services/bufferhub/BufferHubService.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 <array>
 #include <iomanip>
 #include <random>
 #include <sstream>

 #include <android/hardware_buffer.h>
 #include <bufferhub/BufferHubService.h>
 #include <cutils/native_handle.h>
 #include <log/log.h>
 #include <openssl/hmac.h>
 #include <system/graphics-base.h>
 #include <ui/BufferHubDefs.h>

 using ::android::BufferHubDefs::MetadataHeader;
 using ::android::hardware::Void;

 namespace android {
 namespace frameworks {
 namespace bufferhub {
 namespace V1_0 {
 namespace implementation {

 BufferHubService::BufferHubService() {
     std::mt19937_64 randomEngine;
     randomEngine.seed(time(nullptr));

     mKey = randomEngine();
 }

 Return<void> BufferHubService::allocateBuffer(const HardwareBufferDescription& description,
                                               const uint32_t userMetadataSize,
                                               allocateBuffer_cb _hidl_cb) {
     AHardwareBuffer_Desc desc;
     memcpy(&desc, &description, sizeof(AHardwareBuffer_Desc));

     std::shared_ptr<BufferNode> node =
             std::make_shared<BufferNode>(desc.width, desc.height, desc.layers, desc.format,
                                          desc.usage, userMetadataSize,
                                          BufferHubIdGenerator::getInstance().getId());
     if (node == nullptr || !node->isValid()) {
         ALOGE("%s: creating BufferNode failed.", __FUNCTION__);
         _hidl_cb(/*status=*/BufferHubStatus::ALLOCATION_FAILED, /*bufferClient=*/nullptr,
                  /*bufferTraits=*/{});
         return Void();
     }

     sp<BufferClient> client = BufferClient::create(this, node);
     // Add it to list for bookkeeping and dumpsys.
     std::lock_guard<std::mutex> lock(mClientSetMutex);
     mClientSet.emplace(client);

     // Allocate memory for bufferInfo of type hidl_handle on the stack. See
     // http://aosp/286282 for the usage of NATIVE_HANDLE_DECLARE_STORAGE.
     NATIVE_HANDLE_DECLARE_STORAGE(bufferInfoStorage, BufferHubDefs::kBufferInfoNumFds,
                                   BufferHubDefs::kBufferInfoNumInts);
     hidl_handle bufferInfo =
             buildBufferInfo(bufferInfoStorage, node->id(), node->addNewActiveClientsBitToMask(),
                             node->userMetadataSize(), node->metadata().ashmemFd(),
                             node->eventFd().get());
     // During the gralloc allocation carried out by BufferNode, gralloc allocator will populate the
     // fields of its HardwareBufferDescription (i.e. strides) according to the actual
     // gralloc implementation. We need to read those fields back and send them to the client via
     // BufferTraits.
     HardwareBufferDescription allocatedBufferDesc;
     memcpy(&allocatedBufferDesc, &node->bufferDesc(), sizeof(AHardwareBuffer_Desc));
     BufferTraits bufferTraits = {/*bufferDesc=*/allocatedBufferDesc,
                                  /*bufferHandle=*/hidl_handle(node->bufferHandle()),
                                  /*bufferInfo=*/std::move(bufferInfo)};

     _hidl_cb(/*status=*/BufferHubStatus::NO_ERROR, /*bufferClient=*/client,
              /*bufferTraits=*/std::move(bufferTraits));
     return Void();
 }

 Return<void> BufferHubService::importBuffer(const hidl_handle& tokenHandle,
                                             importBuffer_cb _hidl_cb) {
     if (!tokenHandle.getNativeHandle() || tokenHandle->numFds != 0 || tokenHandle->numInts <= 1) {
         // nullptr handle or wrong format
         _hidl_cb(/*status=*/BufferHubStatus::INVALID_TOKEN, /*bufferClient=*/nullptr,
                  /*bufferTraits=*/{});
         return Void();
     }

     int tokenId = tokenHandle->data[0];

