Graydon Hoare [Mon, 3 Apr 2017 18:04:15 +0000 (18:04 +0000)]
[Support] Make printAllJSONValues public, for custom output.
Summary:
This changes the static method TimerGroup::printAllJSONValues from private to
public, to match the static method TimerGroup::printAll. When trying to drive
the reporting machinery by hand, the existing API is _almost_ flexible enough,
but this entrypoint is required to intermix printing timers with other
non-timer output.
The underlying motive here is a Swift change to consolidate the collection of
timers, LLVM statistics and other (non-assert-dependent) counters into JSON
files, which requires a bit of manual intervention in LLVM's stat and timer
output routines. See https://github.com/apple/swift/pull/8477 for details.
Reviewers: MatzeB
Reviewed By: MatzeB
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31566
llvm-svn: 299371
Peter Collingbourne [Mon, 3 Apr 2017 17:58:48 +0000 (17:58 +0000)]
Bitcode: Remove reader support for MODULE_CODE_PURGEVALS.
Support for writing this module code was removed in r73220, which was well
before the LLVM 3.0 release, so we do not need to be able to understand it
for backwards compatibility.
Differential Revision: https://reviews.llvm.org/D31563
llvm-svn: 299370
Craig Topper [Mon, 3 Apr 2017 17:49:15 +0000 (17:49 +0000)]
[InstCombine] Add test cases showing how we fail to fold vector constants into selects the way we do with scalars.
llvm-svn: 299369
Zvi Rackover [Mon, 3 Apr 2017 17:41:19 +0000 (17:41 +0000)]
Revert "[DAGCombine] A shuffle of a splat is always the splat itself"
This reverts commit r299047 which is incorrect because the
simplification may result in incorrect propogation of undefs to users of
the folded shuffle.
Thanks to Andrea Di Biagio for pointing this out.
llvm-svn: 299368
Krzysztof Parzyszek [Mon, 3 Apr 2017 17:26:40 +0000 (17:26 +0000)]
[Hexagon] Factor out some common code in HexagonEarlyIfConv.cpp, NFC
llvm-svn: 299367
Craig Topper [Mon, 3 Apr 2017 17:22:23 +0000 (17:22 +0000)]
Revert r299337 "[InstCombine] Remove redundant combine from visitAnd"
One of the tsan bots started failing at this commit. I don't see anything obviously wrong with the commit so trying this to see if it recovers.
Failing log: http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux-autoconf/builds/6792
llvm-svn: 299366
Sanjay Patel [Mon, 3 Apr 2017 16:53:12 +0000 (16:53 +0000)]
[InstCombine] fix formatting for foldLogOpOfMaskedICmps and related bits; NFCI
1. Improve enum, function, and variable names.
2. Improve comments.
3. Fix variable capitalization.
4. Run clang-format.
As an existing code comment suggests, this should work with vector types / splat constants too,
so making this look right first will reduce the diffs needed for that change.
llvm-svn: 299365
Davide Italiano [Mon, 3 Apr 2017 16:51:39 +0000 (16:51 +0000)]
[TargetInfo] Use llvm::alignOf() instead of rewriting it. NFCI.
llvm-svn: 299364
Alex Lorenz [Mon, 3 Apr 2017 16:43:21 +0000 (16:43 +0000)]
Enhance -Wshadow to warn when shadowing typedefs or type aliases
Enhance -Wshadow to emit a warning when typedefs or type aliases are shadowed.
Fixes bug https://bugs.llvm.org//show_bug.cgi?id=28676.
Patch by Ahmed Asadi.
Differential Revision: https://reviews.llvm.org/D31235
llvm-svn: 299363
Craig Topper [Mon, 3 Apr 2017 16:34:59 +0000 (16:34 +0000)]
[APInt] Move isMask and isShiftedMask out of APIntOps and into the APInt class. Implement them without memory allocation for multiword
This moves the isMask and isShiftedMask functions to be class methods. They now use the MathExtras.h function for single word size and leading/trailing zeros/ones or countPopulation for the multiword size. The previous implementation made multiple temorary memory allocations to do the bitwise arithmetic operations to match the MathExtras.h implementation.
