Move fold of (sext (not i1 x)) -> (add (zext i1 x), -1) from X86 to DAGCombiner to improve codegen on other targets. Differential Revision: https://reviews.llvm.org/D91589
19 KiB
19 KiB