// Copyright 2014 the V8 project authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
#if V8_TARGET_ARCH_PPC
#include "src/codegen.h"
#include "src/ic/ic.h"
#include "src/ic/ic-compiler.h"
#include "src/ic/stub-cache.h"
namespace v8 {
namespace internal {
// ----------------------------------------------------------------------------
// Static IC stub generators.
//
#define __ ACCESS_MASM(masm)
// Helper function used from LoadIC GenerateNormal.
//
// elements: Property dictionary. It is not clobbered if a jump to the miss
// label is done.
// name: Property name. It is not clobbered if a jump to the miss label is
// done
// result: Register for the result. It is only updated if a jump to the miss
// label is not done. Can be the same as elements or name clobbering
// one of these in the case of not jumping to the miss label.
// The two scratch registers need to be different from elements, name and
// result.
// The generated code assumes that the receiver has slow properties,
// is not a global object and does not have interceptors.
static void GenerateDictionaryLoad(MacroAssembler* masm, Label* miss,
Register elements, Register name,
Register result, Register scratch1,
Register scratch2) {
// Main use of the scratch registers.
// scratch1: Used as temporary and to hold the capacity of the property
// dictionary.
// scratch2: Used as temporary.
Label done;
// Probe the dictionary.
NameDictionaryLookupStub::GeneratePositiveLookup(masm, miss, &done, elements,
name, scratch1, scratch2);
// If probing finds an entry check that the value is a normal
// property.
__ bind(&done); // scratch2 == elements + 4 * index
const int kElementsStartOffset =
NameDictionary::kHeaderSize +
NameDictionary::kElementsStartIndex * kPointerSize;
const int kDetailsOffset = kElementsStartOffset + 2 * kPointerSize;
__ LoadP(scratch1, FieldMemOperand(scratch2, kDetailsOffset));
__ mr(r0, scratch2);
__ LoadSmiLiteral(scratch2, Smi::FromInt(PropertyDetails::TypeField::kMask));
__ and_(scratch2, scratch1, scratch2, SetRC);
__ bne(miss, cr0);
__ mr(scratch2, r0);
// Get the value at the masked, scaled index and return.
__ LoadP(result,
FieldMemOperand(scratch2, kElementsStartOffset + 1 * kPointerSize));
}
// Helper function used from StoreIC::GenerateNormal.
//
// elements: Property dictionary. It is not clobbered if a jump to the miss
// label is done.
// name: Property name. It is not clobbered if a jump to the miss label is
// done
// value: The value to store.
// The two scratch registers need to be different from elements, name and
// result.
// The generated code assumes that the receiver has slow properties,
// is not a global object and does not have interceptors.
static void GenerateDictionaryStore(MacroAssembler* masm, Label* miss,
Register elements, Register name,
Register value, Register scratch1,
Register scratch2) {
// Main use of the scratch registers.
// scratch1: Used as temporary and to hold the capacity of the property
// dictionary.
// scratch2: Used as temporary.
Label done;
// Probe the dictionary.
NameDictionaryLookupStub::GeneratePositiveLookup(masm, miss, &done, elements,
name, scratch1, scratch2);
// If probing finds an entry in the dictionary check that the value
// is a normal property that is not read only.
__ bind(&done); // scratch2 == elements + 4 * index
const int kElementsStartOffset =
NameDictionary::kHeaderSize +
NameDictionary::kElementsStartIndex * kPointerSize;
const int kDetailsOffset = kElementsStartOffset + 2 * kPointerSize;
int kTypeAndReadOnlyMask =
PropertyDetails::TypeField::kMask |
PropertyDetails::AttributesField::encode(READ_ONLY);
__ LoadP(scratch1, FieldMemOperand(scratch2, kDetailsOffset));
__ mr(r0, scratch2);
__ LoadSmiLiteral(scratch2, Smi::FromInt(kTypeAndReadOnlyMask));
__ and_(scratch2, scratch1, scratch2, SetRC);
__ bne(miss, cr0);
__ mr(scratch2, r0);
// Store the value at the masked, scaled index and return.
const int kValueOffset = kElementsStartOffset + kPointerSize;
__ addi(scratch2, scratch2, Operand(kValueOffset - kHeapObjectTag));
__ StoreP(value, MemOperand(scratch2));
// Update the write barrier. Make sure not to clobber the value.
