[wasm-simd][x64] Small optimization for i64x2.splat

Movddup can take a memory operand, so we can save a move from gp reg to
xmm reg in that case. No problem with unaligned memory since we are
loading 64 bits (not 128 bits).

Also drive-by comment on i32x4.splat, it uses pshufd, which can also take
a memory operand (saving a mov), but we need aligned memory for that
first.

Bug: v8:9198
Change-Id: I55969888db1debb6ed4d193f767589d0da598386
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2567538Reviewed-by: Bill Budge <bbudge@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#71580}

src/compiler/backend/x64/code-generator-x64.cc

@@ -2719,10 +2719,10 @@ CodeGenerator::CodeGenResult CodeGenerator::AssembleArchInstruction(
       XMMRegister dst = i.OutputSimd128Register();
       if (HasRegisterInput(instr, 0)) {
         __ Movq(dst, i.InputRegister(0));
+        __ Movddup(dst, dst);
       } else {
-        __ Movq(dst, i.InputOperand(0));
+        __ Movddup(dst, i.InputOperand(0));
       }
-      __ Movddup(dst, dst);
       break;
     }
     case kX64I64x2ExtractLane: {
@@ -2815,6 +2815,7 @@ CodeGenerator::CodeGenResult CodeGenerator::AssembleArchInstruction(
       if (HasRegisterInput(instr, 0)) {
         __ Movd(dst, i.InputRegister(0));
       } else {
+        // TODO(v8:9198): Pshufd can load from aligned memory once supported.
         __ Movd(dst, i.InputOperand(0));
       }
       __ Pshufd(dst, dst, uint8_t{0x0});