[all-commits] [llvm/llvm-project] 8943bf: [AMDGPU] Do not scale private alloca size when usi...

[Juan Manuel Martinez Caamaño via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 8943bfbd08202aae1e85181082a1ec337e6a8b0b
      https://github.com/llvm/llvm-project/commit/8943bfbd08202aae1e85181082a1ec337e6a8b0b
  Author: Juan Manuel Martinez Caamaño <jmartinezcaamao at gmail.com>
  Date:   2026-06-03 (Wed, 03 Jun 2026)

  Changed paths:
    M llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp
    M llvm/lib/Target/AMDGPU/SIISelLowering.cpp
    M llvm/test/CodeGen/AMDGPU/GlobalISel/dynamic-alloca-uniform.ll
    M llvm/test/CodeGen/AMDGPU/amdgpu-cs-chain-fp-nosave.ll
    M llvm/test/CodeGen/AMDGPU/dynamic_stackalloc.ll
    M llvm/test/CodeGen/AMDGPU/llvm.sponentry.ll
    M llvm/test/CodeGen/AMDGPU/non-entry-alloca.ll

  Log Message:
  -----------
  [AMDGPU] Do not scale private alloca size when using flat-scratch (#201142)

When using flat-scratch, the `scratch_load/scratch_store` instructions
scale the stack offset by the wavefront size on their own.

Scaling the alloca-size by the wave-front size lead to accesses outside
of the private-memory limit.



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