Add x86 encoding for isub

cranelift/filetests/filetests/isa/x86/simd-arithmetic.clif

@@ -48,3 +48,40 @@ ebb0(v0: i64x2 [%xmm3], v1: i64x2 [%xmm4]):
 [-, %xmm3]  v2 = iadd v0, v1      ; bin: 66 0f d4 dc
             return v2
 }
+
+function %isub_i32x4() -> b1 {
+ebb0:
+[-, %xmm3]    v0 = vconst.i32x4 [1 1 1 1]
+[-, %xmm5]    v1 = vconst.i32x4 [1 2 3 4]
+[-, %xmm3]    v2 = isub v0, v1  ; bin: 66 0f fa dd
+
+    v3 = extractlane v2, 0
+    v4 = icmp_imm eq v3, 0
+
+    v5 = extractlane v2, 1
+    v6 = icmp_imm eq v5, 0xffffffff
+    ; TODO replace extractlanes with vector comparison
+
+    v7 = band v4, v6
+    return v7
+}
+
+; run
+
+function %isub_i64x2(i64x2, i64x2) -> i64x2 {
+ebb0(v0: i64x2 [%xmm0], v1: i64x2 [%xmm1]):
+[-, %xmm0]  v2 = isub v0, v1      ; bin: 66 0f fb c1
+    return v2
+}
+
+function %isub_i16x8(i16x8, i16x8) -> i16x8 {
+ebb0(v0: i16x8 [%xmm3], v1: i16x8 [%xmm4]):
+[-, %xmm3]  v2 = isub v0, v1      ; bin: 66 0f f9 dc
+    return v2
+}
+
+function %isub_i8x16(i8x16, i8x16) -> i8x16 {
+ebb0(v0: i8x16 [%xmm3], v1: i8x16 [%xmm4]):
+[-, %xmm3]  v2 = isub v0, v1      ; bin: 66 0f f8 dc
+    return v2
+}