Files
sdk/runtime/vm/compiler/assembler/assembler_base.cc
T
Tess Strickland 7c55965b76 [vm/compiler] Add initial partial TTSes for implemented types.
Previously, no optimized TTSes were generated for implemented types, and
so they always fell back to the default TTS, which mostly depends on
calling the runtime and cached checks in SubtypeTestCaches. Now,
optimized TTSes are generated that check for certain compatible
implementing classes before falling back on the runtime/STC.

More specifically, the optimized TTSes for implemented types checks for
the following cases:
1) The implemented type is instantiated and the checked class implements
   an instantiated subtype of the implemented type. The only check
   required is a class id match.
2) The instance type arguments of the checked class are compatible with
   the type arguments of the checked type.  That is, given the following
   declarations, where Base, Impl1, and Impl2 have the same number of
   parent type arguments:

```
case Impl1<K, V> implements Base<K, V>
case Impl2<V> implements Base<String, V>
```

   then the generated optimized TTS for Base<S, T>, where S and T are
   either type parameters or instantiated types, checks for instances of
   Base and Impl1, comparing the type arguments of the instance to S and
   T. The generated TTS does not currently check for Impl2, and thus
   when given an instance of Impl2, it falls back to the old runtime
   checking/SubtypeTestCache behavior.

This compatibility restriction allows us to perform the same checks on
the loaded instance type arguments as is done for non-implemented types,
where the checked classes are subclasses and so naturally compatible in
this manner.

Note that two implementing classes whose instance type arguments are
compatible may store their instance type arguments at different field
offsets. Thus, we also split the classes being checked into groups that
share the same instance type arguments field offset, and load the
instance type arguments differently for each checked group.

This CL also removes now-unused code in the HierarchyInfo class.

TEST=vm/cc/TTS_{Generic,}SubtypeRangeCheck

Cq-Include-Trybots: luci.dart.try:vm-kernel-linux-release-x64-try,vm-kernel-precomp-linux-release-x64-try,vm-kernel-precomp-nnbd-linux-release-x64-try,vm-kernel-nnbd-linux-release-x64-try,vm-kernel-linux-product-x64-try,vm-kernel-precomp-linux-product-x64-try,vm-kernel-linux-release-simarm-try,vm-kernel-linux-release-simarm64-try,vm-kernel-precomp-linux-release-simarm64-try,vm-kernel-precomp-linux-release-simarm-try,vm-kernel-precomp-linux-release-simarm_x64-try
Change-Id: I4c3aa23db2e75adbad9c15727b491669b2f3a189
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/209540
Reviewed-by: Martin Kustermann <kustermann@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
2021-08-16 13:02:34 +00:00

