summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* Update DR status list to reflect that Clang 3.5 has been released.Richard Smith2014-09-112-18/+18
| | | | llvm-svn: 217608
* DWARF64 FixesTodd Fiala2014-09-1110-65/+145
| | | | | | | | | | | | | | | | 1. DW_FORM_strp and DW_FORM_sec_offset are 64bits for DWARF64 / 32bits for DWARF32 They are different from DW_FORM_addr, whose size is specified in .debug_info 2. Bump DWARF version support form [2,3] to [2,4] in DWARFDebugLine.cpp 3. Fix DWARFDebugLine to support DWARF64 See http://reviews.llvm.org/D5307 for more details. Reviewed by Greg Clayton and Jason Molenda. Change by Tong Shen. llvm-svn: 217607
* Tests for DR573-580.Richard Smith2014-09-112-8/+134
| | | | llvm-svn: 217606
* [MCJIT] Take the relocation addend into account when applying ARM MachO VANILLALang Hames2014-09-112-1/+5
| | | | | | | | and BR24 relocations. <rdar://problem/18296496> llvm-svn: 217605
* Adding ABI support for __cxa_throw_bad_array_new_length.Aaron Ballman2014-09-114-0/+48
| | | | llvm-svn: 217604
* Added missing LLVM_NOEXCEPT to the definition of _obj2yaml_error_category::nameIsmail Pazarbasi2014-09-111-1/+3
| | | | | | LLVM_NOEXCEPT was added in r210591. llvm-svn: 217603
* [AVX512] Fix miscompile for unpackAdam Nemet2014-09-112-60/+53
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | r189189 implemented AVX512 unpack by essentially performing a 256-bit unpack between the low and the high 256 bits of src1 into the low part of the destination and another unpack of the low and high 256 bits of src2 into the high part of the destination. I don't think that's how unpack works. AVX512 unpack simply has more 128-bit lanes but other than it works the same way as AVX. So in each 128-bit lane, we're always interleaving certain parts of both operands rather different parts of one of the operands. E.g. for this: __v16sf a = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 }; __v16sf b = { 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 }; __v16sf c = __builtin_shufflevector(a, b, 0, 8, 1, 9, 4, 12, 5, 13, 16, 24, 17, 25, 20, 28, 21, 29); we generated punpcklps (notice how the elements of a and b are not interleaved in the shuffle). In turn, c was set to this: 0 16 1 17 4 20 5 21 8 24 9 25 12 28 13 29 Obviously this should have just returned the mask vector of the shuffle vector. I mostly reverted this change and made sure the original AVX code worked for 512-bit vectors as well. Also updated the tests because they matched the logic from the code. llvm-svn: 217602
* Add triple and remove hashes to account for buildbot differences in comment ↵Sanjay Patel2014-09-111-13/+13
| | | | | | strings. llvm-svn: 217601
* Move constant-sized bitvector to the stack.Benjamin Kramer2014-09-111-2/+2
| | | | llvm-svn: 217600
* Combine fmul vector FP constants when unsafe math is allowed.Sanjay Patel2014-09-112-6/+70
| | | | | | | | | | | | | | | | | | | This is an extension of the change made with r215820: http://llvm.org/viewvc/llvm-project?view=revision&revision=215820 That patch allowed combining of splatted vector FP constants that are multiplied. This patch allows combining non-uniform vector FP constants too by relaxing the check on the type of vector. Also, canonicalize a vector fmul in the same way that we already do for scalars - if only one operand of the fmul is a constant, make it operand 1. Otherwise, we miss potential folds. This fold is also done by -instcombine, but it's possible that extra fmuls may have been generated during lowering. Differential Revision: http://reviews.llvm.org/D5254 llvm-svn: 217599
* Merge GetAddrOfCXXConstructor and GetAddrOfCXXDonstructor. NFC.Rafael Espindola2014-09-119-50/+21
| | | | llvm-svn: 217598
* typoSanjay Patel2014-09-111-1/+1
| | | | llvm-svn: 217597
* R600: Add cmpxchg instruction for evergreenAaron Watry2014-09-112-5/+29
| | | | | | | | | | | | | | | | | Refactored the R600_LDS_1A2D class a bit to get it to actually work. It seemed to be previously unused and broken. We also have to disable the conversion to the noret variant for now in R600ISelLowering because the getLDSNoRetOp method only handles 1A1D LDS ops. Someone can feel free to modify the AMDGPU::getLDSNoRetOp method to work for more than 1A1D variants of LDS operations. It's being left as a future TODO for now. Signed-off-by: Aaron Watry <awatry at gmail.com> Reviewed-by: Matt Arsenault <matthew.arsenault@amd.com> llvm-svn: 217596
* R600: Test local atomics for evergreenAaron Watry2014-09-111-0/+27
