[RISCV Add some vsetvli insertion test cases with vmv.s.x+reduction. NFC

These test cases where intended to get a single vsetvli by using
the vmv.s.x intrinsic with the same LMUL as the reduction. This
works for FP, but does not work for integer.

I believe llvm#71501 will break this for FP too. Hopefully the vsetvli
pass can be taught to fix this.

Author: topperc

llvm/test/CodeGen/RISCV/rvv/vsetvli-insert.ll

@@ -627,6 +627,41 @@ define void @add_v16i64(ptr %x, ptr %y) vscale_range(2,2) {
   ret void
 }
 
+define <vscale x 2 x float> @fp_reduction_vfmv_s_f(float %0, <vscale x 8 x float> %1, i64  %2) {
+; CHECK-LABEL: fp_reduction_vfmv_s_f:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, ma
+; CHECK-NEXT:    vfmv.s.f v12, fa0
+; CHECK-NEXT:    vfredusum.vs v8, v8, v12
+; CHECK-NEXT:    ret
+  %4 = tail call <vscale x 8 x float> @llvm.riscv.vfmv.s.f.nxv8f32.i64(<vscale x 8 x float> poison, float %0, i64 %2)
+  %5 = tail call <vscale x 2 x float> @llvm.vector.extract.nxv2f32.nxv8f32(<vscale x 8 x float> %4, i64 0)
+  %6 = tail call <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float> poison, <vscale x 8 x float> %1, <vscale x 2 x float> %5, i64 7, i64 %2)
+  ret <vscale x 2 x float> %6
+}
+
+define dso_local <vscale x 2 x i32> @int_reduction_vmv_s_x(i32 signext %0, <vscale x 8 x i32> %1, i64 %2) {
+; CHECK-LABEL: int_reduction_vmv_s_x:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.s.x v12, a0
+; CHECK-NEXT:    vsetvli zero, a1, e32, m4, ta, ma
+; CHECK-NEXT:    vredsum.vs v8, v8, v12
+; CHECK-NEXT:    ret
+  %4 = tail call <vscale x 8 x i32> @llvm.riscv.vmv.s.x.nxv8i32.i64(<vscale x 8 x i32> poison, i32 %0, i64 %2)
+  %5 = tail call <vscale x 2 x i32> @llvm.vector.extract.nxv2i32.nxv8i32(<vscale x 8 x i32> %4, i64 0)
+  %6 = tail call <vscale x 2 x i32> @llvm.riscv.vredsum.nxv2i32.nxv8i32.i64(<vscale x 2 x i32> poison, <vscale x 8 x i32> %1, <vscale x 2 x i32> %5, i64 %2)
+  ret <vscale x 2 x i32> %6
+}
+
+declare <vscale x 8 x float> @llvm.riscv.vfmv.s.f.nxv8f32.i64(<vscale x 8 x float>, float, i64)
+declare <vscale x 2 x float> @llvm.vector.extract.nxv2f32.nxv8f32(<vscale x 8 x float>, i64)
+declare <vscale x 2 x float> @llvm.riscv.vfredusum.nxv2f32.nxv8f32.i64(<vscale x 2 x float>, <vscale x 8 x float>, <vscale x 2 x float>, i64, i64)
+
+declare <vscale x 8 x i32> @llvm.riscv.vmv.s.x.nxv8i32.i64(<vscale x 8 x i32>, i32, i64) #1
+declare <vscale x 2 x i32> @llvm.vector.extract.nxv2i32.nxv8i32(<vscale x 8 x i32>, i64 immarg) #2
+declare <vscale x 2 x i32> @llvm.riscv.vredsum.nxv2i32.nxv8i32.i64(<vscale x 2 x i32>, <vscale x 8 x i32>, <vscale x 2 x i32>, i64) #1
+
 declare <vscale x 1 x i64> @llvm.riscv.vadd.mask.nxv1i64.nxv1i64(
   <vscale x 1 x i64>,
   <vscale x 1 x i64>,