oota-llvm.git
8 years ago[libFuzzer] avoid build warnings in non-assert build (useful warning in this case)
Kostya Serebryany [Wed, 5 Aug 2015 23:44:42 +0000 (23:44 +0000)]
[libFuzzer] avoid build warnings in non-assert build (useful warning in this case)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244177 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAdd a stat to show how often the limit to decompose GEPs in BasicAA is reached.
Wei Mi [Wed, 5 Aug 2015 23:40:30 +0000 (23:40 +0000)]
Add a stat to show how often the limit to decompose GEPs in BasicAA is reached.

Differential Revision: http://reviews.llvm.org/D9689

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244174 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[dsymutil] Do not create temporary files in -no-output mode.
Frederic Riss [Wed, 5 Aug 2015 23:33:50 +0000 (23:33 +0000)]
[dsymutil] Do not create temporary files in -no-output mode.

The files were never written to and then deleted, but they were created
nonetheless. To prevent that, create a wrapper around the 2 variants of
createUniqueFile and use the one that only does an access(Exists) call
to check for name unicity in -no-output mode.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244172 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[dsymutil] Update source used to generate test binary.
Frederic Riss [Wed, 5 Aug 2015 23:33:45 +0000 (23:33 +0000)]
[dsymutil] Update source used to generate test binary.

Forgot to include that in the last commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244171 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoValueMapper: Simplify remap() helper function, NFC
Duncan P. N. Exon Smith [Wed, 5 Aug 2015 23:22:34 +0000 (23:22 +0000)]
ValueMapper: Simplify remap() helper function, NFC

Rename `remap()` to `remapOperands()`, and restrict its contract to
remapping operands.  Previously, it also called `mapToMetadata()`, but
this logic is hard to reason about externally.  In particular, this
refactors `mapUniquedNode()` to avoid redundant mapping calls, taking
advantage of the RAUWs that are already in place.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244168 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agox86: NFC remove needless InstrCompiler cast
JF Bastien [Wed, 5 Aug 2015 23:15:37 +0000 (23:15 +0000)]
x86: NFC remove needless InstrCompiler cast

Summary: The casts from String to PatFrag weren't needed if we instead provided an SDNode. This fix was suggested by @pete in D11382.

Subscribers: pete, llvm-commits

Differential Revision: http://reviews.llvm.org/D11788

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244167 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[NVPTX] Use LDG for pointer induction variables.
Bjarke Hammersholt Roune [Wed, 5 Aug 2015 23:11:57 +0000 (23:11 +0000)]
[NVPTX] Use LDG for pointer induction variables.

More specifically, make NVPTXISelDAGToDAG able to emit cached loads (LDG) for pointer induction variables.

Also fix latent bug where LDG was not restricted to kernel functions. I believe that this could not be triggered so far since we do not currently infer that a pointer is global outside a kernel function, and only loads of global pointers are considered for cached loads.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244166 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[libFuzzer] in dfsan mode, set labels every time we start recording traces as opposed...
Kostya Serebryany [Wed, 5 Aug 2015 23:02:57 +0000 (23:02 +0000)]
[libFuzzer] in dfsan mode, set labels every time we start recording traces as opposed to doing it at process startup. This ensures that the labels are fresh.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244165 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAdd a TrailingObjects template class.
James Y Knight [Wed, 5 Aug 2015 22:57:34 +0000 (22:57 +0000)]
Add a TrailingObjects template class.

This is intended to help support the idiom of a class that has some
other objects (or multiple arrays of different types of objects)
appended on the end, which is used quite heavily in clang.

Differential Revision: http://reviews.llvm.org/D11272

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244164 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMake sure all temporary files get created under %T.
Artem Belevich [Wed, 5 Aug 2015 22:54:36 +0000 (22:54 +0000)]
Make sure all temporary files get created under %T.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244163 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[dsymutil] Add support for the -arch option.
Frederic Riss [Wed, 5 Aug 2015 22:33:28 +0000 (22:33 +0000)]
[dsymutil] Add support for the -arch option.

This option allows to select a subset of the architectures when
performing a universal binary link. The filter is done completely
in the mach-o specific part of the code.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244160 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoFix Windows test failure with triple instead of using the native OS
Reid Kleckner [Wed, 5 Aug 2015 22:27:08 +0000 (22:27 +0000)]
Fix Windows test failure with triple instead of using the native OS

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244159 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoIf the "CodeView" module flag is set, emit codeview instead of DWARF
Reid Kleckner [Wed, 5 Aug 2015 22:26:20 +0000 (22:26 +0000)]
If the "CodeView" module flag is set, emit codeview instead of DWARF

Summary:
Emit both DWARF and CodeView if "CodeView" and "Dwarf Version" module
flags are set.

Reviewers: majnemer

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D11756

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244158 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMIR Serialization: Serialize the machine operand's offset.
Alex Lorenz [Wed, 5 Aug 2015 22:26:15 +0000 (22:26 +0000)]
MIR Serialization: Serialize the machine operand's offset.

