[llvm] [X86] Fix lower1BitShuffle blend-with-zero shuffles to AND mask (PR #180472)

[via llvm-commits]

================
@@ -250,3 +250,18 @@ define i16 @PR90356(<16 x i1> %a) {
   %2 = bitcast <16 x i1> %1 to i16
   ret i16 %2
 }
+
+define <4 x i1> @ISSUE180426(<4 x i1> %0) {
----------------
woruyu wrote:

Done!

https://github.com/llvm/llvm-project/pull/180472