     wp<BufferClient> originClientWp;
     {
         std::lock_guard<std::mutex> lock(mTokenMutex);
         auto iter = mTokenMap.find(tokenId);
         if (iter == mTokenMap.end()) {
             // Token Id not exist
             ALOGD("%s: token #%d not found.", __FUNCTION__, tokenId);
             _hidl_cb(/*status=*/BufferHubStatus::INVALID_TOKEN, /*bufferClient=*/nullptr,
                      /*bufferTraits=*/{});
             return Void();
         }

         const std::vector<uint8_t>& tokenHMAC = iter->second.first;

         int numIntsForHMAC = (int)ceil(tokenHMAC.size() * sizeof(uint8_t) / (double)sizeof(int));
         if (tokenHandle->numInts - 1 != numIntsForHMAC) {
             // HMAC size not match
             ALOGD("%s: token #%d HMAC size not match. Expected: %d Actual: %d", __FUNCTION__,
                   tokenId, numIntsForHMAC, tokenHandle->numInts - 1);
             _hidl_cb(/*status=*/BufferHubStatus::INVALID_TOKEN, /*bufferClient=*/nullptr,
                      /*bufferTraits=*/{});
             return Void();
         }

         size_t hmacSize = tokenHMAC.size() * sizeof(uint8_t);
         if (memcmp(tokenHMAC.data(), &tokenHandle->data[1], hmacSize) != 0) {
             // HMAC not match
             ALOGD("%s: token #%d HMAC not match.", __FUNCTION__, tokenId);
             _hidl_cb(/*status=*/BufferHubStatus::INVALID_TOKEN, /*bufferClient=*/nullptr,
                      /*bufferTraits=*/{});
             return Void();
         }

         originClientWp = iter->second.second;
         mTokenMap.erase(iter);
     }

     // Check if original client is dead
     sp<BufferClient> originClient = originClientWp.promote();
     if (!originClient) {
         // Should not happen since token should be removed if already gone
         ALOGE("%s: original client %p gone!", __FUNCTION__, originClientWp.unsafe_get());
         _hidl_cb(/*status=*/BufferHubStatus::BUFFER_FREED, /*bufferClient=*/nullptr,
                  /*bufferTraits=*/{});
         return Void();
     }

     sp<BufferClient> client = new BufferClient(*originClient);
     uint32_t clientStateMask = client->getBufferNode()->addNewActiveClientsBitToMask();
     if (clientStateMask == 0U) {
         // Reach max client count
         ALOGE("%s: import failed, BufferNode#%u reached maximum clients.", __FUNCTION__,
               client->getBufferNode()->id());
         _hidl_cb(/*status=*/BufferHubStatus::MAX_CLIENT, /*bufferClient=*/nullptr,
                  /*bufferTraits=*/{});
         return Void();
     }

     std::lock_guard<std::mutex> lock(mClientSetMutex);
     mClientSet.emplace(client);

     std::shared_ptr<BufferNode> node = client->getBufferNode();

     HardwareBufferDescription bufferDesc;
     memcpy(&bufferDesc, &node->bufferDesc(), sizeof(HardwareBufferDescription));

     // Allocate memory for bufferInfo of type hidl_handle on the stack. See
     // http://aosp/286282 for the usage of NATIVE_HANDLE_DECLARE_STORAGE.
     NATIVE_HANDLE_DECLARE_STORAGE(bufferInfoStorage, BufferHubDefs::kBufferInfoNumFds,
                                   BufferHubDefs::kBufferInfoNumInts);
     hidl_handle bufferInfo = buildBufferInfo(bufferInfoStorage, node->id(), clientStateMask,
                                              node->userMetadataSize(), node->metadata().ashmemFd(),
                                              node->eventFd().get());
     BufferTraits bufferTraits = {/*bufferDesc=*/bufferDesc,
                                  /*bufferHandle=*/hidl_handle(node->bufferHandle()),
                                  /*bufferInfo=*/std::move(bufferInfo)};

     _hidl_cb(/*status=*/BufferHubStatus::NO_ERROR, /*bufferClient=*/client,
              /*bufferTraits=*/std::move(bufferTraits));
     return Void();
 }

 Return<void> BufferHubService::debug(const hidl_handle& fd, const hidl_vec<hidl_string>& args) {
     if (fd.getNativeHandle() == nullptr || fd->numFds < 1) {
         ALOGE("%s: missing fd for writing.", __FUNCTION__);
         return Void();
     }

