aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* Support C99 hexadecimal floating-point literals in assemblyTim Northover2013-08-143-2/+90
| | | | | | | | It's useful to be able to write down floating-point numbers without having to worry about what they'll be rounded to (as C99 discovered), this extends that ability to the MC assembly parsers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188370 91177308-0d34-0410-b5e6-96231b3b80d8
* Fix a really terrifying but improbable bug in mem2reg. If you have seenChandler Carruth2013-08-141-1/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | extremely subtle miscompilations (such as a load getting replaced with the value stored *below* the load within a basic block) related to promoting an alloca to an SSA value, there is the dim possibility that you hit this. Please let me know if you won this unfortunate lottery. The first half of mem2reg's core logic (as it is used both in the standalone mem2reg pass and in SROA) builds up a mapping from 'Instruction *' to the index of that instruction within its basic block. This allows quickly establishing which store dominate a particular load even for large basic blocks. We cache this information throughout the run of mem2reg over a function in order to amortize the cost of computing it. This is not in and of itself a strange pattern in LLVM. However, it introduces a very important constraint: absolutely no instruction can be deleted from the program without updating the mapping. Otherwise a newly allocated instruction might get the same pointer address, and then end up with a wrong index. Yes, LLVM routinely suffers from a *single threaded* variant of the ABA problem. Most places in LLVM don't find avoiding this an imposition because they don't both delete and create new instructions iteratively, but mem2reg *loves* to do this... All the time. Fortunately, the mem2reg code was really careful about updating this cache to handle this eventuallity... except when it comes to the debug declare intrinsic. Oops. The fix is to invalidate that pointer in the cache when we delete it, the same as we do when deleting alloca instructions and other instructions. I've also caused the same bug in new code while working on a fix to PR16867, so this seems to be a really unfortunate pattern. Hopefully in subsequent patches the deletion of dead instructions can be consolidated sufficiently to make it less likely that we'll see future occurences of this bug. Sorry for not having a test case, but I have literally no idea how to reliably trigger this kind of thing. It may be single-threaded, but it remains an ABA problem. It would require a really amazing number of stars to align. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188367 91177308-0d34-0410-b5e6-96231b3b80d8
* Make more helper methods into static functions.Craig Topper2013-08-142-17/+17
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188366 91177308-0d34-0410-b5e6-96231b3b80d8
* Remove tab characters.Craig Topper2013-08-141-1/+1
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188365 91177308-0d34-0410-b5e6-96231b3b80d8
* Make some helper methods static.Craig Topper2013-08-142-34/+27
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188364 91177308-0d34-0410-b5e6-96231b3b80d8
* Use MVT in more lowering code.Craig Topper2013-08-141-6/+6
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188363 91177308-0d34-0410-b5e6-96231b3b80d8
* Replace EVT with MVT in isVectorShift. Keeps compiler from generating ↵Craig Topper2013-08-141-3/+5
| | | | | | unneeded checks and handling for extended types. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188362 91177308-0d34-0410-b5e6-96231b3b80d8
* Replace EVT with MVT in many of the shuffle lowering functions. Keeps ↵Craig Topper2013-08-141-37/+46
| | | | | | compiler from generating unneeded checks and handling for extended types. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188361 91177308-0d34-0410-b5e6-96231b3b80d8
* [lit] Fix a relative import issue I missed earlier.Daniel Dunbar2013-08-141-2/+4
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188360 91177308-0d34-0410-b5e6-96231b3b80d8
* [lit] Avoid StringIO.Daniel Dunbar2013-08-141-19/+25
| | | | | | | - We barely used it, and it is very hard to use in a 2.5-3 compatible way because of changing expectations for its input types. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188359 91177308-0d34-0410-b5e6-96231b3b80d8
