[all-commits] [llvm/llvm-project] c6672a: [InstCombine] Fix invalid fshl -> lshr fold when s...

[Benji Smith via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: c6672a6fdf1469d818d0ac8aaa2c31a0f18e729a
      https://github.com/llvm/llvm-project/commit/c6672a6fdf1469d818d0ac8aaa2c31a0f18e729a
  Author: Benji Smith <6193112+Benjins at users.noreply.github.com>
  Date:   2026-07-19 (Sun, 19 Jul 2026)

  Changed paths:
    M llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
    M llvm/test/Transforms/InstCombine/fsh.ll

  Log Message:
  -----------
  [InstCombine] Fix invalid fshl -> lshr fold when shift is vector with constant zero lane (#210606)

For vectors, this is not valid if a subset of the lanes are constant
zero. Doing this fold introduces poison in those lanes. A similar fix
was done for fshr in 46957a138dea339799059ea5bf032e40042ff185 however
the same issue seems to apply to fshl

Fixes https://github.com/llvm/llvm-project/issues/210605



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