     FILE* out = fdopen(dup(fd->data[0]), "w");

     if (args.size() != 0) {
         fprintf(out,
                 "Note: lshal bufferhub currently does not support args. Input arguments are "
                 "ignored.\n");
     }

     std::ostringstream stream;

     // Get the number of clients of each buffer.
     // Map from bufferId to bufferNode_clientCount pair.
     std::map<int, std::pair<const std::shared_ptr<BufferNode>, uint32_t>> clientCount;
     {
         std::lock_guard<std::mutex> lock(mClientSetMutex);
         for (auto iter = mClientSet.begin(); iter != mClientSet.end(); ++iter) {
             sp<BufferClient> client = iter->promote();
             if (client != nullptr) {
                 const std::shared_ptr<BufferNode> node = client->getBufferNode();
                 auto mapIter = clientCount.find(node->id());
                 if (mapIter != clientCount.end()) {
                     ++mapIter->second.second;
                 } else {
                     clientCount.emplace(node->id(),
                                         std::pair<std::shared_ptr<BufferNode>, uint32_t>(node, 1U));
                 }
             }
         }
     }

     stream << "Active Buffers:\n";
     stream << std::right;
     stream << std::setw(6) << "Id";
     stream << " ";
     stream << std::setw(9) << "#Clients";
     stream << " ";
     stream << std::setw(14) << "Geometry";
     stream << " ";
     stream << std::setw(6) << "Format";
     stream << " ";
     stream << std::setw(10) << "Usage";
     stream << " ";
     stream << std::setw(10) << "State";
     stream << " ";
     stream << std::setw(8) << "Index";
     stream << std::endl;

     for (auto iter = clientCount.begin(); iter != clientCount.end(); ++iter) {
         const std::shared_ptr<BufferNode> node = std::move(iter->second.first);
         const uint32_t clientCount = iter->second.second;
         AHardwareBuffer_Desc desc = node->bufferDesc();

         MetadataHeader* metadataHeader =
                 const_cast<BufferHubMetadata*>(&node->metadata())->metadataHeader();
         const uint32_t state = metadataHeader->bufferState.load(std::memory_order_acquire);
         const uint64_t index = metadataHeader->queueIndex;

         stream << std::right;
         stream << std::setw(6) << /*Id=*/node->id();
         stream << " ";
         stream << std::setw(9) << /*#Clients=*/clientCount;
         stream << " ";
         if (desc.format == HAL_PIXEL_FORMAT_BLOB) {
             std::string size = std::to_string(desc.width) + " B";
             stream << std::setw(14) << /*Geometry=*/size;
         } else {
             std::string dimensions = std::to_string(desc.width) + "x" +
                     std::to_string(desc.height) + "x" + std::to_string(desc.layers);
             stream << std::setw(14) << /*Geometry=*/dimensions;
         }
         stream << " ";
         stream << std::setw(6) << /*Format=*/desc.format;
         stream << " ";
         stream << "0x" << std::hex << std::setfill('0');
         stream << std::setw(8) << /*Usage=*/desc.usage;
         stream << std::dec << std::setfill(' ');
         stream << " ";
         stream << "0x" << std::hex << std::setfill('0');
         stream << std::setw(8) << /*State=*/state;
         stream << std::dec << std::setfill(' ');
         stream << " ";
         stream << std::setw(8) << /*Index=*/index;
         stream << std::endl;
     }

     stream << std::endl;

     // Get the number of tokens of each buffer.
     // Map from bufferId to tokenCount
     std::map<int, uint32_t> tokenCount;
     {
         std::lock_guard<std::mutex> lock(mTokenMutex);
         for (auto iter = mTokenMap.begin(); iter != mTokenMap.end(); ++iter) {
             sp<BufferClient> client = iter->second.second.promote();
             if (client != nullptr) {
                 const std::shared_ptr<BufferNode> node = client->getBufferNode();
                 auto mapIter = tokenCount.find(node->id());
                 if (mapIter != tokenCount.end()) {
                     ++mapIter->second;
                 } else {
                     tokenCount.emplace(node->id(), 1U);
                 }
             }
         }
     }

