[PATCH] D131959: [AMDGPU] Fix SDST operand of V_DIV_SCALE to always be VCC

[Stanislav Mekhanoshin via Phabricator via llvm-commits]

rampitec added a comment.

This is really an overkill to have a separate pass to fix vcc copy of just 2 instructions. What if you change the lowering instead? I believe it shall be possible to w/a it by not reading the second result of the lowered DIV_SCALE and glue a CopyFromReg to it with a correct register, i.e. VCC or VCC_LO.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D131959/new/

https://reviews.llvm.org/D131959