SIInstrInfo::FoldImmediate tried to delete move-immediate instructions
after folding them into their only use. This did not work because it was
checking hasOneNonDBGUse after doing the fold, at which point there
should be no uses. This seems to have no effect on codegen, it just
means less stuff for DCE to clean up later.
Details
Details
Diff Detail
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
Comment Actions
Machine DCE is run right after Peephole Optimizer, but that's done in generic code (TargetPassConfig::addMachineSSAOptimization). I don't see how this would affect any of the other runs of Machine DCE, that are added by the AMDGPU backend.