summaryrefslogtreecommitdiffstats
path: root/llvm/lib
Commit message (Collapse)AuthorAgeFilesLines
* [InstSimplify] add helper function for SimplifyICmpInst(); NFCISanjay Patel2016-08-221-133/+143
| | | | | | | | | And add a FIXME because the helper excludes folds for vectors. It's not clear yet how many of these are actually testable (and therefore necessary?) because later analysis uses computeKnownBits and other methods to catch many of these cases. llvm-svn: 279492
* Fix crash from assert in r279466.Pete Cooper2016-08-221-1/+1
| | | | | | | | | | The assert in r279466 checks that we call the correct version of Intrinsic::getName. The version which accepts only an ID should not be used for intrinsics with overloaded types. The global-isel code was calling the wrong version. The test CodeGen/AArch64/GlobalISel/arm64-irtranslator.ll will ensure that we call the correct version from now on. llvm-svn: 279487
* [InstCombine] change param type from Instruction to BinaryOperator for icmp ↵Sanjay Patel2016-08-222-97/+109
| | | | | | | | helpers; NFCI This saves some casting in the helper functions and eases some further refactoring. llvm-svn: 279478
* [GraphTraits] Replace all NodeType usage with NodeRefTim Shen2016-08-224-28/+13
| | | | | | | | This should finish the GraphTraits migration. Differential Revision: http://reviews.llvm.org/D23730 llvm-svn: 279475
* ADT: Remove ilist_*sentinel_traits, NFCDuncan P. N. Exon Smith2016-08-222-8/+0
| | | | | | | | | | Remove all the dead code around ilist_*sentinel_traits. This is a follow-up to gutting them as part of r279314 (originally r278974), staged to prevent broken builds in sub-projects. Uses were removed from clang in r279457 and lld in r279458. llvm-svn: 279473
* [InstCombine] use m_APInt to allow icmp (shr exact X, Y), 0 folds for splat ↵Sanjay Patel2016-08-221-14/+13
| | | | | | constant vectors llvm-svn: 279472
* Add ADT headers to the cmake headers directory for LLVMSupport. NFC.Pete Cooper2016-08-221-0/+1
| | | | | | | | | Xcode and MSVC list the headers and source files for each library. LLVMSupport lists included the source files for ADT but not the headers. This add the ADT headers so that they are browsable by the UI. llvm-svn: 279470
* Add comments and an assert to follow-up on r279113. NFC.Pete Cooper2016-08-221-0/+2
| | | | | | | | | | | Philip commented on r279113 to ask for better comments as to when to use the different versions of getName. Its also possible to assert in the simple case that we aren't an overloaded intrinsic as those have to use the more capable version of getName. Thanks for the comments Philip. llvm-svn: 279466
* AMDGPU: Split SILowerControlFlow into two piecesMatt Arsenault2016-08-226-353/+518
| | | | | | | | | | | | | | Do most of the lowering in a pre-RA pass. Keep the skip jump insertion late, plus a few other things that require more work to move out. One concern I have is now there may be COPY instructions which do not have the necessary implicit exec uses if they will be lowered to v_mov_b32. This has a positive effect on SGPR usage in shader-db. llvm-svn: 279464
* MSSA: Factor out phi node placementDaniel Berlin2016-08-221-17/+22
| | | | llvm-svn: 279462
* MSSA: Only rename accesses whose defining access is nullptrDaniel Berlin2016-08-221-14/+6
| | | | llvm-svn: 279461
* [SimplifyCFG] Rewrite SinkThenElseCodeToEndJames Molloy2016-08-221-150/+236
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | [Recommitting now an unrelated assertion in SROA is sorted out] The new version has several advantages: 1) IMSHO it's more readable and neater 2) It handles loads and stores properly 3) It can handle any number of incoming blocks rather than just two. I'll be taking advantage of this in a followup patch. With this change we can now finally sink load-modify-store idioms such as: if (a) return *b += 3; else return *b += 4; => %z = load i32, i32* %y %.sink = select i1 %a, i32 5, i32 7 %b = add i32 %z, %.sink store i32 %b, i32* %y ret i32 %b When this works for switches it'll be even more powerful. Round 4. This time we should handle all instructions correctly, and not replace any operands that need to be constant with variables. This was really hard to determine safely, so the helper function should be put into the Instruction API. I'll do that as a followup. llvm-svn: 279460