__ mr(scratch1, value);
__ RecordWrite(elements, scratch2, scratch1, kLRHasNotBeenSaved,
kDontSaveFPRegs);
}
void LoadIC::GenerateNormal(MacroAssembler* masm) {
Register dictionary = r3;
DCHECK(!dictionary.is(LoadDescriptor::ReceiverRegister()));
DCHECK(!dictionary.is(LoadDescriptor::NameRegister()));
Label slow;
__ LoadP(dictionary, FieldMemOperand(LoadDescriptor::ReceiverRegister(),
JSObject::kPropertiesOffset));
GenerateDictionaryLoad(masm, &slow, dictionary,
LoadDescriptor::NameRegister(), r3, r6, r7);
__ Ret();
// Dictionary load failed, go slow (but don't miss).
__ bind(&slow);
GenerateRuntimeGetProperty(masm);
}
// A register that isn't one of the parameters to the load ic.
static const Register LoadIC_TempRegister() { return r6; }
static void LoadIC_PushArgs(MacroAssembler* masm) {
Register receiver = LoadDescriptor::ReceiverRegister();
Register name = LoadDescriptor::NameRegister();
Register slot = LoadDescriptor::SlotRegister();
Register vector = LoadWithVectorDescriptor::VectorRegister();
__ Push(receiver, name, slot, vector);
}
void LoadIC::GenerateMiss(MacroAssembler* masm) {
// The return address is in lr.
Isolate* isolate = masm->isolate();
DCHECK(!AreAliased(r7, r8, LoadWithVectorDescriptor::SlotRegister(),
LoadWithVectorDescriptor::VectorRegister()));
__ IncrementCounter(isolate->counters()->ic_load_miss(), 1, r7, r8);
LoadIC_PushArgs(masm);
// Perform tail call to the entry.
__ TailCallRuntime(Runtime::kLoadIC_Miss);
}
void LoadIC::GenerateRuntimeGetProperty(MacroAssembler* masm) {
// The return address is in lr.
__ mr(LoadIC_TempRegister(), LoadDescriptor::ReceiverRegister());
__ Push(LoadIC_TempRegister(), LoadDescriptor::NameRegister());
// Do tail-call to runtime routine.
__ TailCallRuntime(Runtime::kGetProperty);
}
void KeyedLoadIC::GenerateMiss(MacroAssembler* masm) {
// The return address is in lr.
Isolate* isolate = masm->isolate();
DCHECK(!AreAliased(r7, r8, LoadWithVectorDescriptor::SlotRegister(),
LoadWithVectorDescriptor::VectorRegister()));
__ IncrementCounter(isolate->counters()->ic_keyed_load_miss(), 1, r7, r8);
LoadIC_PushArgs(masm);
// Perform tail call to the entry.
__ TailCallRuntime(Runtime::kKeyedLoadIC_Miss);
}
void KeyedLoadIC::GenerateRuntimeGetProperty(MacroAssembler* masm) {
// The return address is in lr.
__ Push(LoadDescriptor::ReceiverRegister(), LoadDescriptor::NameRegister());
// Do tail-call to runtime routine.
__ TailCallRuntime(Runtime::kKeyedGetProperty);
}
static void StoreIC_PushArgs(MacroAssembler* masm) {
__ Push(StoreWithVectorDescriptor::ValueRegister(),
StoreWithVectorDescriptor::SlotRegister(),
StoreWithVectorDescriptor::VectorRegister(),
StoreWithVectorDescriptor::ReceiverRegister(),
StoreWithVectorDescriptor::NameRegister());
}
void KeyedStoreIC::GenerateMiss(MacroAssembler* masm) {
StoreIC_PushArgs(masm);
__ TailCallRuntime(Runtime::kKeyedStoreIC_Miss);
}
void KeyedStoreIC::GenerateSlow(MacroAssembler* masm) {
StoreIC_PushArgs(masm);
// The slow case calls into the runtime to complete the store without causing
// an IC miss that would otherwise cause a transition to the generic stub.
__ TailCallRuntime(Runtime::kKeyedStoreIC_Slow);
}
static void KeyedStoreGenerateMegamorphicHelper(
MacroAssembler* masm, Label* fast_object, Label* fast_double, Label* slow,
KeyedStoreCheckMap check_map, KeyedStoreIncrementLength increment_length,
Register value, Register key, Register receiver, Register receiver_map,
Register elements_map, Register elements) {
Label transition_smi_elements;
Label finish_object_store, non_double_value, transition_double_elements;
Label fast_double_without_map_check;
// Fast case: Do the store, could be either Object or double.
