deps: V8: cherry-pick 6771d3e31883

Original commit message:

    [wasm][liftoff][cleanup] Remove default parameter of GetUnusedRegister

    This CL removes the default parameter of GetUnusedRegister to avoid bugs
    where the default parameter is used accidentially. With "{}" the default
    value of the parameter is easy to write, and also not much more difficult to read.

    R=clemensb@chromium.org

    Bug: v8:10506
    Change-Id: I3debe5eb91578c82abdac81dc6c252435fdf30d6
    Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2202991
    Reviewed-by: Clemens Backes <clemensb@chromium.org>
    Commit-Queue: Andreas Haas <ahaas@chromium.org>
    Cr-Commit-Position: refs/heads/master@{#67822}

Refs: v8/v8@6771d3e

common.gypi

@@ -36,7 +36,7 @@
 
     # Reset this number to 0 on major V8 upgrades.
     # Increment by one for each non-official patch applied to deps/v8.
-    'v8_embedder_string': '-node.57',
+    'v8_embedder_string': '-node.58',
 
     ##### V8 defaults for Node.js #####
 

deps/v8/src/wasm/baseline/arm/liftoff-assembler-arm.h

@@ -437,7 +437,7 @@ void LiftoffAssembler::LoadConstant(LiftoffRegister reg, WasmValue value,
       vmov(liftoff::GetFloatRegister(reg.fp()), value.to_f32_boxed());
       break;
     case ValueType::kF64: {
-      Register extra_scratch = GetUnusedRegister(kGpReg).gp();
+      Register extra_scratch = GetUnusedRegister(kGpReg, {}).gp();
       vmov(reg.fp(), Double(value.to_f64_boxed().get_bits()), extra_scratch);
       break;
     }
@@ -1173,7 +1173,7 @@ void LiftoffAssembler::StoreCallerFrameSlot(LiftoffRegister src,
 void LiftoffAssembler::MoveStackValue(uint32_t dst_offset, uint32_t src_offset,
                                       ValueType type) {
   DCHECK_NE(dst_offset, src_offset);
-  LiftoffRegister reg = GetUnusedRegister(reg_class_for(type));
+  LiftoffRegister reg = GetUnusedRegister(reg_class_for(type), {});
   Fill(reg, src_offset, type);
   Spill(dst_offset, reg, type);
 }
@@ -1216,7 +1216,7 @@ void LiftoffAssembler::Spill(int offset, WasmValue value) {
   // The scratch register will be required by str if multiple instructions
   // are required to encode the offset, and so we cannot use it in that case.
   if (!ImmediateFitsAddrMode2Instruction(dst.offset())) {
-    src = GetUnusedRegister(kGpReg).gp();
+    src = GetUnusedRegister(kGpReg, {}).gp();
   } else {
     src = temps.Acquire();
   }
@@ -1758,7 +1758,7 @@ void LiftoffAssembler::emit_f32_copysign(DoubleRegister dst, DoubleRegister lhs,
                                          DoubleRegister rhs) {
   constexpr uint32_t kF32SignBit = uint32_t{1} << 31;
   UseScratchRegisterScope temps(this);
-  Register scratch = GetUnusedRegister(kGpReg).gp();
+  Register scratch = GetUnusedRegister(kGpReg, {}).gp();
   Register scratch2 = temps.Acquire();
   VmovLow(scratch, lhs);
   // Clear sign bit in {scratch}.
@@ -1777,7 +1777,7 @@ void LiftoffAssembler::emit_f64_copysign(DoubleRegister dst, DoubleRegister lhs,
   // On arm, we cannot hold the whole f64 value in a gp register, so we just
   // operate on the upper half (UH).
   UseScratchRegisterScope temps(this);
-  Register scratch = GetUnusedRegister(kGpReg).gp();
+  Register scratch = GetUnusedRegister(kGpReg, {}).gp();
   Register scratch2 = temps.Acquire();
   VmovHigh(scratch, lhs);
   // Clear sign bit in {scratch}.