* [SROA] Remove incorrect assertionJames Molloy2016-08-221-3/+0
| | | | | | | | | Confirmed with aprantl, this assertion is incorrect - code can get here (for example 80-bit FP types) and if it does it's benign. This is exposed by a completely unrelated patch of mine, so stop the compiler falling over. Original differential: http://reviews.llvm.org/D16187 aprantl's advice to remove assertion: http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20160815/382129.html llvm-svn: 279454
* [SSP] Do not set __guard_local to hidden for OpenBSD SSPTim Shen2016-08-221-3/+1
| | | | | | | | | | | | | | | | | | | | | | | | __guard_local is defined as long on OpenBSD. If the source file contains a definition of __guard_local, it mismatches with the int8 pointer type used in LLVM. In that case, Module::getOrInsertGlobal() returns a cast operation instead of a GlobalVariable. Trying to set the visibility on the cast operation leads to random segfaults (seen when compiling the OpenBSD kernel, which also runs with stack protection). In the kernel, the hidden attribute does not matter. For userspace code, __guard_local is defined as hidden in the startup code. If a program re-defines __guard_local, the definition from the startup code will either win or the linker complains about multiple definitions (depending on whether the re-defined __guard_local is placed in the common segment or not). It also matches what gcc on OpenBSD does. Thanks Stefan Kempf <sisnkemp@gmail.com> for the patch! Differential Revision: http://reviews.llvm.org/D23674 llvm-svn: 279449
* [InstCombine] Allow sinking from unique predecessor with multiple edgesJun Bum Lim2016-08-221-1/+1
| | | | | | | | | | | | Summary: We can allow sinking if the single user block has only one unique predecessor, regardless of the number of edges. Note that a switch statement with multiple cases can have the same destination. Reviewers: mcrosier, majnemer, spatel, reames Subscribers: reames, mcrosier, llvm-commits Differential Revision: https://reviews.llvm.org/D23722 llvm-svn: 279448
* Revert "[SimplifyCFG] Rewrite SinkThenElseCodeToEnd"James Molloy2016-08-221-236/+150
| | | | | | This reverts commit r279443. It caused buildbot failures. llvm-svn: 279447
* [SimplifyCFG] Rewrite SinkThenElseCodeToEndJames Molloy2016-08-221-150/+236
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The new version has several advantages: 1) IMSHO it's more readable and neater 2) It handles loads and stores properly 3) It can handle any number of incoming blocks rather than just two. I'll be taking advantage of this in a followup patch. With this change we can now finally sink load-modify-store idioms such as: if (a) return *b += 3; else return *b += 4; => %z = load i32, i32* %y %.sink = select i1 %a, i32 5, i32 7 %b = add i32 %z, %.sink store i32 %b, i32* %y ret i32 %b When this works for switches it'll be even more powerful. Round 4. This time we should handle all instructions correctly, and not replace any operands that need to be constant with variables. This was really hard to determine safely, so the helper function should be put into the Instruction API. I'll do that as a followup. llvm-svn: 279443
* [X86][AVX] Don't use SubVectorBroadcast if there are additional users of the ↵Simon Pilgrim2016-08-221-4/+6
| | | | | | | | chain (PR29088) We could improve on this by making X86SubVBroadcast a full memory intrinsic similar to X86vzload llvm-svn: 279441
* [mips][ias] Support .dtprel[d]word and .tprel[d]word directivesSimon Atanasyan2016-08-228-2/+207
| | | | | | | | | | | | | Assembler directives .dtprelword, .dtpreldword, .tprelword, and .tpreldword generates relocations R_MIPS_TLS_DTPREL32, R_MIPS_TLS_DTPREL64, R_MIPS_TLS_TPREL32, and R_MIPS_TLS_TPREL64 respectively. The main motivation for this patch is to be able to write test cases for checking correctness of the LLD linker's behaviour. Differential Revision: https://reviews.llvm.org/D23669 llvm-svn: 279439
* [LTO] Constify the Module Hook function (NFC)Mehdi Amini2016-08-221-1/+1
| | | | | | | | It use to be non-const for the sole purpose of custom handling of commons symbol. This is moved now in the regular LTO handling now and such we can constify the callback. llvm-svn: 279438
* Reset isUndef when removing subreg from a def operandKrzysztof Parzyszek2016-08-221-0/+2
