- 0a9371d Remove unnecessary <sstream> includes. by Dan Gohman · 16 years ago
- 4a9a3e5 Minor whitespace and comment cleanups. by Dan Gohman · 16 years ago
- 6bc41e8 Revert r49614. As Dan pointed out, some of these aren't correct. by Owen Anderson · 16 years ago
- 386ea35 Replace calls of the form V1->setName(V2->getName()) with V1->takeName(V2), by Owen Anderson · 17 years ago
- 8f46c78 Fix PR2213 by simultaneously making GVN more aggressive with the return values by Owen Anderson · 17 years ago
- eee962e Teach InstCombine's ComputeMaskedBits to handle pointer expressions by Dan Gohman · 17 years ago
- 0c19630 Disable an xform we've had for a long time, pow(x,0.5) -> sqrt. by Chris Lattner · 17 years ago
- cc7f168 Generalize getUnaryFloatFunction to handle any FP unary function, automatically by Chris Lattner · 17 years ago
- 8bb9a4f use the new ConstantFP::get method to make this work with by Chris Lattner · 17 years ago
- d4e0af9 Be conservative if getresult operand is neither call nor invoke. by Devang Patel · 17 years ago
- a723d1e Factor a bunch of functionality related to memcpy and memset transforms out of by Owen Anderson · 17 years ago
- 82a6629 Remove accidentally duplicated code. by Owen Anderson · 17 years ago
- 77d00b0 many cleanups to the pow optimizer. Allow it to handle powf, by Chris Lattner · 17 years ago
- 34437f3 Fix insert point handling for multiple return values. by Devang Patel · 17 years ago
- 4128212 Add operator= implementations to SparseBitVector, allowing it to be used in GVN. This results by Owen Anderson · 17 years ago
- e2c4304 Use Intrinsic::getDeclaration in more places. by Duncan Sands · 17 years ago
- 3d292ac The "stacksave is not nounwind problem" no longer by Duncan Sands · 17 years ago
- a122043 Use Intrinsic::getDeclaration to get hold of by Duncan Sands · 17 years ago
- e5ffa90 Make GVN more memory efficient, particularly on code that contains a large number of by Owen Anderson · 17 years ago
- 11b80a8 Mark calls to llvm.stacksave, llvm.stackrestore as by Dale Johannesen · 17 years ago
- 0c80c75 silence a warning when assertions are disabled. by Chris Lattner · 17 years ago
- 051a950 API changes for class Use size reduction, wave 1. by Gabor Greif · 17 years ago
- a022e3f by David Greene · 17 years ago
- 7c3becd 1. Drop default inline threshold back down to 200. by Evan Cheng · 17 years ago
- 48afd9f Reverting 49056 due to the build being broken. by Tanya Lattner · 17 years ago
- da2903f by David Greene · 17 years ago
- b6d5b14 Revert 49006 for the moment. by Dale Johannesen · 17 years ago
- 1544e47 Emit exception handling info for functions which are by Dale Johannesen · 17 years ago
- 83ad90a Don't eliminate bitcast instructions that change the type of a pointer by Nate Begeman · 17 years ago
- d27c991 Fix "Control reaches the end of non-void function" warnings, by Chris Lattner · 17 years ago
- 97ae12a change iterator invalidation avoidance to just move the iterator backward by Chris Lattner · 17 years ago
- d22fe2b make the common case of a single store (which clearly shouldn't be turned by Chris Lattner · 17 years ago
- 9f8a6a7 give form-memset a significantly more sane heuristic, enable it by default. by Chris Lattner · 17 years ago
- 7f0965e make memset inference significantly more powerful: it can now handle by Chris Lattner · 17 years ago
- 4a3c0ac PHI->removeIncomingValue may remove PHInode. Increment iterator in advance. by Devang Patel · 17 years ago
- d34af78 Handle a special case xor undef, undef -> 0. Technically this should be transformed to undef. But this is such a common idiom (misuse) we are going to handle it. by Evan Cheng · 17 years ago
- 60a1290 Add incoming value from header only if phi node has any use inside the loop. by Devang Patel · 17 years ago
- 8d84d5b Increasing the inline limit from (overly conservative) 200 to 300. Given each BB costs 20 and each instruction costs 5, 200 means a 4 BB function + 24 instructions (actually less because caller's size also contributes to it). by Evan Cheng · 17 years ago
- 88ffddd Temporarily disabling memset forming optimization. Add an option. by Evan Cheng · 17 years ago
- b98a10e Transform (zext (or (icmp), (icmp))) to (or (zext (cimp), (zext icmp))) if at least one of the (zext icmp) can be transformed to eliminate an icmp. by Evan Cheng · 17 years ago
- 9e49f1b Preserve calling convention during function cloning by Anton Korobeynikov · 17 years ago
- b017c9e implement an initial hack at a straight-line store -> memset optimization. by Chris Lattner · 17 years ago
- d27290d implement the logic for memset insertion and store deletion. by Chris Lattner · 17 years ago
- 641dae1 This is a partially implemented and currently disabled start of a store by Chris Lattner · 17 years ago
- c9235d2 Don't include <map> in Pass.h, which doesn't need it. This requires by Dan Gohman · 17 years ago