* [lit] Fix tests to execute lit with same python as invoked with.Daniel Dunbar2013-08-143-3/+6
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188358 91177308-0d34-0410-b5e6-96231b3b80d8
* [lit] Factor ShTest format script command parsing from other processing.Daniel Dunbar2013-08-141-17/+32
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188357 91177308-0d34-0410-b5e6-96231b3b80d8
* [lit] Move executeCommand() into lit.util.Daniel Dunbar2013-08-144-23/+23
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188356 91177308-0d34-0410-b5e6-96231b3b80d8
* [lit] Move formats into their own subpackage.Daniel Dunbar2013-08-144-117/+130
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188355 91177308-0d34-0410-b5e6-96231b3b80d8
* [CMake] add_llvm_library: Specify explicit suffix .imp to import library to ↵NAKAMURA Takumi2013-08-141-0/+5
| | | | | | | | avoid a warning between profile_rt-static and profile_rt-shared with lib/profile_rt.lib. FIXME: It seems MS version of profile_rt.dll doesn't contain any export symbols. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188351 91177308-0d34-0410-b5e6-96231b3b80d8
* Lit: Introduce "%/[STpst] into parseIntegratedTestScript(), to normalize ↵NAKAMURA Takumi2013-08-141-0/+9
| | | | | | substitutions. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188348 91177308-0d34-0410-b5e6-96231b3b80d8
* [mips] Fix bug in parsing accumulator registers.Akira Hatanaka2013-08-142-2/+16
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188344 91177308-0d34-0410-b5e6-96231b3b80d8
* [mips] Use register operands instead of register classes in DSP instructionAkira Hatanaka2013-08-143-209/+309
| | | | | | | | definitions. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188343 91177308-0d34-0410-b5e6-96231b3b80d8
* [mips] Rename DSPRegs.Akira Hatanaka2013-08-145-113/+117
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188342 91177308-0d34-0410-b5e6-96231b3b80d8
* [mips] Rename HIRegs and LORegs.Akira Hatanaka2013-08-1410-84/+83
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188341 91177308-0d34-0410-b5e6-96231b3b80d8
* llvm/test/CodeGen/X86/setcc-sentinals.ll: Relax expressions for x86_64-win32.NAKAMURA Takumi2013-08-141-2/+2
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188340 91177308-0d34-0410-b5e6-96231b3b80d8
* Fix always creating GEP with i32 indicesMatt Arsenault2013-08-143-13/+38
| | | | | | | | | | | | | | | | Use the pointer size if datalayout is available. Use i64 if it's not, which is consistent with what other places do when the pointer size is unknown. The test doesn't really test this in a useful way since it will be transformed to that later anyway, but this now tests it for non-zero arrays and when datalayout isn't available. The cases in visitGetElementPtrInst should save an extra re-visit to the newly created GEP since it won't need to cleanup after itself. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188339 91177308-0d34-0410-b5e6-96231b3b80d8
* Use type helper functions instead of castMatt Arsenault2013-08-142-11/+8
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188338 91177308-0d34-0410-b5e6-96231b3b80d8
* Use array initializer, space around operatorMatt Arsenault2013-08-141-5/+3
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188337 91177308-0d34-0410-b5e6-96231b3b80d8
* [mips] Properly parse registers that appear in inline-asm constraints.Akira Hatanaka2013-08-143-0/+104
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188336 91177308-0d34-0410-b5e6-96231b3b80d8
* BBVectorize: Add initial stores to the write set when tracking usesHal Finkel2013-08-132-0/+153
| | | | | | | | | | | | | | | | When computing the use set of a store, we need to add the store to the write set prior to iterating over later instructions. Otherwise, if there is a later aliasing load of that store, that load will not be tagged as a use, and bad things will happen. trackUsesOfI still adds later dependent stores of an instruction to that instruction's write set, but it never sees the original instruction, and so when tracking uses of a store, the store must be added to the write set by the caller. Fixes PR16834. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188329 91177308-0d34-0410-b5e6-96231b3b80d8
