]
IntelJITEventListener updates to fix breaks by recent changes to EngineBuilder and DIContext.
By Arch Robison.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216159 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRepace SmallPtrSet with SmallPtrSetImpl in function arguments to avoid needing to...
Craig Topper [Thu, 21 Aug 2014 05:55:13 +0000 (05:55 +0000)]
Repace SmallPtrSet with SmallPtrSetImpl in function arguments to avoid needing to mention the size.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216158 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoInstCombine: Fold ((A | B) & C1) ^ (B & C2) -> (A & C1) ^ B if C1^C2=-1
David Majnemer [Thu, 21 Aug 2014 05:14:48 +0000 (05:14 +0000)]
InstCombine: Fold ((A | B) & C1) ^ (B & C2) -> (A & C1) ^ B if C1^C2=-1

Adapted from a patch by Richard Smith, test-case written by me.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216157 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRemove custom implementations of max/min in StringRef that was originally added to...
Craig Topper [Thu, 21 Aug 2014 04:31:10 +0000 (04:31 +0000)]
Remove custom implementations of max/min in StringRef that was originally added to work an old gcc bug. I believe its been fixed by now.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216156 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoadd self to credits
Eric Fiselier [Thu, 21 Aug 2014 04:27:11 +0000 (04:27 +0000)]
add self to credits

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216155 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoFix a bug around truncating vector in const prop.
Jiangning Liu [Thu, 21 Aug 2014 02:12:35 +0000 (02:12 +0000)]
Fix a bug around truncating vector in const prop.

In constant folding stage, "TRUNC" can't handle vector data type.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216149 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoRevert r216066, "Optimize ZERO_EXTEND and SIGN_EXTEND in both SelectionDAG Builder...
Jiangning Liu [Thu, 21 Aug 2014 01:59:30 +0000 (01:59 +0000)]
Revert r216066, "Optimize ZERO_EXTEND and SIGN_EXTEND in both SelectionDAG Builder and type".

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216147 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PeepholeOptimizer] Take advantage of the isInsertSubreg property in the
Quentin Colombet [Thu, 21 Aug 2014 00:19:16 +0000 (00:19 +0000)]
[PeepholeOptimizer] Take advantage of the isInsertSubreg property in the
advanced copy optimization.

This is the final step patch toward transforming:
udiv    r0, r0, r2
udiv    r1, r1, r3
vmov.32 d16[0], r0
vmov.32 d16[1], r1
vmov    r0, r1, d16
bx      lr

into:
udiv    r0, r0, r2
udiv    r1, r1, r3
bx      lr

Indeed, thanks to this patch, this optimization is able to look through
vmov.32 d16[0], r0
vmov.32 d16[1], r1

and is able to rewrite the following sequence:
vmov.32 d16[0], r0
vmov.32 d16[1], r1
vmov    r0, r1, d16

into simple generic GPR copies that the coalescer managed to remove.

<rdar://problem/12702965>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216144 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[ARM] Mark VSETLNi32 with the InsertSubreg property and implement the related
Quentin Colombet [Thu, 21 Aug 2014 00:10:52 +0000 (00:10 +0000)]
[ARM] Mark VSETLNi32 with the InsertSubreg property and implement the related
target hook.

This patch teaches the compiler that:
dX = VSETLNi32 dY, rZ, imm
is the same as:
dX = INSERT_SUBREG dY, rZ, translateImmToSubIdx(imm)

<rdar://problem/12702965>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216143 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[LoopVectorize] Up the maximum unroll factor to 4 for AArch64
James Molloy [Thu, 21 Aug 2014 00:02:51 +0000 (00:02 +0000)]
[LoopVectorize] Up the maximum unroll factor to 4 for AArch64

Only for Cortex-A57 and Cyclone for now, where it has shown wins.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216141 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[LoopVectorizer] Limit unroll factor in the presence of nested reductions.
James Molloy [Wed, 20 Aug 2014 23:53:52 +0000 (23:53 +0000)]
[LoopVectorizer] Limit unroll factor in the presence of nested reductions.