| | | | llvm-svn: 279437
* [X86] Only accept SM_SentinelUndef (-1) as an undefined shuffle mask in rangeSimon Pilgrim2016-08-221-3/+3
| | | | | | As discussed on D23027 we should be trying to be more strict on what is an undefined mask value. llvm-svn: 279435
* Revert -r278267 [ValueTracking] An improvement to IR ValueTracking on ↵Artur Pilipenko2016-08-221-37/+1
| | | | | | | | | | Non-negative Integers This change cause performance regression on MultiSource/Benchmarks/TSVC/Symbolics-flt/Symbolics-flt from LNT and some other bechmarks. See https://reviews.llvm.org/D18777 for details. llvm-svn: 279433
* Revert -r278269 [IndVarSimplify] Eliminate zext of a signed IV when the IV ↵Artur Pilipenko2016-08-221-7/+2
| | | | | | | | | | is known to be non-negative This change needs to be reverted in order to revert -r278267 which cause performance regression on MultiSource/Benchmarks/TSVC/Symbolics-flt/Symbolics-flt from LNT and some other bechmarks. See comments on https://reviews.llvm.org/D18777 for details. llvm-svn: 279432
* [X86][SSE] Avoid specifying unused arguments in SHUFPD loweringSimon Pilgrim2016-08-222-2/+17
| | | | | | | | | | | | | | | | As discussed on PR26491, we are missing the opportunity to make use of the smaller MOVHLPS instruction because we set both arguments of a SHUFPD when using it to lower a single input shuffle. This patch sets the lowered argument to UNDEF if that shuffle element is undefined. This in turn makes it easier for target shuffle combining to decode UNDEF shuffle elements, allowing combines to MOVHLPS to occur. A fix to match against MOVHPD stores was necessary as well. This builds on the improved MOVLHPS/MOVHLPS lowering and memory folding support added in D16956 Adding similar support for SHUFPS will have to wait until have better support for target combining of binary shuffles. Differential Revision: https://reviews.llvm.org/D23027 llvm-svn: 279430
* [mips][microMIPS] Implement BLTZC, BLEZC, BGEZC and BGTZC instructions, fix ↵Hrvoje Varga2016-08-227-96/+306
| | | | | | | | disassembly and add operand checking to existing B<cond>C implementations Differential Revision: https://reviews.llvm.org/D22667 llvm-svn: 279429
* [MC] Remove guard(s). NFCI.Davide Italiano2016-08-229-20/+0
| | | | | | | All the methods are already marked with LLVM_DUMP_METHOD. llvm-svn: 279428
* [X86] Create a new instruction format to handle 4VOp3 encoding. This saves ↵Craig Topper2016-08-227-64/+103
| | | | | | one bit in TSFlags and simplifies MRMSrcMem/MRMSrcReg format handling. llvm-svn: 279424
* [X86] Create a new instruction format to handle MemOp4 encoding. This saves ↵Craig Topper2016-08-225-68/+123
| | | | | | one bit in TSFlags and simplifies MRMSrcMem/MRMSrcReg format handling. llvm-svn: 279423
* [X86] Space out the encodings of X86 instruction formats. I plan to add some ↵Craig Topper2016-08-222-92/+104
| | | | | | | | new encodings in future commits and this will reduce the size of those commits. NFC This tries to keep all the ModRM memory and register forms in their own regions of the encodings. Hoping to make it simple on some of the switch statements that operate on these encodings. llvm-svn: 279422
* [LTO] Handles commons in monolithic LTOMehdi Amini2016-08-221-0/+33
| | | | | | | | | The gold-plugin was doing this internally, now the API is handling commons correctly based on the given resolution. Differential Revision: https://reviews.llvm.org/D23739 llvm-svn: 279417
* [LTO] Add a "CodeGenOnly" option. Allows the client to skip the optimizer.Mehdi Amini2016-08-222-21/+29
| | | | | | | | | | | | Summary: Slowly getting on par with libLTO Reviewers: tejohnson Subscribers: llvm-commits, mehdi_amini Differential Revision: https://reviews.llvm.org/D23615 llvm-svn: 279416
* [asan] Use 1 byte aligned stores to poison shadow memoryVitaly Buka2016-08-221-2/+2
| | | | | | | | | | | | Summary: r279379 introduced crash on arm 32bit bot. I suspect this is alignment issue. Reviewers: eugenis Subscribers: llvm-commits, aemerson Differential Revision: https://reviews.llvm.org/D23762 llvm-svn: 279413
* [X86] Merge hasVEX_i8ImmReg into the ImmFormat type which had extra unused ↵Craig Topper2016-08-225-84/+91
| | | | | | encodings. This saves one bit in TSFlags. NFC llvm-svn: 279412