__ bind(fast_object);
Register scratch = r7;
Register address = r8;
DCHECK(!AreAliased(value, key, receiver, receiver_map, elements_map, elements,
scratch, address));
if (check_map == kCheckMap) {
__ LoadP(elements_map, FieldMemOperand(elements, HeapObject::kMapOffset));
__ mov(scratch, Operand(masm->isolate()->factory()->fixed_array_map()));
__ cmp(elements_map, scratch);
__ bne(fast_double);
}
// HOLECHECK: guards "A[i] = V"
// We have to go to the runtime if the current value is the hole because
// there may be a callback on the element
Label holecheck_passed1;
__ addi(address, elements, Operand(FixedArray::kHeaderSize - kHeapObjectTag));
__ SmiToPtrArrayOffset(scratch, key);
__ LoadPX(scratch, MemOperand(address, scratch));
__ Cmpi(scratch, Operand(masm->isolate()->factory()->the_hole_value()), r0);
__ bne(&holecheck_passed1);
__ JumpIfDictionaryInPrototypeChain(receiver, elements_map, scratch, slow);
__ bind(&holecheck_passed1);
// Smi stores don't require further checks.
Label non_smi_value;
__ JumpIfNotSmi(value, &non_smi_value);
if (increment_length == kIncrementLength) {
// Add 1 to receiver->length.
__ AddSmiLiteral(scratch, key, Smi::FromInt(1), r0);
__ StoreP(scratch, FieldMemOperand(receiver, JSArray::kLengthOffset), r0);
}
// It's irrelevant whether array is smi-only or not when writing a smi.
__ addi(address, elements, Operand(FixedArray::kHeaderSize - kHeapObjectTag));
__ SmiToPtrArrayOffset(scratch, key);
__ StorePX(value, MemOperand(address, scratch));
__ Ret();
__ bind(&non_smi_value);
// Escape to elements kind transition case.
__ CheckFastObjectElements(receiver_map, scratch, &transition_smi_elements);
// Fast elements array, store the value to the elements backing store.
__ bind(&finish_object_store);
if (increment_length == kIncrementLength) {
// Add 1 to receiver->length.
__ AddSmiLiteral(scratch, key, Smi::FromInt(1), r0);
__ StoreP(scratch, FieldMemOperand(receiver, JSArray::kLengthOffset), r0);
}
__ addi(address, elements, Operand(FixedArray::kHeaderSize - kHeapObjectTag));
__ SmiToPtrArrayOffset(scratch, key);
__ StorePUX(value, MemOperand(address, scratch));
// Update write barrier for the elements array address.
__ mr(scratch, value); // Preserve the value which is returned.
__ RecordWrite(elements, address, scratch, kLRHasNotBeenSaved,
kDontSaveFPRegs, EMIT_REMEMBERED_SET, OMIT_SMI_CHECK);
__ Ret();
__ bind(fast_double);
if (check_map == kCheckMap) {
// Check for fast double array case. If this fails, call through to the
// runtime.
__ CompareRoot(elements_map, Heap::kFixedDoubleArrayMapRootIndex);
__ bne(slow);
}
// HOLECHECK: guards "A[i] double hole?"
// We have to see if the double version of the hole is present. If so
// go to the runtime.
__ addi(address, elements,
Operand((FixedDoubleArray::kHeaderSize + Register::kExponentOffset -
kHeapObjectTag)));
__ SmiToDoubleArrayOffset(scratch, key);
__ lwzx(scratch, MemOperand(address, scratch));
__ Cmpi(scratch, Operand(kHoleNanUpper32), r0);
__ bne(&fast_double_without_map_check);
__ JumpIfDictionaryInPrototypeChain(receiver, elements_map, scratch, slow);
__ bind(&fast_double_without_map_check);
__ StoreNumberToDoubleElements(value, key, elements, scratch, d0,
&transition_double_elements);
if (increment_length == kIncrementLength) {
// Add 1 to receiver->length.
__ AddSmiLiteral(scratch, key, Smi::FromInt(1), r0);
__ StoreP(scratch, FieldMemOperand(receiver, JSArray::kLengthOffset), r0);
}
__ Ret();
__ bind(&transition_smi_elements);
// Transition the array appropriately depending on the value type.