440 lines
14 KiB
C++

// Copyright (c) 2012, the Dart project authors. Please see the AUTHORS file
// for details. All rights reserved. Use of this source code is governed by a
// BSD-style license that can be found in the LICENSE file.
#include "vm/compiler/assembler/assembler_base.h"
#include "platform/utils.h"
#include "vm/compiler/backend/slot.h"
#include "vm/cpu.h"
#include "vm/heap/heap.h"
#include "vm/memory_region.h"
#include "vm/os.h"
#include "vm/zone.h"
namespace dart {
DEFINE_FLAG(bool,
check_code_pointer,
false,
"Verify instructions offset in code object."
"NOTE: This breaks the profiler.");
#if defined(TARGET_ARCH_ARM)
DEFINE_FLAG(bool, use_far_branches, false, "Enable far branches for ARM.");
#endif
namespace compiler {
AssemblerBase::~AssemblerBase() {}
void AssemblerBase::LoadFromSlot(Register dst,
Register base,
const Slot& slot) {
auto const rep = slot.representation();
const FieldAddress address(base, slot.offset_in_bytes());
if (rep != kTagged) {
auto const sz = RepresentationUtils::OperandSize(rep);
return LoadFromOffset(dst, address, sz);
}
if (slot.is_compressed()) {
return LoadCompressedField(dst, address);
}
return LoadField(dst, address);
}
void AssemblerBase::StoreToSlot(Register src, Register base, const Slot& slot) {
auto const rep = slot.representation();
const FieldAddress address(base, slot.offset_in_bytes());
if (rep != kTagged) {
auto const sz = RepresentationUtils::OperandSize(rep);
return StoreToOffset(src, address, sz);
}
if (slot.is_compressed()) {
return StoreCompressedIntoObject(
base, address, src,
slot.ComputeCompileType().CanBeSmi() ? kValueCanBeSmi : kValueIsNotSmi);
}
return StoreIntoObject(
base, address, src,
slot.ComputeCompileType().CanBeSmi() ? kValueCanBeSmi : kValueIsNotSmi);
}
void AssemblerBase::StoreToSlotNoBarrier(Register src,
Register base,
const Slot& slot) {
auto const rep = slot.representation();
const FieldAddress address(base, slot.offset_in_bytes());
if (rep != kTagged) {
auto const sz = RepresentationUtils::OperandSize(rep);
return StoreToOffset(src, address, sz);
}
if (slot.is_compressed()) {
return StoreCompressedIntoObjectNoBarrier(base, address, src);
}
return StoreIntoObjectNoBarrier(base, address, src);
}
intptr_t AssemblerBase::InsertAlignedRelocation(BSS::Relocation reloc) {
// We cannot put a relocation at the very start (it's not a valid
// instruction)!
ASSERT(CodeSize() != 0);
// Align to a target word boundary.
const intptr_t offset =
Utils::RoundUp(CodeSize(), compiler::target::kWordSize);
while (CodeSize() < offset) {
Breakpoint();
}
ASSERT(CodeSize() == offset);
AssemblerBuffer::EnsureCapacity ensured(&buffer_);
buffer_.Emit<compiler::target::word>(BSS::RelocationIndex(reloc) *
compiler::target::kWordSize);
ASSERT(CodeSize() == (offset + compiler::target::kWordSize));
return offset;
}
#if defined(DEBUG)
static void InitializeMemoryWithBreakpoints(uword data, intptr_t length) {
#if defined(TARGET_ARCH_ARM) || defined(TARGET_ARCH_ARM64)
ASSERT(Utils::IsAligned(data, 4));
ASSERT(Utils::IsAligned(length, 4));
const uword end = data + length;
while (data < end) {
*reinterpret_cast<int32_t*>(data) = Instr::kBreakPointInstruction;
data += 4;
}
#else
memset(reinterpret_cast<void*>(data), Instr::kBreakPointInstruction, length);
#endif
}
#endif
static uword NewContents(intptr_t capacity) {
Zone* zone = Thread::Current()->zone();
uword result = zone->AllocUnsafe(capacity);
#if defined(DEBUG)
// Initialize the buffer with kBreakPointInstruction to force a break
// point if we ever execute an uninitialized part of the code buffer.
InitializeMemoryWithBreakpoints(result, capacity);
#endif
return result;
}
#if defined(DEBUG)
AssemblerBuffer::EnsureCapacity::EnsureCapacity(AssemblerBuffer* buffer) {
if (buffer->cursor() >= buffer->limit()) buffer->ExtendCapacity();
// In debug mode, we save the assembler buffer along with the gap
// size before we start emitting to the buffer. This allows us to
// check that any single generated instruction doesn't overflow the
// limit implied by the minimum gap size.
buffer_ = buffer;
gap_ = ComputeGap();
// Make sure that extending the capacity leaves a big enough gap
// for any kind of instruction.
ASSERT(gap_ >= kMinimumGap);
// Mark the buffer as having ensured the capacity.
ASSERT(!buffer->HasEnsuredCapacity()); // Cannot nest.
buffer->has_ensured_capacity_ = true;
