[mips][wasm] Implement "atomic.fence" operator

port 4ca8b4dd https://crrev.com/c/1701856

Original Commit Message:
    This adds decoding and compilation of the "atomic.fence" operator, which
    is intended to preserve the synchronization guarantees of higher-level
    languages.

    Unlike other atomic operators, it does not target a particular linear
    memory. It may occur in modules which declare no memory, or a non-shared
    memory, without causing a validation error.

    See proposal: https://github.com/WebAssembly/threads/pull/141
    See discussion: https://github.com/WebAssembly/threads/issues/140

Change-Id: Ia60d58a6bf58e8236591d515d30184418cee47c5
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1710337Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Yu Yin <xwafish@gmail.com>
Cr-Commit-Position: refs/heads/master@{#62843}

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

@@ -1620,6 +1620,10 @@ CodeGenerator::CodeGenResult CodeGenerator::AssembleArchInstruction(
       __ Usdc1(ft, i.MemoryOperand(), kScratchReg);
       break;
     }
+    case kMipsSync: {
+      __ sync();
+      break;
+    }
     case kMipsPush:
       if (instr->InputAt(0)->IsFPRegister()) {
         LocationOperand* op = LocationOperand::cast(instr->InputAt(0));

src/compiler/backend/mips/instruction-codes-mips.h

@@ -134,6 +134,7 @@ namespace compiler {
   V(MipsStackClaim)                \
   V(MipsSeb)                       \
   V(MipsSeh)                       \
+  V(MipsSync)                      \
   V(MipsS128Zero)                  \
   V(MipsI32x4Splat)                \
   V(MipsI32x4ExtractLane)          \

src/compiler/backend/mips/instruction-scheduler-mips.cc

@@ -284,6 +284,7 @@ int InstructionScheduler::GetTargetInstructionFlags(
     case kMipsUsh:
     case kMipsUsw:
     case kMipsUswc1:
+    case kMipsSync:
     case kMipsWord32AtomicPairStore:
     case kMipsWord32AtomicPairAdd:
     case kMipsWord32AtomicPairSub:

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

@@ -1775,6 +1775,11 @@ void InstructionSelector::VisitFloat64SilenceNaN(Node* node) {
        arraysize(temps), temps);
 }
 
+void InstructionSelector::VisitMemoryBarrier(Node* node) {
+  MipsOperandGenerator g(this);
+  Emit(kMipsSync, g.NoOutput());
+}
+
 void InstructionSelector::VisitWord32AtomicLoad(Node* node) {
   LoadRepresentation load_rep = LoadRepresentationOf(node->op());
   MipsOperandGenerator g(this);

src/compiler/backend/mips64/code-generator-mips64.cc

@@ -1795,6 +1795,10 @@ CodeGenerator::CodeGenResult CodeGenerator::AssembleArchInstruction(
       __ Usdc1(ft, i.MemoryOperand(), kScratchReg);
       break;
     }
+    case kMips64Sync: {
+      __ sync();
+      break;
+    }
     case kMips64Push:
       if (instr->InputAt(0)->IsFPRegister()) {
         __ Sdc1(i.InputDoubleRegister(0), MemOperand(sp, -kDoubleSize));

src/compiler/backend/mips64/instruction-codes-mips64.h

@@ -163,6 +163,7 @@ namespace compiler {
   V(Mips64StackClaim)                        \
   V(Mips64Seb)                               \
   V(Mips64Seh)                               \
+  V(Mips64Sync)                              \
   V(Mips64AssertEqual)                       \
   V(Mips64S128Zero)                          \
   V(Mips64I32x4Splat)                        \

src/compiler/backend/mips64/instruction-scheduler-mips64.cc

@@ -318,6 +318,7 @@ int InstructionScheduler::GetTargetInstructionFlags(
     case kMips64Ush:
     case kMips64Usw:
     case kMips64Uswc1:
+    case kMips64Sync:
     case kMips64Word64AtomicStoreWord8:
     case kMips64Word64AtomicStoreWord16:
     case kMips64Word64AtomicStoreWord32:

src/compiler/backend/mips64/instruction-selector-mips64.cc

@@ -2408,6 +2408,11 @@ void InstructionSelector::VisitFloat64InsertHighWord32(Node* node) {
        g.UseRegister(left), g.UseRegister(right));
 }
 
+void InstructionSelector::VisitMemoryBarrier(Node* node) {
+  Mips64OperandGenerator g(this);
+  Emit(kMips64Sync, g.NoOutput());
+}
+
 void InstructionSelector::VisitWord32AtomicLoad(Node* node) {
   LoadRepresentation load_rep = LoadRepresentationOf(node->op());
   ArchOpcode opcode = kArchNop;