libs/ui/BufferHubBuffer.cpp - platform/frameworks/native - Git at Google

 /*
  * Copyright (C) 2018 The Android Open Source Project
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
  * You may obtain a copy of the License at
  *
  *      http://www.apache.org/licenses/LICENSE-2.0
  *
  * Unless required by applicable law or agreed to in writing, software
  * distributed under the License is distributed on an "AS IS" BASIS,
  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  * See the License for the specific language governing permissions and
  * limitations under the License.
  */

 #include <poll.h>

 #include <android-base/unique_fd.h>
 #include <android/frameworks/bufferhub/1.0/IBufferHub.h>
 #include <log/log.h>
 #include <ui/BufferHubBuffer.h>
 #include <ui/BufferHubDefs.h>
 #include <utils/Trace.h>

 using ::android::base::unique_fd;
 using ::android::BufferHubDefs::AnyClientAcquired;
 using ::android::BufferHubDefs::AnyClientGained;
 using ::android::BufferHubDefs::IsClientAcquired;
 using ::android::BufferHubDefs::IsClientGained;
 using ::android::BufferHubDefs::IsClientPosted;
 using ::android::BufferHubDefs::IsClientReleased;
 using ::android::frameworks::bufferhub::V1_0::BufferHubStatus;
 using ::android::frameworks::bufferhub::V1_0::BufferTraits;
 using ::android::frameworks::bufferhub::V1_0::IBufferClient;
 using ::android::frameworks::bufferhub::V1_0::IBufferHub;
 using ::android::hardware::hidl_handle;
 using ::android::hardware::graphics::common::V1_2::HardwareBufferDescription;

 namespace android {

 std::unique_ptr<BufferHubBuffer> BufferHubBuffer::Create(uint32_t width, uint32_t height,
                                                          uint32_t layerCount, uint32_t format,
                                                          uint64_t usage, size_t userMetadataSize) {
     auto buffer = std::unique_ptr<BufferHubBuffer>(
             new BufferHubBuffer(width, height, layerCount, format, usage, userMetadataSize));
     return buffer->IsValid() ? std::move(buffer) : nullptr;
 }

 std::unique_ptr<BufferHubBuffer> BufferHubBuffer::Import(const native_handle_t* token) {
     if (token == nullptr) {
         ALOGE("%s: token cannot be nullptr!", __FUNCTION__);
         return nullptr;
     }

     auto buffer = std::unique_ptr<BufferHubBuffer>(new BufferHubBuffer(token));
     return buffer->IsValid() ? std::move(buffer) : nullptr;
 }

 BufferHubBuffer::BufferHubBuffer(uint32_t width, uint32_t height, uint32_t layerCount,
                                  uint32_t format, uint64_t usage, size_t userMetadataSize) {
     ATRACE_CALL();
     ALOGD("%s: width=%u height=%u layerCount=%u, format=%u "
           "usage=%" PRIx64 " mUserMetadataSize=%zu",
           __FUNCTION__, width, height, layerCount, format, usage, userMetadataSize);

     sp<IBufferHub> bufferhub = IBufferHub::getService();
     if (bufferhub.get() == nullptr) {
         ALOGE("%s: BufferHub service not found!", __FUNCTION__);
         return;
     }

     AHardwareBuffer_Desc aDesc = {width, height,         layerCount,   format,
                                   usage, /*stride=*/0UL, /*rfu0=*/0UL, /*rfu1=*/0ULL};
     HardwareBufferDescription desc;
     memcpy(&desc, &aDesc, sizeof(HardwareBufferDescription));