deps/v8/src/wasm/baseline/ia32/liftoff-assembler-ia32.h

@@ -130,7 +130,7 @@ inline Register GetTmpByteRegister(LiftoffAssembler* assm, Register candidate) {
   if (candidate.is_byte_register()) return candidate;
   // {GetUnusedRegister()} may insert move instructions to spill registers to
   // the stack. This is OK because {mov} does not change the status flags.
-  return assm->GetUnusedRegister(liftoff::kByteRegs).gp();
+  return assm->GetUnusedRegister(liftoff::kByteRegs, {}).gp();
 }
 
 inline void MoveStackValue(LiftoffAssembler* assm, const Operand& src,
@@ -1462,7 +1462,7 @@ inline void EmitFloatMinOrMax(LiftoffAssembler* assm, DoubleRegister dst,
 
   // We need one tmp register to extract the sign bit. Get it right at the
   // beginning, such that the spilling code is not accidentially jumped over.
-  Register tmp = assm->GetUnusedRegister(kGpReg).gp();
+  Register tmp = assm->GetUnusedRegister(kGpReg, {}).gp();
 
 #define dop(name, ...)            \
   do {                            \
@@ -1525,9 +1525,9 @@ void LiftoffAssembler::emit_f32_max(DoubleRegister dst, DoubleRegister lhs,
 void LiftoffAssembler::emit_f32_copysign(DoubleRegister dst, DoubleRegister lhs,
                                          DoubleRegister rhs) {
   static constexpr int kF32SignBit = 1 << 31;
-  Register scratch = GetUnusedRegister(kGpReg).gp();
-  Register scratch2 =
-      GetUnusedRegister(kGpReg, LiftoffRegList::ForRegs(scratch)).gp();
+  LiftoffRegList pinned;
+  Register scratch = pinned.set(GetUnusedRegister(kGpReg, pinned)).gp();
+  Register scratch2 = GetUnusedRegister(kGpReg, pinned).gp();
   Movd(scratch, lhs);                      // move {lhs} into {scratch}.
   and_(scratch, Immediate(~kF32SignBit));  // clear sign bit in {scratch}.
   Movd(scratch2, rhs);                     // move {rhs} into {scratch2}.
@@ -1654,9 +1654,9 @@ void LiftoffAssembler::emit_f64_copysign(DoubleRegister dst, DoubleRegister lhs,
   static constexpr int kF32SignBit = 1 << 31;
   // On ia32, we cannot hold the whole f64 value in a gp register, so we just
   // operate on the upper half (UH).
-  Register scratch = GetUnusedRegister(kGpReg).gp();
-  Register scratch2 =
-      GetUnusedRegister(kGpReg, LiftoffRegList::ForRegs(scratch)).gp();
+  LiftoffRegList pinned;
+  Register scratch = pinned.set(GetUnusedRegister(kGpReg, pinned)).gp();
+  Register scratch2 = GetUnusedRegister(kGpReg, pinned).gp();
 
   Pextrd(scratch, lhs, 1);                 // move UH of {lhs} into {scratch}.
   and_(scratch, Immediate(~kF32SignBit));  // clear sign bit in {scratch}.
@@ -2494,7 +2494,7 @@ void LiftoffAssembler::emit_i8x16_shl(LiftoffRegister dst, LiftoffRegister lhs,
 void LiftoffAssembler::emit_i8x16_shli(LiftoffRegister dst, LiftoffRegister lhs,
                                        int32_t rhs) {
   static constexpr RegClass tmp_rc = reg_class_for(ValueType::kI32);
-  LiftoffRegister tmp = GetUnusedRegister(tmp_rc);
+  LiftoffRegister tmp = GetUnusedRegister(tmp_rc, {});
   byte shift = static_cast<byte>(rhs & 0x7);
   if (CpuFeatures::IsSupported(AVX)) {
     CpuFeatureScope scope(this, AVX);
@@ -3383,7 +3383,7 @@ void LiftoffAssembler::StackCheck(Label* ool_code, Register limit_address) {
 }
 
 void LiftoffAssembler::CallTrapCallbackForTesting() {
-  PrepareCallCFunction(0, GetUnusedRegister(kGpReg).gp());
+  PrepareCallCFunction(0, GetUnusedRegister(kGpReg, {}).gp());
   CallCFunction(ExternalReference::wasm_call_trap_callback_for_testing(), 0);
 }
 