| | | | | | | | Now that the operations are all implemented, we can test this sub-arch here. Signed-off-by: Aaron Watry <awatry@gmail.com> Reviewed-by: Matt Arsenault <matthew.arsenault@amd.com> llvm-svn: 217595
* R600: Add LDS_WRXCHG[_RET] instructions for Evergreen.Aaron Watry2014-09-111-0/+4
| | | | | | Signed-off-by: Aaron Watry <awatry@gmail.com> Reviewed-by: Matt Arsenault <matthew.arsenault@amd.com> llvm-svn: 217594
* R600: Add LDS_MIN_[U]INT[_RET] instructions for EvergreenAaron Watry2014-09-111-0/+8
| | | | | | Signed-off-by: Aaron Watry <awatry@gmail.com> Reviewed-by: Matt Arsenault <matthew.arsenault@amd.com> llvm-svn: 217593
* R600: Add LDS_XOR[_RET] instructions for EvergreenAaron Watry2014-09-111-0/+4
| | | | | | Signed-off-by: Aaron Watry <awatry@gmail.com> Reviewed-by: Matt Arsenault <matthew.arsenault@amd.com> llvm-svn: 217592
* R600: Add LDS_OR[_RET] instructions for EvergreenAaron Watry2014-09-111-0/+4
| | | | | | Signed-off-by: Aaron Watry <awatry@gmail.com> Reviewed-by: Matt Arsenault <matthew.arsenault@amd.com> llvm-svn: 217591
* R600: Add LDS_AND[_RET] instructions for EvergreenAaron Watry2014-09-111-0/+4
| | | | | | Signed-off-by: Aaron Watry <awatry@gmail.com> Reviewed-by: Matt Arsenault <matthew.arsenault@amd.com> llvm-svn: 217590
* R600: Add LDS_MAX_[U]INT[_RET] instructions for EvergreenAaron Watry2014-09-111-0/+8
| | | | | | | | | | | | | This was only present for SI before. Cayman may still be missing, but I am unable to test that currently. v2: Don't create atomicrmw max tests in separate file Signed-off-by: Aaron Watry <awatry@gmail.com> Reviewed-by: Matt Arsenault <matthew.arsenault@amd.com> CC: Tom Stellard <thomas.stellard@amd.com> llvm-svn: 217589
* Fix bug 20892 - clang-format does not handle C-style commentsRoman Kashitsyn2014-09-112-2/+19
| | | | | | | | | | | | | | | | | | | | | | Summary: http://llvm.org/bugs/show_bug.cgi?id=20892 Add support of C-style formatting enabling/disabling directives. Now the following two styles are supported: // clang-format on /* clang-format on */ The flexibility in comments (support of extra spaces and/or slashes, etc.) is deliberately avoided to simplify search in large code bases. Reviewers: djasper Reviewed By: djasper Subscribers: cfe-commits, curdeius, klimek Differential Revision: http://reviews.llvm.org/D5309 llvm-svn: 217588
* Remove executable bit on all header filesTobias Grosser2014-09-117-0/+0
| | | | | | Some header files had been marked executable by accident. llvm-svn: 217587
* Avoid some unnecessary SmallVector copies.Benjamin Kramer2014-09-115-13/+11
| | | | | | No functionality change. llvm-svn: 217586
* ARM Unwind syntaxRenato Golin2014-09-112-6/+30
| | | | | | | | | | | This patch fixes the bad argument that GAS accepted but the IAS didn't, ie. {#0x20}, moving it to {0x20} which both accept. It also makes the ARMv7+ save/restore correct by using VFP instructions rather than old co-processor ones. Fixes PR20529. llvm-svn: 217585
* [asan] Preserve existing LD_PRELOAD setting on Android.Evgeniy Stepanov2014-09-111-1/+1
| | | | llvm-svn: 217584
* [docs] Mention character array constants in docs/LangRef.rstDaniel Sanders2014-09-111-1/+3
| | | | | | | | | | | | | | | | | | Summary: They were used in the 'Module Structure' example but weren't otherwise documented. Credit to Reed Kotler for noticing. Reviewers: hans Reviewed By: hans Subscribers: hans, llvm-commits Differential Revision: http://reviews.llvm.org/D5191 llvm-svn: 217583
* [ARM] Add Thumb-2 code size optimization regression test for LSR (register).Tilmann Scheller2014-09-111-0/+9
| | | | llvm-svn: 217582
* [ARM] Add Thumb-2 code size optimization regression test for LSR (immediate).Tilmann Scheller2014-09-111-0/+9
| | | | llvm-svn: 217581
* [AArch64] Reenable the PBQP test now that the leak issue has been fixed.Arnaud A. de Grandmaison2014-09-111-0/+14
| | | | | | | | | | | David Blaikie's commits r217563 & r217564, which added shared_ptr to the CostPool have fixed some memory leak issues exposed by the PBQP with coalescing constraints. The sanitizer bot was failing because of those leaks. Now that the leaks are gone, we can reenable the aarch64/pbqp test. llvm-svn: 217580
* [ARM] Add Thumb-2 code size optimization regression test for LSL (register).Tilmann Scheller2014-09-111-0/+9
| | | | llvm-svn: 217579
* [mach-o]: support optional "0x" prefix for -image_baseTim Northover2014-09-112-2/+6
| | | | llvm-svn: 217578
* [mach-o]: tighten up diagnostics for -image_base optionTim Northover2014-09-112-2/+19
