[PATCH] D79304: [DAG] Fold (mul(abs(x),abs(x))) -> (mul(x,x)) (PR39476)

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon marked an inline comment as done.
RKSimon added a comment.

In D79304#2016813 <https://reviews.llvm.org/D79304#2016813>, @lebedev.ri wrote:

> Echoing what @craig.topper asked, do we somehow end up with that pattern after instcombine had a chance to do this fold?


Where did he ask that? https://godbolt.org/z/4vIPwf (from PR39476) shows that this IR currently comes through to the backend.



================
Comment at: llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp:5059
+// on success.
+SDValue DAGCombiner::MatchABS(SDNode *N) {
+  if (N->getOpcode() == ISD::ABS)
----------------
lebedev.ri wrote:
> Precommit NFC cleanup?
np - if people are happy with the patch I'll pull this out as a NFC precommit


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D79304/new/

https://reviews.llvm.org/D79304