summaryrefslogtreecommitdiffstats
path: root/llvm
Commit message (Collapse)AuthorAgeFilesLines
* [GlobalISel] LegalizerInfo: Enable legalization of non-power-of-2 typesVolkan Keles2017-04-113-6/+46
| | | | | | | | | | | | | | Summary: Legalize only if the type is marked as Legal or Custom. If not, return Unsupported as LegalizerHelper is not able to handle non-power-of-2 types right now. Reviewers: qcolombet, aditya_nandakumar, dsanders, t.p.northover, kristof.beyls, javed.absar, ab Reviewed By: kristof.beyls, ab Subscribers: dberris, rovka, igorb, llvm-commits Differential Revision: https://reviews.llvm.org/D31711 llvm-svn: 299929
* Revert "Turn some C-style vararg into variadic templates"Diana Picus2017-04-1124-132/+166
| | | | | | | This reverts commit r299925 because it broke the buildbots. See e.g. http://lab.llvm.org:8011/builders/clang-cmake-armv7-a15/builds/6008 llvm-svn: 299928
* [SelectionDAG] Check CALLSEQ_BEGIN nodes in DelayForLiveRegsSam Parker2017-04-112-1/+138
| | | | | | | | | | | | | | | | | A fix for the bug reported in PR30911. The issue arises when multiple CALLSEQ_BEGIN nodes are unscheduled as the last node to be unscheduled will gain access to the CallResource register. But when a node is being picked, only CALLSEQ_END nodes are checked against the CallResource and have their chains evaluated. This then means that other CALLSEQ_BEGIN nodes can be scheduled before the existing call sequence has been finalised. This patch adds a check against the FrameSetup nodes in DelayForLiveRegs to prevent this from happening. Differential Revision: https://reviews.llvm.org/D31536 llvm-svn: 299926
* Turn some C-style vararg into variadic templatesSerge Guelton2017-04-1124-166/+132
| | | | | | | | | | | | Module::getOrInsertFunction is using C-style vararg instead of variadic templates. From a user prospective, it forces the use of an annoying nullptr to mark the end of the vararg, and there's not type checking on the arguments. The variadic template is an obvious solution to both issues. llvm-svn: 299925
* Simplify the code and remove dead codeSylvestre Ledru2017-04-111-5/+3
| | | | | | | | | | | | Summary: Fix coverity cid 1374240 Reviewers: dberlin Reviewed By: dberlin Differential Revision: https://reviews.llvm.org/D31928 llvm-svn: 299924
* [IR] Fix a typo in a comment. NFCCraig Topper2017-04-111-1/+1
| | | | llvm-svn: 299918
* [InstCombine] Refinement of r299915. Only consider a ConstantVector for Neg ↵Craig Topper2017-04-111-3/+14
| | | | | | if all the elements are Undef or ConstantInt. llvm-svn: 299917
* [InstCombine] Support weird size element types in dyn_castNegVal.Craig Topper2017-04-113-4/+13
| | | | llvm-svn: 299915
* [LoopUnswitch] Fix a test caseSanjoy Das2017-04-111-2/+2
| | | | | | | | | | | | | | | (h/t to Chandler for pointing this out) The test in question was not at all testing what it was supposed to test. We do not //care// about placing `!make.implicit` in inner constant branch (since it will be folded away anyway). We care about placing `!make.implicit` in the outer branch that switches between either version of the loop. Having said that, it is _correct_ to leave behind the `!make.implicit` in the inner branch, but there is no need to do so. llvm-svn: 299912
* [LICM] Hoist fp division from the loops and replace by a reciprocalHal Finkel2017-04-112-0/+57
| | | | | | | | | | | When allowed, we can hoist a division out of a loop in favor of a multiplication by the reciprocal. Fixes PR32157. Patch by vit9696! Differential Revision: https://reviews.llvm.org/D30819 llvm-svn: 299911
* [PowerPC] multiply-with-overflow might use the CTR registerHal Finkel2017-04-112-9/+45
| | | | | | | | | | | | Check the legality of ISD::[US]MULO to see whether Intrinsic::[us]mul_with_overflow will legalize into a function call (and, thus, will use the CTR register). Fixes PR32485. Patch by Tim Neumann! Differential Revision: https://reviews.llvm.org/D31790 llvm-svn: 299910
* [bugpoint] Also remove comdat's from externalized GVsHal Finkel2017-04-111-0/+1
| | | | | | | | We were removing comdats from externalized functions (function declarations can't be comdat), but were not doing the same for variable. Failure to do this would cause bugpoint to fail ("Declaration may not be in a Comdat!"). llvm-svn: 299908
