MemorySSA keeps internal pointers of AA and DT.
If these get invalidated, so should MemorySSA.
Details
Details
- Reviewers
george.burgess.iv chandlerc RKSimon - Commits
- rGb46832031337: [MemorySSA] Invalidate MemorySSA if AA or DT are invalidated.
rL359627: [MemorySSA] Invalidate MemorySSA if AA or DT are invalidated.
rGf5e8f222d6bd: Revert rL359519 : [MemorySSA] Invalidate MemorySSA if AA or DT are invalidated.
rL359555: Revert rL359519 : [MemorySSA] Invalidate MemorySSA if AA or DT are invalidated.
rG9a1edd14a220: [MemorySSA] Invalidate MemorySSA if AA or DT are invalidated.
rL359519: [MemorySSA] Invalidate MemorySSA if AA or DT are invalidated.
Diff Detail
Diff Detail
- Repository
- rL LLVM
Event Timeline
Comment Actions
Oh, you can add a test for this by using the new PM and an invalidate<aa> pass or invalidate<domtree>.
Comment Actions
Excellent suggestion! The invalidate needs to go into the Result internal to the analysis.
Updated and added test. PTAL.
Comment Actions
@asbirlea I'm sorry but I had to revert this at rL359555 as it was causing windows buildbot failures: http://lab.llvm.org:8011/builders/llvm-clang-lld-x86_64-scei-ps4-windows10pro-fast/builds/25459
Comment Actions
Updated test MemorySSA/invalidate.ll: What follows "InvalidateAnalysisPass" looks different on Windows.