- 2c755ba Re-apply 97040 with fix. This survives a ppc self-host llvm-gcc bootstrap. by Evan Cheng · 15 years ago
- cfe30ef Speculatively revert r97011, "Re-apply 96540 and 96556 with fixes.", again in by Daniel Dunbar · 15 years ago
- 97e6b83 Re-apply 96540 and 96556 with fixes. by Evan Cheng · 15 years ago
- 3eba667 Revert commits 96556 and 96640, because commit 96556 breaks the by Duncan Sands · 15 years ago
- 89bb7b5 Some dag combiner goodness: by Evan Cheng · 15 years ago
- 1e55944 by David Greene · 15 years ago
- f451cb8 Fix "the the" and similar typos. by Dan Gohman · 15 years ago
- 87c46d8 Improve EXTRACT_VECTOR_ELT patch based on comments from Duncan by Mon P Wang · 16 years ago
- c6654ec4 Fixed a couple of optimization with EXTRACT_VECTOR_ELT that assumes the result by Mon P Wang · 16 years ago
- 8c7ecaf Implement cond ? -1 : 0 with sbb. by Evan Cheng · 16 years ago
- cd9e155 Fold (add x, shl(0 - y, n)) -> sub(x, shl(y, n)), to simplify some code by Dan Gohman · 16 years ago
- 1ad0e8b Canonicalize -1 - x to ~x. by Evan Cheng · 16 years ago
- ff00a55 Fix a codegen abort seen in 483.xalancbmk. by Dan Gohman · 16 years ago
- fe240b1 Disable transformation of select of two loads to a select of address and then a load if the by Mon P Wang · 16 years ago
- d199636 Revert an earlier change to SIGN_EXTEND_INREG for vectors. The VTSDNode by Dan Gohman · 16 years ago
- bd1fccf Fix rdar://7517201, a regression introduced by r92849. by Chris Lattner · 16 years ago
- ef7634c factor this code better and reduce nesting at the same by Chris Lattner · 16 years ago
- d40d03e Teach dag combine to fold the following transformation more aggressively: by Evan Cheng · 16 years ago
- 9f7c5c0 Don't assign the shift the same type as the variable being shifted. This could by Bill Wendling · 16 years ago
- f109029 Change errs() to dbgs(). by David Greene · 16 years ago
- 9ab2b98 Increase opportunities to optimize (brcond (srl (and c1), c2)). by Evan Cheng · 16 years ago
- 4c2b001 Revert this dag combine change: Fold (zext (and x, cst)) -> (and (zext x), cst) by Evan Cheng · 16 years ago
- 9818c04 Make 91378 more conservative. by Evan Cheng · 16 years ago
- 8acb310 Fold (zext (and x, cst)) -> (and (zext x), cst). by Evan Cheng · 16 years ago
- 99b653c Propagate zest through logical shift. by Evan Cheng · 16 years ago
- 2e141d7 Fix integer cast code to handle vector types. by Dan Gohman · 16 years ago
- 87862e7 Implement vector widening, splitting, and scalarizing for SIGN_EXTEND_INREG. by Dan Gohman · 16 years ago
- 64fa4a9 Move isConsecutiveLoad to SelectionDAG. It's not target dependent and it's primary used by selectdag passes. by Evan Cheng · 16 years ago
- f2dc5c7 Refactor InferAlignment out of DAGCombine. by Evan Cheng · 16 years ago
- 93e0ed3 Don't pull vector sext through both hands of a logical operation, since doing so prevents the fusion of vector sext and setcc into vsetcc. by Nate Begeman · 16 years ago
- 9f0d4e6 Don't call getValueType() on a null SDValue by Jakob Stoklund Olesen · 16 years ago
- e0f06c7 Remove the optimizations that convert BRCOND and BR_CC into by Dan Gohman · 16 years ago
- ba565bf Remove an unneeded #include. by Dan Gohman · 16 years ago
- 75832d7 When discarding SrcValue information, discard all of it so that code by Dan Gohman · 16 years ago
- 1507cd0 Don't call SDNode::isPredecessorOf when it isn't necessary. If the load's by Dan Gohman · 16 years ago
- f5a86f4 Remove includes of Support/Compiler.h that are no longer needed after the by Nick Lewycky · 16 years ago
