- 51f40a7 Make the unreachable probability much much heavier. The previous by Chandler Carruth · 13 years ago
- ddbc274 Manually upgrade the test suite to specify the flag to cttz and ctlz. by Chandler Carruth · 13 years ago
- ecb35ec SCEV fix. In general, Add/Mul expressions should not inherit NSW/NUW. by Andrew Trick · 13 years ago
- d2b5e2d Filecheckize. by Andrew Trick · 13 years ago
- d2bf432 Upgrade syntax of tests using volatile instructions to use 'load volatile' instead of 'volatile load', which is archaic. by Chris Lattner · 13 years ago
- 89d093d Don't forget to check FlagNW when determining whether an AddRecExpr will wrap by Nick Lewycky · 13 years ago
- 588d84c 2>&1 doesn't work here, it just creates an empty file called "&1" by Benjamin Kramer · 13 years ago
- f8a9eb1 Restore commits 142790 and 142843 - they weren't breaking the build by Duncan Sands · 13 years ago
- 45baf6b Fix the API usage in loop probability heuristics. It was incorrectly by Chandler Carruth · 13 years ago
- 526c80b Speculatively revert commits 142790 and 142843 to see if it fixes by Duncan Sands · 13 years ago
- 3846163 Now that we look at all the header PHIs, we need to consider all the header PHIs by Nick Lewycky · 13 years ago
- de1c9bb Remove return heuristics from the static branch probabilities, and by Chandler Carruth · 13 years ago
- 7c3fc57 Reapply r142781 with fix. Original message: by Nick Lewycky · 13 years ago
- 38af3d5 Speculatively revert r142781. Bots are showing by Nick Lewycky · 13 years ago
- 795cb48 Enhance SCEV's brute force loop analysis to handle multiple PHI nodes in the by Nick Lewycky · 13 years ago
- 14edd31 Teach the BranchProbabilityInfo pass to print its results, and use that by Chandler Carruth · 13 years ago
- 614fef6 Make SCEV's brute force analysis stronger in two ways. Firstly, we should be by Nick Lewycky · 13 years ago
- 941aa7b Generalize the reading of probability metadata to work for both branches by Chandler Carruth · 13 years ago
- 99d01c5 Teach the BranchProbabilityInfo analysis pass to read any metadata by Chandler Carruth · 13 years ago
- 6aa5c26 Add pass printing support to BlockFrequencyInfo pass. The implementation by Chandler Carruth · 13 years ago
- 0c38858 Missing test case for r141164. by Andrew Trick · 13 years ago
- 1cbae18 Reapply r140979 with fix! We never did get a testcase, but careful review of the by Nick Lewycky · 13 years ago
- 4fcc80a Revert r140979 due to reports of bootstrap failure. by Nick Lewycky · 13 years ago
- 8fde4f5 Add one more case we compute a max trip count. by Nick Lewycky · 13 years ago
- e6fadce PR10628: Fix getModRefInfo so it queries the underlying alias() implementation correctly while checking nocapture calls. by Eli Friedman · 13 years ago
- 46cb5af Enhance alias analysis for atomic instructions a bit. Upgrade a couple alias-analysis tests to the new atomic instructions. by Eli Friedman · 13 years ago
- 2e95d76 This test only makes sense with -enable-iv-rewrite. by Andrew Trick · 13 years ago
- 81ac8dd Fix the logic in BasicAliasAnalysis::aliasGEP for comparing GEP's with variable differences so that it actually does something sane. Fixes PR10881. by Eli Friedman · 13 years ago
- 69acc93 Teach BasicAA about the aliasing properties of memset_pattern16. by Owen Anderson · 13 years ago
- c103a08 This transform only handles two-operand AddRec's. Prevent it from trying to by Nick Lewycky · 13 years ago
- fa151a7 The logic inside getMulExpr to simplify {a,+,b}*{c,+,d} was wrong, which was by Nick Lewycky · 13 years ago
- b2840fd Revert r139126 due to selfhost failures reported by buildbots. by Nick Lewycky · 13 years ago
- 77b7352 Teach SCEV to report a max backedge count in one interesting case in by Nick Lewycky · 13 years ago
- aecfdd3 Move the loads after the calls so that the fix for by Rafael Espindola · 13 years ago
- 4fe5094 Use CHECK-NEXT. by Rafael Espindola · 13 years ago
- a53616d Remove support for parsing the "type i32" syntax for defining a numbered by Chris Lattner · 13 years ago
- b85e4eb rip out a ton of intrinsic modernization logic from AutoUpgrade.cpp, which is by Chris Lattner · 13 years ago
- d589099 make the asmparser reject function and type redefinitions. 'Merging' hasn't been by Chris Lattner · 13 years ago
- 7a1b9bd Remove support for using "foo" as symbols instead of %"foo". This is ancient by Chris Lattner · 13 years ago
- 26b0000 manually upgrade a bunch of tests to modern syntax, and remove some that by Chris Lattner · 13 years ago
- 410eac5 Test case for r132797. by John McCall · 13 years ago
- 5f1312c Reapply r131781, now that the GVN bug with partially-aliasing loads is disabled. by Dan Gohman · 13 years ago
- baddf6c Remove this test, which should have been reverted along with r131781. by Dan Gohman · 13 years ago
- 9a9d218 Revert r131781 again. Apparently there is more going on here. by Dan Gohman · 13 years ago
- c122c62 Reapply r131781 (revert r131809), now that some BasicAA shortcomings by Dan Gohman · 13 years ago
- 1fc18d7 Fix BasicAA's recursion detection so that it doesn't pessimize by Dan Gohman · 13 years ago
