s390: [wasm-simd] Simplify S8x16Shuffle on s390

Change-Id: I52fc8025a4a878e4494ef9134a8c2bd1cec25fe1
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2289903Reviewed-by: Junliang Yan <jyan@ca.ibm.com>
Commit-Queue: Milad Farazmand <miladfar@ca.ibm.com>
Cr-Commit-Position: refs/heads/master@{#68763}

src/compiler/backend/s390/code-generator-s390.cc

@@ -4173,14 +4173,14 @@ CodeGenerator::CodeGenResult CodeGenerator::AssembleArchInstruction(
       // create 2 * 8 byte inputs indicating new indices
       for (int i = 0, j = 0; i < 2; i++, j = +2) {
 #ifdef V8_TARGET_BIG_ENDIAN
-        __ lgfi(i < 1 ? ip : r0, Operand(k8x16_indices[j + 1]));
-        __ iihf(i < 1 ? ip : r0, Operand(k8x16_indices[j]));
-#else
         __ lgfi(i < 1 ? ip : r0, Operand(k8x16_indices[j]));
         __ iihf(i < 1 ? ip : r0, Operand(k8x16_indices[j + 1]));
+#else
+        __ lgfi(i < 1 ? r0 : ip, Operand(k8x16_indices[j]));
+        __ iihf(i < 1 ? r0 : ip, Operand(k8x16_indices[j + 1]));
 #endif
       }
-      __ vlvgp(kScratchDoubleReg, ip, r0);
+      __ vlvgp(kScratchDoubleReg, r0, ip);
       __ vperm(dst, src0, src1, kScratchDoubleReg, Condition(0), Condition(0));
       break;
     }

src/compiler/backend/s390/instruction-selector-s390.cc

@@ -2834,6 +2834,7 @@ SIMD_VISIT_PMIN_MAX(F32x4Pmax)
 
 void InstructionSelector::VisitS8x16Shuffle(Node* node) {
   uint8_t shuffle[kSimd128Size];
+  uint8_t* shuffle_p = &shuffle[0];
   bool is_swizzle;
   CanonicalizeShuffle(node, shuffle, &is_swizzle);
   S390OperandGenerator g(this);
@@ -2851,22 +2852,14 @@ void InstructionSelector::VisitS8x16Shuffle(Node* node) {
                                ? max_index - current_index
                                : total_lane_count - current_index + max_index);
   }
-  Emit(kS390_S8x16Shuffle, g.DefineAsRegister(node), g.UseRegister(input0),
-       g.UseRegister(input1),
-       // Pack4Lanes reverses the bytes, therefore we will need to pass it in
-       // reverse
-       g.UseImmediate(Pack4Lanes(shuffle_remapped + 12)),
-       g.UseImmediate(Pack4Lanes(shuffle_remapped + 8)),
-       g.UseImmediate(Pack4Lanes(shuffle_remapped + 4)),
-       g.UseImmediate(Pack4Lanes(shuffle_remapped)));
-#else
+  shuffle_p = &shuffle_remapped[0];
+#endif
   Emit(kS390_S8x16Shuffle, g.DefineAsRegister(node),
        g.UseUniqueRegister(input0), g.UseUniqueRegister(input1),
-       g.UseImmediate(Pack4Lanes(shuffle)),
-       g.UseImmediate(Pack4Lanes(shuffle + 4)),
-       g.UseImmediate(Pack4Lanes(shuffle + 8)),
-       g.UseImmediate(Pack4Lanes(shuffle + 12)));
-#endif
+       g.UseImmediate(Pack4Lanes(shuffle_p)),
+       g.UseImmediate(Pack4Lanes(shuffle_p + 4)),
+       g.UseImmediate(Pack4Lanes(shuffle_p + 8)),
+       g.UseImmediate(Pack4Lanes(shuffle_p + 12)));
 }
 
 void InstructionSelector::VisitS8x16Swizzle(Node* node) {