* [IR] Sink some AttributeListImpl methods out of headers NFCReid Kleckner2017-04-112-41/+49
| | | | llvm-svn: 299906
* Revert "NewGVN: Don't propagate over phi backedges where undef causes us to ↵Daniel Berlin2017-04-111-28/+6
| | | | | | | | | | have >1 value." It's not ready yet this was an accidental commit :( This reverts r299903 llvm-svn: 299904
* NewGVN: Don't propagate over phi backedges where undef causes us to have >1 ↵Daniel Berlin2017-04-111-6/+28
| | | | | | | | value. Fixes PR 32607. llvm-svn: 299903
* Remove AttributeSetNode::get(AttributeList, unsigned) and sink constructorReid Kleckner2017-04-103-21/+19
| | | | | | | | | | | The getter was equivalent to AttributeList::getAttributes(unsigned), which seems like a better way to express getting the AttributeSet for a given index. This static helper was only used in one place anyway. The constructor doesn't benefit from inlining and doesn't need to be in a header. llvm-svn: 299900
* Reland "[IR] Make AttributeSetNode public, avoid temporary AttributeList copies"Reid Kleckner2017-04-1015-340/+343
| | | | | | | | | | | | | | | | | | | | | | | | | This re-lands r299875. I introduced a bug in Clang code responsible for replacing K&R, no prototype declarations with a real function definition with a prototype. The bug was here: // Collect any return attributes from the call. - if (oldAttrs.hasAttributes(llvm::AttributeList::ReturnIndex)) - newAttrs.push_back(llvm::AttributeList::get(newFn->getContext(), - oldAttrs.getRetAttributes())); + newAttrs.push_back(oldAttrs.getRetAttributes()); Previously getRetAttributes() carried AttributeList::ReturnIndex in its AttributeList. Now that we return the AttributeSetNode* directly, it no longer carries that index, and we call this overload with a single node: AttributeList::get(LLVMContext&, ArrayRef<AttributeSetNode*>) That aborted with an assertion on x86_32 targets. I added an explicit triple to the test and added CHECKs to help find issues like this in the future sooner. llvm-svn: 299899
* [ARM, x86] add tests to show possible improvement for bool math; NFCSanjay Patel2017-04-102-0/+64
| | | | llvm-svn: 299897
* [NewGVN] Surround with parens to clarify allegedly ambiguous precedence.Davide Italiano2017-04-101-16/+15
| | | | | | | This Placates GCC7 with -Werror. Also, clang-format the assertions while I'm here. llvm-svn: 299895
* [MemorySSA] We don't need to compute dominator levels anymore.Davide Italiano2017-04-102-8/+0
| | | | | | Differential Revision: https://reviews.llvm.org/D31818 llvm-svn: 299893
* CodeGen: BlockPlacement: Don't always tail-duplicate with no other successor.Kyle Butt2017-04-103-7/+56
| | | | | | | | | | The math works out where it can actually be counter-productive. The probability calculations correctly handle the case where the alternative is 0 probability, rely on those calculations. Includes a test case that demonstrates the problem. llvm-svn: 299892
* CodeGen: BlockPlacement: Minor probability changes.Kyle Butt2017-04-103-25/+74
| | | | | | | Qin may be large, and Succ may be more frequent than BB. Take these both into account when deciding if tail-duplication is profitable. llvm-svn: 299891
* CodeGen: BranchFolding: Merge identical blocks, even if they are short.Kyle Butt2017-04-102-0/+57
| | | | | | | | Merging identical blocks when it doesn't reduce fallthrough. It is common for the blocks created from critical edge splitting to be identical. We would like to merge these blocks whenever doing so would not reduce fallthrough. llvm-svn: 299890
* Allow DataLayout to specify addrspace for allocas.Matt Arsenault2017-04-1034-72/+382
| | | | | | | | | | | | | | | | | | | | | | | LLVM makes several assumptions about address space 0. However, alloca is presently constrained to always return this address space. There's no real way to avoid using alloca, so without this there is no way to opt out of these assumptions. The problematic assumptions include: - That the pointer size used for the stack is the same size as the code size pointer, which is also the maximum sized pointer. - That 0 is an invalid, non-dereferencable pointer value. These are problems for AMDGPU because alloca is used to implement the private address space, which uses a 32-bit index as the pointer value. Other pointers are 64-bit and behave more like LLVM's notion of generic address space. By changing the address space used for allocas, we can change our generic pointer type to be LLVM's generic pointer type which does have similar properties. llvm-svn: 299888