deps/v8/src/wasm/baseline/liftoff-assembler.cc

@@ -542,7 +542,7 @@ LiftoffRegister LiftoffAssembler::LoadI64HalfIntoRegister(VarState slot,
   if (slot.is_reg()) {
     return half == kLowWord ? slot.reg().low() : slot.reg().high();
   }
-  LiftoffRegister dst = GetUnusedRegister(kGpReg);
+  LiftoffRegister dst = GetUnusedRegister(kGpReg, {});
   if (slot.is_stack()) {
     FillI64Half(dst.gp(), slot.offset(), half);
     return dst;
@@ -581,7 +581,7 @@ void LiftoffAssembler::PrepareLoopArgs(int num) {
     if (!slot.is_const()) continue;
     RegClass rc =
         kNeedI64RegPair && slot.type() == kWasmI64 ? kGpRegPair : kGpReg;
-    LiftoffRegister reg = GetUnusedRegister(rc);
+    LiftoffRegister reg = GetUnusedRegister(rc, {});
     LoadConstant(reg, slot.constant());
     slot.MakeRegister(reg);
     cache_state_.inc_used(reg);

deps/v8/src/wasm/baseline/liftoff-assembler.h

@@ -340,7 +340,7 @@ class LiftoffAssembler : public TurboAssembler {
   // possible.
   LiftoffRegister GetUnusedRegister(
       RegClass rc, std::initializer_list<LiftoffRegister> try_first,
-      LiftoffRegList pinned = {}) {
+      LiftoffRegList pinned) {
     for (LiftoffRegister reg : try_first) {
       DCHECK_EQ(reg.reg_class(), rc);
       if (cache_state_.is_free(reg)) return reg;
@@ -349,7 +349,7 @@ class LiftoffAssembler : public TurboAssembler {
   }
 
   // Get an unused register for class {rc}, potentially spilling to free one.
-  LiftoffRegister GetUnusedRegister(RegClass rc, LiftoffRegList pinned = {}) {
+  LiftoffRegister GetUnusedRegister(RegClass rc, LiftoffRegList pinned) {
     if (kNeedI64RegPair && rc == kGpRegPair) {
       LiftoffRegList candidates = kGpCacheRegList;
       Register low = pinned.set(GetUnusedRegister(candidates, pinned)).gp();

deps/v8/src/wasm/baseline/liftoff-compiler.cc

@@ -495,7 +495,7 @@ class LiftoffCompiler {
         position, __ cache_state()->used_registers,
         RegisterDebugSideTableEntry(DebugSideTableBuilder::kAssumeSpilling)));
     OutOfLineCode& ool = out_of_line_code_.back();
-    Register limit_address = __ GetUnusedRegister(kGpReg).gp();
+    Register limit_address = __ GetUnusedRegister(kGpReg, {}).gp();
     LOAD_INSTANCE_FIELD(limit_address, StackLimitAddress, kSystemPointerSize);
     __ StackCheck(ool.label.get(), limit_address);
     __ bind(ool.continuation.get());
@@ -604,7 +604,7 @@ class LiftoffCompiler {
                              *next_breakpoint_ptr_ == decoder->position());
       if (!has_breakpoint) {
         DEBUG_CODE_COMMENT("check hook on function call");
-        Register flag = __ GetUnusedRegister(kGpReg).gp();
+        Register flag = __ GetUnusedRegister(kGpReg, {}).gp();
         LOAD_INSTANCE_FIELD(flag, HookOnFunctionCallAddress,
                             kSystemPointerSize);
         Label no_break;
@@ -923,8 +923,8 @@ class LiftoffCompiler {
     constexpr RegClass result_rc = reg_class_for(result_type);
     LiftoffRegister src = __ PopToRegister();
     LiftoffRegister dst = src_rc == result_rc
-                              ? __ GetUnusedRegister(result_rc, {src})
-                              : __ GetUnusedRegister(result_rc);
+                              ? __ GetUnusedRegister(result_rc, {src}, {})
+                              : __ GetUnusedRegister(result_rc, {});
     CallEmitFn(fn, dst, src);
     __ PushRegister(ValueType(result_type), dst);
   }
@@ -951,8 +951,9 @@ class LiftoffCompiler {
     static constexpr RegClass src_rc = reg_class_for(src_type);
     static constexpr RegClass dst_rc = reg_class_for(dst_type);
     LiftoffRegister src = __ PopToRegister();
-    LiftoffRegister dst = src_rc == dst_rc ? __ GetUnusedRegister(dst_rc, {src})
-                                           : __ GetUnusedRegister(dst_rc);
+    LiftoffRegister dst = src_rc == dst_rc
+                              ? __ GetUnusedRegister(dst_rc, {src}, {})
+                              : __ GetUnusedRegister(dst_rc, {});
     DCHECK_EQ(!!can_trap, trap_position > 0);
     Label* trap = can_trap ? AddOutOfLineTrap(
                                  trap_position,
@@ -1122,8 +1123,8 @@ class LiftoffCompiler {
 