}
AssemblerBuffer::EnsureCapacity::~EnsureCapacity() {
// Unmark the buffer, so we cannot emit after this.
buffer_->has_ensured_capacity_ = false;
// Make sure the generated instruction doesn't take up more
// space than the minimum gap.
intptr_t delta = gap_ - ComputeGap();
ASSERT(delta <= kMinimumGap);
}
#endif
AssemblerBuffer::AssemblerBuffer()
: pointer_offsets_(new ZoneGrowableArray<intptr_t>(16)) {
static const intptr_t kInitialBufferCapacity = 4 * KB;
contents_ = NewContents(kInitialBufferCapacity);
cursor_ = contents_;
limit_ = ComputeLimit(contents_, kInitialBufferCapacity);
fixup_ = NULL;
#if defined(DEBUG)
has_ensured_capacity_ = false;
fixups_processed_ = false;
#endif
// Verify internal state.
ASSERT(Capacity() == kInitialBufferCapacity);
ASSERT(Size() == 0);
}
AssemblerBuffer::~AssemblerBuffer() {}
void AssemblerBuffer::ProcessFixups(const MemoryRegion& region) {
AssemblerFixup* fixup = fixup_;
while (fixup != NULL) {
fixup->Process(region, fixup->position());
fixup = fixup->previous();
}
}
void AssemblerBuffer::FinalizeInstructions(const MemoryRegion& instructions) {
// Copy the instructions from the buffer.
MemoryRegion from(reinterpret_cast<void*>(contents()), Size());
instructions.CopyFrom(0, from);
// Process fixups in the instructions.
ProcessFixups(instructions);
#if defined(DEBUG)
fixups_processed_ = true;
#endif
}
void AssemblerBuffer::ExtendCapacity() {
intptr_t old_size = Size();
intptr_t old_capacity = Capacity();
intptr_t new_capacity =
Utils::Minimum(old_capacity * 2, old_capacity + 1 * MB);
if (new_capacity < old_capacity) {
FATAL("Unexpected overflow in AssemblerBuffer::ExtendCapacity");
}
// Allocate the new data area and copy contents of the old one to it.
uword new_contents = NewContents(new_capacity);
memmove(reinterpret_cast<void*>(new_contents),
reinterpret_cast<void*>(contents_), old_size);
// Compute the relocation delta and switch to the new contents area.
intptr_t delta = new_contents - contents_;
contents_ = new_contents;
// Update the cursor and recompute the limit.
cursor_ += delta;
limit_ = ComputeLimit(new_contents, new_capacity);
// Verify internal state.
ASSERT(Capacity() == new_capacity);
ASSERT(Size() == old_size);
}
class PatchCodeWithHandle : public AssemblerFixup {
public:
PatchCodeWithHandle(ZoneGrowableArray<intptr_t>* pointer_offsets,
const Object& object)
: pointer_offsets_(pointer_offsets), object_(object) {}
void Process(const MemoryRegion& region, intptr_t position) {
// Patch the handle into the code. Once the instructions are installed into
// a raw code object and the pointer offsets are setup, the handle is
// resolved.
region.StoreUnaligned<const Object*>(position, &object_);
pointer_offsets_->Add(position);
}
virtual bool IsPointerOffset() const { return true; }
private:
ZoneGrowableArray<intptr_t>* pointer_offsets_;
const Object& object_;
};
intptr_t AssemblerBuffer::CountPointerOffsets() const {
intptr_t count = 0;
AssemblerFixup* current = fixup_;
while (current != NULL) {
if (current->IsPointerOffset()) ++count;
current = current->previous_;
}
return count;
}
#if defined(TARGET_ARCH_IA32)
void AssemblerBuffer::EmitObject(const Object& object) {
// Since we are going to store the handle as part of the fixup information
// the handle needs to be a zone handle.
ASSERT(IsNotTemporaryScopedHandle(object));
ASSERT(IsInOldSpace(object));
EmitFixup(new PatchCodeWithHandle(pointer_offsets_, object));
cursor_ += target::kWordSize; // Reserve space for pointer.
}
#endif
// Shared macros are implemented here.
void AssemblerBase::Unimplemented(const char* message) {
const char* format = "Unimplemented: %s";
const intptr_t len = Utils::SNPrint(NULL, 0, format, message);
char* buffer = reinterpret_cast<char*>(malloc(len + 1));
Utils::SNPrint(buffer, len + 1, format, message);
Stop(buffer);
}
void AssemblerBase::Untested(const char* message) {
const char* format = "Untested: %s";
const intptr_t len = Utils::SNPrint(NULL, 0, format, message);
char* buffer = reinterpret_cast<char*>(malloc(len + 1));
Utils::SNPrint(buffer, len + 1, format, message);
Stop(buffer);
}
void AssemblerBase::Unreachable(const char* message) {
const char* format = "Unreachable: %s";
const intptr_t len = Utils::SNPrint(NULL, 0, format, message);
char* buffer = reinterpret_cast<char*>(malloc(len + 1));
Utils::SNPrint(buffer, len + 1, format, message);
