[all-commits] [llvm/llvm-project] f7ce52: [AMDGPU] Fix FDOT2 fold for non-constant lane indi...

[Arseniy Obolenskiy via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: f7ce52f0f4ec943bfdc749bc67de00a1e3ece6bc
      https://github.com/llvm/llvm-project/commit/f7ce52f0f4ec943bfdc749bc67de00a1e3ece6bc
  Author: Arseniy Obolenskiy <arseniy.obolenskiy at amd.com>
  Date:   2026-09-07 (Mon, 07 Sep 2026)

  Changed paths:
    M llvm/lib/Target/AMDGPU/SIISelLowering.cpp
    M llvm/test/CodeGen/AMDGPU/fdot2.ll

  Log Message:
  -----------
  [AMDGPU] Fix FDOT2 fold for non-constant lane indices (#217646)

The fold only checked that the two lane index nodes were not the same
SDValue, but distinct nodes can still be equal at runtime, silently
picking the wrong lanes

Require both indices to be constant and distinct



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