     BufferHubStatus ret;
     sp<IBufferClient> client;
     BufferTraits bufferTraits;
     IBufferHub::allocateBuffer_cb alloc_cb = [&](const auto& status, const auto& outClient,
                                                  const auto& traits) {
         ret = status;
         client = std::move(outClient);
         bufferTraits = std::move(traits);
     };

     if (!bufferhub->allocateBuffer(desc, static_cast<uint32_t>(userMetadataSize), alloc_cb)
                  .isOk()) {
         ALOGE("%s: allocateBuffer transaction failed!", __FUNCTION__);
         return;
     } else if (ret != BufferHubStatus::NO_ERROR) {
         ALOGE("%s: allocateBuffer failed with error %u.", __FUNCTION__, ret);
         return;
     } else if (client == nullptr) {
         ALOGE("%s: allocateBuffer got null BufferClient.", __FUNCTION__);
         return;
     }

     const int importRet = initWithBufferTraits(bufferTraits);
     if (importRet < 0) {
         ALOGE("%s: Failed to import buffer: %s", __FUNCTION__, strerror(-importRet));
         client->close();
     }
     mBufferClient = std::move(client);
 }

 BufferHubBuffer::BufferHubBuffer(const native_handle_t* token) {
     sp<IBufferHub> bufferhub = IBufferHub::getService();
     if (bufferhub.get() == nullptr) {
         ALOGE("%s: BufferHub service not found!", __FUNCTION__);
         return;
     }

     BufferHubStatus ret;
     sp<IBufferClient> client;
     BufferTraits bufferTraits;
     IBufferHub::importBuffer_cb import_cb = [&](const auto& status, const auto& outClient,
                                                 const auto& traits) {
         ret = status;
         client = std::move(outClient);
         bufferTraits = std::move(traits);
     };

     // hidl_handle(native_handle_t*) simply creates a raw pointer reference withouth ownership
     // transfer.
     if (!bufferhub->importBuffer(hidl_handle(token), import_cb).isOk()) {
         ALOGE("%s: importBuffer transaction failed!", __FUNCTION__);
         return;
     } else if (ret != BufferHubStatus::NO_ERROR) {
         ALOGE("%s: importBuffer failed with error %u.", __FUNCTION__, ret);
         return;
     } else if (client == nullptr) {
         ALOGE("%s: importBuffer got null BufferClient.", __FUNCTION__);
         return;
     }

     const int importRet = initWithBufferTraits(bufferTraits);
     if (importRet < 0) {
         ALOGE("%s: Failed to import buffer: %s", __FUNCTION__, strerror(-importRet));
         client->close();
     }
     mBufferClient = std::move(client);
 }

 BufferHubBuffer::~BufferHubBuffer() {
     // Close buffer client to avoid possible race condition: user could first duplicate and hold
     // token with the original buffer gone, and then try to import the token. The close function
     // will explicitly invalidate the token to avoid this.
     if (mBufferClient != nullptr) {
         if (!mBufferClient->close().isOk()) {
             ALOGE("%s: close BufferClient transaction failed!", __FUNCTION__);
         }
     }
 }

 int BufferHubBuffer::initWithBufferTraits(const BufferTraits& bufferTraits) {
     ATRACE_CALL();

     if (bufferTraits.bufferInfo.getNativeHandle() == nullptr) {
         ALOGE("%s: missing buffer info handle.", __FUNCTION__);
         return -EINVAL;
     }

     if (bufferTraits.bufferHandle.getNativeHandle() == nullptr) {
         ALOGE("%s: missing gralloc handle.", __FUNCTION__);
         return -EINVAL;
     }

     // Import fds. Dup fds because hidl_handle owns the fds.
     unique_fd ashmemFd(fcntl(bufferTraits.bufferInfo->data[0], F_DUPFD_CLOEXEC, 0));
     mMetadata = BufferHubMetadata::Import(std::move(ashmemFd));
     if (!mMetadata.IsValid()) {
         ALOGE("%s: Received an invalid metadata.", __FUNCTION__);
         return -EINVAL;
     }

     mEventFd = BufferHubEventFd(fcntl(bufferTraits.bufferInfo->data[1], F_DUPFD_CLOEXEC, 0));
     if (!mEventFd.isValid()) {
         ALOGE("%s: Received ad invalid event fd.", __FUNCTION__);
         return -EINVAL;
     }

     int bufferId = bufferTraits.bufferInfo->data[2];
     if (bufferId < 0) {
         ALOGE("%s: Received an invalid (negative) id.", __FUNCTION__);
         return -EINVAL;
     }