This commit serializes the offset for the following operands: target index,
global address, external symbol, constant pool index, and block address.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244157 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAdd <numeric> header for use of std::accumulate().
Cong Hou [Wed, 5 Aug 2015 22:13:43 +0000 (22:13 +0000)]
Add <numeric> header for use of std::accumulate().

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244156 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoDivide the primitive size in bits by eight so the initial load's alignment is in
Richard Diamond [Wed, 5 Aug 2015 22:10:57 +0000 (22:10 +0000)]
Divide the primitive size in bits by eight so the initial load's alignment is in
bytes as expected. Tested with the included unit test.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244155 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRecord whether the weights on out-edges from a MBB are normalized.
Cong Hou [Wed, 5 Aug 2015 22:01:20 +0000 (22:01 +0000)]
Record whether the weights on out-edges from a MBB are normalized.

1. Create a utility function normalizeEdgeWeights() in MachineBranchProbabilityInfo that normalizes a list of edge weights so that the sum of then can fit in uint32_t.
2. Provide an interface in MachineBasicBlock to normalize its successors' weights.
3. Add a flag in MachineBasicBlock that tracks whether its successors' weights are normalized.
4. Provide an overload of getSumForBlock that accepts a non-const pointer to a MBB so that it can force normalizing this MBB's successors' weights.
5. Update several uses of getSumForBlock() by eliminating the once needed weight scale.

Differential Revision: http://reviews.llvm.org/D11442

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244154 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[libFuzzer] add option -report_slow_units=Nsec to control when slow units are printed
Kostya Serebryany [Wed, 5 Aug 2015 21:43:48 +0000 (21:43 +0000)]
[libFuzzer] add option -report_slow_units=Nsec to control when slow units are printed

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244152 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[libFuzzer] add a missing test file
Kostya Serebryany [Wed, 5 Aug 2015 21:32:13 +0000 (21:32 +0000)]
[libFuzzer] add a missing test file

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244151 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoDrop the defaulted CallGraph move ctor, since it's unused and MSVC doesn't support...
David Blaikie [Wed, 5 Aug 2015 21:16:46 +0000 (21:16 +0000)]
Drop the defaulted CallGraph move ctor, since it's unused and MSVC doesn't support defaulted move ops

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244135 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoReplace &vector[0] with vector.data() to avoid invalid dereference caught by debug...
Nick Lewycky [Wed, 5 Aug 2015 21:16:02 +0000 (21:16 +0000)]
Replace &vector[0] with vector.data() to avoid invalid dereference caught by debug STL. Also move a '*' for consistency and fix an 80-col violation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244134 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago-Wdeprecated: Remove some dead code that was relying on a questionable (rule-of-3...
David Blaikie [Wed, 5 Aug 2015 21:15:48 +0000 (21:15 +0000)]
-Wdeprecated: Remove some dead code that was relying on a questionable (rule-of-3-violating) copy ctor in MCInstPrinter

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244133 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[LoopUnswitch] Preserve make.implicit metadata for unswitched conditions
Chen Li [Wed, 5 Aug 2015 21:13:26 +0000 (21:13 +0000)]
[LoopUnswitch] Preserve make.implicit metadata for unswitched conditions

Summary: This patch adds support to preserve make.implicit metadata for unswitched conditions in loop pre-header.

Reviewers: sanjoy, weimingz

Subscribers: mcrosier, llvm-commits

Differential Revision: http://reviews.llvm.org/D11769

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244132 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[Hexagon] Edit a comment. NFC
Krzysztof Parzyszek [Wed, 5 Aug 2015 21:08:26 +0000 (21:08 +0000)]
[Hexagon] Edit a comment. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244130 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago-Wdeprecated clean by making LogBuilder move constructible so it can be returned...
David Blaikie [Wed, 5 Aug 2015 21:06:50 +0000 (21:06 +0000)]
-Wdeprecated clean by making LogBuilder move constructible so it can be returned by value (in DifferenceEngine::logf)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244129 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agox86 atomic: optimize a.store(reg op a.load(acquire), release)
JF Bastien [Wed, 5 Aug 2015 21:04:59 +0000 (21:04 +0000)]
x86 atomic: optimize a.store(reg op a.load(acquire), release)

Summary: PR24191 finds that the expected memory-register operations aren't generated when relaxed { load ; modify ; store } is used. This is similar to PR17281 which was addressed in D4796, but only for memory-immediate operations (and for memory orderings up to acquire and release). This patch also handles some floating-point operations.

Reviewers: reames, kcc, dvyukov, nadav, morisset, chandlerc, t.p.northover, pete

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D11382

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244128 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[PM] Remove a failed attempt to port the CallGraph analysis to the new
Chandler Carruth [Wed, 5 Aug 2015 21:04:31 +0000 (21:04 +0000)]
[PM] Remove a failed attempt to port the CallGraph analysis to the new
pass manager.