- 965fefa When merging MustAlias and PartialAlias, chose PartialAlias instead by Dan Gohman · 13 years ago
- 9adf151 Make DecomposeGEPExpression check SimplifyInstruction only by Dan Gohman · 13 years ago
- 4d92550 I missed a checking with my GVN change. by Chris Lattner · 13 years ago
- d22920a Revert commit 131781, to see if it fixes the x86-64 dragonegg buildbot. by Duncan Sands · 13 years ago
- 424b777 When BasicAA can determine that two pointers have the same base but by Dan Gohman · 13 years ago
- 1d7e818 Teach BasicAA about arm.neon.vld1 and vst1. by Dan Gohman · 14 years ago
- bddc1ca When analyzing functions known to only access argument pointees, by Dan Gohman · 14 years ago
- 104f4ad Test case and comment for PR9633. by Andrew Trick · 14 years ago
- a42a757 Make tests more useful. by Benjamin Kramer · 14 years ago
- 0e38219 PR9634: Don't unconditionally tell the AliasSetTracker that the PreheaderLoad by Eli Friedman · 14 years ago
- 1a8b9dd remove postdom frontiers, because it is dead. Forward dom frontiers are by Chris Lattner · 14 years ago
- fa4ebd3 Revert r128140 for now. by Anders Carlsson · 14 years ago
- 90af342 A global variable with internal linkage where all uses are in one function and whose address is never taken is a non-escaping local object and can't alias anything else. by Anders Carlsson · 14 years ago
- c343c1e Propagate SCEV no-wrap flags whenever possible. This needs review. by Andrew Trick · 14 years ago
- e62289b When SCEV can determine the loop test is X < X, set ExactBECount=0. by Andrew Trick · 14 years ago
- 8ebaf90 teach SCEV that the scale and addition of an inbounds gep don't NSW. by Chris Lattner · 14 years ago
- 1021236 Per discussion with Dan G, inbounds geps *certainly* can have by Chris Lattner · 14 years ago
- 640b9c8 Add another rdar number. by Dan Gohman · 14 years ago
- 76167af Simplify some code with no functionality change. Make the test a lot more by Nick Lewycky · 14 years ago
- 630d85a Use value ranges to fold ext(trunc) in SCEV when possible. by Nick Lewycky · 14 years ago
- e906921 Implement requiredTransitive by Tobias Grosser · 14 years ago
- 5c6fc1c Similarly, analyze truncate through multiply. by Nick Lewycky · 14 years ago
- 30aa8b1 Add a missed SCEV fold that is required to continue analyzing the IR produced by Nick Lewycky · 14 years ago
- 73f565e Add a missing SCEV simplification sext(zext x) --> zext x. by Nick Lewycky · 14 years ago
- 615da1a Teach BasicAA to return PartialAlias in cases where both pointers by Dan Gohman · 14 years ago
- 48eec43 Revert the testcase from the previous reverted commit. by Eric Christopher · 14 years ago
- e70d7c2 add a testcase I missed in previous commit. by Chris Lattner · 14 years ago
- 6d5a241 teach SCEV analysis of PHI nodes that PHI recurences formed by Chris Lattner · 14 years ago
- f185989 reduce indentation. Print <nuw> and <nsw> when dumping SCEV AddRec's by Chris Lattner · 14 years ago
- 895ace0 fix rdar://8813415 - a miscompilation of 164.gzip that loop-idiom by Chris Lattner · 14 years ago
- 135bf42 filecheckize by Chris Lattner · 14 years ago
- a4bdf2e -enable-tbaa is on by default now. by Dan Gohman · 14 years ago
- 387f28a Make memcpyopt TBAA-aware. by Dan Gohman · 14 years ago
- fea3b21 Move Sub simplifications and additional Add simplifications out of by Duncan Sands · 14 years ago
- 0f7f194 Reapply r121520, PartialAlias implementation for BasicAA, now that by Dan Gohman · 14 years ago
- 839c092 Revert r121520, which may have introduced miscompilations. by Dan Gohman · 14 years ago
- 3bd5e52 Implement PartialAlias checking in BasicAA. by Dan Gohman · 14 years ago
- 9fd9f9f remove fixme comment too. by Chris Lattner · 14 years ago
- 7fd16ce check in *all* files. This is now handled by my previous DSE commit. by Chris Lattner · 14 years ago
- 86c817e test: Check the feature 'loadable_module' with load modules in %llvmshlibdir. by NAKAMURA Takumi · 14 years ago
- 32d1fe5 Delete unneeded ssp attributes. by Dan Gohman · 14 years ago
- 56653f0 TBAA-enable ArgumentPromotion. by Dan Gohman · 14 years ago
- 5aae3dc Make Sink tbaa-aware. by Dan Gohman · 14 years ago
- f00833c Add a testcase which demonstrates alias analysis pass precedence. by Dan Gohman · 14 years ago
- 733c54d Fully invalidate cached results when a prior query's size or by Dan Gohman · 14 years ago
- 4cf0dcf Teach FunctionAttrs about the VAArg instruction. by Dan Gohman · 14 years ago
- e5dd662 Add a testcase for a call which BasicAA says only accesses memory through by Dan Gohman · 14 years ago
- ea8900f Make FunctionAttrs TBAA-aware. by Dan Gohman · 14 years ago
- cd5c123 Teach memdep to use pointsToConstantMemory to determine that loads by Dan Gohman · 14 years ago
- b338bcb Add a basic testcase for TBAA-aware DSE. by Dan Gohman · 14 years ago
- 561ac17 Add some comments. by Dan Gohman · 14 years ago
- fb8096d Don't pass the raw invalid pointer used to represent conflicting by Dan Gohman · 14 years ago
- 992ac3e Add a basic testcase for TBAA-aware LICM. by Dan Gohman · 14 years ago
- bdcbf61 Run tbaa before basicaa, since that's how it's expected to be used. by Dan Gohman · 14 years ago