     uint32_t clientBitMask;
     memcpy(&clientBitMask, &bufferTraits.bufferInfo->data[3], sizeof(clientBitMask));
     if (clientBitMask == 0U) {
         ALOGE("%s: Received an invalid client state mask.", __FUNCTION__);
         return -EINVAL;
     }

     uint32_t userMetadataSize;
     memcpy(&userMetadataSize, &bufferTraits.bufferInfo->data[4], sizeof(userMetadataSize));
     if (mMetadata.user_metadata_size() != userMetadataSize) {
         ALOGE("%s: user metadata size not match: expected %u, actual %zu.", __FUNCTION__,
               userMetadataSize, mMetadata.user_metadata_size());
         return -EINVAL;
     }

     size_t metadataSize = static_cast<size_t>(mMetadata.metadata_size());
     if (metadataSize < BufferHubDefs::kMetadataHeaderSize) {
         ALOGE("%s: metadata too small: %zu", __FUNCTION__, metadataSize);
         return -EINVAL;
     }

     // Populate shortcuts to the atomics in metadata.
     auto metadata_header = mMetadata.metadata_header();
     buffer_state_ = &metadata_header->buffer_state;
     fence_state_ = &metadata_header->fence_state;
     active_clients_bit_mask_ = &metadata_header->active_clients_bit_mask;
     // The C++ standard recommends (but does not require) that lock-free atomic operations are
     // also address-free, that is, suitable for communication between processes using shared
     // memory.
     LOG_ALWAYS_FATAL_IF(!std::atomic_is_lock_free(buffer_state_) ||
                                 !std::atomic_is_lock_free(fence_state_) ||
                                 !std::atomic_is_lock_free(active_clients_bit_mask_),
                         "Atomic variables in ashmen are not lock free.");

     // Import the buffer: We only need to hold on the native_handle_t here so that
     // GraphicBuffer instance can be created in future.
     mBufferHandle = std::move(bufferTraits.bufferHandle);
     memcpy(&mBufferDesc, &bufferTraits.bufferDesc, sizeof(AHardwareBuffer_Desc));

     mId = bufferId;
     mClientStateMask = clientBitMask;

     // TODO(b/112012161) Set up shared fences.
     ALOGD("%s: id=%d, buffer_state=%" PRIx32 ".", __FUNCTION__, mId,
           buffer_state_->load(std::memory_order_acquire));
     return 0;
 }

 int BufferHubBuffer::Gain() {
     uint32_t current_buffer_state = buffer_state_->load(std::memory_order_acquire);
     if (IsClientGained(current_buffer_state, mClientStateMask)) {
         ALOGV("%s: Buffer is already gained by this client %" PRIx32 ".", __FUNCTION__,
               mClientStateMask);
         return 0;
     }
     do {
         if (AnyClientGained(current_buffer_state & (~mClientStateMask)) ||
             AnyClientAcquired(current_buffer_state)) {
             ALOGE("%s: Buffer is in use, id=%d mClientStateMask=%" PRIx32 " state=%" PRIx32 ".",
                   __FUNCTION__, mId, mClientStateMask, current_buffer_state);
             return -EBUSY;
         }
         // Change the buffer state to gained state, whose value happens to be the same as
         // mClientStateMask.
     } while (!buffer_state_->compare_exchange_weak(current_buffer_state, mClientStateMask,
                                                    std::memory_order_acq_rel,
                                                    std::memory_order_acquire));
     // TODO(b/119837586): Update fence state and return GPU fence.
     return 0;
 }