- 6726b6d Remove VISIBILITY_HIDDEN from class/struct found inside anonymous namespaces. by Nick Lewycky · 16 years ago
- 2bcf60a Fix invalid for vector types fneg(bitconvert(x)) => bitconvert(x ^ sign) by Anton Korobeynikov · 16 years ago
- 677c89d More heuristics for Combiner-AA. Still catches all important cases, but by Nate Begeman · 16 years ago
- cc66cdd Fix combiner-aa issue with bases which are different, but can alias. by Nate Begeman · 16 years ago
- 4e918b2 Use getStoreSize() instead of getStoreSizeInBits()/8. by Dan Gohman · 16 years ago
- 8a55ce4 Rename several variables from EVT to more descriptive names, now that EVT by Dan Gohman · 16 years ago
- b6aef5c Substantially speed up combiner-aa in the following ways: by Nate Begeman · 16 years ago
- 92ad363 Don't swap the operands of a subtraction when trying to create a by Bob Wilson · 16 years ago
- b7c5bdf Remove some unused variables and methods warned about by by Duncan Sands · 16 years ago
- bbbfa99 remove a few DOUTs here and there. by Chris Lattner · 16 years ago
- 5aba5c0 Add check for completeness. Note that this doesn't actually have any by Eli Friedman · 16 years ago
- d68eea2 PR4737: Fix a nasty bug in load narrowing with non-power-of-two types. by Eli Friedman · 16 years ago
- 23b9b19 Add contexts to some of the MVT APIs. No functionality change yet, just the infrastructure work needed to get the contexts to where they need to be first. by Owen Anderson · 16 years ago
- 825b72b Split EVT into MVT and EVT, the former representing _just_ a primitive type, while by Owen Anderson · 16 years ago
- 00edf39 Fix a bug in the DAGCombiner's handling of multiple linked by Dan Gohman · 16 years ago
- a407ca1 Fix a bug where DAGCombine was producing an illegal ConstantFP by Dan Gohman · 16 years ago
- e50ed30 Rename MVT to EVT, in preparation for splitting SimpleValueType out into its own struct type. by Owen Anderson · 16 years ago
- eb1fedc Make this comment more closely reflect the code. by Dan Gohman · 16 years ago
- 17421d8 Don't build illegal ops in DAGCombiner::SimplifyBinOpWithSameOpcodeHands(). by Jakob Stoklund Olesen · 16 years ago
- 5cbd37e Fix a few places in DAGCombiner that were creating all-ones-bits by Dan Gohman · 16 years ago
- 4ea4804 Avoid forming a SELECT_CC in a type that the target doesn't by Dan Gohman · 16 years ago
- debcb01 Move types back to the 2.5 API. by Owen Anderson · 16 years ago
- 1fd7096 Change ConstantArray to 2.5 API. by Owen Anderson · 16 years ago
- 78d1264 Add support for promoting SETCC operations. by Jakob Stoklund Olesen · 16 years ago
- d101a72 Fix a dagga combiner bug: avoid creating illegal constant. by Evan Cheng · 16 years ago
- a90b3dc Move a few more convenience factory functions from Constant to LLVMContext. by Owen Anderson · 16 years ago
- c23197a llvm_unreachable->llvm_unreachable(0), LLVM_UNREACHABLE->llvm_unreachable. by Torok Edwin · 16 years ago
- c25e758 assert(0) -> LLVM_UNREACHABLE. by Torok Edwin · 16 years ago
- 7d696d8 Convert more assert(0)+abort() -> LLVM_UNREACHABLE, by Torok Edwin · 16 years ago
- d1474d0 Thread LLVMContext through MVT and related parts of SDISel. by Owen Anderson · 16 years ago
- 2b7a271 dag combine sext(setcc) -> vsetcc before legalize. To make this safe, by Chris Lattner · 16 years ago
- abc0199 Adapt the x86 build_vector dagcombine to the current state of the legalizer. by Nate Begeman · 16 years ago
- 77b81fe Don't do the X * 0.0 -> 0.0 transformation in instcombine, because by Dan Gohman · 16 years ago