       LiftoffRegister lhs = __ PopToRegister();
       LiftoffRegister dst = src_rc == result_rc
-                                ? __ GetUnusedRegister(result_rc, {lhs})
-                                : __ GetUnusedRegister(result_rc);
+                                ? __ GetUnusedRegister(result_rc, {lhs}, {})
+                                : __ GetUnusedRegister(result_rc, {});
 
       CallEmitFn(fnImm, dst, lhs, imm);
       __ PushRegister(ValueType(result_type), dst);
@@ -1141,8 +1142,8 @@ class LiftoffCompiler {
     LiftoffRegister rhs = __ PopToRegister();
     LiftoffRegister lhs = __ PopToRegister(LiftoffRegList::ForRegs(rhs));
     LiftoffRegister dst = src_rc == result_rc
-                              ? __ GetUnusedRegister(result_rc, {lhs, rhs})
-                              : __ GetUnusedRegister(result_rc);
+                              ? __ GetUnusedRegister(result_rc, {lhs, rhs}, {})
+                              : __ GetUnusedRegister(result_rc, {});
 
     if (swap_lhs_rhs) std::swap(lhs, rhs);
 
@@ -1483,20 +1484,20 @@ class LiftoffCompiler {
     if (value_i32 == value) {
       __ PushConstant(kWasmI64, value_i32);
     } else {
-      LiftoffRegister reg = __ GetUnusedRegister(reg_class_for(kWasmI64));
+      LiftoffRegister reg = __ GetUnusedRegister(reg_class_for(kWasmI64), {});
       __ LoadConstant(reg, WasmValue(value));
       __ PushRegister(kWasmI64, reg);
     }
   }
 
   void F32Const(FullDecoder* decoder, Value* result, float value) {
-    LiftoffRegister reg = __ GetUnusedRegister(kFpReg);
+    LiftoffRegister reg = __ GetUnusedRegister(kFpReg, {});
     __ LoadConstant(reg, WasmValue(value));
     __ PushRegister(kWasmF32, reg);
   }
 
   void F64Const(FullDecoder* decoder, Value* result, double value) {
-    LiftoffRegister reg = __ GetUnusedRegister(kFpReg);
+    LiftoffRegister reg = __ GetUnusedRegister(kFpReg, {});
     __ LoadConstant(reg, WasmValue(value));
     __ PushRegister(kWasmF64, reg);
   }
@@ -1546,7 +1547,7 @@ class LiftoffCompiler {
         break;
       case kStack: {
         auto rc = reg_class_for(imm.type);
-        LiftoffRegister reg = __ GetUnusedRegister(rc);
+        LiftoffRegister reg = __ GetUnusedRegister(rc, {});
         __ Fill(reg, slot.offset(), imm.type);
         __ PushRegister(slot.type(), reg);
         break;
@@ -1570,7 +1571,7 @@ class LiftoffCompiler {
     }
     DCHECK_EQ(type, __ local_type(local_index));
     RegClass rc = reg_class_for(type);
-    LiftoffRegister dst_reg = __ GetUnusedRegister(rc);
+    LiftoffRegister dst_reg = __ GetUnusedRegister(rc, {});
     __ Fill(dst_reg, src_slot.offset(), type);
     *dst_slot = LiftoffAssembler::VarState(type, dst_reg, dst_slot->offset());
     __ cache_state()->inc_used(dst_reg);
@@ -1609,7 +1610,7 @@ class LiftoffCompiler {
 