 int BufferHubBuffer::Post() {
     uint32_t current_buffer_state = buffer_state_->load(std::memory_order_acquire);
     uint32_t updated_buffer_state = (~mClientStateMask) & BufferHubDefs::kHighBitsMask;
     do {
         if (!IsClientGained(current_buffer_state, mClientStateMask)) {
             ALOGE("%s: Cannot post a buffer that is not gained by this client. buffer_id=%d "
                   "mClientStateMask=%" PRIx32 " state=%" PRIx32 ".",
                   __FUNCTION__, mId, mClientStateMask, current_buffer_state);
             return -EBUSY;
         }
         // Set the producer client buffer state to released, other clients' buffer state to posted.
         // Post to all existing and non-existing clients.
     } while (!buffer_state_->compare_exchange_weak(current_buffer_state, updated_buffer_state,
                                                    std::memory_order_acq_rel,
                                                    std::memory_order_acquire));
     // TODO(b/119837586): Update fence state and return GPU fence if needed.
     return 0;
 }

 int BufferHubBuffer::Acquire() {
     uint32_t current_buffer_state = buffer_state_->load(std::memory_order_acquire);
     if (IsClientAcquired(current_buffer_state, mClientStateMask)) {
         ALOGV("%s: Buffer is already acquired by this client %" PRIx32 ".", __FUNCTION__,
               mClientStateMask);
         return 0;
     }
     uint32_t updated_buffer_state = 0U;
     do {
         if (!IsClientPosted(current_buffer_state, mClientStateMask)) {
             ALOGE("%s: Cannot acquire a buffer that is not in posted state. buffer_id=%d "
                   "mClientStateMask=%" PRIx32 " state=%" PRIx32 ".",
                   __FUNCTION__, mId, mClientStateMask, current_buffer_state);
             return -EBUSY;
         }
         // Change the buffer state for this consumer from posted to acquired.
         updated_buffer_state = current_buffer_state ^ mClientStateMask;
     } while (!buffer_state_->compare_exchange_weak(current_buffer_state, updated_buffer_state,
                                                    std::memory_order_acq_rel,
                                                    std::memory_order_acquire));
     // TODO(b/119837586): Update fence state and return GPU fence.
     return 0;
 }

 int BufferHubBuffer::Release() {
     uint32_t current_buffer_state = buffer_state_->load(std::memory_order_acquire);
     if (IsClientReleased(current_buffer_state, mClientStateMask)) {
         ALOGV("%s: Buffer is already released by this client %" PRIx32 ".", __FUNCTION__,
               mClientStateMask);
         return 0;
     }
     uint32_t updated_buffer_state = 0U;
     do {
         updated_buffer_state = current_buffer_state & (~mClientStateMask);
     } while (!buffer_state_->compare_exchange_weak(current_buffer_state, updated_buffer_state,
                                                    std::memory_order_acq_rel,
                                                    std::memory_order_acquire));
     // TODO(b/119837586): Update fence state and return GPU fence if needed.
     return 0;
 }

 bool BufferHubBuffer::IsReleased() const {
     return (buffer_state_->load(std::memory_order_acquire) &
             active_clients_bit_mask_->load(std::memory_order_acquire)) == 0;
 }

 bool BufferHubBuffer::IsValid() const {
     return mBufferHandle.getNativeHandle() != nullptr && mId >= 0 && mClientStateMask != 0U &&
             mEventFd.get() >= 0 && mMetadata.IsValid() && mBufferClient != nullptr;
 }

 native_handle_t* BufferHubBuffer::Duplicate() {
     if (mBufferClient == nullptr) {
         ALOGE("%s: missing BufferClient!", __FUNCTION__);
         return nullptr;
     }

     hidl_handle token;
     BufferHubStatus ret;
     IBufferClient::duplicate_cb dup_cb = [&](const auto& outToken, const auto& status) {
         token = std::move(outToken);
         ret = status;
     };

     if (!mBufferClient->duplicate(dup_cb).isOk()) {
         ALOGE("%s: duplicate transaction failed!", __FUNCTION__);
         return nullptr;
     } else if (ret != BufferHubStatus::NO_ERROR) {
         ALOGE("%s: duplicate failed with error %u.", __FUNCTION__, ret);
         return nullptr;
     } else if (token.getNativeHandle() == nullptr) {
         ALOGE("%s: duplicate got null token.", __FUNCTION__);
         return nullptr;
     }

     return native_handle_clone(token.getNativeHandle());
 }

 } // namespace android
	/*
	* Copyright (C) 2018 The Android Open Source Project
	*
	* Licensed under the Apache License, Version 2.0 (the "License");
	* you may not use this file except in compliance with the License.
	* You may obtain a copy of the License at
	*
	* http://www.apache.org/licenses/LICENSE-2.0
	*
	* Unless required by applicable law or agreed to in writing, software
	* distributed under the License is distributed on an "AS IS" BASIS,
	* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
	* See the License for the specific language governing permissions and
	* limitations under the License.
	*/

