Line data    Source code

       1              : /**
       2              :  * Copyright (c) 2025 Huawei Technologies Co., Ltd.
       3              :  * This program is free software, you can redistribute it and/or modify it under the terms and conditions of
       4              :  * CANN Open Software License Agreement Version 2.0 (the "License").
       5              :  * Please refer to the License for details. You may not use this file except in compliance with the License.
       6              :  * THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
       7              :  * INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
       8              :  * See LICENSE in the root of the software repository for the full text of the License.
       9              :  */
      10              : #ifndef HCCLV2_SOCKET_MANAGER_H
      11              : #define HCCLV2_SOCKET_MANAGER_H
      12              : 
      13              : #include <string>
      14              : #include <vector>
      15              : #include <set>
      16              : #include <unordered_map>
      17              : #include <memory>
      18              : 
      19              : #include "../../unified_platform/resource/socket/socket.h"
      20              : #include "virtual_topo.h"
      21              : #include "socket_config.h"
      22              : #include "env_func.h"
      23              : #include "orion_adapter_hccp.h"
      24              : 
      25              : namespace Hccl {
      26              : 
      27              : using RankIpPortMap = std::unordered_map<u32, std::unordered_map<IpAddress, u32>>;
      28              : using RankIpPortMapPtr = std::shared_ptr<RankIpPortMap>;
      29              : 
      30              : class CommunicatorImpl;
      31              : class SocketManager {
      32              : public:
      33            1 :     SocketManager() = default;
      34              :     SocketManager(const CommunicatorImpl &communicator, u32 localRank, u32 devicePhyId, u32 deviceLogicId,
      35              :                   std::function<shared_ptr<Socket>(IpAddress &localIpAddress, IpAddress &remoteIpAddress,
      36              :                                                    u32 listenPort, SocketHandle socketHandle, const std::string &tag,
      37              :                                                    SocketRole socketRole, NicType nicType)>
      38              :                       socketProducer
      39              :                   = nullptr);
      40              : 
      41              :     SocketManager(u32 localRank, u32 devicePhyId, u32 deviceLogicId, const std::string &socketTag);
      42              : 
      43              :     void SetDeviceServerListenPortMap(const std::unordered_map<u32, std::unordered_map<IpAddress, u32>> &rankListenPortMap);
      44              : 
      45              :     std::unordered_map<u32, std::unordered_map<IpAddress, u32>> GetSubCommDeviceServerListenPortMap(const std::vector<u32> &rankIds) const;
      46              : 
      47              :     u32 GetDeviceListenPort(const u32 &rankId, const IpAddress &ipAddress);
      48              : 
      49              :     void BatchCreateSockets(const vector<LinkData> &links);
      50              :     void ServerListen(const SocketConfig &socketConfig);
      51              :     void ConnectSockets(const SocketConfig &socketConfig);
      52              : 
      53              :     void BatchCreateSockets(const SocketConfig &socketConfig);
      54              : 
      55              :     void ServerInit(PortData &localPort);
      56              : 
      57              :     static void ServerInitAll(NewRankInfo &rankInfo);
      58              : 
      59              :     bool ServerDeInit(PortData &localPort) const;
      60              : 
      61              :     Socket *CreateConnectedSocket(const SocketConfig &socketConfig);
      62              : 
      63              :     Socket *GetConnectedSocket(const SocketConfig &socketConfig) const;
      64              : 
      65              :     bool CheckServerPortListening(const PortData &portData, const uint32_t port) const;
      66              : 
      67              :     void DestroyAll();
      68              : 
      69              :     void AddWhiteList(PortData &localPort, vector<RaSocketWhitelist> &wlistInfoVec) const;
      70              : 
      71              :     bool DelWhiteList(PortData &localPort, vector<RaSocketWhitelist> &wlistInfoVec) const;
      72              : 
      73              :     ~SocketManager();
      74              : 
      75              :     SocketManager(const SocketManager &socketManager) = delete;
      76              : 
      77              :     SocketManager &operator=(const SocketManager &socketManager) = delete;
      78              : 
      79              : private:
      80              :     void PrepareLinkAndServerInit(const SocketConfig &socketConfig);
      81              :     void BatchServerInit(const vector<LinkData> &links);
      82              :     void BatchAddWhiteList(const vector<LinkData> &links);
      83              :     void BatchCreateConnectedSockets(const vector<LinkData> &links);
      84              :     void AddWhiteList(const SocketConfig &socketConfig);
      85              :     const CommunicatorImpl *comm;
      86              :     static std::unordered_map<PortData, shared_ptr<Socket>>& GetServerSocketMap();
      87              :     u32               localRank;
      88              :     u32               devicePhyId;
      89              :     u32               deviceLogicId_;
      90              :     std::unordered_map<u32, std::unordered_map<IpAddress, u32>> rankListenPortMap_{};
      91              :     std::function<shared_ptr<Socket>(IpAddress &localIpAddress, IpAddress &remoteIpAddress, u32 listenPort,
      92              :                                      SocketHandle socketHandle, const std::string &tag, SocketRole socketRole,
      93              :                                      NicType nicType)>
      94              :         socketProducer
      95           12 :         = [](IpAddress &localIpAddress, IpAddress &remoteIpAddress, u32 listenPort, SocketHandle socketHandle,
      96              :              const std::string &tag, SocketRole socketRole, NicType nicType) -> shared_ptr<Socket> {
      97              :         auto tmpSocket = std::make_shared<Socket>(socketHandle, localIpAddress, listenPort, remoteIpAddress, tag,
      98           12 :                                                   socketRole, nicType);
      99           36 :         HCCL_INFO("create socket with role %u", static_cast<u32>(socketRole));
     100           12 :         return tmpSocket;
     101            0 :     };
     102              : 
     103              :     std::unordered_map<SocketConfig, shared_ptr<Socket>> connectedSocketMap;
     104              :     std::unordered_map<PortData, vector<RaSocketWhitelist>> socketWlistMap{};
     105              : 
     106              :     Socket *GetServerListenSocket(const PortData &localPort) const;
     107              :     std::set<LinkData>      availableLinks;
     108              : 
     109              :     std::string socketTag_{};
     110              :     static std::mutex socketLock;
     111              : };
     112              : 
     113              : } // namespace Hccl
     114              : 
     115              : #endif // HCCLV2_SOCKET_MANAGER_H