Commits · 64634c36dd8770ca85bf09137ce3a453dcd39b1f · Roger Ferrer / llvm-epi-0.8 · GitLab

This is an archived project. Repository and other project resources are read-only.

Nov 12, 2010
- tidy up. · 64634c36
  Chris Lattner authored Nov 12, 2010
  64634c36
Nov 11, 2010
- Remove the memmove->memcpy optimization from CodeGen. MemCpyOpt does this. · 6cf9bb45
  Dan Gohman authored Nov 11, 2010
  6cf9bb45
Nov 09, 2010
- Fix DAGCombiner to avoid folding a sext-in-reg or similar through a shl · 5db89214
  Dan Gohman authored Nov 09, 2010
  5db89214
- Fix an inline asm pasto from 117667; was preventing · f11ea9ce
  Dale Johannesen authored Nov 09, 2010
  f11ea9ce
Nov 05, 2010
- When passing a parameter using the 'byval' mechanism, inline code needs to be used · 6c25ca4f
  Duncan Sands authored Nov 05, 2010
  6c25ca4f
Nov 03, 2010
- Just return undef for invalid masks or elts, and since we're doing that, · c6418b10
  Eric Christopher authored Nov 03, 2010
  c6418b10
- Simplify uses of MVT and EVT. An MVT can be compared directly · 14627770
  Duncan Sands authored Nov 03, 2010
  14627770
- Inside the calling convention logic LocVT is always a simple · f5dda01f
  Duncan Sands authored Nov 03, 2010
  f5dda01f
- If we have an undef mask our Elt will be -1 for our access, handle · fcc9e684
  Eric Christopher authored Nov 03, 2010
  fcc9e684
- Fix DAGCombiner to avoid going into an infinite loop when it · 68fb0046
  Dan Gohman authored Nov 03, 2010
  68fb0046
- Two sets of changes. Sorry they are intermingled. · debf9c50
  Evan Cheng authored Nov 03, 2010
  debf9c50
Nov 02, 2010
- If value map does not have register for an argument then try to find frame index before giving up. · bc741405
  Devang Patel authored Nov 02, 2010
  bc741405
- Use frameindex, if available, as a last resort to emit debug info for a parameter. · 94f2a257
  Devang Patel authored Nov 02, 2010
  94f2a257
Oct 30, 2010
- Remove DAG combiner patch to fold vector splats. Instcombiner does it now. · 08882be8
  Bob Wilson authored Oct 29, 2010
  08882be8
Oct 29, 2010
- Avoiding overly aggressive latency scheduling. If the two nodes share an · 6c1414f9
  Evan Cheng authored Oct 29, 2010
  6c1414f9
- Inline asm multiple alternative constraints development phase 2 - improved... · e8360b71
  John Thompson authored Oct 29, 2010
  e8360b71
Oct 28, 2010
- Teach the DAG combiner to fold a splat of a splat. Radar 8597790. · f63da12b
  Bob Wilson authored Oct 28, 2010
  f63da12b
- Re-commit 117518 and 117519 now that ARM MC test failures are out of the way. · ff310737
  Evan Cheng authored Oct 28, 2010
  ff310737
- Revert 117518 and 117519 for now. They changed scheduling and cause MC tests to fail. Ugh. · e2c211c1
  Evan Cheng authored Oct 28, 2010
  e2c211c1
- Fix a major bug in operand latency computation. The use index must be adjusted · 523fa3a2
  Evan Cheng authored Oct 28, 2010
  523fa3a2
Oct 27, 2010
- Use a MemIntrinsicSDNode for ISD::PREFETCH, which touches · e660f4d0
  Dale Johannesen authored Oct 26, 2010
  e660f4d0
- Assign source ordering to nodes created for StoreInst. · 05561e8b
  Devang Patel authored Oct 26, 2010
  05561e8b
Oct 26, 2010
- For statistics that are only used in functions declared in !NDEBUG, wrap the · 90b2ac26
  Nick Lewycky authored Oct 26, 2010
  90b2ac26
Oct 25, 2010
- Simplify. · 43c3f4b6
  Devang Patel authored Oct 25, 2010
  43c3f4b6
- Add counters to count basic blocks and machine basic blocks with out of order line number info. · 3bc6d198
  Devang Patel authored Oct 25, 2010
  3bc6d198
Oct 23, 2010
- Move the remaining attribute macros to systematic names based on the attribute · 82058c05
  Chandler Carruth authored Oct 23, 2010
  82058c05
Oct 21, 2010
- X86: Base _fltused on the FunctionType of the called value instead of the... · 0e36e034
  Michael J. Spencer authored Oct 21, 2010
  0e36e034
- CodeGen-Windows: Only emit _fltused if a VarArg function is called with floating point args. · 83ce5f18
  Michael J. Spencer authored Oct 21, 2010
  83ce5f18
Oct 20, 2010
- Remove Synthesizable from the Type system; as MMX vector · 320a5533
  Dale Johannesen authored Oct 20, 2010
  320a5533
- Make CodeGen TBAA-aware. · a94cc6df
  Dan Gohman authored Oct 20, 2010
  a94cc6df
- Add a pre-dispatch SjLj EH hook on the unwind edge for targets to do any · bbdc5d2e
  Jim Grosbach authored Oct 19, 2010
  bbdc5d2e
Oct 19, 2010
- Get rid of static constructors for pass registration. Instead, every pass... · 6c18d1aa
  Owen Anderson authored Oct 19, 2010
  6c18d1aa
Oct 16, 2010
- X86-Windows: Emit an undefined global __fltused symbol when targeting Windows · 5e683250
  Michael J. Spencer authored Oct 16, 2010
  5e683250
- Whitespace! · d3ea25e6
  Michael J. Spencer authored Oct 16, 2010
  d3ea25e6
Oct 10, 2010
- fix the default va_arg expansion (in the realignment case) to not implicitly · eb313a46
  Chris Lattner authored Oct 10, 2010
  eb313a46
Oct 06, 2010
- ComputeLinearIndex doesn't need its TLI argument. · aadc5596
  Dan Gohman authored Oct 06, 2010
  aadc5596
- - Add TargetInstrInfo::getOperandLatency() to compute operand latencies. This · 49d4c0bd
  Evan Cheng authored Oct 06, 2010
  49d4c0bd
Oct 05, 2010
- Use a more efficient lowering of uint64_t --> float that can take advantage of... · d8d1dcc0
  Owen Anderson authored Oct 05, 2010
  d8d1dcc0
- This DAG combine BRCOND transformation can look pass truncate of the operand: · c8d6cfd7
  Evan Cheng authored Oct 04, 2010
  c8d6cfd7
Oct 01, 2010
- Fix code gen crash reported in PR 8235. We still lose debug info for the... · d3fe5fa5
  Devang Patel authored Oct 01, 2010
  d3fe5fa5