Differential Revision: https://reviews.llvm.org/D31565
llvm-svn: 299362
Simon Pilgrim [Mon, 3 Apr 2017 15:27:49 +0000 (15:27 +0000)]
[DAGCombiner] Check limits before accessing array element (PR32502)
llvm-svn: 299361
Tobias Grosser [Mon, 3 Apr 2017 15:23:08 +0000 (15:23 +0000)]
[PerfMonitor] Use Intrinsics::getDeclaration
Instead of creating the declaration ourselves, we obtain it directly from the
LLVM intrinsic definitions. This addresses a post-review comment for r299359.
Suggested-by: Hongzing Zheng <etherzhhb@gmail.com>
llvm-svn: 299360
Tobias Grosser [Mon, 3 Apr 2017 14:55:37 +0000 (14:55 +0000)]
[CodeGen] Add Performance Monitor
Add support for -polly-codegen-perf-monitoring. When performance monitoring
is enabled, we emit performance monitoring code during code generation that
prints after program exit statistics about the total number of cycles executed
as well as the number of cycles spent in scops. This gives an estimate on how
useful polyhedral optimizations might be for a given program.
Example output:
Polly runtime information
-------------------------
Total:
783110081637
Scops:
663718949365
In the future, we might also add functionality to measure how much time is spent
in optimized scops and how many cycles are spent in the fallback code.
Reviewers: bollu,sebpop
Tags: #polly
Differential Revision: https://reviews.llvm.org/D31599
llvm-svn: 299359
Sjoerd Meijer [Mon, 3 Apr 2017 14:50:04 +0000 (14:50 +0000)]
ARMAsmParser: clean up of isImmediate functions
- we are now using immediate AsmOperands so that the range check functions are
tablegen'ed.
- Big bonus is that error messages become much more accurate, i.e. instead of a
useless "invalid operand" error message it will not say that the immediate
operand must in range [x,y], which is why regression tests needed updating.
More tablegen operand descriptions could probably benefit from using
immediateAsmOperand, but this is a first good step to get rid of most of the
nearly identical range check functions. I will address the remaining immediate
operands in next clean ups.
Differential Revision: https://reviews.llvm.org/D31333
llvm-svn: 299358
Michael Kruse [Mon, 3 Apr 2017 12:37:10 +0000 (12:37 +0000)]
[test] Fix two testcases. NFC.
Trivial fix for two testcases. When Polly isn't linked into opt,
independent of whether it's built in-tree or not, these testcases forget
to load the appropriate library.
Contributed-by: Philip Pfaffe <philip.pfaffe@gmail.com>
Differential Revision: https://reviews.llvm.org/D31596
llvm-svn: 299357
Michael Kruse [Mon, 3 Apr 2017 12:03:38 +0000 (12:03 +0000)]
[ScopInfo] Fix typos in option description.
llvm-svn: 299356
Gabor Horvath [Mon, 3 Apr 2017 11:57:11 +0000 (11:57 +0000)]
[ASTImporter] Fix for importing unnamed structs
Patch by Peter Szecsi!
Differential Revision: https://reviews.llvm.org/D30876
llvm-svn: 299355
Gabor Horvath [Mon, 3 Apr 2017 09:30:20 +0000 (09:30 +0000)]
[ASTImporter] Lookup SearchName instead of Name
When the SearchName is already calculated we should use that for the lookup.
Patch by Peter Szecsi!
Differential Revision: https://reviews.llvm.org/D30877
llvm-svn: 299354
Tobias Grosser [Mon, 3 Apr 2017 07:44:23 +0000 (07:44 +0000)]
Add test case for r299352.
llvm-svn: 299353
Tobias Grosser [Mon, 3 Apr 2017 07:42:50 +0000 (07:42 +0000)]
[PollyIRBuilder] Bound size of alias metadata
No-alias metadata grows quadratic in the size of arrays involved, which can
become very costly for large programs. This commit bounds the number of arrays
for which we construct no-alias information to ten. This is conservatively
correct, as we just provide less information to LLVM and speeds up the compile
time of one of my internal test cases from 'does-not-terminate' to
'finishes-in-less-than-a-minute'. In the future we might try to be more clever
here, but this change should provide a good baseline.
llvm-svn: 299352
Craig Topper [Mon, 3 Apr 2017 07:08:08 +0000 (07:08 +0000)]
[InstCombine] Make foldOpWithConstantIntoOperand take a BinaryOperator instead of a generic Instruction.
