Commits · f82dab1a7a68b76c54f670c960bfd16acbb70c43 · Roger Ferrer / llvm-epi · GitLab

Sep 05, 2008
- For whatever the reason, x86 CallingConv::Fast (i.e. fastcc) was not passing... · 6c94b99c
  Evan Cheng authored Sep 04, 2008
  6c94b99c
Sep 04, 2008
- If function notes say optimize for size, then adjust alignment. · b9d5e028
  Devang Patel authored Sep 04, 2008
  b9d5e028
- Tidy up several unbeseeming casts from pointer to intptr_t. · a79db30d
  Dan Gohman authored Sep 04, 2008
  a79db30d
- Fix the ordering of operands to the store (inverted relative to LLVM IR), and fix the testcase. · b8c7ba22
  Owen Anderson authored Sep 04, 2008
  b8c7ba22
- Add a first attempt at implementing stores for X86 fast isel using target hooks. · 4f948bd8
  Owen Anderson authored Sep 04, 2008
  4f948bd8
- Load from GV stub should be locally CSE'd. · 8d8f47d5
  Evan Cheng authored Sep 04, 2008
  8d8f47d5
- Remove code that pad number of bytes to pop for X86_FastCall CC. The code... · 3152edf4
  Evan Cheng authored Sep 04, 2008
  3152edf4
- Add intrinsics for log, log2, log10, exp, exp2. · da2d8068
  Dale Johannesen authored Sep 04, 2008
  da2d8068
- Create HandlePHINodesInSuccessorBlocksFast, a version of · 7bda51f5
  Dan Gohman authored Sep 03, 2008
  7bda51f5
Sep 03, 2008
- Add X86 target hook to implement load (even from GlobalAddress). · a41ee297
  Evan Cheng authored Sep 03, 2008
  a41ee297
- Fix capitalization in #include of FastISel.h. This unbreaks the build on... · 2175b55d
  Ted Kremenek authored Sep 03, 2008
  2175b55d
- Unbreak fast isel. · 8f23ec96
  Evan Cheng authored Sep 03, 2008
  8f23ec96
- Let tblgen only generate fastisel routines, not the class definition. This... · 24422d49
  Evan Cheng authored Sep 03, 2008
  24422d49
Aug 31, 2008
- fix a bunch of 80-col violations · 81d6a384
  Gabor Greif authored Aug 31, 2008
  81d6a384
Aug 30, 2008
- For now, can't mark XOR64rr isAsCheapAsAMove. It's technically correct. But... · b8d588d8
  Evan Cheng authored Aug 30, 2008
  b8d588d8
- Transform (x << (y&31)) -> (x << y). This takes advantage of the fact x86... · cfb7f3ab
  Evan Cheng authored Aug 30, 2008
  cfb7f3ab
- Swap fp comparison operands and change predicate to allow load folding (safely this time). · 3fddc7e9
  Evan Cheng authored Aug 29, 2008
  3fddc7e9
- Use static_cast instead of C style cast. · f93bc7f7
  Evan Cheng authored Aug 29, 2008
  f93bc7f7
- Backing out 55521. Not safe. · b3ed0970
  Evan Cheng authored Aug 29, 2008
  b3ed0970
Aug 29, 2008
- Add initial support for fast isel of instructions that have inputs pinned to physical registers. · 0673a8af
  Owen Anderson authored Aug 29, 2008
  0673a8af
- Swap fp comparison operands and change predicate to allow load folding. · 960b17a3
  Evan Cheng authored Aug 28, 2008
  960b17a3
- Add a target callback for FastISel. · d58f3e36
  Dan Gohman authored Aug 28, 2008
  d58f3e36
- remove tabs, fix > 80 cols · 95d77f54
  Gabor Greif authored Aug 28, 2008
  95d77f54
Aug 28, 2008
- erect abstraction boundaries for accessing SDValue members, rename Val -> Node to reflect semantics · f304a7aa
  Gabor Greif authored Aug 28, 2008
  f304a7aa
- Use resize instead of reserve. Reserve doesn't change size(). · 26d54b3e
  Rafael Espindola authored Aug 28, 2008
  26d54b3e
- FsFLD0S{S|D} and V_SETALLONES are as cheap as moves. · 97af20f8
  Evan Cheng authored Aug 28, 2008
  97af20f8
- Split the ATOMIC NodeType's to include the size, e.g. · 41be0d44
  Dale Johannesen authored Aug 28, 2008
  41be0d44
Aug 27, 2008
- Make "movdq2q" and "movq2dq" dependent upon having SSE2 because they use the · 76105a4a
  Bill Wendling authored Aug 27, 2008
  76105a4a
- Reinstate the x86-64 portion of r55190. When doing extloads into · 7f2777e8
  Dan Gohman authored Aug 27, 2008
  7f2777e8
- disallow direct access to SDValue::ResNo, provide a getter instead · abfdf928
  Gabor Greif authored Aug 26, 2008
  abfdf928
Aug 26, 2008
- These assertions should be return false's instead, allowing the client to detect the failure. · 42ccd396
  Owen Anderson authored Aug 26, 2008
  42ccd396
- Make TargetInstrInfo::copyRegToReg return a bool indicating whether the copy requested · 27fb3dcb
  Owen Anderson authored Aug 26, 2008
  27fb3dcb
- If an xmm register is referenced explicitly in an inline asm, make sure to · 09f8cef5
  Chris Lattner authored Aug 26, 2008
  09f8cef5
- This is done. · 4884f878
  Evan Cheng authored Aug 26, 2008
  4884f878
Aug 25, 2008
- 80 col. violations. · 2da2e65d
  Evan Cheng authored Aug 25, 2008
  2da2e65d
- Try approach to moving call address load inside of callseq_start. Now it's... · f00f1e50
  Evan Cheng authored Aug 25, 2008
  f00f1e50
- Nevermind. This broke the bootstrap (?!). · 6cfd3830
  Bill Wendling authored Aug 25, 2008
  6cfd3830
- MOVQ2DQ and MOVQ2DQ use SSE2. We should conditionalize the use of these · dd6759ae
  Bill Wendling authored Aug 25, 2008
  dd6759ae
- Fix asm printing of MOVSDto64mr and MOV64toSDrm. · e4146813
  Evan Cheng authored Aug 25, 2008
  e4146813
Aug 24, 2008
- Temporarily reverting r55292. It's causing a bootstraping failure: · 5b836c5f
  Bill Wendling authored Aug 24, 2008
  5b836c5f