__ LoadP(scratch, FieldMemOperand(value, HeapObject::kMapOffset));
__ CompareRoot(scratch, Heap::kHeapNumberMapRootIndex);
__ bne(&non_double_value);
// Value is a double. Transition FAST_SMI_ELEMENTS ->
// FAST_DOUBLE_ELEMENTS and complete the store.
__ LoadTransitionedArrayMapConditional(
FAST_SMI_ELEMENTS, FAST_DOUBLE_ELEMENTS, receiver_map, scratch, slow);
AllocationSiteMode mode =
AllocationSite::GetMode(FAST_SMI_ELEMENTS, FAST_DOUBLE_ELEMENTS);
ElementsTransitionGenerator::GenerateSmiToDouble(masm, receiver, key, value,
receiver_map, mode, slow);
__ LoadP(elements, FieldMemOperand(receiver, JSObject::kElementsOffset));
__ b(&fast_double_without_map_check);
__ bind(&non_double_value);
// Value is not a double, FAST_SMI_ELEMENTS -> FAST_ELEMENTS
__ LoadTransitionedArrayMapConditional(FAST_SMI_ELEMENTS, FAST_ELEMENTS,
receiver_map, scratch, slow);
mode = AllocationSite::GetMode(FAST_SMI_ELEMENTS, FAST_ELEMENTS);
ElementsTransitionGenerator::GenerateMapChangeElementsTransition(
masm, receiver, key, value, receiver_map, mode, slow);
__ LoadP(elements, FieldMemOperand(receiver, JSObject::kElementsOffset));
__ b(&finish_object_store);
__ bind(&transition_double_elements);
// Elements are FAST_DOUBLE_ELEMENTS, but value is an Object that's not a
// HeapNumber. Make sure that the receiver is a Array with FAST_ELEMENTS and
// transition array from FAST_DOUBLE_ELEMENTS to FAST_ELEMENTS
__ LoadTransitionedArrayMapConditional(FAST_DOUBLE_ELEMENTS, FAST_ELEMENTS,
receiver_map, scratch, slow);
mode = AllocationSite::GetMode(FAST_DOUBLE_ELEMENTS, FAST_ELEMENTS);
ElementsTransitionGenerator::GenerateDoubleToObject(
masm, receiver, key, value, receiver_map, mode, slow);
__ LoadP(elements, FieldMemOperand(receiver, JSObject::kElementsOffset));
__ b(&finish_object_store);
}
void KeyedStoreIC::GenerateMegamorphic(MacroAssembler* masm,
LanguageMode language_mode) {
// ---------- S t a t e --------------
// -- r3 : value
// -- r4 : key
// -- r5 : receiver
// -- lr : return address
// -----------------------------------
Label slow, fast_object, fast_object_grow;
Label fast_double, fast_double_grow;
Label array, extra, check_if_double_array, maybe_name_key, miss;
// Register usage.
Register value = StoreDescriptor::ValueRegister();
Register key = StoreDescriptor::NameRegister();
Register receiver = StoreDescriptor::ReceiverRegister();
DCHECK(receiver.is(r4));
DCHECK(key.is(r5));
DCHECK(value.is(r3));
Register receiver_map = r6;
Register elements_map = r9;
Register elements = r10; // Elements array of the receiver.
// r7 and r8 are used as general scratch registers.
// Check that the key is a smi.
__ JumpIfNotSmi(key, &maybe_name_key);
// Check that the object isn't a smi.
__ JumpIfSmi(receiver, &slow);
// Get the map of the object.
__ LoadP(receiver_map, FieldMemOperand(receiver, HeapObject::kMapOffset));
// Check that the receiver does not require access checks.
// The generic stub does not perform map checks.
__ lbz(ip, FieldMemOperand(receiver_map, Map::kBitFieldOffset));
__ andi(r0, ip, Operand(1 << Map::kIsAccessCheckNeeded));
__ bne(&slow, cr0);
// Check if the object is a JS array or not.
__ lbz(r7, FieldMemOperand(receiver_map, Map::kInstanceTypeOffset));
__ cmpi(r7, Operand(JS_ARRAY_TYPE));
__ beq(&array);
// Check that the object is some kind of JSObject.
