[PATCH] D141557: [AMDGPU] Further reduce attaching of implicit operands to spills / copies

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: llvm/test/CodeGen/AMDGPU/accvgpr-copy.mir:554
     ; GFX908-NEXT: {{  $}}
-    ; GFX908-NEXT: $vgpr255 = V_MOV_B32_e32 $sgpr0, implicit $exec, implicit $sgpr0_sgpr1
+    ; GFX908-NEXT: $vgpr255 = V_MOV_B32_e32 $sgpr0, implicit $exec, implicit-def $sgpr0_sgpr1
     ; GFX908-NEXT: $agpr0 = V_ACCVGPR_WRITE_B32_e64 killed $vgpr255, implicit $exec, implicit-def $agpr0_agpr1
----------------
I don't understand what happened here, a use became a def and is breaking the value for sgpr1. I also doubt we needed this implicit use in the first place 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D141557/new/

https://reviews.llvm.org/D141557