[PATCH] D89234: [AMDGPU] Avoid offset register in MUBUF for direct stack object accesses

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: llvm/test/CodeGen/AMDGPU/stack-pointer-offset-relative-frameindex.ll:43-46
+; GCN-NEXT:    v_mov_b32_e32 v2, 0x4000
+; GCN-NEXT:    buffer_load_dword v0, v2, s[36:39], 0 offen offset:4
+; GCN-NEXT:    v_mov_b32_e32 v2, 0x4000
+; GCN-NEXT:    buffer_load_dword v1, v2, s[36:39], 0 offen offset:8
----------------
This is worse. Why did this break the local stack offset optimization?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D89234/new/

https://reviews.llvm.org/D89234