[PATCH] D109888: Bitcast wrapped in a call obscures function attributes, pessimizing MemorySSA

[Johannes Doerfert via Phabricator via llvm-commits]

jdoerfert added a comment.

I'm slightly worried about the inconsistency but I am convincing myself slowly this is fine to do. I also imagine a single bitcast is the most important case to catch. The tests for MSSA is a bit obscure but I guess there is sufficient test coverage across the 4 tests.

The Attributor tests need to be updated properly though. I left comments.



================
Comment at: llvm/test/Transforms/Attributor/IPConstantProp/arg-count-mismatch.ll:1
-; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-attributes --check-globals
 ; RUN: opt -attributor -enable-new-pm=0 -attributor-manifest-internal  -attributor-max-iterations-verify -attributor-annotate-decl-cs -attributor-max-iterations=1 -S < %s | FileCheck %s --check-prefixes=CHECK,NOT_CGSCC_NPM,NOT_CGSCC_OPM,NOT_TUNIT_NPM,IS__TUNIT____,IS________OPM,IS__TUNIT_OPM
----------------
Don't change the options.


================
Comment at: llvm/test/Transforms/Attributor/IPConstantProp/arg-count-mismatch.ll:135
-; IS__CGSCC____: attributes #[[ATTR0]] = { nofree norecurse nosync nounwind readnone willreturn }
-;.
----------------
Run the update script so these things don't go away. 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D109888/new/

https://reviews.llvm.org/D109888