Stop(buffer);
}
void AssemblerBase::Comment(const char* format, ...) {
if (EmittingComments()) {
char buffer[1024];
va_list args;
va_start(args, format);
Utils::VSNPrint(buffer, sizeof(buffer), format, args);
va_end(args);
comments_.Add(
new CodeComment(buffer_.GetPosition(), AllocateString(buffer)));
}
}
bool AssemblerBase::EmittingComments() {
return FLAG_code_comments || FLAG_disassemble || FLAG_disassemble_optimized ||
FLAG_disassemble_stubs;
}
void AssemblerBase::Stop(const char* message) {
Comment("Stop: %s", message);
Breakpoint();
}
uword ObjIndexPair::Hash(Key key) {
if (key.type() != ObjectPoolBuilderEntry::kTaggedObject) {
return key.raw_value_;
}
return ObjectHash(*key.obj_);
}
void ObjectPoolBuilder::Reset() {
// Null out the handles we've accumulated.
for (intptr_t i = 0; i < object_pool_.length(); ++i) {
if (object_pool_[i].type() == ObjectPoolBuilderEntry::kTaggedObject) {
SetToNull(const_cast<Object*>(object_pool_[i].obj_));
SetToNull(const_cast<Object*>(object_pool_[i].equivalence_));
}
}
object_pool_.Clear();
object_pool_index_table_.Clear();
}
intptr_t ObjectPoolBuilder::AddObject(
const Object& obj,
ObjectPoolBuilderEntry::Patchability patchable) {
ASSERT(IsNotTemporaryScopedHandle(obj));
return AddObject(ObjectPoolBuilderEntry(&obj, patchable));
}
intptr_t ObjectPoolBuilder::AddImmediate(uword imm) {
return AddObject(
ObjectPoolBuilderEntry(imm, ObjectPoolBuilderEntry::kImmediate,
ObjectPoolBuilderEntry::kNotPatchable));
}
intptr_t ObjectPoolBuilder::AddObject(ObjectPoolBuilderEntry entry) {
ASSERT((entry.type() != ObjectPoolBuilderEntry::kTaggedObject) ||
(IsNotTemporaryScopedHandle(*entry.obj_) &&
(entry.equivalence_ == NULL ||
IsNotTemporaryScopedHandle(*entry.equivalence_))));
if (entry.type() == ObjectPoolBuilderEntry::kTaggedObject) {
// If the owner of the object pool wrapper specified a specific zone we
// should use we'll do so.
if (zone_ != NULL) {
entry.obj_ = &NewZoneHandle(zone_, *entry.obj_);
if (entry.equivalence_ != NULL) {
entry.equivalence_ = &NewZoneHandle(zone_, *entry.equivalence_);
}
}
}
const intptr_t idx = base_index_ + object_pool_.length();
object_pool_.Add(entry);
if (entry.patchable() == ObjectPoolBuilderEntry::kNotPatchable) {
// The object isn't patchable. Record the index for fast lookup.
object_pool_index_table_.Insert(ObjIndexPair(entry, idx));
}
return idx;
}
intptr_t ObjectPoolBuilder::FindObject(ObjectPoolBuilderEntry entry) {
// If the object is not patchable, check if we've already got it in the
// object pool.
if (entry.patchable() == ObjectPoolBuilderEntry::kNotPatchable) {
// First check in the parent pool if we have one.
if (parent_ != nullptr) {
const intptr_t idx = parent_->object_pool_index_table_.LookupValue(entry);
if (idx != ObjIndexPair::kNoIndex) {
used_from_parent_.Add(idx);
return idx;
}
}
const intptr_t idx = object_pool_index_table_.LookupValue(entry);
if (idx != ObjIndexPair::kNoIndex) {
return idx;
}
}
return AddObject(entry);
}
intptr_t ObjectPoolBuilder::FindObject(
const Object& obj,
ObjectPoolBuilderEntry::Patchability patchable) {
return FindObject(ObjectPoolBuilderEntry(&obj, patchable));
}
intptr_t ObjectPoolBuilder::FindObject(const Object& obj,
const Object& equivalence) {
return FindObject(ObjectPoolBuilderEntry(
&obj, &equivalence, ObjectPoolBuilderEntry::kNotPatchable));
}
intptr_t ObjectPoolBuilder::FindImmediate(uword imm) {
return FindObject(
ObjectPoolBuilderEntry(imm, ObjectPoolBuilderEntry::kImmediate,
ObjectPoolBuilderEntry::kNotPatchable));
}
intptr_t ObjectPoolBuilder::FindNativeFunction(
const ExternalLabel* label,
ObjectPoolBuilderEntry::Patchability patchable) {
return FindObject(ObjectPoolBuilderEntry(
label->address(), ObjectPoolBuilderEntry::kNativeFunction, patchable));
}
intptr_t ObjectPoolBuilder::FindNativeFunctionWrapper(
const ExternalLabel* label,
ObjectPoolBuilderEntry::Patchability patchable) {
return FindObject(ObjectPoolBuilderEntry(
label->address(), ObjectPoolBuilderEntry::kNativeFunctionWrapper,
patchable));
}
bool ObjectPoolBuilder::TryCommitToParent() {
ASSERT(parent_ != nullptr);
if (parent_->CurrentLength() != base_index_) {
return false;
}
for (intptr_t i = 0; i < object_pool_.length(); i++) {
intptr_t idx = parent_->AddObject(object_pool_[i]);
ASSERT(idx == (base_index_ + i));
}
return true;
}
} // namespace compiler
} // namespace dart