- 5aa3fa6 llvm.dbg.declare intrinsic does not use any llvm::Values. It's magic! by Devang Patel · 14 years ago
- dcf7757 Reorder comments to put them the right way around. by Nick Lewycky · 14 years ago
- c14bc77 Add more analysis of the sign bit of an srem instruction. If the LHS is negative by Nick Lewycky · 14 years ago
- d01f50f ConstantInt has some getters which return ConstantInt's or ConstantVector's of by Nick Lewycky · 14 years ago
- 103391d InstCombine: We know the number of items initially added to the worklist map, reserve space early to avoid rehashing. by Benjamin Kramer · 14 years ago
- 58bfcdb Thread comparisons over udiv/sdiv/ashr/lshr exact and lshr nuw/nsw whenever by Nick Lewycky · 14 years ago
- 9feda17 Try once again to optimize "icmp (srem X, Y), Y" by turning the comparison into by Nick Lewycky · 14 years ago
- d70be0b Make InstCombiner::FoldAndOfICmps create a ConstantRange that's the by Anders Carlsson · 14 years ago
- 3dc7e49 srem doesn't actually have the same resulting sign as its numerator, you could by Nick Lewycky · 14 years ago
- b042f8e Teach InstCombine to fold "(shr exact X, Y) == 0" --> X == 0, fixing #1 from by Nick Lewycky · 14 years ago
- d8d1584 The sign of an srem instruction is the sign of its dividend (the first by Nick Lewycky · 14 years ago
- 091b1e3 change instcombine to not turn a call to non-varargs bitcast of by Chris Lattner · 14 years ago
- 38f7f66 Move "A | ~(A & ?) -> -1" from InstCombine to InstructionSimplify. by Benjamin Kramer · 14 years ago
- 91e37ef InstCombine: Add a bunch of combines of the form x | (y ^ z). by Benjamin Kramer · 14 years ago
- 1347623 PR9218: SimplifyDemandedVectorElts can return a non-null value that is not by Eli Friedman · 14 years ago
- 39a7de7 Add some transforms of the kind X-Y>X -> 0>Y which are valid when there is no by Duncan Sands · 14 years ago
- c8cb8ef have instcombine preserve nsw/nuw/exact when sinking by Chris Lattner · 14 years ago
- 1521e91 fix typo by Chris Lattner · 14 years ago
- 4bd8217 fix instcombine merging GEPs through a PHI to only make the by Chris Lattner · 14 years ago
- bd9f6bf add is always integer, thanks to Frits for noticing this. by Chris Lattner · 14 years ago
- a772433 Transform "A + B >= A + C" into "B >= C" if the adds do not wrap. Likewise for some by Duncan Sands · 14 years ago
- 41429e3 preserve NUW/NSW when transforming add x,x by Chris Lattner · 14 years ago
- ab4c366 Spelling fix: consequtive -> consecutive. by Duncan Sands · 14 years ago
- d9e087b by Nadav Rotem · 14 years ago
- e7394ea Do not forget DebugLoc! by Devang Patel · 14 years ago
- da1d660 tidy up a bit. by Chris Lattner · 14 years ago
- 2ca5c86 convert ConstantVector::get to use ArrayRef. by Chris Lattner · 14 years ago
- 7583190 revert my ConstantVector patch, it seems to have made the llvm-gcc by Chris Lattner · 14 years ago
- 283c8ca Switch ConstantVector::get to use ArrayRef instead of a pointer+size by Chris Lattner · 14 years ago
- e5116f8 remove a now-unneccesary cast. by Chris Lattner · 14 years ago
- bb75d33 implement instcombine folding for things like (x >> c) < 42. by Chris Lattner · 14 years ago
- 74542aa refactor some code out into a helper method. by Chris Lattner · 14 years ago
- b6c8cb4 Also fold (A+B) == A -> B == 0 when the add is commuted. by Benjamin Kramer · 14 years ago
- 6aa68a7 When lowering an inbounds gep, the intermediate adds can have by Chris Lattner · 14 years ago