It blindly assumes there are two operands so make it explicit.
llvm-svn: 299351
Tobias Grosser [Mon, 3 Apr 2017 06:46:16 +0000 (06:46 +0000)]
Update to isl-0.18-410-gc253447
This is a regular maintenance update to ensure latest isl changes are tested
in our buildbots.
llvm-svn: 299350
Craig Topper [Mon, 3 Apr 2017 06:02:09 +0000 (06:02 +0000)]
[InstCombine] Remove a And transform that should be handled by SimplifyDemandedInstructionBits. NFCI
llvm-svn: 299349
Shoaib Meenai [Mon, 3 Apr 2017 04:04:24 +0000 (04:04 +0000)]
[libc++] Explicitly mark specializations as dllexport
Method specializations don't get exported even if there's an exported
extern template instantiation on Windows. Explicitly mark the methods
for export. They're already exported on Linux and Darwin, so there's no
ABI change on those platforms.
llvm-svn: 299348
Craig Topper [Mon, 3 Apr 2017 03:51:57 +0000 (03:51 +0000)]
[AVX-512] Fix a couple more intrinsic macros I missed in r299346.
llvm-svn: 299347
Craig Topper [Mon, 3 Apr 2017 03:41:29 +0000 (03:41 +0000)]
[AVX-512] Fix some intrinsic macros that use the wrong macro parameter names and don't have parentheses around them.
Thanks to Matthew Barr for reporting this issue.
llvm-svn: 299346
Jason Molenda [Mon, 3 Apr 2017 00:59:47 +0000 (00:59 +0000)]
Back out temporary masking of EXC_SYSCALL mach exceptions.
<rdar://problem/
31359720>
llvm-svn: 299345
NAKAMURA Takumi [Sun, 2 Apr 2017 23:57:17 +0000 (23:57 +0000)]
Trailing whitespace.
llvm-svn: 299344
NAKAMURA Takumi [Sun, 2 Apr 2017 23:57:10 +0000 (23:57 +0000)]
Reformat.
llvm-svn: 299343
Craig Topper [Sun, 2 Apr 2017 19:35:18 +0000 (19:35 +0000)]
[APInt] Make use of whichWord and maskBit to simplify some code. NFC
llvm-svn: 299342
Craig Topper [Sun, 2 Apr 2017 19:17:22 +0000 (19:17 +0000)]
[APInt] Add a public typedef for the internal type of APInt use it instead of integerPart. Make APINT_BITS_PER_WORD and APINT_WORD_SIZE public.
This patch is one step to attempt to unify the main APInt interface and the tc functions used by APFloat.
This patch adds a WordType to APInt and uses that in all the tc functions. I've added temporary typedefs to APFloat to alias it to integerPart to keep the patch size down. I'll work on removing that in a future patch.
In future patches I hope to reuse the tc functions to implement some of the main APInt functionality.
I may remove APINT_ from BITS_PER_WORD and WORD_SIZE constants so that we don't have the repetitive APInt::APINT_ externally.
Differential Revision: https://reviews.llvm.org/D31523
llvm-svn: 299341
Krystyna Gajczyk [Sun, 2 Apr 2017 19:12:20 +0000 (19:12 +0000)]
Fixes for modernize-use-using check:
- removed unnessacary namespaces
- added option to print warning in macros
- no fix for typedef with array
- removed "void" word from functions with 0 parameters
Differential Revision: https://reviews.llvm.org/D29262
llvm-svn: 299340
Craig Topper [Sun, 2 Apr 2017 17:57:30 +0000 (17:57 +0000)]
[InstCombine] Make InstCombiner::OptAndOp take a BinaryOperator instead of an Instruction.
The callers have already performed the necessary cast before calling. This allows us to remove a comment that says the instruction must be a BinaryOperator and make it explicit in the argument type.