This never worked, and won't ever work. It was actually why I ended up
building the LazyCallGraph set of code which is more more effectively
wired up to the new pass manager. This accidentally got committed when
I was trying to land a cleanup of the code organization in the other
parts of this file. =[ My bad, but fortunately Dave was keen eyed enough
to spot that this code couldn't possibly work. =]

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244127 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoUpdate GettingStarted docs list of LLVM_TARGETS_TO_BUILD to match cmake.
Pete Cooper [Wed, 5 Aug 2015 20:55:53 +0000 (20:55 +0000)]
Update GettingStarted docs list of LLVM_TARGETS_TO_BUILD to match cmake.

Since the docs were written, we've added the BPF backend to the list.
Updating the docs to take this in to account.  Also sorted them to
match cmake while I was changing these lines.

Reviewed by Chris B.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244123 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago-Wdeprecated cleanup: Make CallGraph movable by default by using unique_ptr members...
David Blaikie [Wed, 5 Aug 2015 20:55:50 +0000 (20:55 +0000)]
-Wdeprecated cleanup: Make CallGraph movable by default by using unique_ptr members rather than raw pointers.

The only place that tries to return a CallGraph by value
(CallGraphAnalysis::run) doesn't seem to be used right now, but it's a
reasonable bit of cleanup anyway.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244122 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRevert "Fix MO's analyzePhysReg, it was confusing sub- and super-registers. Problem...
JF Bastien [Wed, 5 Aug 2015 20:53:56 +0000 (20:53 +0000)]
Revert "Fix MO's analyzePhysReg, it was confusing sub- and super-registers. Problem pointed out by Michael Hordijk."

I mistakenly committed the patch for D6629, and was trying to commit another. Reverting until it gets proper signoff.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244121 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoFix MO's analyzePhysReg, it was confusing sub- and super-registers. Problem pointed...
JF Bastien [Wed, 5 Aug 2015 20:49:46 +0000 (20:49 +0000)]
Fix MO's analyzePhysReg, it was confusing sub- and super-registers. Problem pointed out by Michael Hordijk.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244120 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago-Wdeprecated-clean: Remove uses of throw() in favor of noexcept
David Blaikie [Wed, 5 Aug 2015 20:38:57 +0000 (20:38 +0000)]
-Wdeprecated-clean: Remove uses of throw() in favor of noexcept

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244115 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoNFC.
Cong Hou [Wed, 5 Aug 2015 20:30:11 +0000 (20:30 +0000)]
NFC.

Convert comments in MachineBasicBlock.h into new style.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244113 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago-Wdeprecated-clean: Fix cases of violating the rule of 5 in ways that are deprecated...
David Blaikie [Wed, 5 Aug 2015 20:20:29 +0000 (20:20 +0000)]
-Wdeprecated-clean: Fix cases of violating the rule of 5 in ways that are deprecated in C++11

LoadedObjectInfo was depending on the implicit copy ctor in the presence
of a user-declared dtor. Default (and protect) it in the base class and
make the devired classes final to avoid any risk of a public API that
would enable slicing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244112 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRevert "Fix `llvm-config` to emit the linker flag for the combined shared object...
Richard Diamond [Wed, 5 Aug 2015 20:03:26 +0000 (20:03 +0000)]
Revert "Fix `llvm-config` to emit the linker flag for the combined shared object built by autoconfig/make instead of the individual components."

It seems I was wrong thinking `autoconf`/`make` only installed shared libraries if configured with `--enable-shared`, even if `--disable-static` is present. I'll re-address with a followup patch.

This reverts commit r243297 for causing PR#24154.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244108 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoWrite access test.
Richard Diamond [Wed, 5 Aug 2015 19:40:39 +0000 (19:40 +0000)]
Write access test.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244103 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMIR Parser: Report an error when parsing large immediate operands.
Alex Lorenz [Wed, 5 Aug 2015 19:03:42 +0000 (19:03 +0000)]
MIR Parser: Report an error when parsing large immediate operands.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244100 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMIR Serialization: Serialize the typed immediate integer machine operands.
Alex Lorenz [Wed, 5 Aug 2015 18:52:21 +0000 (18:52 +0000)]
MIR Serialization: Serialize the typed immediate integer machine operands.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244098 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[Unroll] Switch to using 'int' cost types in preparation for a somewhat
Chandler Carruth [Wed, 5 Aug 2015 18:46:21 +0000 (18:46 +0000)]
[Unroll] Switch to using 'int' cost types in preparation for a somewhat
more involved change to the cost computation pattern.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244095 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[dsymutil] Fix test patterns.
Frederic Riss [Wed, 5 Aug 2015 18:45:13 +0000 (18:45 +0000)]
[dsymutil] Fix test patterns.

Depending on the filesystem paths, the YAML dump might quote paths.
Account for that in the regex patterns.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244094 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMIR Parser: Extract the IR constant parsing code into a new method. NFC.
Alex Lorenz [Wed, 5 Aug 2015 18:44:00 +0000 (18:44 +0000)]
MIR Parser: Extract the IR constant parsing code into a new method. NFC.