- 6cdf2ea implement the first part of PR8882: when lowering an inbounds by Chris Lattner · 14 years ago
- 7a6aa1a Enhance a bunch of transformations in instcombine to start generating by Chris Lattner · 14 years ago
- b20c0b5 Enhance the "compare with shift" and "compare with div" by Chris Lattner · 14 years ago
- 44cc997 more cleanups, notably bitcast isn't used for "signed to unsigned type by Chris Lattner · 14 years ago
- b9b9044 A bunch of cleanups and simplifications using the new PatternMatch predicates by Chris Lattner · 14 years ago
- a81556f teach SimplifyDemandedBits that exact shifts demand the bits they by Chris Lattner · 14 years ago
- 81a0dc9 Teach instsimplify some tricks about exact/nuw/nsw shifts. by Chris Lattner · 14 years ago
- aeaf3d4 Rework InstrTypes.h so to reduce the repetition around the NSW/NUW/Exact by Chris Lattner · 14 years ago
- 35bda89 enhance vmcore to know that udiv's can be exact, and add a trivial by Chris Lattner · 14 years ago
- 5195b71 Conservatively, clear optional flags, such as nsw, when performing by Dan Gohman · 14 years ago
- 77bc49e Recognize and simplify by Anders Carlsson · 14 years ago
- 31726c1 Call SimplifyFDivInst() in InstCombiner::visitFDiv(). by Frits van Bommel · 14 years ago
- 1fca2c3 Move InstCombine's knowledge of fdiv to SimplifyInstruction(). by Frits van Bommel · 14 years ago
- 593faa5 My auto-simplifier noticed that ((X/Y)*Y)/Y occurs several times in SPEC by Duncan Sands · 14 years ago
- df3bfae Fold select + select where both selects are on the same condition. by Nick Lewycky · 14 years ago
- 584520e Null initialize a few variables flagged by by Ted Kremenek · 14 years ago
- 5d2e188 Just because we have determined that an (fcmp | fcmp) is true for A < B, by Owen Anderson · 14 years ago
- cd151d2 fix PR9013, an infinite loop in instcombine. by Chris Lattner · 14 years ago
- 084fe62 update obsolete comment. by Chris Lattner · 14 years ago
- acf4a7c Don't try to pull vector bitcasts that change the number of elements through by Nick Lewycky · 14 years ago
- 9d32f60 At -O123 the early-cse pass is run before instcombine has run. According to my by Duncan Sands · 14 years ago
- 6eb6116 remove a dead check, this was needed before we had an explicit veto on uses of phis. by Chris Lattner · 14 years ago
- 192228e enhance FoldOpIntoPhi in instcombine to try harder when a phi has by Chris Lattner · 14 years ago
- 9922ccf remove the AllowAggressive argument to FoldOpIntoPhi. It is forced to false in the by Chris Lattner · 14 years ago
- 7dfe8fd more cleanups: use the IR builder. by Chris Lattner · 14 years ago
- 5aac832 tidy up code. by Chris Lattner · 14 years ago
- 62fe406 implement an instcombine xform that canonicalizes casts outside of and-with-constant operations. by Chris Lattner · 14 years ago
- 6792032 one more instcombine variant that is needed to work with future changes, by Chris Lattner · 14 years ago
- 27a9848 fix typo by Chris Lattner · 14 years ago
- fdb5b01 Catch ~x < cst just like ~x < ~y, we currently handle this through by Chris Lattner · 14 years ago
- 28621cb reduce indentation by Chris Lattner · 14 years ago
- c43cee3 Move some shift transforms out of instcombine and into InstructionSimplify. by Duncan Sands · 14 years ago
- a33b625 Remove dead variable, const-ref-ize an APInt. by Owen Anderson · 14 years ago
- da1c122 Fix a random missed optimization by making InstCombine more aggressive when determining which bits are demanded by by Owen Anderson · 14 years ago
