[PATCH] D114879: [AArch64][CodeGen] Emit alignment "Max Skip" operand for AArch64 loops

[Dave Green via Phabricator via llvm-commits]

dmgreen added inline comments.


================
Comment at: llvm/lib/Target/AArch64/AArch64ISelLowering.cpp:8529
 
+Align AArch64TargetLowering::getPrefLoopAlignment(MachineLoop *ML) const {
+  // TODO Investigate how to make this better, this is not ready to commit
----------------
The alignments in AArch64 are usually set in MachineBlockPlacement::alignBlocks, which is probably the more general place for this. The alignments can then be set in the same places as setPrefLoopAlignment for each backend.

Maybe at the moment, before we've fixed on proper values for AArch64 cpus, we could use an option as a way of testing that the values are getting propagated correctly.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D114879/new/

https://reviews.llvm.org/D114879