[PATCH] D63566: AMDGPU: Fix folding immediate into readfirstlane through reg_sequence

[Matt Arsenault via Phabricator via llvm-commits]

arsenm marked an inline comment as done.
arsenm added inline comments.


================
Comment at: lib/Target/AMDGPU/SIInstrInfo.cpp:6082-6084
+  const MachineInstr *DefMI = MRI.getVRegDef(VReg);
   auto *TRI = MRI.getTargetRegisterInfo();
+  auto *DefBB = DefMI->getParent();
----------------
Actually it would probably be more correct to delete the assert. We care about the original value def point that we're going to fold, not the actual def instruction


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D63566/new/

https://reviews.llvm.org/D63566