     stream << "Unused Tokens:\n";
     stream << std::right;
     stream << std::setw(8) << "Buffer Id";
     stream << " ";
     stream << std::setw(7) << "#Tokens";
     stream << std::endl;

     for (auto iter = tokenCount.begin(); iter != tokenCount.end(); ++iter) {
         stream << std::right;
         stream << std::setw(8) << /*Buffer Id=*/iter->first;
         stream << " ";
         stream << std::setw(7) << /*#Tokens=*/iter->second;
         stream << std::endl;
     }

     fprintf(out, "%s", stream.str().c_str());

     fclose(out);
     return Void();
 }

 hidl_handle BufferHubService::registerToken(const wp<BufferClient>& client) {
     // Find next available token id
     std::lock_guard<std::mutex> lock(mTokenMutex);
     do {
         ++mLastTokenId;
     } while (mTokenMap.find(mLastTokenId) != mTokenMap.end());

     std::array<uint8_t, EVP_MAX_MD_SIZE> hmac;
     uint32_t hmacSize = 0U;

     HMAC(/*evp_md=*/EVP_sha256(), /*key=*/&mKey, /*key_len=*/kKeyLen,
          /*data=*/(uint8_t*)&mLastTokenId, /*data_len=*/mTokenIdSize,
          /*out=*/hmac.data(), /*out_len=*/&hmacSize);

     int numIntsForHMAC = (int)ceil(hmacSize / (double)sizeof(int));
     native_handle_t* handle = native_handle_create(/*numFds=*/0, /*numInts=*/1 + numIntsForHMAC);
     handle->data[0] = mLastTokenId;
     // Set all the the bits of last int to 0 since it might not be fully overwritten
     handle->data[numIntsForHMAC] = 0;
     memcpy(&handle->data[1], hmac.data(), hmacSize);

     // returnToken owns the native_handle_t* thus doing lifecycle management
     hidl_handle returnToken;
     returnToken.setTo(handle, /*shoudOwn=*/true);

     std::vector<uint8_t> hmacVec;
     hmacVec.resize(hmacSize);
     memcpy(hmacVec.data(), hmac.data(), hmacSize);
     mTokenMap.emplace(mLastTokenId, std::pair(hmacVec, client));

     return returnToken;
 }

 void BufferHubService::onClientClosed(const BufferClient* client) {
     removeTokenByClient(client);

     std::lock_guard<std::mutex> lock(mClientSetMutex);
     auto iter = std::find(mClientSet.begin(), mClientSet.end(), client);
     if (iter != mClientSet.end()) {
         mClientSet.erase(iter);
     }
 }

 // Implementation of this function should be consistent with the definition of bufferInfo handle in
 // ui/BufferHubDefs.h.
 hidl_handle BufferHubService::buildBufferInfo(char* bufferInfoStorage, int bufferId,
                                               uint32_t clientBitMask, uint32_t userMetadataSize,
                                               int metadataFd, int eventFd) {
     native_handle_t* infoHandle =
             native_handle_init(bufferInfoStorage, BufferHubDefs::kBufferInfoNumFds,
                                BufferHubDefs::kBufferInfoNumInts);

     infoHandle->data[0] = metadataFd;
     infoHandle->data[1] = eventFd;
     infoHandle->data[2] = bufferId;
     // Use memcpy to convert to int without missing digit.
     // TOOD(b/121345852): use bit_cast to unpack bufferInfo when C++20 becomes available.
     memcpy(&infoHandle->data[3], &clientBitMask, sizeof(clientBitMask));
     memcpy(&infoHandle->data[4], &userMetadataSize, sizeof(userMetadataSize));

     hidl_handle bufferInfo;
     bufferInfo.setTo(infoHandle, /*shouldOwn=*/false);

     return bufferInfo;
 }

 void BufferHubService::removeTokenByClient(const BufferClient* client) {
     std::lock_guard<std::mutex> lock(mTokenMutex);
     auto iter = mTokenMap.begin();
     while (iter != mTokenMap.end()) {
         if (iter->second.second == client) {
             auto oldIter = iter;
             ++iter;
             mTokenMap.erase(oldIter);
         } else {
             ++iter;
         }
     }
 }