This commit extracts the code that parses the IR constant values into a new
method named 'parseIRConstant' in the 'MIParser' class. The new method will
be reused by the code that parses the typed integer immediate machine operands.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244093 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[Hexagon] Implement TargetTransformInfo for Hexagon
Krzysztof Parzyszek [Wed, 5 Aug 2015 18:35:37 +0000 (18:35 +0000)]
[Hexagon] Implement TargetTransformInfo for Hexagon

Author: Brendon Cahoon <bcahoon@codeaurora.org>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244089 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[dsymutil] Implement support for handling mach-o universal binaries as main input...
Frederic Riss [Wed, 5 Aug 2015 18:27:44 +0000 (18:27 +0000)]
[dsymutil] Implement support for handling mach-o universal binaries as main input/output.

The DWARF linker isn't touched by this, the implementation links
individual files and merges them together into a fat binary by
calling out to the 'lipo' utility.

The main change is that the MachODebugMapParser can now return
multiple debug maps for a single binary.

The test just verifies that lipo would be invoked correctly, but
doesn't actually generate a binary. This mimics the way clang
tests its external iplatform tools integration.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244087 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[dsymutil] Introduce exit helper. NFC.
Frederic Riss [Wed, 5 Aug 2015 18:27:38 +0000 (18:27 +0000)]
[dsymutil] Introduce exit helper. NFC.

llvm-dsymutil will start creating temporary files in a followup
commit. To ease the correct cleanup of this files, introduce a
helper called to exit dsymutil.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244086 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[dsymutil] Split some logic into a helper function. NFC
Frederic Riss [Wed, 5 Aug 2015 18:27:34 +0000 (18:27 +0000)]
[dsymutil] Split some logic into a helper function. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244085 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[libFuzzer] use data-flow feedback from strcmp
Kostya Serebryany [Wed, 5 Aug 2015 18:23:01 +0000 (18:23 +0000)]
[libFuzzer] use data-flow feedback from strcmp

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244084 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMIR Parser: Report an error when parsing duplicate register flags.
Alex Lorenz [Wed, 5 Aug 2015 18:09:03 +0000 (18:09 +0000)]
MIR Parser: Report an error when parsing duplicate register flags.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244081 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[TTI] Make the cost APIs in TargetTransformInfo consistently use 'int'
Chandler Carruth [Wed, 5 Aug 2015 18:08:10 +0000 (18:08 +0000)]
[TTI] Make the cost APIs in TargetTransformInfo consistently use 'int'
rather than 'unsigned' for their costs.

For something like costs in particular there is a natural "negative"
value, that of savings or saved cost. As a consequence, there is a lot
of code that subtracts or creates negative values based on cost, all of
which is prone to awkwardness or bugs when dealing with an unsigned
type. Similarly, we *never* want these values to wrap, as that would
cause Very Bad code generation (likely percieved as an infinite loop as
we try to emit over 2^32 instructions or some such insanity).

All around 'int' seems a much better fit for these basic metrics. I've
added asserts to ensure that at least the TTI interface never returns
negative numbers here. If we ever have a use case for negative numbers,
we can remove this, but this way a bug where someone used '-1' to
produce a 'very large' cost will be caught by the assert.

This passes all tests, and is also UBSan clean.

No functional change intended.

Differential Revision: http://reviews.llvm.org/D11741

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244080 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[GMR] Teach the conservative path of GMR to catch even more easy cases.
Chandler Carruth [Wed, 5 Aug 2015 17:58:30 +0000 (17:58 +0000)]
[GMR] Teach the conservative path of GMR to catch even more easy cases.

In PR24288 it was pointed out that the easy case of a non-escaping
global and something that *obviously* required an escape sometimes is
hidden behind PHIs (or selects in theory). Because we have this binary
test, we can easily just check that all possible input values satisfy
the requirement. This is done with a (very small) recursion through PHIs
and selects. With this, the specific example from the PR is correctly
folded by GVN.

Differential Revision: http://reviews.llvm.org/D11707

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244078 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMIR Serialization: Serialize the 'early-clobber' register operand flag.
Alex Lorenz [Wed, 5 Aug 2015 17:49:03 +0000 (17:49 +0000)]
MIR Serialization: Serialize the 'early-clobber' register operand flag.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244075 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMove BB succ_iterator to be inside TerminatorInst. NFC.
Pete Cooper [Wed, 5 Aug 2015 17:43:01 +0000 (17:43 +0000)]
Move BB succ_iterator to be inside TerminatorInst.  NFC.

To get the successors of a BB we currently do successors(BB) which
ultimately walks the successors of the BB's terminator.

This moves the iterator to TerminatorInst as thats what we're actually
using to do the iteration, and adds a member function to TerminatorInst
to allow us to iterate directly over successors given an instruction.

For example, we can now do

  for (auto *Succ : BI->successors())

instead of

  for (unsigned i = 0, e = BI->getNumSuccessors(); i != e; ++i)

Reviewed by Tobias Grosser.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244074 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMIR Serialization: Serialize the 'debug-use' register operand flag.
Alex Lorenz [Wed, 5 Aug 2015 17:41:17 +0000 (17:41 +0000)]
MIR Serialization: Serialize the 'debug-use' register operand flag.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244071 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMIR Parser: Simplify the handling of quoted tokens. NFC.
Alex Lorenz [Wed, 5 Aug 2015 17:35:55 +0000 (17:35 +0000)]
MIR Parser: Simplify the handling of quoted tokens. NFC.