* [X86] Remove ignoreVEX_L from TSFlags. Only the disassembler needs it and ↵Craig Topper2016-08-222-23/+9
| | | | | | the disassembler doesn't use TSFlags. NFC llvm-svn: 279411
* Reformat.NAKAMURA Takumi2016-08-222-22/+22
| | | | llvm-svn: 279409
* Untabify.NAKAMURA Takumi2016-08-223-23/+20
| | | | llvm-svn: 279408
* [InstCombine] use m_APInt to allow icmp (shl X, Y), C folds for splat ↵Sanjay Patel2016-08-211-13/+10
| | | | | | | | | | | constant vectors, part 4 This concludes the fixes for icmp+shl in this series: https://reviews.llvm.org/rL279339 https://reviews.llvm.org/rL279398 https://reviews.llvm.org/rL279399 llvm-svn: 279401
* [InstCombine] use m_APInt to allow icmp (shl X, Y), C folds for splat ↵Sanjay Patel2016-08-211-6/+6
| | | | | | | | constant vectors, part 3 This is a partial enablement (move the ConstantInt guard down). llvm-svn: 279399
* [InstCombine] use m_APInt to allow icmp (shl X, Y), C folds for splat ↵Sanjay Patel2016-08-211-6/+6
| | | | | | | | constant vectors, part 2 This is a partial enablement (move the ConstantInt guard down). llvm-svn: 279398
* [X86][AVX] Dropped combineShuffle256 - this can now be performed by ↵Simon Pilgrim2016-08-211-82/+0
| | | | | | EltsFromConsecutiveLoads llvm-svn: 279397
* [InstCombine] use APInt instead of ConstantInt in isSignBitCheck(); NFCISanjay Patel2016-08-211-8/+8
| | | | | | | | The callers still have ConstantInt guards, so there is no functional change intended from this change. But relaxing the callers will allow more folds for vector types. llvm-svn: 279396
* [AVX512][FastISel] Do not use K registers in TEST instructionsGuy Blank2016-08-211-6/+31
| | | | | | | | | In some cases, FastIsel was emitting TEST instruction with K reg input, which is illegal. Changed to using KORTEST when dealing with K regs. Differential Revision: https://reviews.llvm.org/D23163 llvm-svn: 279393
* ARM: Avoid dereferencing end() in ARMFrameLowering::emitEpilogueDuncan P. N. Exon Smith2016-08-211-2/+2
| | | | | | | | | | | | | | | | | | This fixes the crash from PR29072, where the MachineBasicBlock::iterator wasn't being properly checked against MachineBasicBlock::end() before iterating. This was another bug exposed by the new ilist::iterator::operator*() assertion from r279314. This testcase is poor quality. bugpoint couldn't reduce any further, and I haven't had time to dig into what's going on so I can't invent a better one. I didn't even get good CHECK lines in: this is just a crasher. I'm committing anyway since this is a real crash with an obvious fix, but I'll leave PR29072 open and ask an ARM maintainer to help improve the testcase. llvm-svn: 279391
* [asan] Minimize code size by using __asan_set_shadow_* for large blocksVitaly Buka2016-08-201-5/+48
| | | | | | | | | | | | | Summary: We can insert function call instead of multiple store operation. Current default is blocks larger than 64 bytes. Changes are hidden behind -asan-experimental-poisoning flag. PR27453 Differential Revision: https://reviews.llvm.org/D23711 llvm-svn: 279383
* Use SDValue::getOpcode() helper instead of via SDValue::getNode()Simon Pilgrim2016-08-201-2/+2
| | | | llvm-svn: 279381
* [asan] Initialize __asan_set_shadow_* callbacksVitaly Buka2016-08-201-0/+19
| | | | | | | | | | | | | | | Summary: Callbacks are not being used yet. PR27453 Reviewers: kcc, eugenis Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D23634 llvm-svn: 279380
* [asan] Optimize store size in FunctionStackPoisoner::poisonRedZonesVitaly Buka2016-08-201-47/+60
| | | | | | | | | | | | Summary: Reduce store size to avoid leading and trailing zeros. Reviewers: kcc, eugenis Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D23648 llvm-svn: 279379
* [asan] Cleanup instrumentation of dynamic allocasVitaly Buka2016-08-201-24/+44
| | | | | | | | | | Summary: Extract instrumenting dynamic allocas into separate method. Rename asan-instrument-allocas -> asan-instrument-dynamic-allocas Differential Revision: https://reviews.llvm.org/D23707 llvm-svn: 279376
* [asan] Add support of lifetime poisoning into ComputeASanStackFrameLayoutVitaly Buka2016-08-202-4/+14
| | | | | | | | | | | | | | | Summary: We are going to combine poisoning of red zones and scope poisoning. PR27453 Reviewers: kcc, eugenis Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D23623 llvm-svn: 279373
OpenPOWER on IntegriCloud