* Remove duplicate copy of testcase in r188327.Nick Lewycky2013-08-131-26/+0
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188328 91177308-0d34-0410-b5e6-96231b3b80d8
* Revert r187191, which broke opt -mem2reg on the testcases included in PR16867.Nick Lewycky2013-08-137-325/+168
| | | | | | | | | | | | | However, opt -O2 doesn't run mem2reg directly so nobody noticed until r188146 when SROA started sending more things directly down the PromoteMemToReg path. In order to revert r187191, I also revert dependent revisions r187296, r187322 and r188146. Fixes PR16867. Does not add the testcases from that PR, but both of them should get added for both mem2reg and sroa when this revert gets unreverted. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188327 91177308-0d34-0410-b5e6-96231b3b80d8
* [Mips][msa] Value types for MSA support.Jack Carter2013-08-134-24/+33
| | | | | | | | | | Added v8f16 to ValueTypes.h, ValueTypes.cpp, ValueTypes.td, and CodeGenTarget.cpp Patch by Daniel Sanders git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188326 91177308-0d34-0410-b5e6-96231b3b80d8
* Options: explicit handling of --Hans Wennborg2013-08-132-30/+1
| | | | | | | | | | | Clients of the option parsing library should handle it explicitly using a KIND_REMAINING_ARGS option. Clang and lld have been updated in r188316 and r188318, respectively. Also fix -Wsign-compare warning in the option parsing test. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188323 91177308-0d34-0410-b5e6-96231b3b80d8
* DAG: Combine (and (setne X, 0), (setne X, -1)) -> (setuge (add X, 1), 2)Jim Grosbach2013-08-133-0/+40
| | | | | | | | | | | | | | | | | | | | | | | A common idiom is to use zero and all-ones as sentinal values and to check for both in a single conditional ("x != 0 && x != (unsigned)-1"). That generates code, for i32, like: testl %edi, %edi setne %al cmpl $-1, %edi setne %cl andb %al, %cl With this transform, we generate the simpler: incl %edi cmpl $1, %edi seta %al Similar improvements for other integer sizes and on other platforms. In general, combining the two setcc instructions into one is better. rdar://14689217 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188315 91177308-0d34-0410-b5e6-96231b3b80d8
* Options: Add new option kind that consumes remaining argumentsHans Wennborg2013-08-136-0/+47
| | | | | | | | | | | | | | | | This adds KIND_REMAINING_ARGS, a class of options that consume all remaining arguments on the command line. This will be used to support /link in clang-cl, which is used to forward all remaining arguments to the linker. It also allows us to remove the hard-coded handling of "--", allowing clients (clang and lld) to implement that functionality themselves with this new option class. Differential Revision: http://llvm-reviews.chandlerc.com/D1387 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188314 91177308-0d34-0410-b5e6-96231b3b80d8
* [Mips][msa] Added initial MSA support.Jack Carter2013-08-138-1/+176
| | | | | | | | | | | | | * msa SubtargetFeature * registers * ld.[bhwd], and st.[bhwd] instructions Does not correctly prohibit use of both 32-bit FPU registers and MSA together. Patch by Daniel Sanders git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188313 91177308-0d34-0410-b5e6-96231b3b80d8
* [Mips] Support for unaligned load/store microMips instructionsJack Carter2013-08-1310-3/+157
| | | | | | | | | This includes instructions lwl, lwr, swl and swr. Patch by Zoran Jovnovic git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188312 91177308-0d34-0410-b5e6-96231b3b80d8
* [lit] Support use of setup.py from other directories.Daniel Dunbar2013-08-131-1/+8
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188309 91177308-0d34-0410-b5e6-96231b3b80d8
* Update makeLibCall to return both the call and the chain associated with the ↵Michael Gottesman2013-08-136-67/+80
| | | | | | | | | | | | | | | libcall instead of just the call. This allows us to specify libcalls that return void. LowerCallTo returns a pair with the return value of the call as the first element and the chain associated with the return value as the second element. If we lower a call that has a void return value, LowerCallTo returns an SDValue with a NULL SDNode and the chain for the call. Thus makeLibCall by just returning the first value makes it impossible for you to set up the chain so that the call is not eliminated as dead code. I also updated all references to makeLibCall to reflect the new return type. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188300 91177308-0d34-0410-b5e6-96231b3b80d8