	#include <poll.h>

	#include <android-base/unique_fd.h>
	#include <android/frameworks/bufferhub/1.0/IBufferHub.h>
	#include <log/log.h>
	#include <ui/BufferHubBuffer.h>
	#include <ui/BufferHubDefs.h>
	#include <utils/Trace.h>

	using ::android::base::unique_fd;
	using ::android::BufferHubDefs::AnyClientAcquired;
	using ::android::BufferHubDefs::AnyClientGained;
	using ::android::BufferHubDefs::IsClientAcquired;
	using ::android::BufferHubDefs::IsClientGained;
	using ::android::BufferHubDefs::IsClientPosted;
	using ::android::BufferHubDefs::IsClientReleased;
	using ::android::frameworks::bufferhub::V1_0::BufferHubStatus;
	using ::android::frameworks::bufferhub::V1_0::BufferTraits;
	using ::android::frameworks::bufferhub::V1_0::IBufferClient;
	using ::android::frameworks::bufferhub::V1_0::IBufferHub;
	using ::android::hardware::hidl_handle;
	using ::android::hardware::graphics::common::V1_2::HardwareBufferDescription;

	namespace android {

	std::unique_ptr<BufferHubBuffer> BufferHubBuffer::Create(uint32_t width, uint32_t height,
	uint32_t layerCount, uint32_t format,
	uint64_t usage, size_t userMetadataSize) {
	auto buffer = std::unique_ptr<BufferHubBuffer>(
	new BufferHubBuffer(width, height, layerCount, format, usage, userMetadataSize));
	return buffer->IsValid() ? std::move(buffer) : nullptr;
	}

	std::unique_ptr<BufferHubBuffer> BufferHubBuffer::Import(const native_handle_t* token) {
	if (token == nullptr) {
	ALOGE("%s: token cannot be nullptr!", __FUNCTION__);
	return nullptr;
	}

	auto buffer = std::unique_ptr<BufferHubBuffer>(new BufferHubBuffer(token));
	return buffer->IsValid() ? std::move(buffer) : nullptr;
	}

	BufferHubBuffer::BufferHubBuffer(uint32_t width, uint32_t height, uint32_t layerCount,
	uint32_t format, uint64_t usage, size_t userMetadataSize) {
	ATRACE_CALL();
	ALOGD("%s: width=%u height=%u layerCount=%u, format=%u "
	"usage=%" PRIx64 " mUserMetadataSize=%zu",
	__FUNCTION__, width, height, layerCount, format, usage, userMetadataSize);

	sp<IBufferHub> bufferhub = IBufferHub::getService();
	if (bufferhub.get() == nullptr) {
	ALOGE("%s: BufferHub service not found!", __FUNCTION__);
	return;
	}

	AHardwareBuffer_Desc aDesc = {width, height, layerCount, format,
	usage, /stride=/0UL, /rfu0=/0UL, /rfu1=/0ULL};
	HardwareBufferDescription desc;
	memcpy(&desc, &aDesc, sizeof(HardwareBufferDescription));