- 874ae25 Revert 72707 and 72709, for the moment. by Dale Johannesen · 16 years ago
- 4150d83 Make the implicit inputs and outputs of target-independent by Dale Johannesen · 16 years ago
- d3c76bb Do not try to create a MVT type of width 0. by Evan Cheng · 16 years ago
- cdcecc0 Incorporate patch feedbacks. by Evan Cheng · 16 years ago
- 8b944d3 Added optimization that narrow load / op / store and the 'op' is a bit twiddling instruction and its second operand is an immediate. If bits that are touched by 'op' can be done with a narrower instruction, reduce the width of the load and store as well. This happens a lot with bitfield manipulation code. by Evan Cheng · 16 years ago
- 6bb4958 Fix PR4254. by Torok Edwin · 16 years ago
- 8c562e2 Silence Release-Asserts warnings. by Daniel Dunbar · 16 years ago
- 777d230 Rename PaddedSize to AllocSize, in the hope that this by Duncan Sands · 16 years ago
- caab129 Do not use register as base ptr of pre- and post- inc/dec load / store nodes. by Evan Cheng · 16 years ago
- 98a366d Instead of passing in an unsigned value for the optimization level, use an enum, by Bill Wendling · 16 years ago
- 5a5ca15 Implement review feedback for vector shuffle work. by Nate Begeman · 16 years ago
- be8cc2a Second attempt: by Bill Wendling · 16 years ago
- c69d56f r70270 isn't ready yet. Back this out. Sorry for the noise. by Bill Wendling · 16 years ago
- 2e9d5f9 Massive check in. This changes the "-fast" flag to "-O#" in llc. If you want to by Bill Wendling · 16 years ago
- 9008ca6 2nd attempt, fixing SSE4.1 issues and implementing feedback from duncan. by Nate Begeman · 16 years ago
- c7b3444 When transforming sext(trunc(load(x))) into sext(smaller load(x)), by Dan Gohman · 16 years ago
- 4128700 Add a top-level comment about DAGCombiner's role in the compiler. by Dan Gohman · 16 years ago
- 15684b2 Revert 69952. Causes testsuite failures on linux x86-64. by Rafael Espindola · 16 years ago
- b706d29 PR2957 by Nate Begeman · 16 years ago
- c885165 Revise my previous change 68996 as suggested by Duncan. by Bob Wilson · 16 years ago
- b10b5ac Don't try to make BUILD_VECTOR operands have the same by Duncan Sands · 16 years ago
- b1303d0 Change SelectionDAG type legalization to allow BUILD_VECTOR operands to be by Bob Wilson · 16 years ago
- 57fc82d Generalize ExtendUsesToFormExtLoad to be usable for ANY_EXTEND, by Dan Gohman · 16 years ago
- 97121ba Implement support for using modeling implicit-zero-extension on x86-64 by Dan Gohman · 16 years ago
- 0b0cd91 Optimize some 64-bit multiplication by constants into two lea's or one lea + shl since imulq is slow (latency 5). e.g. by Evan Cheng · 16 years ago
- a02a3dd Pull transform from target-dependent code into target-independent code. by Bill Wendling · 16 years ago
- 93b7415 Fix a problem with DAGCombine where we were building an illegal build by Mon P Wang · 16 years ago
- 0b7a786 Avoid doing the transformation c ? 1.0 : 2.0 as load { 2.0, 1.0 } + c*4 by Mon P Wang · 16 years ago
- 1606e8e Fix some significant problems with constant pools that resulted in unnecessary paddings between constant pool entries, larger than necessary alignments (e.g. 8 byte alignment for .literal4 sections), and potentially other issues. by Evan Cheng · 16 years ago
- d1980a5 Move 3 "(add (select cc, 0, c), x) -> (select cc, x, (add, x, c))" by Chris Lattner · 16 years ago
- 600fec3 reapply my previous patch (r66358) with a tweak to set the by Chris Lattner · 16 years ago