   Register GetGlobalBaseAndOffset(const WasmGlobal* global,
                                   LiftoffRegList* pinned, uint32_t* offset) {
-    Register addr = pinned->set(__ GetUnusedRegister(kGpReg)).gp();
+    Register addr = pinned->set(__ GetUnusedRegister(kGpReg, {})).gp();
     if (global->mutability && global->imported) {
       LOAD_INSTANCE_FIELD(addr, ImportedMutableGlobals, kSystemPointerSize);
       __ Load(LiftoffRegister(addr), addr, no_reg,
@@ -1675,8 +1676,8 @@ class LiftoffCompiler {
     DCHECK_EQ(type, __ cache_state()->stack_state.end()[-2].type());
     LiftoffRegister false_value = pinned.set(__ PopToRegister(pinned));
     LiftoffRegister true_value = __ PopToRegister(pinned);
-    LiftoffRegister dst =
-        __ GetUnusedRegister(true_value.reg_class(), {true_value, false_value});
+    LiftoffRegister dst = __ GetUnusedRegister(true_value.reg_class(),
+                                               {true_value, false_value}, {});
     __ PushRegister(type, dst);
 
     // Now emit the actual code to move either {true_value} or {false_value}
@@ -2075,7 +2076,7 @@ class LiftoffCompiler {
   }
 
   void CurrentMemoryPages(FullDecoder* decoder, Value* result) {
-    Register mem_size = __ GetUnusedRegister(kGpReg).gp();
+    Register mem_size = __ GetUnusedRegister(kGpReg, {}).gp();
     LOAD_INSTANCE_FIELD(mem_size, MemorySize, kSystemPointerSize);
     __ emit_ptrsize_shri(mem_size, mem_size, kWasmPageSizeLog2);
     __ PushRegister(kWasmI32, LiftoffRegister(mem_size));
@@ -2344,7 +2345,7 @@ class LiftoffCompiler {
         src_rc == result_rc
             ? __ GetUnusedRegister(result_rc, {src3},
                                    LiftoffRegList::ForRegs(src1, src2))
-            : __ GetUnusedRegister(result_rc);
+            : __ GetUnusedRegister(result_rc, {});
     CallEmitFn(fn, dst, src1, src2, src3);
     __ PushRegister(ValueType(result_type), dst);
   }
@@ -2360,14 +2361,14 @@ class LiftoffCompiler {
       int32_t imm = rhs_slot.i32_const();
 
       LiftoffRegister operand = __ PopToRegister();
-      LiftoffRegister dst = __ GetUnusedRegister(result_rc, {operand});
+      LiftoffRegister dst = __ GetUnusedRegister(result_rc, {operand}, {});
 
       CallEmitFn(fnImm, dst, operand, imm);
       __ PushRegister(kWasmS128, dst);
     } else {
       LiftoffRegister count = __ PopToRegister();
       LiftoffRegister operand = __ PopToRegister();
-      LiftoffRegister dst = __ GetUnusedRegister(result_rc, {operand});
+      LiftoffRegister dst = __ GetUnusedRegister(result_rc, {operand}, {});
 
       CallEmitFn(fn, dst, operand, count);
       __ PushRegister(kWasmS128, dst);
@@ -2689,8 +2690,8 @@ class LiftoffCompiler {
     static constexpr RegClass result_rc = reg_class_for(result_type);
     LiftoffRegister lhs = __ PopToRegister();
     LiftoffRegister dst = src_rc == result_rc
-                              ? __ GetUnusedRegister(result_rc, {lhs})
-                              : __ GetUnusedRegister(result_rc);
+                              ? __ GetUnusedRegister(result_rc, {lhs}, {})
+                              : __ GetUnusedRegister(result_rc, {});
     fn(dst, lhs, imm.lane);
     __ PushRegister(ValueType(result_type), dst);
   }
@@ -2716,7 +2717,7 @@ class LiftoffCompiler {
         (src2_rc == result_rc || pin_src2)
             ? __ GetUnusedRegister(result_rc, {src1},
                                    LiftoffRegList::ForRegs(src2))
-            : __ GetUnusedRegister(result_rc, {src1});
+            : __ GetUnusedRegister(result_rc, {src1}, {});
     fn(dst, src1, src2, imm.lane);
     __ PushRegister(kWasmS128, dst);
   }