__ cmpi(r7, Operand(FIRST_JS_OBJECT_TYPE));
__ blt(&slow);
// Object case: Check key against length in the elements array.
__ LoadP(elements, FieldMemOperand(receiver, JSObject::kElementsOffset));
// Check array bounds. Both the key and the length of FixedArray are smis.
__ LoadP(ip, FieldMemOperand(elements, FixedArray::kLengthOffset));
__ cmpl(key, ip);
__ blt(&fast_object);
// Slow case, handle jump to runtime.
__ bind(&slow);
// Entry registers are intact.
// r3: value.
// r4: key.
// r5: receiver.
PropertyICCompiler::GenerateRuntimeSetProperty(masm, language_mode);
// Never returns to here.
__ bind(&maybe_name_key);
__ LoadP(r7, FieldMemOperand(key, HeapObject::kMapOffset));
__ lbz(r7, FieldMemOperand(r7, Map::kInstanceTypeOffset));
__ JumpIfNotUniqueNameInstanceType(r7, &slow);
// The handlers in the stub cache expect a vector and slot. Since we won't
// change the IC from any downstream misses, a dummy vector can be used.
Register vector = StoreWithVectorDescriptor::VectorRegister();
Register slot = StoreWithVectorDescriptor::SlotRegister();
DCHECK(!AreAliased(vector, slot, r8, r9, r10, r11));
Handle<TypeFeedbackVector> dummy_vector =
TypeFeedbackVector::DummyVector(masm->isolate());
int slot_index = dummy_vector->GetIndex(
FeedbackVectorSlot(TypeFeedbackVector::kDummyKeyedStoreICSlot));
__ LoadRoot(vector, Heap::kDummyVectorRootIndex);
__ LoadSmiLiteral(slot, Smi::FromInt(slot_index));
masm->isolate()->store_stub_cache()->GenerateProbe(masm, receiver, key, r8,
r9, r10, r11);
// Cache miss.
__ b(&miss);
// Extra capacity case: Check if there is extra capacity to
// perform the store and update the length. Used for adding one
// element to the array by writing to array[array.length].
__ bind(&extra);
// Condition code from comparing key and array length is still available.
__ bne(&slow); // Only support writing to writing to array[array.length].
// Check for room in the elements backing store.
// Both the key and the length of FixedArray are smis.
__ LoadP(ip, FieldMemOperand(elements, FixedArray::kLengthOffset));
__ cmpl(key, ip);
__ bge(&slow);
__ LoadP(elements_map, FieldMemOperand(elements, HeapObject::kMapOffset));
__ mov(ip, Operand(masm->isolate()->factory()->fixed_array_map()));
__ cmp(elements_map, ip); // PPC - I think I can re-use ip here
__ bne(&check_if_double_array);
__ b(&fast_object_grow);
__ bind(&check_if_double_array);
__ mov(ip, Operand(masm->isolate()->factory()->fixed_double_array_map()));
__ cmp(elements_map, ip); // PPC - another ip re-use
__ bne(&slow);
__ b(&fast_double_grow);
// Array case: Get the length and the elements array from the JS
// array. Check that the array is in fast mode (and writable); if it
// is the length is always a smi.
__ bind(&array);
__ LoadP(elements, FieldMemOperand(receiver, JSObject::kElementsOffset));
// Check the key against the length in the array.
__ LoadP(ip, FieldMemOperand(receiver, JSArray::kLengthOffset));
__ cmpl(key, ip);
__ bge(&extra);
KeyedStoreGenerateMegamorphicHelper(
masm, &fast_object, &fast_double, &slow, kCheckMap, kDontIncrementLength,
value, key, receiver, receiver_map, elements_map, elements);
KeyedStoreGenerateMegamorphicHelper(masm, &fast_object_grow,
&fast_double_grow, &slow, kDontCheckMap,
kIncrementLength, value, key, receiver,
receiver_map, elements_map, elements);
__ bind(&miss);
GenerateMiss(masm);
}
void StoreIC::GenerateMiss(MacroAssembler* masm) {
StoreIC_PushArgs(masm);
// Perform tail call to the entry.