Had to add a default case to the switch because BinaryOperator::getOpcode() returns a BinaryOps enum.
llvm-svn: 299339
Simon Pilgrim [Sun, 2 Apr 2017 17:45:41 +0000 (17:45 +0000)]
[X86][MMX] Improve support for folding fptosi from XMM to MMX
llvm-svn: 299338
Craig Topper [Sun, 2 Apr 2017 17:34:30 +0000 (17:34 +0000)]
[InstCombine] Remove redundant combine from visitAnd
As far as I can tell this combine is fully handled by SimplifyDemandedInstructionBits.
I was only looking at this because it is the only user of APIntOps::isShiftedMask which is itself broken. As demonstrated by r299187. I was going to fix isShiftedMask and needed to make sure we had coverage for the new cases it would expose to this combine. But looks like we can nuke it instead.
Differential Revision: https://reviews.llvm.org/D31543
llvm-svn: 299337
Simon Pilgrim [Sun, 2 Apr 2017 16:20:34 +0000 (16:20 +0000)]
[X86][MMX] Simplify tablegen patterns by always combining MOVDQ2Q from v2i64
llvm-svn: 299336
Simon Pilgrim [Sun, 2 Apr 2017 15:52:28 +0000 (15:52 +0000)]
[X86][MMX] Added support for subvector extraction to MMX register
llvm-svn: 299335
NAKAMURA Takumi [Sun, 2 Apr 2017 15:05:18 +0000 (15:05 +0000)]
APInt.h: Prune \param(s) in \returns. [-Wdocumentation]
llvm-svn: 299334
Simon Pilgrim [Sun, 2 Apr 2017 14:21:14 +0000 (14:21 +0000)]
Regenerate test with codegen. NFCI.
llvm-svn: 299333
Simon Pilgrim [Sun, 2 Apr 2017 13:59:37 +0000 (13:59 +0000)]
Regenerate test with codegen. NFCI.
llvm-svn: 299332
Simon Pilgrim [Sun, 2 Apr 2017 13:50:44 +0000 (13:50 +0000)]
Regenerate test. NFCI.
llvm-svn: 299331
Daniel Berlin [Sun, 2 Apr 2017 13:23:44 +0000 (13:23 +0000)]
NewGVN: Handle coercion of constant stores, loads, memory insts.
Summary:
Depends on D30928.
This adds support for coercion of stores and memory instructions that do not require insertion to process.
Another few tests down.
I added the relevant tests from rle.ll
Reviewers: davide
Subscribers: llvm-commits, Prazek
Differential Revision: https://reviews.llvm.org/D30929
llvm-svn: 299330
Nikolai Bozhenov [Sun, 2 Apr 2017 13:14:30 +0000 (13:14 +0000)]
[BypassSlowDivision] Do not bypass division of hash-like values
Disable bypassing if one of the operands looks like a hash value. Slow
division often occurs in hashtable implementations and fast division is
never taken there because a hash value is extremely unlikely to have
enough upper bits set to zero.
A value is considered to be hash-like if it is produced by
1) XOR operation
2) Multiplication by a constant wider than the shorter type
3) PHI node with all incoming values being hash-like
Differential Revision: https://reviews.llvm.org/D28200
llvm-svn: 299329
Simon Pilgrim [Sun, 2 Apr 2017 13:10:20 +0000 (13:10 +0000)]
[X86][MMX] Add generic fptosi 4f32-4i32 test
llvm-svn: 299328
Zvi Rackover [Sun, 2 Apr 2017 10:42:21 +0000 (10:42 +0000)]
Add another interesting shufflevector test case for InstSimplify. NFC.
Test case shows opportunity to constant fold a shuffle with one variable
input vector operand.
llvm-svn: 299327
Craig Topper [Sun, 2 Apr 2017 06:59:43 +0000 (06:59 +0000)]
[X86] Use tcAdd/tcSubtract to implement the slow case of operator+=/operator-=.
llvm-svn: 299326
Craig Topper [Sun, 2 Apr 2017 06:59:41 +0000 (06:59 +0000)]
[APInt] Combine declaration and initialization. NFC
llvm-svn: 299325
Craig Topper [Sun, 2 Apr 2017 06:59:38 +0000 (06:59 +0000)]
[APInt] Simplify some code by using operator+=(uint64_t) instead of doing a more complex assignment into a temporary APInt just to use the APInt operator+=.
llvm-svn: 299324
Craig Topper [Sun, 2 Apr 2017 06:59:36 +0000 (06:59 +0000)]
[APInt] Fix typo in comment. NFC
llvm-svn: 299323
Daniel Berlin [Sun, 2 Apr 2017 05:09:15 +0000 (05:09 +0000)]
MemorySSA: Add support for caching clobbering access in stores
Summary:
This enables us to cache the clobbering access for stores, despite the
fact that we can't rewrite the use-def chains themselves.