 } // namespace implementation
 } // namespace V1_0
 } // namespace bufferhub
 } // namespace frameworks
 } // 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 <array>
	#include <iomanip>
	#include <random>
	#include <sstream>

	#include <android/hardware_buffer.h>
	#include <bufferhub/BufferHubService.h>
	#include <cutils/native_handle.h>
	#include <log/log.h>
	#include <openssl/hmac.h>
	#include <system/graphics-base.h>
	#include <ui/BufferHubDefs.h>

	using ::android::BufferHubDefs::MetadataHeader;
	using ::android::hardware::Void;

	namespace android {
	namespace frameworks {
	namespace bufferhub {
	namespace V1_0 {
	namespace implementation {

	BufferHubService::BufferHubService() {
	std::mt19937_64 randomEngine;
	randomEngine.seed(time(nullptr));

	mKey = randomEngine();
	}

	Return<void> BufferHubService::allocateBuffer(const HardwareBufferDescription& description,
	const uint32_t userMetadataSize,
	allocateBuffer_cb _hidl_cb) {
	AHardwareBuffer_Desc desc;
	memcpy(&desc, &description, sizeof(AHardwareBuffer_Desc));

	std::shared_ptr<BufferNode> node =
	std::make_shared<BufferNode>(desc.width, desc.height, desc.layers, desc.format,
	desc.usage, userMetadataSize,
	BufferHubIdGenerator::getInstance().getId());
	if (node == nullptr \|\| !node->isValid()) {
	ALOGE("%s: creating BufferNode failed.", __FUNCTION__);
	_hidl_cb(/status=/BufferHubStatus::ALLOCATION_FAILED, /bufferClient=/nullptr,
	/bufferTraits=/{});
	return Void();
	}

	sp<BufferClient> client = BufferClient::create(this, node);
	// Add it to list for bookkeeping and dumpsys.
	std::lock_guard<std::mutex> lock(mClientSetMutex);
	mClientSet.emplace(client);

	// Allocate memory for bufferInfo of type hidl_handle on the stack. See
	// http://aosp/286282 for the usage of NATIVE_HANDLE_DECLARE_STORAGE.
	NATIVE_HANDLE_DECLARE_STORAGE(bufferInfoStorage, BufferHubDefs::kBufferInfoNumFds,
	BufferHubDefs::kBufferInfoNumInts);
	hidl_handle bufferInfo =
	buildBufferInfo(bufferInfoStorage, node->id(), node->addNewActiveClientsBitToMask(),
	node->userMetadataSize(), node->metadata().ashmemFd(),
	node->eventFd().get());
	// During the gralloc allocation carried out by BufferNode, gralloc allocator will populate the
	// fields of its HardwareBufferDescription (i.e. strides) according to the actual
	// gralloc implementation. We need to read those fields back and send them to the client via
	// BufferTraits.
	HardwareBufferDescription allocatedBufferDesc;
	memcpy(&allocatedBufferDesc, &node->bufferDesc(), sizeof(AHardwareBuffer_Desc));
	BufferTraits bufferTraits = {/bufferDesc=/allocatedBufferDesc,
	/bufferHandle=/hidl_handle(node->bufferHandle()),
	/bufferInfo=/std::move(bufferInfo)};

	_hidl_cb(/status=/BufferHubStatus::NO_ERROR, /bufferClient=/client,
	/bufferTraits=/std::move(bufferTraits));
	return Void();
	}

	Return<void> BufferHubService::importBuffer(const hidl_handle& tokenHandle,
	importBuffer_cb _hidl_cb) {
	if (!tokenHandle.getNativeHandle() \|\| tokenHandle->numFds != 0 \|\| tokenHandle->numInts <= 1) {
	// nullptr handle or wrong format
	_hidl_cb(/status=/BufferHubStatus::INVALID_TOKEN, /bufferClient=/nullptr,
	/bufferTraits=/{});
	return Void();
	}

	int tokenId = tokenHandle->data[0];

