On Nov 30, 2007, at 2:59 PM, Chris Lattner wrote: > > On Nov 30, 2007, at 1:23 PM, Evan Cheng wrote: > >> + // FIXME: It may be possible to fold load when there are >> multiple uses. >> + // e.g. On x86, TEST32rr r, r -> CMP32rm [mem], 0 >> + if (NumUses > 1) >> + return false; > > Maybe it would be better for the target foldMemoryOperand hook to > handle this?
Yeah, I have added that. Not yet checked in. > > Is there any hope of autogenerating foldMemoryOperand from the .td > files? If so, that would be easy to add. IMO, that's tricky enough to not be worthwhile. Evan > > -Chris > _______________________________________________ > llvm-commits mailing list > llvm-commits@cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits _______________________________________________ llvm-commits mailing list llvm-commits@cs.uiuc.edu http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits