- 851bfff In addition to arguments passed to it, memcpy (and all other calls) can ModRef pointers that alias their arguments as well. This fixes PR2057. by Owen Anderson · 16 years ago
- 61c24e9 Major improvements to yesterday's return slot optimization. Remove some unneccessary constraints, by Owen Anderson · 16 years ago
- 46acf85 Fix some minor issues folding undef, PR2052 by Chris Lattner · 16 years ago
- 50b2ca4 fdiv/frem of undef can produce undef, because the undef operand by Chris Lattner · 16 years ago
- 3890561 regenerate by Chris Lattner · 16 years ago
- 689e8b2 Fix PR2060 by rejecting invalid types for integer constants. by Chris Lattner · 16 years ago
- fa113f8 Factor the profitability check for return slot optimization out into a static function. by Owen Anderson · 16 years ago
- c0755a6 An sret parameter is required to be the first parameter, so there's no need to loop by Owen Anderson · 16 years ago
- 7ad3920 Don't spew stats to stderr. by Nick Lewycky · 16 years ago
- 7acc0af Cleanup some of my patches from yesterday. Refactor the check for which xform by Owen Anderson · 16 years ago
- 9103ba1 Fix up the run line for this new test. by Nick Lewycky · 16 years ago
- c2580ab Fix Transforms/GVN/memcpy.ll, which Chris broke in r47275 by reordering the branches. memcpy's are a kind of CallInst. by Owen Anderson · 16 years ago
- acfab64 New test. by Evan Cheng · 16 years ago
- 31886db PR1909: Tail merging pass ran wild. It makes no sense to merge blocks in order to save a single instruction since a branch will be inserted for each BB. by Evan Cheng · 16 years ago
- 422cba6 Me not like duplicated comments. by Evan Cheng · 16 years ago
- 2ad38be Remove llvm-upgrade and update tests. by Tanya Lattner · 16 years ago
- 53a66d1 Remove llvm-upgrade and update tests. by Tanya Lattner · 16 years ago
- efec751 - When DAG combiner is folding a bit convert into a BUILD_VECTOR, it should check if it's essentially a SCALAR_TO_VECTOR. Avoid turning (v8i16) <10, u, u, u> to <10, 0, u, u, u, u, u, u>. Instead, simply convert it to a SCALAR_TO_VECTOR of the proper type. by Evan Cheng · 16 years ago
- e0cfecf Correctly fold divide-by-constant, even when faced with overflow. by Nick Lewycky · 16 years ago
- b00ee21 Chris pointed out that it's not necessary to set i64 MUL to Expand by Dan Gohman · 16 years ago
- a9407f5 - Remove the previous check which broke coalescer-commute3.ll by Evan Cheng · 16 years ago
- 7c4e9a4 upgrade some tests. by Chris Lattner · 16 years ago
- e9fe65c Add a note by Nate Begeman · 16 years ago
- 92b416f Add a note about sext from i1 plus flags use. by Chris Lattner · 16 years ago
- 339ffed Don't mark scalar integer multiplication as Expand on x86, since x86 by Dan Gohman · 16 years ago
- e10c0b9 Transforming -A + -B --> -(A + B) isn't safe for FP, thanks by Chris Lattner · 16 years ago
- 913b19f minor code simplification, no functionality change. by Chris Lattner · 16 years ago
- ea04159 make this just a bit more strict. by Chris Lattner · 16 years ago
- 548448a Simplify caller updating using a CallSite, as by Duncan Sands · 16 years ago
- bb494bc don't bother calling getUnderlyingObject for non-pointers. by Chris Lattner · 16 years ago
- 3956449 Fix a missing space in the description of the find-bugs option. by Dan Gohman · 16 years ago
- ef56247 Since we're not checking for the more general AllocationInst first, we need to explicitly check by Owen Anderson · 16 years ago
- 8dd2528 New helper function getMBBFromIndex() that given an index in any instruction of an MBB returns a pointer the MBB. Reviewed by Evan. by Roman Levenstein · 16 years ago
- 5aa4f2a Add support to GVN for performing sret return slot optimization. This means that, if an sret function tail calls by Owen Anderson · 16 years ago
- 874a892 I got the predicate backwards in my last patch. The comment is correct, the code was not. by Owen Anderson · 16 years ago
- 26d914a This check is not correct for mallocs, so exclude them earlier. by Owen Anderson · 16 years ago
- 96cfff0 For now, avoid commuting def MI for copy MI's whose source is not killed. That simply trade a live interval for another and because only the non-two-address operands can be folded into loads, may end up pessimising code. by Evan Cheng · 16 years ago
- 8244243 switch simplifycfg from using vectors for most things to smallvectors, by Chris Lattner · 16 years ago
- bf1d8a7 optimize away stackrestore calls that have no intervening alloca or call. by Chris Lattner · 16 years ago
- 28cb205 upgrade this test. by Chris Lattner · 16 years ago
- e26dec6 Clarify that 'sret' only applies to pointers, and by Duncan Sands · 16 years ago
- 117bbd3 Make the definition of the noalias attribute clearer. by Owen Anderson · 16 years ago
- 9eb948b Duncan pointed out that we can fast fail here, because the sret parameter of by Owen Anderson · 16 years ago
- ae708a3 Fix a comment, and a bug where we weren't applying the tail call logic in cases that failed the first test. by Owen Anderson · 16 years ago
- 9aa7c35 Fix bugs that Chris noticed in my last patch. by Owen Anderson · 16 years ago
- d5b8d92 simplify some code, BreakUpSubtract always returns nonnull now. by Chris Lattner · 16 years ago