| | | | | | | | | The provided base must also be a multiple of the system's page size, which is a reasonable enough demand. Also check the other diagnostics more thoroughly. llvm-svn: 217577
* [ARM] Add Thumb2 code size optimization regression test for LSL (immediate).Tilmann Scheller2014-09-111-0/+9
| | | | llvm-svn: 217576
* [x86] Fixup r217565 which baked in an assumption about the functionChandler Carruth2014-09-111-1/+1
| | | | | | | name that breaks on some platforms. This part of the test just doesn't matter... llvm-svn: 217575
* [AlignmentFromAssumptions] Don't crash just because the target is 32-bitHal Finkel2014-09-112-0/+219
| | | | | | | | | We used to crash processing any relevant @llvm.assume on a 32-bit target (because we'd ask SE to subtract expressions of differing types). I've copied our 'simple.ll' test, but with the data layout from arm-linux-gnueabihf to get some meaningful test coverage here. llvm-svn: 217574
* NULL->nullptrAlexander Musman2014-09-111-1/+1
| | | | llvm-svn: 217573
* [mach-o]: Support deprecated -seg1addr alias for -image_baseTim Northover2014-09-111-0/+2
| | | | | | | Because NO LINKER MAY CHANGE. EVER. Even if it's a complete rewrite from scratch. llvm-svn: 217572
* Build correct vector filled with undef nodesDavid Xu2014-09-112-4/+62
| | | | llvm-svn: 217570
* Fix a couple of -Wsign-compare warnings introduced in r217556Justin Bogner2014-09-111-3/+2
| | | | llvm-svn: 217569
* Make getFlavor function.Rui Ueyama2014-09-111-24/+28
| | | | | | | The dangling "else" at the end of #if looked a bit error-prone. Make it a separate function. No functionality change. llvm-svn: 217568
* utils: Teach lldbDataFormatters how to format ArrayRefsJustin Bogner2014-09-111-0/+33
| | | | llvm-svn: 217567
* If lld is renamed (or symlinked) to "ld" automatically pick the right flavor.Nick Kledzik2014-09-114-6/+74
| | | | | | | | | | | | | The existing system linkers on Darwin and Linux are called "ld". We'd like to eventually drop in lld as "ld" and have it just work. But lld is a universal linker that requires the first option to be -flavor to know which command line mode to emulate (gnu or darwin). This change tests if argv[0] is "ld" and if so, if the tool was built on MacOSX then assume the darwin flavor otherwise the gnu flavor. There are two test cases which copy lld to "ld" and then run it. One for darwin and one for linux. llvm-svn: 217566
* [x86] FileCheck-ize this test.Chandler Carruth2014-09-111-5/+24
| | | | llvm-svn: 217565
* Use CostPool::PoolRef typedef some moreDavid Blaikie2014-09-111-5/+3
| | | | | | Cleanup to 217563 suggested by Lang Hames in post-commit review. llvm-svn: 217564
* shared_ptrify ownershp of PoolEntries in PBQP's CostPoolDavid Blaikie2014-09-101-46/+10
| | | | | | | | | | | | | | Leveraging both intrusive shared_ptr-ing (std::enable_shared_from_this) and shared_ptr<T>-owning-U (to allow external users to hold std::shared_ptr<CostT> while keeping the underlying PoolEntry alive). The intrusiveness could be removed if we had a weak_set that implicitly removed items from the set when their underlying data went away. This /might/ fix an existing memory leak reported by LeakSanitizer in r217504. llvm-svn: 217563
* R600/SI: Fix losing chain when fixing reg class of loads.Matt Arsenault2014-09-102-6/+40
| | | | | | | The lost chain resulting in earlier side effecting nodes being deleted. llvm-svn: 217561
* R600/SI: Report offset in correct units for st64 DS instructionsMatt Arsenault2014-09-101-0/+15
| | | | | | | | | | | Need to convert the 64 element offset into bytes, not just the element size like the normal case instructions. Noticed by inspection. This can't be hit now because st64 instructions aren't emitted during instruction selection, and the post-RA scheduler isn't enabled. llvm-svn: 217560
* [TSan] Use common flags in the same way as all the other sanitizersAlexey Samsonov2014-09-108-116/+37
| | | | llvm-svn: 217559
* [Sanitizer] Get rid of Symbolizer::Get() and Symbolizer::GetOrNull().Alexey Samsonov2014-09-1012-52/+27
| | | | | | | | | We may as well just use Symbolizer::GetOrInit() in all the cases. Don't call Symbolizer::Get() early in tools initialization: these days it doesn't do any important setup work, and we may as well create the symbolizer the first time it's actually needed. llvm-svn: 217558
* Add LLVMgold target to test dependencies.Peter Collingbourne2014-09-101-0/+4
| | | | llvm-svn: 217557
OpenPOWER on IntegriCloud