Early testing shows that, after this change, for larger testcases, it will be a significant net positive (memory and time) to remove the walker caching.
Reviewers: george.burgess.iv, davide
Subscribers: Prazek, llvm-commits
Differential Revision: https://reviews.llvm.org/D31567
llvm-svn: 299322
Craig Topper [Sun, 2 Apr 2017 03:02:53 +0000 (03:02 +0000)]
[X86] Add __extension__ to f16c macro intrinsics to suppress warnings about compound literals when compiled for with earlier language standards enabled.
Fixes PR32491.
llvm-svn: 299321
Craig Topper [Sat, 1 Apr 2017 21:50:10 +0000 (21:50 +0000)]
[APInt] Use conditional operator to simplify some code. NFC
llvm-svn: 299320
Craig Topper [Sat, 1 Apr 2017 21:50:08 +0000 (21:50 +0000)]
[APInt] Implement flipAllBitsSlowCase with tcComplement. NFCI
llvm-svn: 299319
Craig Topper [Sat, 1 Apr 2017 21:50:06 +0000 (21:50 +0000)]
[APInt] Fix indentation. NFC
llvm-svn: 299318
Craig Topper [Sat, 1 Apr 2017 21:50:03 +0000 (21:50 +0000)]
[APInt] Implement AndAssignSlowCase using tcAnd. Do the same for Or and Xor. NFCI
llvm-svn: 299317
Faisal Vali [Sat, 1 Apr 2017 21:30:49 +0000 (21:30 +0000)]
[NFC, Scoped Enum] Convert Sema::ExpressionEvaluationContext into a scoped Enum
- also replace direct equality checks against the ConstantEvaluated enumerator with isConstantEvaluted(), in anticipation of adding finer granularity to the various ConstantEvaluated contexts and reinstating certain restrictions on where lambda expressions can occur in C++17.
- update the clang tablegen backend that uses these Enumerators, and add the relevant scope where needed.
llvm-svn: 299316
Davide Italiano [Sat, 1 Apr 2017 21:07:07 +0000 (21:07 +0000)]
[Driver] Don't crash on invalid values of -mrelocation-model=.
This is handled in a similar way we handle invalid -mcode-model.
PR: 31840
llvm-svn: 299315
Craig Topper [Sat, 1 Apr 2017 20:30:57 +0000 (20:30 +0000)]
[APInt] Allow GreatestCommonDivisor to take rvalue inputs efficiently. Use moves instead of copies in the loop.
Summary:
GreatestComonDivisor currently makes a copy of both its inputs. Then in the loop we do one move and two copies, plus any allocation the urem call does.
This patch changes it to take its inputs by value so that we can do a move of any rvalue inputs instead of copying. Then in the loop we do 3 move assignments and no copies. This way the only possible allocations we have in the loop is from the urem call.
Reviewers: dblaikie, RKSimon, hans
Reviewed By: dblaikie
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31572
llvm-svn: 299314
Davide Italiano [Sat, 1 Apr 2017 19:47:52 +0000 (19:47 +0000)]
[WASM] Remove other comparison of unsigned expression >= 0.