The machine instructions lexer should not expose the difference between quoted
and unquoted tokens to the parser.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244068 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AArch64] Register AArch64DeadRegisterDefinition pass with LLVM pass manager.
Chad Rosier [Wed, 5 Aug 2015 17:35:34 +0000 (17:35 +0000)]
[AArch64] Register AArch64DeadRegisterDefinition pass with LLVM pass manager.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244067 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[Sparc] Fix disassembly of popc instruction.
James Y Knight [Wed, 5 Aug 2015 17:00:30 +0000 (17:00 +0000)]
[Sparc] Fix disassembly of popc instruction.

And add tests.

Patch by David Wiberg!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244064 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAMDGPU/SI: Remove EXECReg
Matt Arsenault [Wed, 5 Aug 2015 16:42:57 +0000 (16:42 +0000)]
AMDGPU/SI: Remove EXECReg

For the same reasons as the other physical registers.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244062 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAMDGPU: Remove SCCReg.
Matt Arsenault [Wed, 5 Aug 2015 16:42:54 +0000 (16:42 +0000)]
AMDGPU: Remove SCCReg.

These should be handled as a physical register rather
than a virtual register class with one member.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244061 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AArch64] Register (existing) AArch64BranchRelaxation pass with LLVM pass manager.
Chad Rosier [Wed, 5 Aug 2015 16:12:10 +0000 (16:12 +0000)]
[AArch64] Register (existing) AArch64BranchRelaxation pass with LLVM pass manager.

Summary: Among other things, this allows -print-after-all/-print-before-all to
dump IR around this pass.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244060 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoForce the MachO generated for Darwin to have VERSION_MIN load command
Steven Wu [Wed, 5 Aug 2015 15:36:38 +0000 (15:36 +0000)]
Force the MachO generated for Darwin to have VERSION_MIN load command

On Darwin, it is required to stamp the object file with VERSION_MIN load
command. This commit will provide a VERSRION_MIN load command to the
MachO file that doesn't specify the version itself by inferring from
Target Triple.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244059 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agotest-release.sh: Fix naming of OpenMP runtime tarball
Hans Wennborg [Wed, 5 Aug 2015 15:36:07 +0000 (15:36 +0000)]
test-release.sh: Fix naming of OpenMP runtime tarball

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244058 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AArch64] Make the naming of the Address Type Promotion pass consistent.
Chad Rosier [Wed, 5 Aug 2015 15:32:23 +0000 (15:32 +0000)]
[AArch64] Make the naming of the Address Type Promotion pass consistent.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244057 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AArch64] Register (existing) AArch64AdvSIMDScalar pass with LLVM pass manager.
Chad Rosier [Wed, 5 Aug 2015 15:18:58 +0000 (15:18 +0000)]
[AArch64] Register (existing) AArch64AdvSIMDScalar pass with LLVM pass manager.

Summary: Among other things, this allows -print-after-all/-print-before-all to
dump IR around this pass.

IIRC, this pass is off by default, but it's still helpful when debugging.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244056 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agorevert r243687: enable fast-math-flag propagation to DAG nodes
Sanjay Patel [Wed, 5 Aug 2015 15:12:03 +0000 (15:12 +0000)]
revert r243687: enable fast-math-flag propagation to DAG nodes

We can't propagate FMF partially without breaking DAG-level CSE. We either need to
relax CSE to account for mismatched FMF as a temporary work-around or fully propagate
FMF throughout the DAG.

Surprisingly, there are no existing regression tests for this, but here's an example:

  define float @fmf(float %a, float %b) {
    %mul1 = fmul fast float %a, %b
    %nega = fsub fast float 0.0, %a
    %mul2 = fmul fast float %nega, %b
    %abx2 = fsub fast float %mul1, %mul2
    ret float %abx2
  }

$ llc -o - badflags.ll -march=x86-64 -mattr=fma -enable-unsafe-fp-math -enable-fmf-dag=0
...
    vmulss    %xmm1, %xmm0, %xmm0
    vaddss    %xmm0, %xmm0, %xmm0
    retq

$ llc -o - badflags.ll -march=x86-64 -mattr=fma -enable-unsafe-fp-math -enable-fmf-dag=1
...
    vmulss    %xmm1, %xmm0, %xmm2
    vfmadd213ss    %xmm2, %xmm1, %xmm0  <--- failed to recognize that (a * b) was already calculated
    retq

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244053 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMake this less error prone by using a #define. NFC.
Chad Rosier [Wed, 5 Aug 2015 14:48:44 +0000 (14:48 +0000)]
Make this less error prone by using a #define. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244048 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AArch64] Register (existing) AArch64ExpandPseudo pass with LLVM pass manager.
Chad Rosier [Wed, 5 Aug 2015 14:22:53 +0000 (14:22 +0000)]
[AArch64] Register (existing) AArch64ExpandPseudo pass with LLVM pass manager.