	wp<BufferClient> originClientWp;
	{
	std::lock_guard<std::mutex> lock(mTokenMutex);
	auto iter = mTokenMap.find(tokenId);
	if (iter == mTokenMap.end()) {
	// Token Id not exist
	ALOGD("%s: token #%d not found.", __FUNCTION__, tokenId);
	_hidl_cb(/status=/BufferHubStatus::INVALID_TOKEN, /bufferClient=/nullptr,
	/bufferTraits=/{});
	return Void();
	}

	const std::vector<uint8_t>& tokenHMAC = iter->second.first;

	int numIntsForHMAC = (int)ceil(tokenHMAC.size() * sizeof(uint8_t) / (double)sizeof(int));
	if (tokenHandle->numInts - 1 != numIntsForHMAC) {
	// HMAC size not match
	ALOGD("%s: token #%d HMAC size not match. Expected: %d Actual: %d", __FUNCTION__,
	tokenId, numIntsForHMAC, tokenHandle->numInts - 1);
	_hidl_cb(/status=/BufferHubStatus::INVALID_TOKEN, /bufferClient=/nullptr,
	/bufferTraits=/{});
	return Void();
	}

	size_t hmacSize = tokenHMAC.size() * sizeof(uint8_t);
	if (memcmp(tokenHMAC.data(), &tokenHandle->data[1], hmacSize) != 0) {
	// HMAC not match
	ALOGD("%s: token #%d HMAC not match.", __FUNCTION__, tokenId);
	_hidl_cb(/status=/BufferHubStatus::INVALID_TOKEN, /bufferClient=/nullptr,
	/bufferTraits=/{});
	return Void();
	}

	originClientWp = iter->second.second;
	mTokenMap.erase(iter);
	}

	// Check if original client is dead
	sp<BufferClient> originClient = originClientWp.promote();
	if (!originClient) {
	// Should not happen since token should be removed if already gone
	ALOGE("%s: original client %p gone!", __FUNCTION__, originClientWp.unsafe_get());
	_hidl_cb(/status=/BufferHubStatus::BUFFER_FREED, /bufferClient=/nullptr,
	/bufferTraits=/{});
	return Void();
	}

	sp<BufferClient> client = new BufferClient(*originClient);
	uint32_t clientStateMask = client->getBufferNode()->addNewActiveClientsBitToMask();
	if (clientStateMask == 0U) {
	// Reach max client count
	ALOGE("%s: import failed, BufferNode#%u reached maximum clients.", __FUNCTION__,
	client->getBufferNode()->id());
	_hidl_cb(/status=/BufferHubStatus::MAX_CLIENT, /bufferClient=/nullptr,
	/bufferTraits=/{});
	return Void();
	}

	std::lock_guard<std::mutex> lock(mClientSetMutex);
	mClientSet.emplace(client);

	std::shared_ptr<BufferNode> node = client->getBufferNode();

	HardwareBufferDescription bufferDesc;
	memcpy(&bufferDesc, &node->bufferDesc(), sizeof(HardwareBufferDescription));

	// Allocate memory for bufferInfo of type hidl_handle on the stack. See
	// http://aosp/286282 for the usage of NATIVE_HANDLE_DECLARE_STORAGE.
	NATIVE_HANDLE_DECLARE_STORAGE(bufferInfoStorage, BufferHubDefs::kBufferInfoNumFds,
	BufferHubDefs::kBufferInfoNumInts);
	hidl_handle bufferInfo = buildBufferInfo(bufferInfoStorage, node->id(), clientStateMask,
	node->userMetadataSize(), node->metadata().ashmemFd(),
	node->eventFd().get());
	BufferTraits bufferTraits = {/bufferDesc=/bufferDesc,
	/bufferHandle=/hidl_handle(node->bufferHandle()),
	/bufferInfo=/std::move(bufferInfo)};

	_hidl_cb(/status=/BufferHubStatus::NO_ERROR, /bufferClient=/client,
	/bufferTraits=/std::move(bufferTraits));
	return Void();
	}

	Return<void> BufferHubService::debug(const hidl_handle& fd, const hidl_vec<hidl_string>& args) {
	if (fd.getNativeHandle() == nullptr \|\| fd->numFds < 1) {
	ALOGE("%s: missing fd for writing.", __FUNCTION__);
	return Void();
	}