deps/v8/src/wasm/baseline/mips/liftoff-assembler-mips.h

@@ -603,7 +603,7 @@ void LiftoffAssembler::StoreCallerFrameSlot(LiftoffRegister src,
 void LiftoffAssembler::MoveStackValue(uint32_t dst_offset, uint32_t src_offset,
                                       ValueType type) {
   DCHECK_NE(dst_offset, src_offset);
-  LiftoffRegister reg = GetUnusedRegister(reg_class_for(type));
+  LiftoffRegister reg = GetUnusedRegister(reg_class_for(type), {});
   Fill(reg, src_offset, type);
   Spill(dst_offset, reg, type);
 }
@@ -646,13 +646,13 @@ void LiftoffAssembler::Spill(int offset, WasmValue value) {
   MemOperand dst = liftoff::GetStackSlot(offset);
   switch (value.type().kind()) {
     case ValueType::kI32: {
-      LiftoffRegister tmp = GetUnusedRegister(kGpReg);
+      LiftoffRegister tmp = GetUnusedRegister(kGpReg, {});
       TurboAssembler::li(tmp.gp(), Operand(value.to_i32()));
       sw(tmp.gp(), dst);
       break;
     }
     case ValueType::kI64: {
-      LiftoffRegister tmp = GetUnusedRegister(kGpRegPair);
+      LiftoffRegister tmp = GetUnusedRegister(kGpRegPair, {});
 
       int32_t low_word = value.to_i64();
       int32_t high_word = value.to_i64() >> 32;
@@ -2251,7 +2251,7 @@ void LiftoffAssembler::StackCheck(Label* ool_code, Register limit_address) {
 }
 
 void LiftoffAssembler::CallTrapCallbackForTesting() {
-  PrepareCallCFunction(0, GetUnusedRegister(kGpReg).gp());
+  PrepareCallCFunction(0, GetUnusedRegister(kGpReg, {}).gp());
   CallCFunction(ExternalReference::wasm_call_trap_callback_for_testing(), 0);
 }
 

deps/v8/src/wasm/baseline/mips64/liftoff-assembler-mips64.h

@@ -532,7 +532,7 @@ void LiftoffAssembler::StoreCallerFrameSlot(LiftoffRegister src,
 void LiftoffAssembler::MoveStackValue(uint32_t dst_offset, uint32_t src_offset,
                                       ValueType type) {
   DCHECK_NE(dst_offset, src_offset);
-  LiftoffRegister reg = GetUnusedRegister(reg_class_for(type));
+  LiftoffRegister reg = GetUnusedRegister(reg_class_for(type), {});
   Fill(reg, src_offset, type);
   Spill(dst_offset, reg, type);
 }
@@ -582,13 +582,13 @@ void LiftoffAssembler::Spill(int offset, WasmValue value) {
   MemOperand dst = liftoff::GetStackSlot(offset);
   switch (value.type().kind()) {
     case ValueType::kI32: {
-      LiftoffRegister tmp = GetUnusedRegister(kGpReg);
+      LiftoffRegister tmp = GetUnusedRegister(kGpReg, {});
       TurboAssembler::li(tmp.gp(), Operand(value.to_i32()));
       sw(tmp.gp(), dst);
       break;
     }
     case ValueType::kI64: {
-      LiftoffRegister tmp = GetUnusedRegister(kGpReg);
+      LiftoffRegister tmp = GetUnusedRegister(kGpReg, {});
       TurboAssembler::li(tmp.gp(), value.to_i64());
       sd(tmp.gp(), dst);
       break;
@@ -2197,7 +2197,7 @@ void LiftoffAssembler::StackCheck(Label* ool_code, Register limit_address) {
 }
 
 void LiftoffAssembler::CallTrapCallbackForTesting() {
-  PrepareCallCFunction(0, GetUnusedRegister(kGpReg).gp());
+  PrepareCallCFunction(0, GetUnusedRegister(kGpReg, {}).gp());
   CallCFunction(ExternalReference::wasm_call_trap_callback_for_testing(), 0);
 }