	BufferHubStatus ret;
	sp<IBufferClient> client;
	BufferTraits bufferTraits;
	IBufferHub::allocateBuffer_cb alloc_cb = [&](const auto& status, const auto& outClient,
	const auto& traits) {
	ret = status;
	client = std::move(outClient);
	bufferTraits = std::move(traits);
	};

	if (!bufferhub->allocateBuffer(desc, static_cast<uint32_t>(userMetadataSize), alloc_cb)
	.isOk()) {
	ALOGE("%s: allocateBuffer transaction failed!", __FUNCTION__);
	return;
	} else if (ret != BufferHubStatus::NO_ERROR) {
	ALOGE("%s: allocateBuffer failed with error %u.", __FUNCTION__, ret);
	return;
	} else if (client == nullptr) {
	ALOGE("%s: allocateBuffer got null BufferClient.", __FUNCTION__);
	return;
	}

	const int importRet = initWithBufferTraits(bufferTraits);
	if (importRet < 0) {
	ALOGE("%s: Failed to import buffer: %s", __FUNCTION__, strerror(-importRet));
	client->close();
	}
	mBufferClient = std::move(client);
	}

	BufferHubBuffer::BufferHubBuffer(const native_handle_t* token) {
	sp<IBufferHub> bufferhub = IBufferHub::getService();
	if (bufferhub.get() == nullptr) {
	ALOGE("%s: BufferHub service not found!", __FUNCTION__);
	return;
	}

	BufferHubStatus ret;
	sp<IBufferClient> client;
	BufferTraits bufferTraits;
	IBufferHub::importBuffer_cb import_cb = [&](const auto& status, const auto& outClient,
	const auto& traits) {
	ret = status;
	client = std::move(outClient);
	bufferTraits = std::move(traits);
	};

	// hidl_handle(native_handle_t*) simply creates a raw pointer reference withouth ownership
	// transfer.
	if (!bufferhub->importBuffer(hidl_handle(token), import_cb).isOk()) {
	ALOGE("%s: importBuffer transaction failed!", __FUNCTION__);
	return;
	} else if (ret != BufferHubStatus::NO_ERROR) {
	ALOGE("%s: importBuffer failed with error %u.", __FUNCTION__, ret);
	return;
	} else if (client == nullptr) {
	ALOGE("%s: importBuffer got null BufferClient.", __FUNCTION__);
	return;
	}

	const int importRet = initWithBufferTraits(bufferTraits);
	if (importRet < 0) {
	ALOGE("%s: Failed to import buffer: %s", __FUNCTION__, strerror(-importRet));
	client->close();
	}
	mBufferClient = std::move(client);
	}

	BufferHubBuffer::~BufferHubBuffer() {
	// Close buffer client to avoid possible race condition: user could first duplicate and hold
	// token with the original buffer gone, and then try to import the token. The close function
	// will explicitly invalidate the token to avoid this.
	if (mBufferClient != nullptr) {
	if (!mBufferClient->close().isOk()) {
	ALOGE("%s: close BufferClient transaction failed!", __FUNCTION__);
	}
	}
	}

	int BufferHubBuffer::initWithBufferTraits(const BufferTraits& bufferTraits) {
	ATRACE_CALL();

	if (bufferTraits.bufferInfo.getNativeHandle() == nullptr) {
	ALOGE("%s: missing buffer info handle.", __FUNCTION__);
	return -EINVAL;
	}

	if (bufferTraits.bufferHandle.getNativeHandle() == nullptr) {
	ALOGE("%s: missing gralloc handle.", __FUNCTION__);
	return -EINVAL;
	}

	// Import fds. Dup fds because hidl_handle owns the fds.
	unique_fd ashmemFd(fcntl(bufferTraits.bufferInfo->data[0], F_DUPFD_CLOEXEC, 0));
	mMetadata = BufferHubMetadata::Import(std::move(ashmemFd));
	if (!mMetadata.IsValid()) {
	ALOGE("%s: Received an invalid metadata.", __FUNCTION__);
	return -EINVAL;
	}

	mEventFd = BufferHubEventFd(fcntl(bufferTraits.bufferInfo->data[1], F_DUPFD_CLOEXEC, 0));
	if (!mEventFd.isValid()) {
	ALOGE("%s: Received ad invalid event fd.", __FUNCTION__);
	return -EINVAL;
	}

