Commits · 80e4ac802a93429333eda717bc6f8cfbf02e8be7 · Roger Ferrer / llvm-epi-0.8 · GitLab

This project is archived. Its data is read-only.

Aug 13, 2015
- [AArch64] Provide "too few operands" diags on short-form NEON also. · 80e4ac80
  Ahmed Bougacha authored Aug 13, 2015
  80e4ac80
- MIR Parser: Don't allow negative alignments for memory operands. · 68661040
  Alex Lorenz authored Aug 13, 2015
  68661040
- [X86][SSE] Use the general SMAX/SMIN/UMAX/UMIN pattern matching and remove the X86 implementation · 4a8d6b3b
  Simon Pilgrim authored Aug 13, 2015
  4a8d6b3b
- [SimplifyLibCalls] Correctly set the is_zero_undef flag for llvm.cttz · a195386c
  Davide Italiano authored Aug 13, 2015
  a195386c
- MIR Parser: Extract the code that parses the alignment into a new method. NFC. · 620f8914
  Alex Lorenz authored Aug 13, 2015
  620f8914
- MIR Parser: Rename the method 'diagFromLLVMAssemblyDiag'. NFC. · 9b62cf61
  Alex Lorenz authored Aug 13, 2015
  9b62cf61
- [SeparateConstOffsetFromGEP] strengthen the inbounds attribute · 13a80eac
  Jingyue Wu authored Aug 13, 2015
  13a80eac
- Remove and forbid raw_svector_ostream::flush() calls. · 556b21aa
  Yaron Keren authored Aug 13, 2015
  556b21aa
- Fix GCC warning: extra `;' [-Wpedantic]. · e2f6fb5d
  Nick Lewycky authored Aug 13, 2015
  e2f6fb5d
- Scalar to vector conversions using direct moves · 1c39ca65
  Nemanja Ivanovic authored Aug 13, 2015
  1c39ca65
- [ARM] FMINNAN/FMAXNAN of f64 are not legal. · 31117875
  James Molloy authored Aug 13, 2015
  31117875
- [ARM] Allow vmin/vmax of scalars to be emitted without UseNEONForFP. · c71f78f4
  James Molloy authored Aug 13, 2015
  c71f78f4
- [DeadStoreElimination] remove a redundant store even if the load is in a different block. · 11fc8175
  Erik Eckstein authored Aug 13, 2015
  11fc8175
- [mips][mcjit] Calculate correct addend for HI16 and PCHI16 reloc · d22164dc
  Petar Jovanovic authored Aug 13, 2015
  d22164dc
- [WinEHPrepare] Update demotion logic · c9ff914c
  Joseph Tremoulet authored Aug 13, 2015
  c9ff914c
- [SystemZ] Support large LLVM IR struct return values · a887f062
  Ulrich Weigand authored Aug 13, 2015
  a887f062
- Remove raw_svector_ostream::resync and users. It's no-op after r244870. · a3668a3f
  Yaron Keren authored Aug 13, 2015
  a3668a3f
- [InstCombinePHI] Partial simplification of identity operations. · 6153698f
  Charlie Turner authored Aug 13, 2015
  6153698f
- Revert "[LIR] Start leveraging the fundamental guarantees of a loop..." · 655348f0
  Renato Golin authored Aug 13, 2015
  655348f0
- Revert "[LIR] Handle access to AliasAnalysis the same way as the other... · 4d57906b
  Renato Golin authored Aug 13, 2015
  4d57906b
- Test Commit. · 47802628
  Ashutosh Nema authored Aug 13, 2015
  47802628
- [ARM] Reorganise and simplify thumb-1 load/store selection · 68acdcb4
  John Brawn authored Aug 13, 2015
  68acdcb4
- [LIR] Handle access to AliasAnalysis the same way as the other analysis · c2af0982
  Chandler Carruth authored Aug 13, 2015
  c2af0982
- [LIR] Start leveraging the fundamental guarantees of a loop in · 8ae7b815
  Chandler Carruth authored Aug 13, 2015
  8ae7b815
- [LIR] Handle the LoopInfo the same as all the other analyses. No utility · 18c2669a
  Chandler Carruth authored Aug 13, 2015
  18c2669a
- [InstCombine] SSE/AVX vector shifts demanded shift amount bits · becd5e8a
  Simon Pilgrim authored Aug 13, 2015
  becd5e8a
- Modify raw_svector_ostream to use its SmallString without additional buffering. · 3d1173ba
  Yaron Keren authored Aug 13, 2015
  3d1173ba
- [LoopUnswitch] Check OptimizeForSize before traversing over all basic blocks in current loop · f458c6f3
  Chen Li authored Aug 13, 2015
  f458c6f3
- [CodeGen] Mark the promoted FCOPYSIGN result FP_ROUND as TRUNCating. · a196661b
  Ahmed Bougacha authored Aug 13, 2015
  a196661b
- [AArch64] Also custom-lowering mismatched vector/f16 FCOPYSIGN. · 2a97b1bc
  Ahmed Bougacha authored Aug 13, 2015
  2a97b1bc
- [CodeGen] Assert on getNode(FP_EXTEND) with a smaller dst type. · b5b0cfdf
  Ahmed Bougacha authored Aug 13, 2015
  b5b0cfdf
- [CodeGen] When Promoting, don't extend the 2nd FCOPYSIGN operand. · 40ded502
  Ahmed Bougacha authored Aug 13, 2015
  40ded502
- [CodeGen] Simplify getNode(*EXT/TRUNC) type size assert. NFC. · 31e0d9a2
  Ahmed Bougacha authored Aug 13, 2015
  31e0d9a2
- There is only one saver of strings. · b82455d2
  Rafael Espindola authored Aug 13, 2015
  b82455d2
- [LIR] Make the LoopIdiomRecognize pass get analyses essentially the same · dc298329
  Chandler Carruth authored Aug 13, 2015
  dc298329
- [LIR] Remove the dedicated class for popcount recognition and sink the · 8219a501
  Chandler Carruth authored Aug 13, 2015
  8219a501
- Return ErrorOr from FileOutputBuffer::create. NFC. · 169284a6
  Rafael Espindola authored Aug 13, 2015
  169284a6
- [LIR] Move all the helpers to be private and re-order the methods in · d9c6070c
  Chandler Carruth authored Aug 13, 2015
  d9c6070c
- [LIR] Remove the 'LIRUtils' abstraction which was unnecessary and adding · be158b17
  Chandler Carruth authored Aug 12, 2015
  be158b17
- [LIR] Run clang-format over LoopIdiomRecognize in preparation for · bad690e8
  Chandler Carruth authored Aug 12, 2015
  bad690e8