Commits · 6459bbe14a43313c92cbd6d5a6f6eb58b8775c39 · Roger Ferrer / llvm-epi-0.8 · GitLab

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

May 15, 2010
- sigh · 6459bbe1
  Chris Lattner authored May 15, 2010
  6459bbe1
- fix _mm_shuffle_pd too, thanks to Joel Falcou for pointing this out. · 8b3b1453
  Chris Lattner authored May 15, 2010
  8b3b1453
- When applying the named return value optimization, we still need to · 17012564
  Douglas Gregor authored May 15, 2010
  17012564
- Substantially alter the design of the Objective C type AST by introducing · 8b07ec25
  John McCall authored May 15, 2010
  8b07ec25
- Fix an GCC warning that seems to have actually caught a bug (!!!) in · 75142e6b
  Chandler Carruth authored May 15, 2010
  75142e6b
- Model 128-bit vld lane with REG_SEQUENCE. · 9e688cbc
  Evan Cheng authored May 15, 2010
  9e688cbc
- Implement a simple form of the C++ named return value optimization for · 290c93ec
  Douglas Gregor authored May 15, 2010
  290c93ec
- Calculate liveness on the fly for local registers. · 84ce2908
  Jakob Stoklund Olesen authored May 15, 2010
  84ce2908
- Implement semantic analysis and an AST representation for the named · 6fd1b180
  Douglas Gregor authored May 15, 2010
  6fd1b180
- add type casts to _mm_shuffle_ps so the front-end applies · 6625c702
  Chris Lattner authored May 15, 2010
  6625c702
- Teach the always inliner to release its inline cost estimates, like the basic · b35818eb
  Nick Lewycky authored May 15, 2010
  b35818eb
- Clean up, no functional change. · 002a45eb
  Nick Lewycky authored May 15, 2010
  002a45eb
- Add reg_nodbg_iterator · 3ced97c0
  Jakob Stoklund Olesen authored May 15, 2010
  3ced97c0
- v4i64 and v8i64 are only synthesizable when NEON is available. · 3d214cdf
  Evan Cheng authored May 15, 2010
  3d214cdf
- Allow TargetLowering::getRegClassFor() to be called on illegal types. Also · 4cad68eb
  Evan Cheng authored May 15, 2010
  4cad68eb
- <rdar://problem/7987491 > Libcompiler_rt isn't platform aware · 749dc33d
  Nick Kledzik authored May 15, 2010
  749dc33d
- Model 64-bit lane vld with REG_SEQUENCE. · 0cbd11df
  Evan Cheng authored May 15, 2010
  0cbd11df
- A partial re-def instruction may be a copy. · e26e56e7
  Evan Cheng authored May 15, 2010
  e26e56e7
- Recognize when the named return value optimization applies in a · 222cf0ef
  Douglas Gregor authored May 15, 2010
  222cf0ef
- C++/ABI/x86_64: Member pointers should be classified as INTEGER. · 36d4d154
  Daniel Dunbar authored May 15, 2010
  36d4d154
- C++/ABI/i386: Member function pointers should be passed by value. · 4bd95c65
  Daniel Dunbar authored May 15, 2010
  4bd95c65
- Change install_name for libclang.dylib to '@rpath/libclang.dylib'. Fixes <rdar://problem/7987039 >. · be4556f5
  Ted Kremenek authored May 14, 2010
  be4556f5
- Teach two-address pass to do some coalescing while eliminating REG_SEQUENCE · 8c2d062e
  Evan Cheng authored May 14, 2010
  8c2d062e
- Add some comments about undef. · 62c70559
  Dan Gohman authored May 14, 2010
  62c70559
- Model VST*_UPD and VST*oddUPD pair with REG_SEQUENCE. · cb78e555
  Evan Cheng authored May 14, 2010
  cb78e555
- Fast ISel trivially coalesces away no-op casts, so check for this when · 88fb2535
  Dan Gohman authored May 14, 2010
  88fb2535
- Don't bother spilling before a return · 089e9421
  Jakob Stoklund Olesen authored May 14, 2010
  089e9421
- RegAllocLocal can count copies too · cdef6bc8
  Jakob Stoklund Olesen authored May 14, 2010
  cdef6bc8
- SystemZ really does mean "has calls" and not just "adjusts stack." Go ahead and · 0160e558
  Bill Wendling authored May 14, 2010
  0160e558
- Track allocatable instead of reserved regs, and never take an unallocatable hint. · b1601393
  Jakob Stoklund Olesen authored May 14, 2010
  b1601393
- Don't set kill flags for instructions which the scheduler has cloned. · 2f277c86
  Dan Gohman authored May 14, 2010
  2f277c86
- BR is a barrier. · 062a97f0
  Dan Gohman authored May 14, 2010
  062a97f0
- Driver: Make -fverbose-asm the default for platforms where the integrated · 0d8ca9e9
  Daniel Dunbar authored May 14, 2010
  0d8ca9e9
May 14, 2010
- Several tail call tests apparently rely upon this being "adjusts stack" instead · 1713d958
  Bill Wendling authored May 14, 2010
  1713d958
- Avoid scanning the long tail of physreg operands on calls · e68b814c
  Jakob Stoklund Olesen authored May 14, 2010
  e68b814c
- Do not forget to mark prcessed arguments. · 36debf80
  Devang Patel authored May 14, 2010
  36debf80
- Count coalesced copies · 6c038e33
  Jakob Stoklund Olesen authored May 14, 2010
  6c038e33
- Allow virtreg redefines when verifying for RegAllocFast · 33af4fcd
  Jakob Stoklund Olesen authored May 14, 2010
  33af4fcd
- Tweak test so that it does not require <typeinfo> · 2d15ecad
  Douglas Gregor authored May 14, 2010
  2d15ecad
- [rand.dist.pois.poisson] · 0e675818
  Howard Hinnant authored May 14, 2010
  0e675818