If we have a scalar reduction, we can increase the critical path length if the loop we're unrolling is inside another loop. Limit, by default to 2, so the critical path only gets increased by one reduction operation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216140 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoAdd isInsertSubreg property.
Quentin Colombet [Wed, 20 Aug 2014 23:49:36 +0000 (23:49 +0000)]
Add isInsertSubreg property.

This patch adds a new property: isInsertSubreg and the related target hooks:
TargetIntrInfo::getInsertSubregInputs and
TargetInstrInfo::getInsertSubregLikeInputs to specify that a target specific
instruction is a (kind of) INSERT_SUBREG.

The approach is similar to r215394.

<rdar://problem/12702965>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216139 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoLower thumbv4t & thumbv5 lo->lo copies through a push-pop sequence
Jonathan Roelofs [Wed, 20 Aug 2014 23:38:50 +0000 (23:38 +0000)]
Lower thumbv4t & thumbv5 lo->lo copies through a push-pop sequence

On pre-v6 hardware, 'MOV lo, lo' gives undefined results, so such copies need to
be avoided. This patch trades simplicity for implementation time at the expense
of performance... As they say: correctness first, then performance.

See http://lists.cs.uiuc.edu/pipermail/llvmdev/2014-August/075998.html for a few
ideas on how to make this better.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216138 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoMention the right target hook in the comment on isExtractSubreg property.
Quentin Colombet [Wed, 20 Aug 2014 23:25:28 +0000 (23:25 +0000)]
Mention the right target hook in the comment on isExtractSubreg property.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216137 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[PeepholeOptimizer] Take advantage of the isExtractSubreg property in the
Quentin Colombet [Wed, 20 Aug 2014 23:13:02 +0000 (23:13 +0000)]
[PeepholeOptimizer] Take advantage of the isExtractSubreg property in the
advanced copy optimization.

This patch is a step toward transforming:
udiv r0, r0, r2
udiv r1, r1, r3
vmov.32 d16[0], r0
vmov.32 d16[1], r1
vmov r0, r1, d16
bx lr

into:
udiv r0, r0, r2
udiv r1, r1, r3
bx lr

Indeed, thanks to this patch, this optimization is able to look through
vmov r0, r1, d16
but it does not understand yet
vmov.32 d16[0], r0
vmov.32 d16[1], r1

Comming patches will fix that and update the related test case.

<rdar://problem/12702965>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216136 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoNew InstCombine pattern: (icmp ult/ule (A + C1), C3) | (icmp ult/ule (A + C2), C3...
Yi Jiang [Wed, 20 Aug 2014 22:55:40 +0000 (22:55 +0000)]
New InstCombine pattern: (icmp ult/ule (A + C1), C3) | (icmp ult/ule (A + C2), C3)  to (icmp ult/ule ((A & ~(C1 ^ C2)) + max(C1, C2)), C3) under certain condition

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216135 91177308-0d34-0410-b5e6-96231b3b80d8

10 years agoDon't allow MCStreamer::EmitIntValue to output 0-byte integers.
Alexey Samsonov [Wed, 20 Aug 2014 22:46:38 +0000 (22:46 +0000)]
Don't allow MCStreamer::EmitIntValue to output 0-byte integers.
It makes no sense and can hide bugs. In particular, it lead
to left shift by 64 bits, which is an undefined behavior,
properly reported by UBSan.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216134 91177308-0d34-0410-b5e6-96231b3b80d8

10 years ago[ARM] Mark VMOVRRD with the ExtractSubreg property and implement the related
Quentin Colombet [Wed, 20 Aug 2014 22:16:19 +0000 (22:16 +0000)]
[ARM] Mark VMOVRRD with the ExtractSubreg property and implement the related
target hook.

This patch teaches the compiler that:
rX, rY = VMOVRRD dZ
is the same as:
rX = EXTRACT_SUBREG dZ, ssub_0
rY = EXTRACT_SUBREG dZ, ssub_1

<rdar://problem/12702965>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216132 91177308-0d34-0410-b5e6-96231b3b80d8