[PATCH] D60660: [InstCombine] Prune fshl/fshr with masked operands

[Shawn Landden via Phabricator via llvm-commits]

shawnl marked 2 inline comments as done.
shawnl added inline comments.


================
Comment at: lib/Transforms/InstCombine/InstCombineCalls.cpp:2087
+
+      if (SimplifyDemandedInstructionBits(*II))
+        return &CI;
----------------
nikic wrote:
> lebedev.ri wrote:
> > This should probably be at the top level of whatever function this is.
> > Not sure where in the function though.
> It should probably be right below this block. We already have the demanded bits simplification for the shift amount there (which we should probably move into the fshl handling in InstCombineSimplifyDemanded as a followup).
I think it should go at the top of the match block here. This text is for the shift being a constant, and it can't be simplified unless the shift is a constant.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D60660/new/

https://reviews.llvm.org/D60660