- 8e1e95c the size of a smallvector shouldn't be part of the interface to these methods. by Chris Lattner · 17 years ago
- 2e60701 make gvn marginally faster by reallocating the lastSeenLoad map for by Chris Lattner · 17 years ago
- 88365bb Minor cleanups and shrinkification. by Chris Lattner · 17 years ago
- f1d3d66 Handle getresult instructions in different basic blocks by Dan Gohman · 17 years ago
- 7623716 FunctionExtractorPass has been superceded by GVExtractorPass by Andrew Lenharth · 17 years ago
- 1d57a75 Fix the build for gcc-4.2. by Duncan Sands · 17 years ago
- 41dc0fc Teach masked value is zero about add and sub, and use MVIZ to by Chris Lattner · 17 years ago
- 488b678 Incorporate feedback. by Devang Patel · 17 years ago
- 69cf031 C and Objective Caml bindings for mem2reg and reg2mem. by Gordon Henriksen · 17 years ago
- 2b3407f Take the old function's name. by Zhou Sheng · 17 years ago
- 71d8374 80 col violation. by Evan Cheng · 17 years ago
- ac3844d Add comment. by Devang Patel · 17 years ago
- 55e641b Remove dead options. by Evan Cheng · 17 years ago
- b176038 Update heuritics that estimates cost of call instructions. by Devang Patel · 17 years ago
- 41ba154 C and Objective Caml bindings for several scalar transforms. by Gordon Henriksen · 17 years ago
- b9d4f8d The inst combining of inttoptr into GEP with one index was using the bit size of by Bill Wendling · 17 years ago
- 8a97fdd Fix a bug in GVN that Duncan noticed, where we potentially need to insert a by Owen Anderson · 17 years ago
- 6e7aeb1 Update -mem2reg to use succ_iterator instead of iterating across TerminatorInst by Nick Lewycky · 17 years ago
- 58d7491 Reimplement the parameter attributes support, phase #1. hilights: by Chris Lattner · 17 years ago
- 6bb0bd5 Improve the return slot optimization to be both more aggressive (not limited to sret parameters), and by Owen Anderson · 17 years ago
- 045497a Check multiple return values. by Devang Patel · 17 years ago
- 8f9b551 Fix attribute handling. by Devang Patel · 17 years ago
- 7db30ba Handle multiple ret values. by Devang Patel · 17 years ago
- a023877 Initialize. by Devang Patel · 17 years ago
- 882d87d Check to see if a two-entry PHI block can be simplified by Dan Gohman · 17 years ago
- 75e6f02 Skip functions that return multiple values. by Devang Patel · 17 years ago
- 826c491 Become multiple return value aware. by Devang Patel · 17 years ago
- 9af014f Add TODO reminder. by Devang Patel · 17 years ago
- 7c490d4 Initial multiple return values support. by Devang Patel · 17 years ago
- b8f198a Restore optimization that merges blocks when inline function by Devang Patel · 17 years ago
- 641ca93 Simplify by Devang Patel · 17 years ago
- 7498f90 simplify by Devang Patel · 17 years ago
- 9be3c97 Turn unwind_to into "unwinds to". by Nick Lewycky · 17 years ago
- 58d5e05 Fix some compilation errors on msvc: by Ted Kremenek · 17 years ago
- 3c4a262 And again. by Nick Lewycky · 17 years ago
- f336983 Braces belong here. No functionality change. by Nick Lewycky · 17 years ago
- a66696e SCCP also needs to be taught to follow unwind_to by Nick Lewycky · 17 years ago
- 9143699 Don't eliminate blocks that are only reachable by unwind_to. by Nick Lewycky · 17 years ago
- 3f4cc31 Firstly, having a BranchInst isn't exclusive with having an unwind_to. by Nick Lewycky · 17 years ago
- db96ae9 A BB that unwind_to an "unwind" inst is that same as one that doesn't unwind_to by Nick Lewycky · 17 years ago
- 529de8a Update the block cloner which fixes bugpoint on code using unwind_to (phew!) by Nick Lewycky · 17 years ago
- 6af31aa Update the inliner and simplifycfg to handle unwind_to. by Nick Lewycky · 17 years ago
- c669422 Two things. Preserve the unwind_to when splitting a BB. by Nick Lewycky · 17 years ago
- c0d025f Prune the unwind_to labels on BBs that don't need them. Another step in the by Nick Lewycky · 17 years ago
- 6c36a0f Not all users of a BB are Instructions any more. by Nick Lewycky · 17 years ago
- 12a466b Update inliner to handle functions that return multiple values. by Devang Patel · 17 years ago
- d245a8a add a pass that can extract all kinds of global values, not just functions. Update llvm-extract to use it and optionally extract a global variable if you want it too by Andrew Lenharth · 17 years ago
- 29d8aa7 fix typos by Gabor Greif · 17 years ago
- c1a2a61 Don't try to simplify urem and srem using arithmetic rules that don't work by Nick Lewycky · 17 years ago
- e0a6a3f Skip, for now, callsites where use of sret argument is not dominated by callsite. by Devang Patel · 17 years ago
- c5eb380 Handle 'ret' with multiple values. by Devang Patel · 17 years ago
- 4d51945 Skip functions that return multiple values. by Devang Patel · 17 years ago
- ac3746f Use while loop. by Devang Patel · 17 years ago
- 3796355 Use cast instead of dyn_cast. by Devang Patel · 17 years ago
- a9fe8bb Filter nested structs by Devang Patel · 17 years ago