- 9096028 Fix an overly strict assertion. Source register of a copy may not be killed, it may be killed by an implicit super-register use. by Evan Cheng · 17 years ago
- c2feb5c Remove code that pad number of bytes to pop for X86_FastCall CC. The code doesn't do the "aligning" for Cygwin, Mingw, and Windows. But aligning it on Darwin and Linux breaks gcc compatibility. That ruled out all the platforms we support! by Evan Cheng · 17 years ago
- 7794f2a Add intrinsics for log, log2, log10, exp, exp2. by Dale Johannesen · 17 years ago
- 104e4ce Do trivial local CSE for constants and other non-Instruction values in FastISel. by Dan Gohman · 17 years ago
- 2a7c671 Put RegsForValue in the llvm namespace to avoid warnings about by Dan Gohman · 17 years ago
- 3df24e6 Create HandlePHINodesInSuccessorBlocksFast, a version of by Dan Gohman · 17 years ago
- 7bbb433 Update inline threshold for current function if the notes say, optimize for size. by Devang Patel · 17 years ago
- 0c7f91c Fix a bug that prevented PRE from applying in some cases. by Owen Anderson · 17 years ago
- 72ab076 Avoid extra comma. by Devang Patel · 17 years ago
- 52e89dd Parse and print opt_size note. by Devang Patel · 17 years ago
- cf996d4 Initial version of a Partial Specialization IPO pass. It triggers a couple hundred times on 176.gcc. I don't know the performance impact yet, the heuristic is quite simple still. by Andrew Lenharth · 17 years ago
- 61e6093 Do not emit a UsedDirective for things in the llvm.used by Dale Johannesen · 17 years ago
- 7946e7b Fix typo in a comment. by Devang Patel · 17 years ago
- 3d566dd Add missing decls. by Devang Patel · 17 years ago
- 79cca4f Add parentheses to make code more readable. by Devang Patel · 17 years ago
- 3fb6837 Fix comments. by Devang Patel · 17 years ago
- faf600a Testcase for commits 55700 and 55714. by Duncan Sands · 17 years ago
- 34f2a0c If a SCC has a node without a function, then the SCC by Duncan Sands · 17 years ago
- 22ec199 Add custom inliner that handles only functions that are marked as always_inline. by Devang Patel · 17 years ago
- 6724339 Handle "always inline" note during inline cost analysis. by Devang Patel · 17 years ago
- 5e06010 Check noinline note and ignore other notes. by Devang Patel · 17 years ago
- 910c120 Handle "noinline" note inside the simple inliner. by Devang Patel · 17 years ago
- d5d81a4 Oops, I accidentally broke the fallback case with my last commit. by Owen Anderson · 17 years ago
- 99aaf10 Fix an issue where we were reusing materializations of constants in blocks not dominated by the materialization. This is by Owen Anderson · 17 years ago
- 6bd8aa8 Update test to check call instruction. by Devang Patel · 17 years ago
- f0cbcd4 Split the SelectionDAG-building code, including the FunctionLoweringInfo by Dan Gohman · 17 years ago
- b070bee Fix maxo bado thinko. by Duncan Sands · 17 years ago
- 94b8d7e Separate MachineInstr-emitting routines from actual scheduling by Dan Gohman · 17 years ago
- 3f62940 Fix addRegisterDead and addRegisterKilled to be more thorough by Dan Gohman · 17 years ago
- d0ac373 Since onlyReadsMemory returns true if in fact by Duncan Sands · 17 years ago
- 9a036b9 Cleanup GlobalsModRef a bit. When analysing the by Duncan Sands · 17 years ago
- dced0a3 Testcase for PR1678. by Duncan Sands · 17 years ago
- 8373d38 Try to fold each element of a vector. This is needed to maintain structural by Nick Lewycky · 17 years ago
- 8b19e56 Add X86 target hook to implement load (even from GlobalAddress). by Evan Cheng · 17 years ago
- f45261f If TargetSelectInstruction returns true, move to next instruction. by Evan Cheng · 17 years ago
- ea09f4f Make UpdateValueMap, createResultReg, etc. protected instead of private so they can used by target hooks. by Evan Cheng · 17 years ago
- 23c0430 Don't apply this transform to vectors. Fixes PR2756. by Nick Lewycky · 17 years ago
- 83c05e5 Don't crash when trying to constant fold a vector with some elements that can't by Nick Lewycky · 17 years ago
- b388eb8 Fix capitalization in #include of FastISel.h. This unbreaks the build on case-sensitive filesystems. by Ted Kremenek · 17 years ago
- 88e3041 Unbreak fast isel. by Evan Cheng · 17 years ago
- e925ccc Add additional check to ensure that iv is canonicalized. by Devang Patel · 17 years ago
- a9348dc Check iteration count. by Devang Patel · 17 years ago
- c3f44b0 Let tblgen only generate fastisel routines, not the class definition. This makes it easier for targets to define its own fastisel class. by Evan Cheng · 17 years ago
- c721391 While removing PHI, use basicblock to identify incoming value. by Devang Patel · 17 years ago
- cb7f483 s/FP_AlwaysInline/FN_NOTE_AlwaysInline/g by Devang Patel · 17 years ago
- 108f92e If all IV uses are extending integer IV then change the type of IV itself, if possible. by Devang Patel · 17 years ago
