[PATCH] D90369: [FastRA] Fix handling of bundled MIs

[Pushpinder Singh via Phabricator via llvm-commits]

pdhaliwal added inline comments.


================
Comment at: llvm/lib/CodeGen/RegAllocFast.cpp:1450
 
     allocateInstruction(MI);
+
----------------
qcolombet wrote:
> Could we fix `allocateInstruction` instead of having a somewhat parallel way of assigning registers?
If fixing the allocateInstruction means that for each assignment happening over bundle header,  a corresponding assignment will happen for bundled instruction, then I am not sure if that is the good way. `allocateInstruction` has multiple assignment passes over operands. And for each operand's assignment, I would need to get uses and apply the same assignment. Is that what you are thinking of? Or Should I just move the if condition down below to the allocateInstruction?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D90369/new/

https://reviews.llvm.org/D90369