	FILE* out = fdopen(dup(fd->data[0]), "w");

	if (args.size() != 0) {
	fprintf(out,
	"Note: lshal bufferhub currently does not support args. Input arguments are "
	"ignored.\n");
	}

	std::ostringstream stream;

	// Get the number of clients of each buffer.
	// Map from bufferId to bufferNode_clientCount pair.
	std::map<int, std::pair<const std::shared_ptr<BufferNode>, uint32_t>> clientCount;
	{
	std::lock_guard<std::mutex> lock(mClientSetMutex);
	for (auto iter = mClientSet.begin(); iter != mClientSet.end(); ++iter) {
	sp<BufferClient> client = iter->promote();
	if (client != nullptr) {
	const std::shared_ptr<BufferNode> node = client->getBufferNode();
	auto mapIter = clientCount.find(node->id());
	if (mapIter != clientCount.end()) {
	++mapIter->second.second;
	} else {
	clientCount.emplace(node->id(),
	std::pair<std::shared_ptr<BufferNode>, uint32_t>(node, 1U));
	}
	}
	}
	}

	stream << "Active Buffers:\n";
	stream << std::right;
	stream << std::setw(6) << "Id";
	stream << " ";
	stream << std::setw(9) << "#Clients";
	stream << " ";
	stream << std::setw(14) << "Geometry";
	stream << " ";
	stream << std::setw(6) << "Format";
	stream << " ";
	stream << std::setw(10) << "Usage";
	stream << " ";
	stream << std::setw(10) << "State";
	stream << " ";
	stream << std::setw(8) << "Index";
	stream << std::endl;

	for (auto iter = clientCount.begin(); iter != clientCount.end(); ++iter) {
	const std::shared_ptr<BufferNode> node = std::move(iter->second.first);
	const uint32_t clientCount = iter->second.second;
	AHardwareBuffer_Desc desc = node->bufferDesc();

	MetadataHeader* metadataHeader =
	const_cast<BufferHubMetadata*>(&node->metadata())->metadataHeader();
	const uint32_t state = metadataHeader->bufferState.load(std::memory_order_acquire);
	const uint64_t index = metadataHeader->queueIndex;

	stream << std::right;
	stream << std::setw(6) << /Id=/node->id();
	stream << " ";
	stream << std::setw(9) << /#Clients=/clientCount;
	stream << " ";
	if (desc.format == HAL_PIXEL_FORMAT_BLOB) {
	std::string size = std::to_string(desc.width) + " B";
	stream << std::setw(14) << /Geometry=/size;
	} else {
	std::string dimensions = std::to_string(desc.width) + "x" +
	std::to_string(desc.height) + "x" + std::to_string(desc.layers);
	stream << std::setw(14) << /Geometry=/dimensions;
	}
	stream << " ";
	stream << std::setw(6) << /Format=/desc.format;
	stream << " ";
	stream << "0x" << std::hex << std::setfill('0');
	stream << std::setw(8) << /Usage=/desc.usage;
	stream << std::dec << std::setfill(' ');
	stream << " ";
	stream << "0x" << std::hex << std::setfill('0');
	stream << std::setw(8) << /State=/state;
	stream << std::dec << std::setfill(' ');
	stream << " ";
	stream << std::setw(8) << /Index=/index;
	stream << std::endl;
	}

	stream << std::endl;

	// Get the number of tokens of each buffer.
	// Map from bufferId to tokenCount
	std::map<int, uint32_t> tokenCount;
	{
	std::lock_guard<std::mutex> lock(mTokenMutex);
	for (auto iter = mTokenMap.begin(); iter != mTokenMap.end(); ++iter) {
	sp<BufferClient> client = iter->second.second.promote();
	if (client != nullptr) {
	const std::shared_ptr<BufferNode> node = client->getBufferNode();
	auto mapIter = tokenCount.find(node->id());
	if (mapIter != tokenCount.end()) {
	++mapIter->second;
	} else {
	tokenCount.emplace(node->id(), 1U);
	}
	}
	}
	}