- 9cc9f50 Teach instcombine about the rest of the SSE and SSE2 conversion by Chandler Carruth · 14 years ago
- aa2be84 Instcombine: Fix pattern where the sext did not dominate the icmp using it by Tobias Grosser · 14 years ago
- b686eb9 Fix a bug in r123034 (trying to sext/zext non-integers) and clean up a little. by Frits van Bommel · 14 years ago
- 46431d7 InstCombine: Match min/max hidden by sext/zext by Tobias Grosser · 14 years ago
- 8d088bd Some whitespace fixes by Tobias Grosser · 14 years ago
- eaff66a Revert 122959, it needs more thought. Add it back to README.txt with additional notes. by Benjamin Kramer · 14 years ago
- 8143a84 InstCombine: Turn _chk functions into the "unsafe" variant if length and max langth are equal. by Benjamin Kramer · 14 years ago
- 240d42d InstCombine: If we call llvm.objectsize on a malloc call we can replace it with the size passed to malloc. by Benjamin Kramer · 14 years ago
- 783a5c2 InstCombine: Teach llvm.objectsize folding to look through GEPs. by Benjamin Kramer · 14 years ago
- ae47be1 don't lose TD info by Chris Lattner · 14 years ago
- 687140c Move getOrEnforceKnownAlignment out of instcombine into Transforms/Utils. by Chris Lattner · 14 years ago
- ec3953f When determining if we can fold (x >> C1) << C2, the bits that we need to verify are zero by Owen Anderson · 14 years ago
- 4ac1947 InstCombine: creating selects from -1 and 0 is fine, they combine into a sext from i1. by Benjamin Kramer · 14 years ago
- 37bf92b Add a generic expansion transform: A op (B op' C) -> (A op B) op' (A op C) by Duncan Sands · 14 years ago
- a3c44a5 Add some statistics, good for understanding how much more powerful by Duncan Sands · 14 years ago
- 82fdab3 Pull a few more simplifications out of instcombine (there are still by Duncan Sands · 14 years ago
- b0de244 Add a check missing from my last commit and avoid a potential overflow situation. by Benjamin Kramer · 14 years ago
- 4dc1ac0 Reduce indentation. by Benjamin Kramer · 14 years ago
- 5337f20 Teach InstCombine to merge (icmp ult (X + CA), C1) | (icmp eq X, C2) into (icmp ult (X + CA), C1 + 1) if C2 + CA == C1. by Benjamin Kramer · 14 years ago
- 2b9375e fix PR8807 by making transformConstExprCastCall aware of byval arguments. by Chris Lattner · 14 years ago
- ab215bc various cleanups for transformConstExprCastCall by Chris Lattner · 14 years ago
- 55fb9b0 Avoid dropping the address space when InstCombine optimizes memset by Mon P Wang · 14 years ago
- f2a97ed fix an oversight caught by Frits! by Chris Lattner · 14 years ago
- 74b6461 move a transformation to a more logical place, simplifying it. by Chris Lattner · 14 years ago
- e5cbdca recognize an unsigned add with overflow idiom into uadd. by Chris Lattner · 14 years ago
- 26b482d optimize uadd(x, cst) into a comparison when the normal by Chris Lattner · 14 years ago
- 0fe80bb use IC.ReplaceInstUsesWith instead of a raw RAUW so that uses of by Chris Lattner · 14 years ago
- 0a62474 generalize the sadd creation code to not require that the by Chris Lattner · 14 years ago
- dd7e837 fix another miscompile in the llvm.sadd formation logic: it wasn't by Chris Lattner · 14 years ago
- 368397b fix a bug (possibly 8816) in the sadd forming xform: it isn't by Chris Lattner · 14 years ago
- f0f568b rework the code added in r122072 to pull it out to its own by Chris Lattner · 14 years ago
- 2d0537c Whitespace fixes. No functionality change. by Bill Wendling · 14 years ago