- 691e524 respect inline=never and inline=always notes. by Devang Patel · 17 years ago
- 36fd941 80 col violations. by Evan Cheng · 17 years ago
- 7ad033c Read and write function notes. by Devang Patel · 17 years ago
- 1b47e48 Use bitwise AND. by Devang Patel · 17 years ago
- 09a2ffe New testcase for targets without 64-bit atomics; xfail old test for ppc. by Dale Johannesen · 17 years ago
- 9018f56 Function notes tests. by Devang Patel · 17 years ago
- 94bdae1 Print function notes. by Devang Patel · 17 years ago
- d498081 Parse function notes. by Devang Patel · 17 years ago
- 81b2ab8 Initialize function notes. by Devang Patel · 17 years ago
- a619d01 Fix some bugs in the code sequences for atomics. by Dale Johannesen · 17 years ago
- d2ff647 Ensure that HandlePHINodesInSuccessorBlocks is run for all blocks, by Dan Gohman · 17 years ago
- 21dbb99 Getter and setter for function notes. by Devang Patel · 17 years ago
- ecfe67c Reapply majority of r55557 but with the changes to compilation flags by Daniel Dunbar · 17 years ago
- 5b8dbf8 plug memleak: destroy internal buffer by Nuno Lopes · 17 years ago
- 6e95d90 plug a little memleak in verifyFunction() by Nuno Lopes · 17 years ago
- 0783f0d Revert r55557, it is causing linking failures on 32bit linux. by Matthijs Kooijman · 17 years ago
- 252ddfb MMI may be null. by Evan Cheng · 17 years ago
- 95ce117 Add Mac OS X compatible JIT callback routine. by Evan Cheng · 17 years ago
- 7602e11 Revamp ARM JIT. by Evan Cheng · 17 years ago
- acff339 Change getBinaryCodeForInstr prototype. First operand MachineInstr& should be const. Make corresponding changes. by Evan Cheng · 17 years ago
- ed63214 Provide two overloads of AnalyzeNewNode. by Gabor Greif · 17 years ago
- 1189f3a Even though no caller actually uses the new value by Duncan Sands · 17 years ago
- 14a093a Turn this legalize types test on. by Duncan Sands · 17 years ago
- 3cd652d Add a small pass that sets the readnone/readonly by Duncan Sands · 17 years ago
- 3aac788 Control flow instruction encodings. by Evan Cheng · 17 years ago
- 3c2ee49 ldm / stm instruction encodings. by Evan Cheng · 17 years ago
- 5d2c1cf AXI2 and AXI3 instruction encodings. by Evan Cheng · 17 years ago
- 4bbd5f8 Reorganize instruction formats again; AXI1 encoding. by Evan Cheng · 17 years ago
- 840917b addrmode3 instruction encodings. by Evan Cheng · 17 years ago
- 0d14fc8 Reorganize some instruction format definitions. No functionality change. by Evan Cheng · 17 years ago
- 9391273 Rest of addrmode2 instruction encodings. by Evan Cheng · 17 years ago
- 17222df Addr2 word / byte load encodings. by Evan Cheng · 17 years ago
- b7880ac Addr1 instructions opcodes are encoded in bits 21-24; encode S bit. by Evan Cheng · 17 years ago
- 93c53e5 fix a bunch of 80-col violations by Gabor Greif · 17 years ago
- 9440e35 Revert the "XFAIL" for the rotate_ops.ll testcase. Instead, mark ISD::ROTR by Bill Wendling · 17 years ago
- 3156b62 Expand for ROTR with MVT::i64. by Bill Wendling · 17 years ago
- 70fcb6b CellSPU doesn't appear to support fully the "ISD::ROTR" operation. The DAG by Bill Wendling · 17 years ago
- acb04ec Cosmetic changes to Machine LICM. No functionality change. by Bill Wendling · 17 years ago
- 2692d59 Another situation where ROTR is cheaper than ROTL. by Bill Wendling · 17 years ago
- 353dea2 For this pattern, ROTR is the cheaper option. by Bill Wendling · 17 years ago
- c5cbda1 - Fix comment so that it describes how the code really works: by Bill Wendling · 17 years ago
- dc71563 typo by Gabor Greif · 17 years ago
- 12632d2 fix some 80-col violations by Gabor Greif · 17 years ago
- 07cabf6 PR2731: C and Ocaml bindings for setTailCall and isTailCall. by Gordon Henriksen · 17 years ago
- 699609c Fix ConstantExpr::getInsertElement. by Gordon Henriksen · 17 years ago
- 9236268 fix some 80-col violations by Gabor Greif · 17 years ago
- cd04708 Re-apply 55467 with fix. If copy is being replaced by remat'ed def, transfer the implicit defs onto the remat'ed instruction. by Evan Cheng · 17 years ago
- 3689ff4 Fold isRematerializable checks into isSafeToReMat. by Evan Cheng · 17 years ago
- b18ae3c For now, can't mark XOR64rr isAsCheapAsAMove. It's technically correct. But various passes cannot handle remating these. by Evan Cheng · 17 years ago
- eb9f892 Transform (x << (y&31)) -> (x << y). This takes advantage of the fact x86 shift instructions 2nd operand (shift count) is limited to 0 to 31 (or 63 in the x86-64 case). by Evan Cheng · 17 years ago
- dcf114e Enable -fvisibility-inlines-hidden by default for compilers which support it. by Daniel Dunbar · 17 years ago
- c4db727 Testcases for ppc atomics. by Dale Johannesen · 17 years ago
- cc54e76 Fix an issue where a use might be selected before a def, and then we didn't respect the pre-chosen vreg by Owen Anderson · 17 years ago
- ea9eedb Add ppc partial-word ATOMIC_CMP_SWAP. by Dale Johannesen · 17 years ago