This should finally fix the GCC 7 build with -Werror.
llvm-svn: 299313
Davide Italiano [Sat, 1 Apr 2017 19:40:51 +0000 (19:40 +0000)]
[WASM] Remove a set but never used variable.
llvm-svn: 299312
Davide Italiano [Sat, 1 Apr 2017 19:37:15 +0000 (19:37 +0000)]
[WASM] Remove an assertion that can never fire.
uint* is by definition always >=0.
llvm-svn: 299311
Davide Italiano [Sat, 1 Apr 2017 19:30:17 +0000 (19:30 +0000)]
[AMDGPU] Garbage collect now unused dead code. NFCI.
llvm-svn: 299310
Sanjay Patel [Sat, 1 Apr 2017 19:05:11 +0000 (19:05 +0000)]
[InstSimplify] add constant folding for fdiv/frem
Also, add a helper function so we don't have to repeat this code for each binop.
llvm-svn: 299309
Sanjay Patel [Sat, 1 Apr 2017 18:44:03 +0000 (18:44 +0000)]
[InstSimplify] add tests for missed constant folding; NFC
llvm-svn: 299308
Sanjay Patel [Sat, 1 Apr 2017 18:40:30 +0000 (18:40 +0000)]
fix formatting; NFC
llvm-svn: 299307
Benjamin Kramer [Sat, 1 Apr 2017 17:59:01 +0000 (17:59 +0000)]
[ObjC++] Use the correct EH personality in GNU mode
Previously, it would just always use the ObjC DWARF personality, even with SjLj
or SEH exceptions.
Patch by Jonathan Schleifer, test case by me.
llvm-svn: 299306
Sanjay Patel [Sat, 1 Apr 2017 15:53:12 +0000 (15:53 +0000)]
fix formatting; NFC
llvm-svn: 299305
Sanjay Patel [Sat, 1 Apr 2017 15:05:54 +0000 (15:05 +0000)]
[DAGCombiner] enable vector transforms for any/all {sign} bits set/clear
The code already allowed vector types in via "isInteger" (which might want
a more specific name), so use splat-friendly constant predicates to match
those types.
llvm-svn: 299304
Sanjay Patel [Sat, 1 Apr 2017 14:32:18 +0000 (14:32 +0000)]
[PowerPC, x86] add vector tests for any/all {sign} bits set/clear; NFC
llvm-svn: 299303
Daniel Berlin [Sat, 1 Apr 2017 10:04:28 +0000 (10:04 +0000)]
MemorySSA: Update expensive checking version of def_chain_iterator for templating changes
llvm-svn: 299301
Daniel Berlin [Sat, 1 Apr 2017 09:44:33 +0000 (09:44 +0000)]
NewGVN: Don't try to kill off the stored value of stores when
processing the congruence class of the store.
Because we use the stored value of a store as the def, it isn't dead
just because it appears as a def when it comes from a store.
Note: I have not hit any cases with the memory code as it is where
this breaks anything, just because of what memory congruences we
actually allow. In a followup that improves memory congruence,
this bug actually breaks real stuff (but the verifier catches it).
llvm-svn: 299300
Daniel Berlin [Sat, 1 Apr 2017 09:44:29 +0000 (09:44 +0000)]
NewGVN: Clean up GVNExpression memory hierarchy, restructure hash computation a bit so we don't have to redefine it for loads, stores, and calls
llvm-svn: 299299
Daniel Berlin [Sat, 1 Apr 2017 09:44:24 +0000 (09:44 +0000)]
NewGVN: Use def_chain iterator in singleReachablePhiPath instead of recursion
llvm-svn: 299298
Daniel Berlin [Sat, 1 Apr 2017 09:44:19 +0000 (09:44 +0000)]
Move def_chain iterator to MemorySSA.h so it can be reused
llvm-svn: 299297
Daniel Berlin [Sat, 1 Apr 2017 09:44:14 +0000 (09:44 +0000)]
MemorySSA.h - make clang-format happy
llvm-svn: 299296
Daniel Berlin [Sat, 1 Apr 2017 09:01:12 +0000 (09:01 +0000)]
MemorySSA: Push const correctness further.
llvm-svn: 299295
Daniel Berlin [Sat, 1 Apr 2017 08:59:45 +0000 (08:59 +0000)]
MemorySSA: Kill the WalkTargetCache now that we have getBlockDefs.
llvm-svn: 299294
Craig Topper [Sat, 1 Apr 2017 06:50:00 +0000 (06:50 +0000)]
[APInt] Implement operator! using operator==(uint64_t). NFCI
llvm-svn: 299293
Craig Topper [Sat, 1 Apr 2017 05:08:57 +0000 (05:08 +0000)]
[APInt] Remove the mul/urem/srem/udiv/sdiv functions from the APIntOps namespace. Replace the few usages with calls to the class methods. NFC
llvm-svn: 299292
Craig Topper [Sat, 1 Apr 2017 04:26:20 +0000 (04:26 +0000)]
[DAGCombiner] Fix fold (or (shuf A, V_0, MA), (shuf B, V_0, MB)) -> (shuf A, B, Mask) to explicitly ensure that only one of the inputs of each shuffle is a zero vector.
