[PATCH] D146206: [LAA] Fix transitive analysis invalidation bug by implementing LoopAccessInfoManager::invalidate

[Bjorn Pettersson via Phabricator via llvm-commits]

bjope marked 2 inline comments as done.
bjope added inline comments.


================
Comment at: llvm/test/Analysis/LoopAccessAnalysis/invalidation.ll:8-11
+; CHECK-AA-DAG: Running analysis: ScalarEvolutionAnalysis on foo
+; CHECK-AA-DAG: Running analysis: DominatorTreeAnalysis on foo
+; CHECK-AA-DAG: Running analysis: LoopAnalysis on foo
+; CHECK-AA-DAG: Running analysis: AAManager on foo
----------------
aeubanks wrote:
> can we make this consistent by making the call order consistent?
I don't know why analyses tend to appear in different order (both when running and invalidating), e.g. when comparing an opt built using clang or with gcc. Maybe some set iteration that isn't guaranteeing iteration order.

I've updated this patch to use a bit less DAG checks. I don't think it was important to check everything.

But it is an interesting idea for the future to see if we can get it more consistent (lots of tests that are checking "Running analysis:" and "Invalidating analysis:" seem to use -DAG, which I think is a bit easy to forget when adding new tests).


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D146206/new/

https://reviews.llvm.org/D146206