1 /* Copyright 2017 The TensorFlow Authors. All Rights Reserved. 2 3 Licensed under the Apache License, Version 2.0 (the "License"); 4 you may not use this file except in compliance with the License. 5 You may obtain a copy of the License at 6 7 http://www.apache.org/licenses/LICENSE-2.0 8 9 Unless required by applicable law or agreed to in writing, software 10 distributed under the License is distributed on an "AS IS" BASIS, 11 WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 12 See the License for the specific language governing permissions and 13 limitations under the License. 14 ==============================================================================*/ 15 16 #include "tensorflow/contrib/gdr/gdr_server_lib.h" 17 18 #include "grpc/support/alloc.h" 19 #include "tensorflow/contrib/gdr/gdr_memory_manager.h" 20 #include "tensorflow/contrib/gdr/gdr_rendezvous_mgr.h" 21 #include "tensorflow/contrib/gdr/gdr_worker.h" 22 23 #include "grpc/support/alloc.h" 24 25 namespace tensorflow { 26 27 GdrServer::GdrServer(const ServerDef& server_def, Env* env) 28 : GrpcServer(server_def, env) { 29 string host; 30 string port; 31 for (const auto& job : server_def.cluster().job()) { 32 if (job.name() == server_def.job_name()) { 33 auto iter = job.tasks().find(server_def.task_index()); 34 if (iter != job.tasks().end()) { 35 const std::vector<string> hostname_port = 36 str_util::Split(iter->second, ':'); 37 if (hostname_port.size() == 2) { 38 host = hostname_port[0]; 39 port = hostname_port[1]; 40 } 41 } 42 } 43 } 44 remote_memory_manager_ = std::unique_ptr<RemoteMemoryManager>( 45 CreateRemoteMemoryManager(host, port)); 46 } 47 48 GdrServer::~GdrServer() {} 49 50 Status GdrServer::Init() { 51 RendezvousMgrCreationFunction rendezvous_mgr_func = 52 [this](const WorkerEnv* env) { 53 return new GdrRendezvousMgr(env, remote_memory_manager_.get()); 54 }; 55 WorkerCreationFunction worker_func = [this](WorkerEnv* env) { 56 return std::unique_ptr<GdrWorker>( 57 new GdrWorker(env, remote_memory_manager_.get())); 58 }; 59 TF_RETURN_IF_ERROR( 60 GrpcServer::Init(nullptr, rendezvous_mgr_func, worker_func)); 61 62 return remote_memory_manager_->Init(); 63 } 64 65 Status GdrServer::Start() { 66 { 67 mutex_lock l(mu_); 68 gdr_thread_.reset(worker_env()->env->StartThread( 69 ThreadOptions(), "TF_gdr_service", 70 [this] { remote_memory_manager_->Run(); })); 71 } 72 return GrpcServer::Start(); 73 } 74 75 Status GdrServer::Stop() { 76 TF_RETURN_IF_ERROR(GrpcServer::Stop()); 77 remote_memory_manager_->Stop(); 78 return Status::OK(); 79 } 80 81 Status GdrServer::Join() { 82 { 83 mutex_lock l(mu_); 84 gdr_thread_.reset(); 85 } 86 return GrpcServer::Join(); 87 } 88 89 /* static */ 90 Status GdrServer::Create(const ServerDef& server_def, Env* env, 91 std::unique_ptr<ServerInterface>* out_server) { 92 std::unique_ptr<GdrServer> ret( 93 new GdrServer(server_def, env == nullptr ? Env::Default() : env)); 94 TF_RETURN_IF_ERROR(ret->Init()); 95 *out_server = std::move(ret); 96 return Status::OK(); 97 } 98 99 namespace { 100 101 class GdrServerFactory : public ServerFactory { 102 public: 103 bool AcceptsOptions(const ServerDef& server_def) override { 104 return server_def.protocol() == "grpc+gdr"; 105 } 106 107 Status NewServer(const ServerDef& server_def, 108 std::unique_ptr<ServerInterface>* out_server) override { 109 return GdrServer::Create(server_def, Env::Default(), out_server); 110 } 111 }; 112 113 // Registers a `ServerFactory` for `GdrServer` instances. 114 class GdrServerRegistrar { 115 public: 116 GdrServerRegistrar() { 117 gpr_allocation_functions alloc_fns; 118 memset(&alloc_fns, 0, sizeof(alloc_fns)); 119 alloc_fns.malloc_fn = port::Malloc; 120 alloc_fns.realloc_fn = port::Realloc; 121 alloc_fns.free_fn = port::Free; 122 gpr_set_allocation_functions(alloc_fns); 123 ServerFactory::Register("GDR_SERVER", new GdrServerFactory()); 124 } 125 }; 126 static GdrServerRegistrar registrar; 127 128 } // namespace 129 } // namespace tensorflow 130