	stream << "Unused Tokens:\n";
	stream << std::right;
	stream << std::setw(8) << "Buffer Id";
	stream << " ";
	stream << std::setw(7) << "#Tokens";
	stream << std::endl;

	for (auto iter = tokenCount.begin(); iter != tokenCount.end(); ++iter) {
	stream << std::right;
	stream << std::setw(8) << /Buffer Id=/iter->first;
	stream << " ";
	stream << std::setw(7) << /#Tokens=/iter->second;
	stream << std::endl;
	}

	fprintf(out, "%s", stream.str().c_str());

	fclose(out);
	return Void();
	}

	hidl_handle BufferHubService::registerToken(const wp<BufferClient>& client) {
	// Find next available token id
	std::lock_guard<std::mutex> lock(mTokenMutex);
	do {
	++mLastTokenId;
	} while (mTokenMap.find(mLastTokenId) != mTokenMap.end());

	std::array<uint8_t, EVP_MAX_MD_SIZE> hmac;
	uint32_t hmacSize = 0U;

	HMAC(/evp_md=/EVP_sha256(), /key=/&mKey, /key_len=/kKeyLen,
	/data=/(uint8_t)&mLastTokenId, /data_len=*/mTokenIdSize,
	/out=/hmac.data(), /out_len=/&hmacSize);

	int numIntsForHMAC = (int)ceil(hmacSize / (double)sizeof(int));
	native_handle_t* handle = native_handle_create(/numFds=/0, /numInts=/1 + numIntsForHMAC);
	handle->data[0] = mLastTokenId;
	// Set all the the bits of last int to 0 since it might not be fully overwritten
	handle->data[numIntsForHMAC] = 0;
	memcpy(&handle->data[1], hmac.data(), hmacSize);

	// returnToken owns the native_handle_t* thus doing lifecycle management
	hidl_handle returnToken;
	returnToken.setTo(handle, /shoudOwn=/true);

	std::vector<uint8_t> hmacVec;
	hmacVec.resize(hmacSize);
	memcpy(hmacVec.data(), hmac.data(), hmacSize);
	mTokenMap.emplace(mLastTokenId, std::pair(hmacVec, client));

	return returnToken;
	}

	void BufferHubService::onClientClosed(const BufferClient* client) {
	removeTokenByClient(client);

	std::lock_guard<std::mutex> lock(mClientSetMutex);
	auto iter = std::find(mClientSet.begin(), mClientSet.end(), client);
	if (iter != mClientSet.end()) {
	mClientSet.erase(iter);
	}
	}

	// Implementation of this function should be consistent with the definition of bufferInfo handle in
	// ui/BufferHubDefs.h.
	hidl_handle BufferHubService::buildBufferInfo(char* bufferInfoStorage, int bufferId,
	uint32_t clientBitMask, uint32_t userMetadataSize,
	int metadataFd, int eventFd) {
	native_handle_t* infoHandle =
	native_handle_init(bufferInfoStorage, BufferHubDefs::kBufferInfoNumFds,
	BufferHubDefs::kBufferInfoNumInts);

	infoHandle->data[0] = metadataFd;
	infoHandle->data[1] = eventFd;
	infoHandle->data[2] = bufferId;
	// Use memcpy to convert to int without missing digit.
	// TOOD(b/121345852): use bit_cast to unpack bufferInfo when C++20 becomes available.
	memcpy(&infoHandle->data[3], &clientBitMask, sizeof(clientBitMask));
	memcpy(&infoHandle->data[4], &userMetadataSize, sizeof(userMetadataSize));

	hidl_handle bufferInfo;
	bufferInfo.setTo(infoHandle, /shouldOwn=/false);

	return bufferInfo;
	}

	void BufferHubService::removeTokenByClient(const BufferClient* client) {
	std::lock_guard<std::mutex> lock(mTokenMutex);
	auto iter = mTokenMap.begin();
	while (iter != mTokenMap.end()) {
	if (iter->second.second == client) {
	auto oldIter = iter;
	++iter;
	mTokenMap.erase(oldIter);
	} else {
	++iter;
	}
	}
	}

	} // namespace implementation
	} // namespace V1_0
	} // namespace bufferhub
	} // namespace frameworks
	} // namespace android