* Output DW_AT_stmt_list dwarf debug info as DW_FORM_sec_offset instead of ↵Carlo Kok2013-08-132-1/+41
| | | | | | DW_FORM_data4 as it is a section offset (fixes the coff/dwarf debug info statement locations) git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188297 91177308-0d34-0410-b5e6-96231b3b80d8
* For COFF only: dwarf debug info output a label reference as a section ↵Carlo Kok2013-08-133-6/+10
| | | | | | relative item only when it's one of dw_from strp, sec_offset, ref_addr or op_call_ref instead of going by size. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188296 91177308-0d34-0410-b5e6-96231b3b80d8
* dfsan: fix lint warningsDmitry Vyukov2013-08-131-2/+2
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188293 91177308-0d34-0410-b5e6-96231b3b80d8
* ARMv8: SWP and SWPB are obsoleted on ARMv8.Joey Gouly2013-08-132-2/+13
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188288 91177308-0d34-0410-b5e6-96231b3b80d8
* Also remove logic in LateVectorizeArnold Schwaighofer2013-08-131-1/+1
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188285 91177308-0d34-0410-b5e6-96231b3b80d8
* Remove logic that decides whether to vectorize or not depending on O-levelsArnold Schwaighofer2013-08-132-1/+3
| | | | | | I have moved this logic into clang and opt. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188281 91177308-0d34-0410-b5e6-96231b3b80d8
* Fix compiler warnings.Evgeniy Stepanov2013-08-131-6/+4
| | | | | | | | | | | | ../lib/Target/X86/X86ISelLowering.cpp:9715:7: error: unused variable 'OpVT' [-Werror,-Wunused-variable] EVT OpVT = Op0.getValueType(); ^ ../lib/Target/X86/X86ISelLowering.cpp:9763:14: error: unused variable 'NumElems' [-Werror,-Wunused-variable] unsigned NumElems = VT.getVectorNumElements(); git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188269 91177308-0d34-0410-b5e6-96231b3b80d8
* Fix signed overflow in when computing encodings for ADR instructionsMihai Popa2013-08-132-2/+3
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188268 91177308-0d34-0410-b5e6-96231b3b80d8
* AVX-512: Added CMP and BLEND instructions.Elena Demikhovsky2013-08-135-9/+406
| | | | | | | Lowering for SETCC. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188265 91177308-0d34-0410-b5e6-96231b3b80d8
* This patch introduces changes to MipsAsmParser register parsing routines. ↵Vladimir Medic2013-08-131-176/+146
| | | | | | The code now follows more deterministic path and makes the code more efficient and easier to maintain. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188264 91177308-0d34-0410-b5e6-96231b3b80d8
* GCC warns about removing const with a c-style cast.Benjamin Kramer2013-08-131-1/+1
| | | | git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188259 91177308-0d34-0410-b5e6-96231b3b80d8
* Pass DIEHash::collectAttributes output argument by-pointer instead of by-value.Evgeniy Stepanov2013-08-133-7/+7
| | | | | | | Before this, collectAttributes() was operating on a local object. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188254 91177308-0d34-0410-b5e6-96231b3b80d8
* [-cxx-abi microsoft] Stick zero initialized symbols into the .bss section ↵David Majnemer2013-08-133-4/+19
| | | | | | | | | | | | | | | | | | | | | | | for COFF Summary: We need to do two things: - Initialize BSSSection in MCObjectFileInfo::InitCOFFMCObjectFileInfo - Teach TargetLoweringObjectFileCOFF::SelectSectionForGlobal what to do with it This fixes PR16861. Reviewers: rnk Reviewed By: rnk CC: llvm-commits Differential Revision: http://llvm-reviews.chandlerc.com/D1361 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188244 91177308-0d34-0410-b5e6-96231b3b80d8
* Add the start of DIE hashing for DWARF4 type units and split dwarfEric Christopher2013-08-134-5/+164
| | | | | | | | | | CUs. Currently only hashes the name of CUs and the names of any children, but it's an obvious first step to show the framework. The testcase should continue to be correct, however, as it's an empty TU. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188243 91177308-0d34-0410-b5e6-96231b3b80d8
generated by cgit v1.1 at 2025-02-10 07:23:33 +0000