[PATCH] D38276: [X86] Remove dead code from X86ISelDAGToDAG.cpp multiply handling

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added a comment.

What is the test coverage like? Searching for "mul i8" in x86 doesn't find much.....



================
Comment at: lib/Target/X86/X86ISelDAGToDAG.cpp:2733
-      if (!SDValue(Node, 0).use_empty())
-        ReplaceUses(SDValue(Node, 1),
-          CurDAG->getTargetExtractSubreg(X86::sub_8bit, dl, MVT::i8, Result));
----------------
craig.topper wrote:
> Shouldn't this ReplaceUses call have used SDValue(Node, 0)?
Yes, I think so. It looks like there is a lot of a MUL_LOHI ambiguity.


https://reviews.llvm.org/D38276