This can only happen when we have a mix of zero and undef elements and the two vectors have a different arrangement of zeros/undefs. The shuffle should eventually be constant folded to all zeros.
Fixes PR32484.
llvm-svn: 299291
Duncan P. N. Exon Smith [Sat, 1 Apr 2017 03:20:48 +0000 (03:20 +0000)]
string: Remove always_inline from basic_string::__init
This is effectively a partial revert of r278356, which started inlining
basic_string::__init. Even if we want to help the compiler along with
an inlinehint, we shouldn't hamstring it by forcing it to inline all the
time.
Libc++ uses always_inline widely as a limit-the-ABI-hack, but since
__init is already on the dylib boundary, it makes no sense here and just
harms the debugging experience at -O0.
rdar://problem/
31013102
llvm-svn: 299290
Rui Ueyama [Sat, 1 Apr 2017 01:42:20 +0000 (01:42 +0000)]
Change the error format to report corrupted .eh_frame.
llvm-svn: 299289
Quentin Colombet [Sat, 1 Apr 2017 01:26:24 +0000 (01:26 +0000)]
Revert "Feature generic option to setup start/stop-after/before"
This reverts commit r299282.
Didn't intend to commit this :(
llvm-svn: 299288
Quentin Colombet [Sat, 1 Apr 2017 01:26:21 +0000 (01:26 +0000)]
Revert "Localizer fun"
This reverts commit r299283.
Didn't intend to commit this :(
llvm-svn: 299287
Quentin Colombet [Sat, 1 Apr 2017 01:26:17 +0000 (01:26 +0000)]
Revert "Instrument SDISel C++ patterns"
This reverts commit r299284.
Didn't intend to commit this :(
llvm-svn: 299286
Quentin Colombet [Sat, 1 Apr 2017 01:26:14 +0000 (01:26 +0000)]
[RegBankSelect] Support REG_SEQUENCE for generic mapping
REG_SEQUENCE falls into the same category as COPY for operands mapping:
- They don't have MCInstrDesc with register constraints
- The input variable could use whatever register classes
- It is possible to have register class already assigned to the operands
In particular, given REG_SEQUENCE are always target specific because of
the subreg indices. Those indices must apply to the register class of
the definition of the REG_SEQUENCE and therefore, the target must set a
register class to that definition. As a result, the generic code can
always use that register class to derive a valid mapping for a
REG_SEQUENCE.
llvm-svn: 299285
Quentin Colombet [Sat, 1 Apr 2017 01:21:32 +0000 (01:21 +0000)]
Instrument SDISel C++ patterns
llvm-svn: 299284
Quentin Colombet [Sat, 1 Apr 2017 01:21:28 +0000 (01:21 +0000)]
Localizer fun
WIP
llvm-svn: 299283
Quentin Colombet [Sat, 1 Apr 2017 01:21:24 +0000 (01:21 +0000)]
Feature generic option to setup start/stop-after/before
This patch refactors the code used in llc such that all the users of the
addPassesToEmitFile API have access to a homogeneous way of handling
start/stop-after/before options right out of the box.
Previously each user would have needed to duplicate this logic and set
up its own options.
NFC
llvm-svn: 299282
Gor Nishanov [Sat, 1 Apr 2017 00:22:47 +0000 (00:22 +0000)]
[coroutines] Add cleanup for compiler injected objects/allocations in coroutine body
Summary:
* Use pushCleanup to emit freeing coroutine memory on normal and EH exits.
* Surround emitted code with CodeGenFunction::RunCleanupsScope.
Reviewers: rsmith, rnk, EricWF
Reviewed By: rnk
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D31460
llvm-svn: 299281
Rui Ueyama [Fri, 31 Mar 2017 23:40:21 +0000 (23:40 +0000)]
Change the error message format for duplicate symbols.