Summary: Among other things, this allows -print-after-all/-print-before-all to
dump IR around this pass.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244046 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AA] Use CallSite cast idiom. No functionality change.
Benjamin Kramer [Wed, 5 Aug 2015 14:16:44 +0000 (14:16 +0000)]
[AA] Use CallSite cast idiom. No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244045 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[YAMLTraits] Use StringRef::copy. No functionality change.
Benjamin Kramer [Wed, 5 Aug 2015 14:16:38 +0000 (14:16 +0000)]
[YAMLTraits] Use StringRef::copy. No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244044 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[IR] Simplify code with ArrayRef::copy. No functionality change.
Benjamin Kramer [Wed, 5 Aug 2015 14:16:28 +0000 (14:16 +0000)]
[IR] Simplify code with ArrayRef::copy. No functionality change.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244043 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[AArch64] Register (existing) AArch64LoadStoreOpt pass with LLVM pass manager.
Chad Rosier [Wed, 5 Aug 2015 13:44:51 +0000 (13:44 +0000)]
[AArch64] Register (existing) AArch64LoadStoreOpt pass with LLVM pass manager.

Summary: Among other things, this allows -print-after-all/-print-before-all to
dump IR around this pass.

This is the AArch64 version of r243052.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244041 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoUpdate comment. NFC.
Chad Rosier [Wed, 5 Aug 2015 12:40:13 +0000 (12:40 +0000)]
Update comment. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244038 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoDrop unnecessary static_cast.
Rafael Espindola [Wed, 5 Aug 2015 11:53:59 +0000 (11:53 +0000)]
Drop unnecessary static_cast.

ELFT::TargetEndianness already has the correct type.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244032 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoARMISelDAGToDAG.cpp had this self-contradictory code:
Artyom Skrobov [Wed, 5 Aug 2015 11:02:14 +0000 (11:02 +0000)]
ARMISelDAGToDAG.cpp had this self-contradictory code:

  return StringSwitch<int>(Flags)
          .Case("g", 0x1)
          .Case("nzcvq", 0x2)
          .Case("nzcvqg", 0x3)
          .Default(-1);
...

  // The _g and _nzcvqg versions are only valid if the DSP extension is
  // available.
  if (!Subtarget->hasThumb2DSP() && (Mask & 0x2))
    return -1;

ARMARM confirms that the comment is right, and the code was wrong.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244029 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ArrayRefTest] Work around a GCC 4.8 internal compiler error.
Benjamin Kramer [Wed, 5 Aug 2015 09:39:41 +0000 (09:39 +0000)]
[ArrayRefTest] Work around a GCC 4.8 internal compiler error.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244023 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[InstCombine] Added more specific SSE2/AVX2 vector shift tests.
Simon Pilgrim [Wed, 5 Aug 2015 08:21:38 +0000 (08:21 +0000)]
[InstCombine] Added more specific SSE2/AVX2 vector shift tests.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244022 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoFixed line endings.
Simon Pilgrim [Wed, 5 Aug 2015 08:18:00 +0000 (08:18 +0000)]
Fixed line endings.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244021 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[MachineCombiner] Don't use the opcode-only form of computeInstrLatency
Hal Finkel [Wed, 5 Aug 2015 07:45:28 +0000 (07:45 +0000)]
[MachineCombiner] Don't use the opcode-only form of computeInstrLatency

In r242277, I updated the MachineCombiner to work with itineraries, but I
missed a call that is scheduling-model-only (the opcode-only form of
computeInstrLatency). Using the form that takes an MI* allows this to work with
itineraries (and should be NFC for subtargets with scheduling models).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244020 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[llvm-objdump] Call exit(1) on error, i.e. fail early.
Davide Italiano [Wed, 5 Aug 2015 07:18:31 +0000 (07:18 +0000)]
[llvm-objdump] Call exit(1) on error, i.e. fail early.

Previously we kept going on partly corrupted input, which might result
in garbage being printed, or even worse, random crashes.
Rafael mentioned that this is the GNU behavior as well, but after some
discussion we both agreed it's probably better to emit a reasonable
error message and exit. As a side-effect of this commit, now we don't
rely on global state for error codes anymore. objdump was the last tool
in the toolchain which needed to be converted. Hopefully the old behavior
won't sneak into the tree again.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244019 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agounittests/ADT/ArrayRefTest.cpp: Suppress r243995 on g++-4.8 for now to unbreak bots.
NAKAMURA Takumi [Wed, 5 Aug 2015 06:11:23 +0000 (06:11 +0000)]
unittests/ADT/ArrayRefTest.cpp: Suppress r243995 on g++-4.8 for now to unbreak bots.

