[all-commits] [llvm/llvm-project] 6cfe6a: [NFC][AMDGPU] Assert no bad shift operations will ...

[Georgi Mirazchiyski via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 6cfe6a6b3e9578be80120add7fbe19506f747196
      https://github.com/llvm/llvm-project/commit/6cfe6a6b3e9578be80120add7fbe19506f747196
  Author: Georgi Mirazchiyski <georgi.mirazchiyski at codeplay.com>
  Date:   2024-09-24 (Tue, 24 Sep 2024)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPUArgumentUsageInfo.h
    M llvm/lib/Target/AMDGPU/SIInstrInfo.cpp

  Log Message:
  -----------
  [NFC][AMDGPU] Assert no bad shift operations will happen (#108416)

The assumption in the asserts is based on the fact that no SGPR/VGPR
register Arg mask in the ISelLowering and Legalizer can equal zero. They
are implicitly set to ~0 by default (meaning non-masked) or explicitly
to a non-zero value.
The `optimizeCompareInstr` case is different from the above described.
It requires the mask to be a power-of-two because it's a special-case
optimization, hence in this case we still cannot have an invalid shift.
This commit also silences static analysis tools wrt potential bad shifts
that could result from the output of `countr_zero(Mask)`.



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