This patch is intended to improve readability of "duplicate symbol"
error messages.
Without this patch:
/ssd/clang/bin/ld.lld: error: /ssd/llvm-project/lld/ELF/Relocations.cpp:1054: duplicate symbol 'lld::elf::demangle(llvm::StringRef)'
/ssd/clang/bin/ld.lld: error: /ssd/llvm-project/lld/ELF/Strings.cpp:93: previous definition was here
With this patch:
/ssd/clang/bin/ld.lld: error: duplicate symbol: lld::elf::demangle(llvm::StringRef)
>>> defined at Strings.cpp:93 (/ssd/llvm-project/lld/ELF/Strings.cpp:93)
>>> Strings.cpp.o:(lld::elf::demangle(llvm::StringRef)) in archive lib/liblldELF.a
>>> defined at Relocations.cpp:1054 (/ssd/llvm-project/lld/ELF/Relocations.cpp:1054)
>>> Relocations.cpp.o:(.text+0x4C30) in archive lib/liblldELF.a
Discussion thread:
http://lists.llvm.org/pipermail/llvm-dev/2017-March/111459.html
Differential Revision: https://reviews.llvm.org/D31507
llvm-svn: 299280
Peter Collingbourne [Fri, 31 Mar 2017 23:22:19 +0000 (23:22 +0000)]
Fix a test to check assembly output instead of bitcode.
llvm-svn: 299279
Eric Christopher [Fri, 31 Mar 2017 23:12:27 +0000 (23:12 +0000)]
Reduce the number of times we query the subtarget for the same information.
llvm-svn: 299278
Eric Christopher [Fri, 31 Mar 2017 23:12:24 +0000 (23:12 +0000)]
Small cleanup to remove extraneous cast.
llvm-svn: 299277
Jim Ingham [Fri, 31 Mar 2017 22:39:55 +0000 (22:39 +0000)]
DisassembleRange can return an empty DisassemblerSP
check for it.
<rdar://problem/
31379799>
llvm-svn: 299276
Konstantin Zhuravlyov [Fri, 31 Mar 2017 22:36:39 +0000 (22:36 +0000)]
AMDGPU/llvm-readobj: Rename RuntimeMDNoteType -> CodeObjectMetadataNoteType to
match the new metadata. NFC.
llvm-svn: 299275
Teresa Johnson [Fri, 31 Mar 2017 22:35:47 +0000 (22:35 +0000)]
[ThinLTO] Handle -emit-llvm* in ThinLTO backends
Summary:
Use PreCodeGenModuleHook to invoke the correct writer when emitting LLVM
IR, returning false to skip codegen from within thinBackend.
Reviewers: pcc, mehdi_amini
Subscribers: Prazek, cfe-commits
Differential Revision: https://reviews.llvm.org/D31534
llvm-svn: 299274
Craig Topper [Fri, 31 Mar 2017 22:23:42 +0000 (22:23 +0000)]
[APInt] Fix bugs in isShiftedMask to match behavior of the similar function in MathExtras.h
This removes a parameter from the routine that was responsible for a lot of the issue. It was a bit count that had to be set to the BitWidth of the APInt and would get passed to getLowBitsSet. This guaranteed the call to getLowBitsSet would create an all ones value. This was then compared to (V | (V-1)). So the only shifted masks we detected had to have the MSB set.
The one in tree user is a transform in InstCombine that never fires due to earlier transforms covering the case better. I've submitted a patch to remove it completely, but for now I've just adapted it to the new interface for isShiftedMask.
llvm-svn: 299273
Bob Haarman [Fri, 31 Mar 2017 22:17:02 +0000 (22:17 +0000)]
COFF: tests for thinlto with mangled names
Summary: This change adds tests for r299268.
Reviewers: pcc, rnk
Reviewed By: pcc
Subscribers: mehdi_amini, llvm-commits, Prazek
Differential Revision: https://reviews.llvm.org/D31549
llvm-svn: 299272
Konstantin Zhuravlyov [Fri, 31 Mar 2017 22:14:54 +0000 (22:14 +0000)]
[AMDGPU] Fix typo in test filename. NFC.
llvm-svn: 299271