For example of mingw-w64-g++-4.8.1,

  llvm/unittests/ADT/ArrayRefTest.cpp: In member function 'virtual void {anonymous}::ArrayRefTest_AllocatorCopy_Test::TestBody()':
  llvm/unittests/ADT/ArrayRefTest.cpp:56:40: internal compiler error: in count_type_elements, at expr.c:5523
     } Array3Src[] = {{"hello"}, {"world"}};
                                          ^
  Please submit a full bug report,
  with preprocessed source if appropriate.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244017 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoWindows/COM.inc: Fix emacs mode in the first line.
NAKAMURA Takumi [Wed, 5 Aug 2015 06:11:08 +0000 (06:11 +0000)]
Windows/COM.inc: Fix emacs mode in the first line.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244016 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoTemporarily revert r244012 while we see if it's really necessary.
Eric Christopher [Wed, 5 Aug 2015 05:32:22 +0000 (05:32 +0000)]
Temporarily revert r244012 while we see if it's really necessary.
Add a comment explaining the current theory as to why we'd need
the -lole32 on the link line.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244015 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRemove autoconf checks for sigsetjmp and siglongjmp as they appear to
Eric Christopher [Wed, 5 Aug 2015 04:50:46 +0000 (04:50 +0000)]
Remove autoconf checks for sigsetjmp and siglongjmp as they appear to
be unused.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244014 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRemove the apparently unused rand48 configure checks and associated m4.
Eric Christopher [Wed, 5 Aug 2015 04:50:43 +0000 (04:50 +0000)]
Remove the apparently unused rand48 configure checks and associated m4.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244013 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRemove obsolete configure check for libole32 as we don't appear to use
Eric Christopher [Wed, 5 Aug 2015 04:50:40 +0000 (04:50 +0000)]
Remove obsolete configure check for libole32 as we don't appear to use
it anymore.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244012 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRemove the unused header check for utime.h.
Eric Christopher [Wed, 5 Aug 2015 04:50:37 +0000 (04:50 +0000)]
Remove the unused header check for utime.h.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@244011 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRename all references to old mailing lists to new lists.llvm.org address.
Tanya Lattner [Wed, 5 Aug 2015 03:51:17 +0000 (03:51 +0000)]
Rename all references to old mailing lists to new lists.llvm.org address.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243999 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoAvoid passing nullptr to std::equal.
Yaron Keren [Tue, 4 Aug 2015 15:57:04 +0000 (15:57 +0000)]
Avoid passing nullptr to std::equal.
As documented in the LLVM Coding Standards, indeed MSVC incorrectly asserts
on this in Debug mode. This happens when building clang with Visual C++ and
-triple i686-pc-windows-gnu on these clang regression tests:

 clang/test/CodeGen/2011-03-08-ZeroFieldUnionInitializer.c
 clang/test/CodeGen/empty-union-init.c

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243996 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[ArrayRef] Make copy use std::uninitialized_copy.
Benjamin Kramer [Tue, 4 Aug 2015 15:52:56 +0000 (15:52 +0000)]
[ArrayRef] Make copy use std::uninitialized_copy.

std::copy does not work for non-trivially copyable classes when we're
copying into uninitialized memory.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243995 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agowrap OptSize and MinSize attributes for easier and consistent access (NFCI)
Sanjay Patel [Tue, 4 Aug 2015 15:49:57 +0000 (15:49 +0000)]
wrap OptSize and MinSize attributes for easier and consistent access (NFCI)

Create wrapper methods in the Function class for the OptimizeForSize and MinSize
attributes. We want to hide the logic of "or'ing" them together when optimizing
just for size (-Os).

Currently, we are not consistent about this and rely on a front-end to always set
OptimizeForSize (-Os) if MinSize (-Oz) is on. Thus, there are 18 FIXME changes here
that should be added as follow-on patches with regression tests.

This patch is NFC-intended: it just replaces existing direct accesses of the attributes
by the equivalent wrapper call.

Differential Revision: http://reviews.llvm.org/D11734

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243994 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[RuntimeDyld] Adapt PPC64 relocations to PPC32
Hal Finkel [Tue, 4 Aug 2015 15:29:00 +0000 (15:29 +0000)]
[RuntimeDyld] Adapt PPC64 relocations to PPC32

Begin adapting some of the implemented PPC64 relocations for PPC32 (with a
test case).

Patch by Pierre-Andre Saulais!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243991 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[x86] machine combiner reassociation: mark EFLAGS operand as 'dead'
Sanjay Patel [Tue, 4 Aug 2015 15:21:56 +0000 (15:21 +0000)]
[x86] machine combiner reassociation: mark EFLAGS operand as 'dead'

In the commentary for D11660, I wasn't sure if it was alright to create new
integer machine instructions without also creating the implicit EFLAGS operand.
From what I can see, the implicit operand is always created by the MachineInstrBuilder
based on the instruction type, so we don't have to do that explicitly. However, in
reviewing the debug output, I noticed that the operand was not marked as 'dead'.
The machine combiner should do that to preserve future optimization opportunities
that may be checking for that dead EFLAGS operand themselves.

Differential Revision: http://reviews.llvm.org/D11696

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243990 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[mips][FastISel] Disable code generation for unsupported targets through FastISel.
Vasileios Kalintiris [Tue, 4 Aug 2015 14:35:50 +0000 (14:35 +0000)]
[mips][FastISel] Disable code generation for unsupported targets through FastISel.