- 0ab5a4a bitcasts of pointers are always pointers. by Chris Lattner · 16 years ago
- 823391a Add support for setting parameters to CallSite. by Owen Anderson · 16 years ago
- 7d54254 Add a predicate to Argument to check for the StructRet attribute. by Owen Anderson · 16 years ago
- 9b636cb Teach getModRefInfo that memcpy, memmove, and memset don't "capture" memory addresses. by Owen Anderson · 16 years ago
- dd12f96 Fold (-x + -y) -> -(x+y) which promotes better association, fixing by Chris Lattner · 16 years ago
- 5329bb2 fix pasto by Chris Lattner · 16 years ago
- 0b0803a Split up subtracts into add+negate if they have a reassociable use or operand by Chris Lattner · 16 years ago
- f06c3c2 upgrade and simplify this test. by Chris Lattner · 16 years ago
- 9bc5ed7 make the logic for breaking up subtracts more explicit, no by Chris Lattner · 16 years ago
- 6263f94 Remove llvm-upgrade. by Tanya Lattner · 16 years ago
- 456012c move PR2053 to here. by Chris Lattner · 16 years ago
- 5f13724 fix this test. by Chris Lattner · 16 years ago
- d349a4e Remove llvm-upgrade by Tanya Lattner · 16 years ago
- 2d290f9 Remove llvm-upgrade. by Tanya Lattner · 16 years ago
- 26fdb11 this test isn't useful since we added @ notation for globals. by Chris Lattner · 16 years ago
- 8a39257 this line was commented out. by Chris Lattner · 16 years ago
- 709f6ec Remove llvm-upgrade and update tests. by Tanya Lattner · 16 years ago
- 9197493 Removing llvm upgrade, so remove tests specific to llvm-upgrade and update the tests that used it. by Tanya Lattner · 16 years ago
- 3d5378f Remove any 'nest' parameter attributes if the function by Duncan Sands · 16 years ago
- b3a9460 Some micro-optimizations. by Duncan Sands · 16 years ago
- d497d9f I cannot find a libgcc function for this builtin. Therefor expanding it to a noop (which is how it use to be treated). If someone who knows the x86 backend better than me could tell me how to get a lock prefix on an instruction, that would be nice to complete x86 support. by Andrew Lenharth · 16 years ago
- 051bb7b Teach LegalizeTypes how to expand the operands of by Duncan Sands · 16 years ago
- 17bcde9 Try to clarify which compilers can be used for the build. by Duncan Sands · 16 years ago
- 559f422 Refactor some code; check if commuteInstruction is able to commute the instruction. by Evan Cheng · 16 years ago
- 22c5c1b llvm.memory.barrier, and impl for x86 and alpha by Andrew Lenharth · 16 years ago
- 527c250 Fix typos. by Bill Wendling · 16 years ago
- 42a7788 Rename CountMemOperands to ComputeMemOperandsEnd to reflect what by Dan Gohman · 16 years ago
- 94ebde1 Revert 47177, which was incorrect. by Dan Gohman · 16 years ago
- d1bb1c2 upgrade this test, which wasn't testing the right thing since llvm-upgrade came around. by Chris Lattner · 16 years ago
- 0104234 rename llx -> ll by Chris Lattner · 16 years ago
- de7f560 This test is not interesting. by Evan Cheng · 16 years ago
- 0123b7d Make tblgen a little smarter about constants smaller than i32. Currently, by Scott Michel · 16 years ago
- 9f8ec25 simplify this. by Chris Lattner · 16 years ago
- 35e11cd Modify test to expect improved code. by Dale Johannesen · 16 years ago
- bd3e5aa The copy instruction being coalesced will be removed, it is not a kill. by Evan Cheng · 16 years ago
- e3e0957 Fixed bug in FoldingSetIteratorImpl where we did not correctly check if by Ted Kremenek · 16 years ago
- 53c607f Skip over the defs and start at the uses when looking for operands by Dan Gohman · 16 years ago
- 3b66555 Use the TargetInstrDescr to determine the number of operands by Dan Gohman · 16 years ago
- 31fdedd Fix this test on linux, which returns S242 using sret. by Duncan Sands · 16 years ago
- ca3a56f Teach LegalizeTypes how to promote the flags by Duncan Sands · 16 years ago
- ec321b4 Handle \n's in value names for more targets. The asm printers by Chris Lattner · 16 years ago
- 3502d0d If the llvm name contains an unprintable character, don't print it in by Chris Lattner · 16 years ago
- 705da96 targets that support quotes for mangled names still need to escape newlines by Chris Lattner · 16 years ago
- 3b40744 Cosmetics. by Dale Johannesen · 16 years ago
- c8d044e - Removing the infamous r2rMap_ and rep() method. Now the coalescer will update by Evan Cheng · 16 years ago
- f20db15 Added CommuteChangesDestination(). This returns true if commuting the specified by Evan Cheng · 16 years ago
- 63e3cd4 Remove unnecessary #include. by Evan Cheng · 16 years ago
- 21be384 Use StoreSDNode::getValue instead of calling getOperand directly by Dan Gohman · 16 years ago
- ab1a035 Remove warning about 64-bit code on processor by Dale Johannesen · 16 years ago
- 2e78b78 It is not safe to call fork in PrintStackTrace. Sometimes it freezes the program. by Lauro Ramos Venancio · 16 years ago
- 9ae39b5 Change MemOperand's size and offset to be 64-bit. by Dan Gohman · 16 years ago
- 9d9b204 Fix PR2028 by Devang Patel · 16 years ago
- b4c9a68 nocona, core2 and penryn support 64 bit. by Dale Johannesen · 16 years ago
- b6e1dd7 Support vector constant zeros, thanks to Zack Rusin for the testcase. by Chris Lattner · 16 years ago