[wasm-simd][liftoff][x64] Implement v64x2.alltrue and i64x2.ne

Other archs will come later.

Bug: v8:11347,v8:11348
Change-Id: I9ea656b9c7ce03c9dafb631dd67f6e2f7d4346a2
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2686312Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#72674}

src/wasm/baseline/arm/liftoff-assembler-arm.h

@@ -2824,6 +2824,11 @@ void LiftoffAssembler::emit_i64x2_neg(LiftoffRegister dst,
        liftoff::GetSimd128Register(src));
 }
 
+void LiftoffAssembler::emit_v64x2_alltrue(LiftoffRegister dst,
+                                          LiftoffRegister src) {
+  bailout(kSimd, "v64x2_alltrue");
+}
+
 void LiftoffAssembler::emit_i64x2_shl(LiftoffRegister dst, LiftoffRegister lhs,
                                       LiftoffRegister rhs) {
   liftoff::EmitSimdShift<liftoff::kLeft, NeonS64, Neon32>(this, dst, lhs, rhs);
@@ -3730,6 +3735,11 @@ void LiftoffAssembler::emit_i64x2_eq(LiftoffRegister dst, LiftoffRegister lhs,
           liftoff::GetSimd128Register(rhs));
 }
 
+void LiftoffAssembler::emit_i64x2_ne(LiftoffRegister dst, LiftoffRegister lhs,
+                                     LiftoffRegister rhs) {
+  bailout(kSimd, "i64x2_ne");
+}
+
 void LiftoffAssembler::emit_f32x4_eq(LiftoffRegister dst, LiftoffRegister lhs,
                                      LiftoffRegister rhs) {
   vceq(liftoff::GetSimd128Register(dst), liftoff::GetSimd128Register(lhs),

src/wasm/baseline/arm64/liftoff-assembler-arm64.h

@@ -1993,6 +1993,11 @@ void LiftoffAssembler::emit_i64x2_neg(LiftoffRegister dst,
   Neg(dst.fp().V2D(), src.fp().V2D());
 }
 
+void LiftoffAssembler::emit_v64x2_alltrue(LiftoffRegister dst,
+                                          LiftoffRegister src) {
+  bailout(kSimd, "v64x2_alltrue");
+}
+
 void LiftoffAssembler::emit_i64x2_shl(LiftoffRegister dst, LiftoffRegister lhs,
                                       LiftoffRegister rhs) {
   liftoff::EmitSimdShift<liftoff::ShiftDirection::kLeft>(
@@ -2734,6 +2739,11 @@ void LiftoffAssembler::emit_i64x2_eq(LiftoffRegister dst, LiftoffRegister lhs,
   Cmeq(dst.fp().V2D(), lhs.fp().V2D(), rhs.fp().V2D());
 }
 
+void LiftoffAssembler::emit_i64x2_ne(LiftoffRegister dst, LiftoffRegister lhs,
+                                     LiftoffRegister rhs) {
+  bailout(kSimd, "i64x2_ne");
+}
+
 void LiftoffAssembler::emit_f32x4_eq(LiftoffRegister dst, LiftoffRegister lhs,
                                      LiftoffRegister rhs) {
   Fcmeq(dst.fp().V4S(), lhs.fp().V4S(), rhs.fp().V4S());

src/wasm/baseline/ia32/liftoff-assembler-ia32.h

@@ -3111,6 +3111,11 @@ void LiftoffAssembler::emit_i64x2_eq(LiftoffRegister dst, LiftoffRegister lhs,
       this, dst, lhs, rhs, SSE4_1);
 }
 
+void LiftoffAssembler::emit_i64x2_ne(LiftoffRegister dst, LiftoffRegister lhs,
+                                     LiftoffRegister rhs) {
+  bailout(kSimd, "i64x2_ne");
+}
+
 void LiftoffAssembler::emit_f32x4_eq(LiftoffRegister dst, LiftoffRegister lhs,
                                      LiftoffRegister rhs) {
   liftoff::EmitSimdCommutativeBinOp<&Assembler::vcmpeqps, &Assembler::cmpeqps>(
@@ -3819,6 +3824,11 @@ void LiftoffAssembler::emit_i64x2_neg(LiftoffRegister dst,
   }
 }
 