* Get the TOC save offset off of PPCFrameLowering rather than a separate copy ↵Eric Christopher2017-04-101-1/+1
| | | | | | of the same data. llvm-svn: 299887
* Bitcode: Merge the code for parsing global value module codes in the summary ↵Peter Collingbourne2017-04-101-26/+6
| | | | | | | | | | | bitcode reader. NFCI. This code will need to be taught to handle string tables and it's better if there is only one copy of it. Differential Revision: https://reviews.llvm.org/D31829 llvm-svn: 299886
* Emit less compiler optimization remarks in samplepgo to reduce a call to ↵Dehao Chen2017-04-103-5/+3
| | | | | | | | | | | | | | | | findCalleeFunctionSamples which is going to be refactored. Summary: Now the SamplePGO support is more stable, we do not need so many verbose optimization remarks emitted. Reviewers: dnovillo, davidxl Reviewed By: davidxl Subscribers: fhahn, llvm-commits Differential Revision: https://reviews.llvm.org/D31826 llvm-svn: 299883
* [GVNHoist] Call isGuaranteedToTransferExecutionToSuccessor on each instructionGeoff Berry2017-04-102-17/+76
| | | | | | | | | | | w.r.t. https://bugs.llvm.org/show_bug.cgi?id=32153 The consensus seems to be isGuaranteedToTransferExecutionToSuccessor should be called for each function. Patch by Aditya Kumar Differential Revision: https://reviews.llvm.org/D31035 llvm-svn: 299882
* Revert "[asan] Put ctor/dtor in comdat."Evgeniy Stepanov2017-04-103-54/+10
| | | | | | This reverts commit r299696, which is causing mysterious test failures. llvm-svn: 299880
* Revert "[asan] Fix dead stripping of globals on Linux."Evgeniy Stepanov2017-04-107-163/+48
| | | | | | This reverts commit r299697, which caused a big increase in object file size. llvm-svn: 299879
* Revert "[IR] Make AttributeSetNode public, avoid temporary AttributeList copies"Reid Kleckner2017-04-1015-343/+340
| | | | | | | This reverts r299875. A Linux bot came back with a test failure: http://bb.pgr.jp/builders/test-clang-i686-linux-RA/builds/741/steps/test_clang/logs/Clang%20%3A%3A%20CodeGen__2006-05-19-SingleEltReturn.c llvm-svn: 299878
* Add address space mangling to lifetime intrinsicsMatt Arsenault2017-04-10117-813/+839
| | | | | | In preparation for allowing allocas to have non-0 addrspace. llvm-svn: 299876
* [IR] Make AttributeSetNode public, avoid temporary AttributeList copiesReid Kleckner2017-04-1015-340/+343
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: AttributeList::get(Fn|Ret|Param)Attributes no longer creates a temporary AttributeList just to hide the AttributeSetNode type. I've also added a factory method to create AttributeLists from a parallel array of AttributeSetNodes. I think this simplifies construction of AttributeLists when rewriting function prototypes. Previously we would test if a particular index had attributes, and conditionally add a temporary attribute list to a vector. Now the attribute set vector is parallel to the argument vector already that these passes already construct. My long term vision is to wrap AttributeSetNode* inside an AttributeSet type that holds the enum attributes, but that will come in a follow up change. I haven't done any performance measurements for this change because profiling hasn't shown that any of the affected code is hot. Reviewers: pete, chandlerc, sanjoy, hfinkel Reviewed By: pete Subscribers: jfb, llvm-commits Differential Revision: https://reviews.llvm.org/D31198 llvm-svn: 299875
* Fix signed / unsigned comparison warnings.Zachary Turner2017-04-101-2/+2
| | | | llvm-svn: 299873
* [mips] Use Triple::isLittleEndian to check endianness. NFCSimon Atanasyan2017-04-101-3/+1
| | | | llvm-svn: 299872
* [InstCombine] improve variable names; NFCISanjay Patel2017-04-101-34/+34
| | | | llvm-svn: 299871
* [InstSimplify] Use cast instead of dyn_cast after isa<> check. NFCICraig Topper2017-04-101-2/+2
| | | | llvm-svn: 299870
* [llvm-pdbdump] Display padding bytes on record layoutZachary Turner2017-04-1010-64/+145