	int bufferId = bufferTraits.bufferInfo->data[2];
	if (bufferId < 0) {
	ALOGE("%s: Received an invalid (negative) id.", __FUNCTION__);
	return -EINVAL;
	}

	uint32_t clientBitMask;
	memcpy(&clientBitMask, &bufferTraits.bufferInfo->data[3], sizeof(clientBitMask));
	if (clientBitMask == 0U) {
	ALOGE("%s: Received an invalid client state mask.", __FUNCTION__);
	return -EINVAL;
	}

	uint32_t userMetadataSize;
	memcpy(&userMetadataSize, &bufferTraits.bufferInfo->data[4], sizeof(userMetadataSize));
	if (mMetadata.user_metadata_size() != userMetadataSize) {
	ALOGE("%s: user metadata size not match: expected %u, actual %zu.", __FUNCTION__,
	userMetadataSize, mMetadata.user_metadata_size());
	return -EINVAL;
	}

	size_t metadataSize = static_cast<size_t>(mMetadata.metadata_size());
	if (metadataSize < BufferHubDefs::kMetadataHeaderSize) {
	ALOGE("%s: metadata too small: %zu", __FUNCTION__, metadataSize);
	return -EINVAL;
	}

	// Populate shortcuts to the atomics in metadata.
	auto metadata_header = mMetadata.metadata_header();
	buffer_state_ = &metadata_header->buffer_state;
	fence_state_ = &metadata_header->fence_state;
	active_clients_bit_mask_ = &metadata_header->active_clients_bit_mask;
	// The C++ standard recommends (but does not require) that lock-free atomic operations are
	// also address-free, that is, suitable for communication between processes using shared
	// memory.
	LOG_ALWAYS_FATAL_IF(!std::atomic_is_lock_free(buffer_state_) \|\|
	!std::atomic_is_lock_free(fence_state_) \|\|
	!std::atomic_is_lock_free(active_clients_bit_mask_),
	"Atomic variables in ashmen are not lock free.");

	// Import the buffer: We only need to hold on the native_handle_t here so that
	// GraphicBuffer instance can be created in future.
	mBufferHandle = std::move(bufferTraits.bufferHandle);
	memcpy(&mBufferDesc, &bufferTraits.bufferDesc, sizeof(AHardwareBuffer_Desc));

	mId = bufferId;
	mClientStateMask = clientBitMask;

	// TODO(b/112012161) Set up shared fences.
	ALOGD("%s: id=%d, buffer_state=%" PRIx32 ".", __FUNCTION__, mId,
	buffer_state_->load(std::memory_order_acquire));
	return 0;
	}

	int BufferHubBuffer::Gain() {
	uint32_t current_buffer_state = buffer_state_->load(std::memory_order_acquire);
	if (IsClientGained(current_buffer_state, mClientStateMask)) {
	ALOGV("%s: Buffer is already gained by this client %" PRIx32 ".", __FUNCTION__,
	mClientStateMask);
	return 0;
	}
	do {
	if (AnyClientGained(current_buffer_state & (~mClientStateMask)) \|\|
	AnyClientAcquired(current_buffer_state)) {
	ALOGE("%s: Buffer is in use, id=%d mClientStateMask=%" PRIx32 " state=%" PRIx32 ".",
	__FUNCTION__, mId, mClientStateMask, current_buffer_state);
	return -EBUSY;
	}
	// Change the buffer state to gained state, whose value happens to be the same as
	// mClientStateMask.
	} while (!buffer_state_->compare_exchange_weak(current_buffer_state, mClientStateMask,
	std::memory_order_acq_rel,
	std::memory_order_acquire));
	// TODO(b/119837586): Update fence state and return GPU fence.
	return 0;
	}

