[PATCH] D157426: [AMDGPU][True16] Fix ISel for A16 Image Instructions

[Jay Foad via Phabricator via llvm-commits]

foad accepted this revision.
foad added a comment.
This revision is now accepted and ready to land.

LGTM. From reading SITargetLowering::lowerImage I think SelectionDAG already does it this way.



================
Comment at: llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp:5890-5891
 
-      packImage16bitOpsToDwords(B, MI, PackedRegs, ArgOffset, Intr, IsA16,
-                                IsG16);
-
----------------
arsenm wrote:
> This deleted all the code by using a function it was already using?
The code below is just un-indented, not deleted.


================
Comment at: llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.atomic.dim.a16.ll:684
   ; GFX9-NEXT:   SI_RETURN_TO_EPILOG implicit $vgpr0
+  ;
   ; GFX10NSA-LABEL: name: atomic_add_2d
----------------
Could do a pre-commit that just regenerates checks in these files, to avoid this kind of noise showing up in your patch.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D157426/new/

https://reviews.llvm.org/D157426