| | | | | | | | | | | | | | | When dumping classes, show where padding occurs, and at the end of the class print statistics about how many bytes total of padding exist in a class. Since PDB doesn't specifically contain information about padding, we have to mimic this by sort of reversing a small portion of the record layout algorithm (e.g. looking at offsets and sizes and trying to determine whether something is part of the same field or a new field). Differential Revision: https://reviews.llvm.org/D31800 llvm-svn: 299869
* [MemCpyOpt] Only replace memcpy with bitcast if address spaces matchMatt Arsenault2017-04-102-0/+18
| | | | | | Patch by James Price llvm-svn: 299866
* MemorySSA: Make lifetime starts defs for mustaliased pointersDaniel Berlin2017-04-102-3/+5
| | | | | | | | | | | | | | | | | | Summary: While we don't want them aliasing with other pointers, there seems to be no point in not having them clobber must-aliased'd pointers. If some day, we split the aliasing and ordering chains, we'd make this not aliasing but an ordering barrier (IE it doesn't affect it's memory, but we can't hoist it above it). Reviewers: hfinkel, george.burgess.iv Subscribers: Prazek, llvm-commits Differential Revision: https://reviews.llvm.org/D31865 llvm-svn: 299865
* [ARM/AArch64] Ensure valid vector element types for interleaved accessesMatthew Simpson2017-04-108-39/+111
| | | | | | | | | | | This patch refactors and strengthens the type checks performed for interleaved accesses. The primary functional change is to ensure that the interleaved accesses have valid element types. The added test cases previously failed because the element type is f128. Differential Revision: https://reviews.llvm.org/D31817 llvm-svn: 299864
* [InstCombine] Use commutable matchers and m_OneUse in visitSub to shorten ↵Craig Topper2017-04-102-15/+133
| | | | | | | | code. Add missing test cases. In one case I removed commute handling for a multiply with a constant since we'll eventually get the constant on the right hand side. llvm-svn: 299863
* AMDGPU: Fix crash when disassembling VOP3 macMatt Arsenault2017-04-1011-19/+42
| | | | | | | | | | | | The unused dummy src2_modifiers is missing, so it crashes when trying to print it. I tried to fully remove src2_modifiers, but there are some irritations in the places where it is converted to mad since it starts to require modifying use lists while iterating over them. llvm-svn: 299861
* [Support] Add support for finding unset bits in a BitVector.Zachary Turner2017-04-103-0/+108
| | | | | | | | | | | | BitVector had methods for searching for the first and next set bits, but it did not have analagous methods for finding the first and next unset bits. This is useful when your ones and zeros are grouped together and you want to iterate over ranges of ones and zeros. Differential Revision: https://reviews.llvm.org/D31802 llvm-svn: 299857
* Fix line endings.Zachary Turner2017-04-101-592/+592
| | | | llvm-svn: 299856
* Remove eol-style:native from BitVector.hZachary Turner2017-04-101-592/+592
| | | | llvm-svn: 299855
* [InstCombine] Use m_c_Add to shorten some code. Add testcases for this fold ↵Craig Topper2017-04-102-2/+19
| | | | | | since they were missing. NFC llvm-svn: 299853
* [X86][MMX] Add fast-isel support for MMX non-temporal writesSimon Pilgrim2017-04-102-1/+5
| | | | | | Differential Revision: https://reviews.llvm.org/D31754 llvm-svn: 299852
* [InstCombine] fix matching of or-of-icmps constants (PR32524)Sanjay Patel2017-04-102-16/+19
| | | | | | | | | | | Also, make the same change in and-of-icmps and remove a hack for detecting that case. Finally, add some FIXME comments because the code duplication here is awful. This should fix the remaining IR problem noted in: https://bugs.llvm.org/show_bug.cgi?id=32524 llvm-svn: 299851
* Improves pretty printing of variable types in llvm-pdbdumpAdrian McCarthy2017-04-1015-64/+154
| | | | | | | | | | | | | | | | | * Adds support for pointers to arrays, which was missing * Adds some tests * Improves consistency of const and volatile qualifiers * Eliminates non-composable special case code for arrays and function by using a more general recursive approach * Has a hack for getting the calling convention into the right spot for pointer-to-functions Given the rapid changes happenning in llvm-pdbdump, this may be difficult to merge. Differential Revision: https://reviews.llvm.org/D31832 llvm-svn: 299848
OpenPOWER on IntegriCloud