+void LiftoffAssembler::emit_v64x2_alltrue(LiftoffRegister dst,
+                                          LiftoffRegister src) {
+  bailout(kSimd, "v64x2_alltrue");
+}
+
 void LiftoffAssembler::emit_i64x2_shl(LiftoffRegister dst, LiftoffRegister lhs,
                                       LiftoffRegister rhs) {
   liftoff::EmitSimdShiftOp<&Assembler::vpsllq, &Assembler::psllq, 6>(this, dst,

src/wasm/baseline/liftoff-assembler.h

@@ -932,6 +932,8 @@ class LiftoffAssembler : public TurboAssembler {
                               LiftoffRegister rhs);
   inline void emit_i64x2_eq(LiftoffRegister dst, LiftoffRegister lhs,
                             LiftoffRegister rhs);
+  inline void emit_i64x2_ne(LiftoffRegister dst, LiftoffRegister lhs,
+                            LiftoffRegister rhs);
   inline void emit_f32x4_eq(LiftoffRegister dst, LiftoffRegister lhs,
                             LiftoffRegister rhs);
   inline void emit_f32x4_ne(LiftoffRegister dst, LiftoffRegister lhs,
@@ -1092,6 +1094,7 @@ class LiftoffAssembler : public TurboAssembler {
                                              LiftoffRegister src1,
                                              LiftoffRegister src2);
   inline void emit_i64x2_neg(LiftoffRegister dst, LiftoffRegister src);
+  inline void emit_v64x2_alltrue(LiftoffRegister dst, LiftoffRegister src);
   inline void emit_i64x2_shl(LiftoffRegister dst, LiftoffRegister lhs,
                              LiftoffRegister rhs);
   inline void emit_i64x2_shli(LiftoffRegister dst, LiftoffRegister lhs,

src/wasm/baseline/liftoff-compiler.cc

@@ -3205,6 +3205,8 @@ class LiftoffCompiler {
             &LiftoffAssembler::emit_i32x4_extmul_high_i16x8_u);
       case wasm::kExprI64x2Neg:
         return EmitUnOp<kS128, kS128>(&LiftoffAssembler::emit_i64x2_neg);
+      case wasm::kExprV64x2AllTrue:
+        return EmitUnOp<kS128, kI32>(&LiftoffAssembler::emit_v64x2_alltrue);
       case wasm::kExprI64x2Shl:
         return EmitSimdShiftOp(&LiftoffAssembler::emit_i64x2_shl,
                                &LiftoffAssembler::emit_i64x2_shli);

src/wasm/baseline/x64/liftoff-assembler-x64.h

@@ -2319,7 +2319,11 @@ inline void EmitAnyTrue(LiftoffAssembler* assm, LiftoffRegister dst,
 
 template <void (TurboAssembler::*pcmp)(XMMRegister, XMMRegister)>
 inline void EmitAllTrue(LiftoffAssembler* assm, LiftoffRegister dst,
-                        LiftoffRegister src) {
+                        LiftoffRegister src,
+                        base::Optional<CpuFeature> feature = base::nullopt) {
+  base::Optional<CpuFeatureScope> sse_scope;
+  if (feature.has_value()) sse_scope.emplace(assm, *feature);
+
   XMMRegister tmp = kScratchDoubleReg;
   assm->xorq(dst.gp(), dst.gp());
   assm->Pxor(tmp, tmp);
@@ -2696,6 +2700,14 @@ void LiftoffAssembler::emit_i64x2_eq(LiftoffRegister dst, LiftoffRegister lhs,
       this, dst, lhs, rhs, SSE4_1);
 }
 
+void LiftoffAssembler::emit_i64x2_ne(LiftoffRegister dst, LiftoffRegister lhs,
+                                     LiftoffRegister rhs) {
+  liftoff::EmitSimdCommutativeBinOp<&Assembler::vpcmpeqq, &Assembler::pcmpeqq>(
+      this, dst, lhs, rhs, SSE4_1);
+  Pcmpeqq(kScratchDoubleReg, kScratchDoubleReg);
+  Pxor(dst.fp(), kScratchDoubleReg);
+}
+
 void LiftoffAssembler::emit_f32x4_eq(LiftoffRegister dst, LiftoffRegister lhs,
                                      LiftoffRegister rhs) {
   liftoff::EmitSimdCommutativeBinOp<&Assembler::vcmpeqps, &Assembler::cmpeqps>(
@@ -3391,6 +3403,11 @@ void LiftoffAssembler::emit_i64x2_neg(LiftoffRegister dst,
   }
 }
 
+void LiftoffAssembler::emit_v64x2_alltrue(LiftoffRegister dst,
+                                          LiftoffRegister src) {
+  liftoff::EmitAllTrue<&TurboAssembler::Pcmpeqq>(this, dst, src, SSE4_1);
+}
+
 void LiftoffAssembler::emit_i64x2_shl(LiftoffRegister dst, LiftoffRegister lhs,
                                       LiftoffRegister rhs) {
   liftoff::EmitSimdShiftOp<&Assembler::vpsllq, &Assembler::psllq, 6>(this, dst,