[PATCH] D137238: [AMDGPU][MC][GFX11][NFC] Correct VINTERP src operands

Joe Nash via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Wed Nov 2 11:42:58 PDT 2022


Joe_Nash added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/VINTERPInstructions.td:67
   let Outs64 = (outs VGPR_32:$vdst);
-  let Ins64 = (ins Src0Mod:$src0_modifiers, VRegSrc_32:$src0,
-                   Src1Mod:$src1_modifiers, VRegSrc_32:$src1,
----------------
I think setting the src operand type to VGPR_32 here is slightly wrong. The comment in SIRegisterInfo.td says use VRegSrc for operands that take only VGPRs and have enum{9} (are 9 bits in the encoding). The encoding for VINTERP has 9 bit fields. Functionally, it does not appear to make a difference. 


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D137238/new/

https://reviews.llvm.org/D137238



More information about the llvm-commits mailing list