__ TailCallRuntime(Runtime::kStoreIC_Miss);
}
void StoreIC::GenerateNormal(MacroAssembler* masm) {
Label miss;
Register receiver = StoreDescriptor::ReceiverRegister();
Register name = StoreDescriptor::NameRegister();
Register value = StoreDescriptor::ValueRegister();
Register dictionary = r8;
DCHECK(receiver.is(r4));
DCHECK(name.is(r5));
DCHECK(value.is(r3));
DCHECK(StoreWithVectorDescriptor::VectorRegister().is(r6));
DCHECK(StoreWithVectorDescriptor::SlotRegister().is(r7));
__ LoadP(dictionary, FieldMemOperand(receiver, JSObject::kPropertiesOffset));
GenerateDictionaryStore(masm, &miss, dictionary, name, value, r9, r10);
Counters* counters = masm->isolate()->counters();
__ IncrementCounter(counters->ic_store_normal_hit(), 1, r9, r10);
__ Ret();
__ bind(&miss);
__ IncrementCounter(counters->ic_store_normal_miss(), 1, r9, r10);
GenerateMiss(masm);
}
#undef __
Condition CompareIC::ComputeCondition(Token::Value op) {
switch (op) {
case Token::EQ_STRICT:
case Token::EQ:
return eq;
case Token::LT:
return lt;
case Token::GT:
return gt;
case Token::LTE:
return le;
case Token::GTE:
return ge;
default:
UNREACHABLE();
return kNoCondition;
}
}
bool CompareIC::HasInlinedSmiCode(Address address) {
// The address of the instruction following the call.
Address cmp_instruction_address =
Assembler::return_address_from_call_start(address);
// If the instruction following the call is not a cmp rx, #yyy, nothing
// was inlined.
Instr instr = Assembler::instr_at(cmp_instruction_address);
return Assembler::IsCmpImmediate(instr);
}
//
// This code is paired with the JumpPatchSite class in full-codegen-ppc.cc
//
void PatchInlinedSmiCode(Isolate* isolate, Address address,
InlinedSmiCheck check) {
Address cmp_instruction_address =
Assembler::return_address_from_call_start(address);
// If the instruction following the call is not a cmp rx, #yyy, nothing
// was inlined.
Instr instr = Assembler::instr_at(cmp_instruction_address);
if (!Assembler::IsCmpImmediate(instr)) {
return;
}
// The delta to the start of the map check instruction and the
// condition code uses at the patched jump.
int delta = Assembler::GetCmpImmediateRawImmediate(instr);
delta += Assembler::GetCmpImmediateRegister(instr).code() * kOff16Mask;
// If the delta is 0 the instruction is cmp r0, #0 which also signals that
// nothing was inlined.
if (delta == 0) {
return;
}
if (FLAG_trace_ic) {
PrintF("[ patching ic at %p, cmp=%p, delta=%d\n",
static_cast<void*>(address),
static_cast<void*>(cmp_instruction_address), delta);
}
Address patch_address =
cmp_instruction_address - delta * Instruction::kInstrSize;
Instr instr_at_patch = Assembler::instr_at(patch_address);
Instr branch_instr =
Assembler::instr_at(patch_address + Instruction::kInstrSize);
// This is patching a conditional "jump if not smi/jump if smi" site.
// Enabling by changing from
// cmp cr0, rx, rx
// to
// rlwinm(r0, value, 0, 31, 31, SetRC);
// bc(label, BT/BF, 2)
// and vice-versa to be disabled again.
CodePatcher patcher(isolate, patch_address, 2);
Register reg = Assembler::GetRA(instr_at_patch);
if (check == ENABLE_INLINED_SMI_CHECK) {
DCHECK(Assembler::IsCmpRegister(instr_at_patch));
DCHECK_EQ(Assembler::GetRA(instr_at_patch).code(),
Assembler::GetRB(instr_at_patch).code());
patcher.masm()->TestIfSmi(reg, r0);
} else {
DCHECK(check == DISABLE_INLINED_SMI_CHECK);
DCHECK(Assembler::IsAndi(instr_at_patch));
patcher.masm()->cmp(reg, reg, cr0);
}
DCHECK(Assembler::IsBranch(branch_instr));
// Invert the logic of the branch
if (Assembler::GetCondition(branch_instr) == eq) {
patcher.EmitCondition(ne);
} else {
DCHECK(Assembler::GetCondition(branch_instr) == ne);
patcher.EmitCondition(eq);
}
}
} // namespace internal
} // namespace v8
#endif // V8_TARGET_ARCH_PPC