	int BufferHubBuffer::Post() {
	uint32_t current_buffer_state = buffer_state_->load(std::memory_order_acquire);
	uint32_t updated_buffer_state = (~mClientStateMask) & BufferHubDefs::kHighBitsMask;
	do {
	if (!IsClientGained(current_buffer_state, mClientStateMask)) {
	ALOGE("%s: Cannot post a buffer that is not gained by this client. buffer_id=%d "
	"mClientStateMask=%" PRIx32 " state=%" PRIx32 ".",
	__FUNCTION__, mId, mClientStateMask, current_buffer_state);
	return -EBUSY;
	}
	// Set the producer client buffer state to released, other clients' buffer state to posted.
	// Post to all existing and non-existing clients.
	} while (!buffer_state_->compare_exchange_weak(current_buffer_state, updated_buffer_state,
	std::memory_order_acq_rel,
	std::memory_order_acquire));
	// TODO(b/119837586): Update fence state and return GPU fence if needed.
	return 0;
	}

	int BufferHubBuffer::Acquire() {
	uint32_t current_buffer_state = buffer_state_->load(std::memory_order_acquire);
	if (IsClientAcquired(current_buffer_state, mClientStateMask)) {
	ALOGV("%s: Buffer is already acquired by this client %" PRIx32 ".", __FUNCTION__,
	mClientStateMask);
	return 0;
	}
	uint32_t updated_buffer_state = 0U;
	do {
	if (!IsClientPosted(current_buffer_state, mClientStateMask)) {
	ALOGE("%s: Cannot acquire a buffer that is not in posted state. buffer_id=%d "
	"mClientStateMask=%" PRIx32 " state=%" PRIx32 ".",
	__FUNCTION__, mId, mClientStateMask, current_buffer_state);
	return -EBUSY;
	}
	// Change the buffer state for this consumer from posted to acquired.
	updated_buffer_state = current_buffer_state ^ mClientStateMask;
	} while (!buffer_state_->compare_exchange_weak(current_buffer_state, updated_buffer_state,
	std::memory_order_acq_rel,
	std::memory_order_acquire));
	// TODO(b/119837586): Update fence state and return GPU fence.
	return 0;
	}

	int BufferHubBuffer::Release() {
	uint32_t current_buffer_state = buffer_state_->load(std::memory_order_acquire);
	if (IsClientReleased(current_buffer_state, mClientStateMask)) {
	ALOGV("%s: Buffer is already released by this client %" PRIx32 ".", __FUNCTION__,
	mClientStateMask);
	return 0;
	}
	uint32_t updated_buffer_state = 0U;
	do {
	updated_buffer_state = current_buffer_state & (~mClientStateMask);
	} while (!buffer_state_->compare_exchange_weak(current_buffer_state, updated_buffer_state,
	std::memory_order_acq_rel,
	std::memory_order_acquire));
	// TODO(b/119837586): Update fence state and return GPU fence if needed.
	return 0;
	}

	bool BufferHubBuffer::IsReleased() const {
	return (buffer_state_->load(std::memory_order_acquire) &
	active_clients_bit_mask_->load(std::memory_order_acquire)) == 0;
	}

	bool BufferHubBuffer::IsValid() const {
	return mBufferHandle.getNativeHandle() != nullptr && mId >= 0 && mClientStateMask != 0U &&
	mEventFd.get() >= 0 && mMetadata.IsValid() && mBufferClient != nullptr;
	}

	native_handle_t* BufferHubBuffer::Duplicate() {
	if (mBufferClient == nullptr) {
	ALOGE("%s: missing BufferClient!", __FUNCTION__);
	return nullptr;
	}

	hidl_handle token;
	BufferHubStatus ret;
	IBufferClient::duplicate_cb dup_cb = [&](const auto& outToken, const auto& status) {
	token = std::move(outToken);
	ret = status;
	};

	if (!mBufferClient->duplicate(dup_cb).isOk()) {
	ALOGE("%s: duplicate transaction failed!", __FUNCTION__);
	return nullptr;
	} else if (ret != BufferHubStatus::NO_ERROR) {
	ALOGE("%s: duplicate failed with error %u.", __FUNCTION__, ret);
	return nullptr;
	} else if (token.getNativeHandle() == nullptr) {
	ALOGE("%s: duplicate got null token.", __FUNCTION__);
	return nullptr;
	}

	return native_handle_clone(token.getNativeHandle());
	}

	} // namespace android