Summary:
Previously, we would check whether the target is supported or not, only in
fastSelectInstruction(). This means that 64-bit targets could use FastISel too.
We fix this by checking every overridden method of the FastISel class and
by falling back to SelectionDAG if the target isn't supported. This change
should have been committed along with r243638, but somehow I missed it.

Reviewers: dsanders

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D11755

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243986 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoRevert r229675 - [mips] Avoid redundant sign extension of the result of binary bitwis...
Vasileios Kalintiris [Tue, 4 Aug 2015 14:26:35 +0000 (14:26 +0000)]
Revert r229675 - [mips] Avoid redundant sign extension of the result of binary bitwise instructions.

It introduced two regressions on 64-bit big-endian targets running under N32
(MultiSource/Benchmarks/tramp3d-v4/tramp3d-v4, and
MultiSource/Applications/kimwitu++/kc) The issue is that on 64-bit targets
comparisons such as BEQ compare the whole GPR64 but incorrectly tell the
instruction selector that they operate on GPR32's. This leads to the
elimination of i32->i64 extensions that are actually required by
comparisons to work correctly.

There's currently a patch under review that fixes this problem.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243984 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoFix 80-column
Duncan P. N. Exon Smith [Tue, 4 Aug 2015 13:24:26 +0000 (13:24 +0000)]
Fix 80-column

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243977 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoLinker: Fix ASan failure from r243961
Duncan P. N. Exon Smith [Tue, 4 Aug 2015 13:23:30 +0000 (13:23 +0000)]
Linker: Fix ASan failure from r243961

r243883 and r243961 made a use-after-free far more likely:
http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-fast/builds/6041/steps/check-llvm%20asan/logs/stdio

Unresolved nodes get inserted into the `Cycles` array.  If they later
get resolved through RAUW, we need to update the reference.  It's
interesting that this never hit before (maybe an asan-ified clang
bootstrap with `-flto -g` would have hit it, but I admit I haven't tried
anything quite that crazy).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243976 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoDrive-by fixes for LandingPad -> EHPad
David Majnemer [Tue, 4 Aug 2015 08:21:40 +0000 (08:21 +0000)]
Drive-by fixes for LandingPad -> EHPad

This change was done as an audit and is by inspection.  The new EH
system is still very much a work in progress.  NFC for the landingpad
case.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243965 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[InstCombine] Split off SSE2/AVX2 vector shift tests.
Simon Pilgrim [Tue, 4 Aug 2015 08:05:27 +0000 (08:05 +0000)]
[InstCombine] Split off SSE2/AVX2 vector shift tests.

These aren't vector demanded bits tests. More tests to follow.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243963 91177308-0d34-0410-b5e6-96231b3b80d8

8 years ago[InstCombine] Moved SSE vector shift constant folding into its own helper function...
Simon Pilgrim [Tue, 4 Aug 2015 07:49:58 +0000 (07:49 +0000)]
[InstCombine] Moved SSE vector shift constant folding into its own helper function. NFCI.

This will make some upcoming bugfixes + improvements easier to manage.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243962 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoLinker: Fix references to uniqued nodes after r243883
Duncan P. N. Exon Smith [Tue, 4 Aug 2015 06:42:31 +0000 (06:42 +0000)]
Linker: Fix references to uniqued nodes after r243883

r243883 started moving 'distinct' nodes instead of duplicated them in
lib/Linker.  This had the side-effect of sometimes not cloning uniqued
nodes that reference them.  I missed a corner case:

    !named = !{!0}
    !0 = !{!1}
    !1 = distinct !{!0}

!0 is the entry point for "remapping", and a temporary clone (say,
!0-temp) is created and mapped in case we need to model a uniquing
cycle.

    Recursive descent into !1.  !1 is distinct, so we leave it alone,
    but update its operand to !0-temp.

Pop back out to !0.  Its only operand, !1, hasn't changed, so we don't
need to use !0-temp.  !0-temp goes out of scope, and we're finished
remapping, but we're left with:

    !named = !{!0}
    !0 = !{!1}
    !1 = distinct !{null} ; uh oh...

Previously, if !0 and !0-temp ended up with identical operands, then
!0-temp couldn't have been referenced at all.  Now that distinct nodes
don't get duplicated, that assumption is invalid.  We need to
!0-temp->replaceAllUsesWith(!0) before freeing !0-temp.

I found this while running an internal `-flto -g` bootstrap.  Strangely,
there was no case of this in the open source bootstrap I'd done before
commit...

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243961 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoUpdate test suite to make "ninja check" succeed without native backend builtin
Mehdi Amini [Tue, 4 Aug 2015 06:32:54 +0000 (06:32 +0000)]
Update test suite to make "ninja check" succeed without native backend builtin

Requires "native" feature in most places that were failing.

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

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243960 91177308-0d34-0410-b5e6-96231b3b80d8

8 years agoMove generic MIR tests in their own subdir, requires "native" as well
Mehdi Amini [Tue, 4 Aug 2015 06:32:45 +0000 (06:32 +0000)]
Move generic MIR tests in their own subdir, requires "native" as well

These tests rely on the native backend to be built-in.

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

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@243959 91177308-0d34-0410-b5e6-96231b3b80d8