[PATCH] D134862: [AMDGPU][GISel] Update `isCanonicalized`

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SIISelLowering.cpp:10222-10223
+    return isCanonicalized(MI->getOperand(1).getReg(), MF, MaxDepth - 1);
   case AMDGPU::G_FMINNUM_IEEE:
   case AMDGPU::G_FMAXNUM_IEEE: {
     if (Subtarget->supportsMinMaxDenormModes() ||
----------------
Also should handle G_FMINNUM/G_FMAXNUM. Also, we should really stop ascribing target instruction behavior to the generic opcodes. This is a separate patch though since this is a big mess


================
Comment at: llvm/lib/Target/AMDGPU/SIISelLowering.cpp:10266
   default:
     return denormalsEnabledForType(MRI.getType(Reg), MF) &&
            isKnownNeverSNaN(Reg, MRI);
----------------
Should default to return false


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D134862/new/

https://reviews.llvm.org/D134862