[llvm] [RISCV] Fold vector shift of sext/zext to widening multiply (PR #121563)

[Piotr Fusik via llvm-commits]

================
@@ -654,15 +670,17 @@ define <16 x i16> @vwsll_vx_i8_v16i16_zext(<16 x i8> %a, i8 %b) {
 define <16 x i16> @vwsll_vi_v16i16(<16 x i8> %a) {
 ; CHECK-LABEL: vwsll_vi_v16i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, ma
-; CHECK-NEXT:    vzext.vf2 v10, v8
-; CHECK-NEXT:    vsll.vi v8, v10, 2
+; CHECK-NEXT:    li a0, 4
+; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, ma
+; CHECK-NEXT:    vwmulu.vx v10, v8, a0
+; CHECK-NEXT:    vmv2r.v v8, v10
 ; CHECK-NEXT:    ret
 ;
 ; CHECK-ZVBB-LABEL: vwsll_vi_v16i16:
 ; CHECK-ZVBB:       # %bb.0:
+; CHECK-ZVBB-NEXT:    li a0, 4
 ; CHECK-ZVBB-NEXT:    vsetivli zero, 16, e8, m1, ta, ma
-; CHECK-ZVBB-NEXT:    vwsll.vi v10, v8, 2
+; CHECK-ZVBB-NEXT:    vwmulu.vx v10, v8, a0
----------------
pfusik wrote:

Fixed by delaying this transform.

https://github.com/llvm/llvm-project/pull/121563