[all-commits] [llvm/llvm-project] 3969d2: [InstCombine] Disable select known bits fold for v...

[Nikita Popov via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 3969d2c3b5f42e4a180f5205efa780b0f950d733
      https://github.com/llvm/llvm-project/commit/3969d2c3b5f42e4a180f5205efa780b0f950d733
  Author: Nikita Popov <npopov at redhat.com>
  Date:   2024-07-03 (Wed, 03 Jul 2024)

  Changed paths:
    M llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp
    M llvm/test/Transforms/InstCombine/select-binop-cmp.ll

  Log Message:
  -----------
  [InstCombine] Disable select known bits fold for vectors

This is not safe if the simplification ends up looking through
lane-crossing operations. For now, we don't have a good way to
limit this in computeKnownBits(), so just disable vector handling
entirely.

Fixes https://github.com/llvm/llvm-project/issues/97475.



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications