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 UB_MEMORY_TRANSPORT_H
      11              : #define UB_MEMORY_TRANSPORT_H
      12              : #include "virtual_topo.h"
      13              : #include "dev_buffer.h"
      14              : #include "socket.h"
      15              : #include "local_ipc_rma_buffer_v2.h"
      16              : #include "remote_rma_buffer.h"
      17              : #include "base_mem_transport.h"
      18              : 
      19              : namespace Hccl {
      20              : class UbMemoryTransport {
      21              : public:
      22           56 :     MAKE_ENUM(
      23              :         UBTransportStatus, INIT, SOCKET_OK, SEND_MEM_INFO, RECV_MEM_INFO, RECV_MEM_INFO_PROCESS, SEND_NAME, RECV_NAME,
      24              :         CONNECT_FAILED, SOCKET_TIMEOUT, READY)
      25              : 
      26              :     UbMemoryTransport(
      27              :         const std::shared_ptr<Buffer> cclBuffer, const std::shared_ptr<Buffer> aivTagBuffer,
      28              :         const std::shared_ptr<Buffer> aivOffloadTagBuffer, Socket* socket, int32_t deviceLogicId);
      29              : 
      30              :     UbMemoryTransport(const UbMemoryTransport& that) = delete;
      31              :     UbMemoryTransport& operator=(const UbMemoryTransport& other) = delete;
      32              :     HcclResult Init();
      33              :     UBTransportStatus GetStatus();
      34              : 
      35              :     // 给算子开发提供的两个接口
      36              :     LocalIpcRmaBuffer* GetLocMemBuffer(const u32 bufIndex) const;
      37              :     RemoteIpcRmaBuffer* GetRmtMemBuffer(const u32 bufIndex) const;
      38              :     std::string Describe() const;
      39              : 
      40              :     vector<char> GetRmtHandshakeMsg() // 返回握手消息
      41              :     {
      42              :         return rmtHandshakeMsg;
      43              :     }
      44              : 
      45              :     vector<char> GetLocalHandshakeMsg() // 返回握手消息
      46              :     {
      47              :         return localHandshakeMsg;
      48              :     }
      49              : 
      50            1 :     void SetHandshakeMsg(const vector<char>& handshakeMsg) { localHandshakeMsg = handshakeMsg; }
      51              : 
      52              :     AcceleratorState& GetRmtOpAcceState() { return rmtOpAcceState; }
      53              : 
      54              :     AcceleratorState& GetLocalOpAcceState() { return locOpAcceState; }
      55              : 
      56            1 :     void SetLocalOpAcceState(const AcceleratorState& opAcceState) { locOpAcceState = opAcceState; }
      57              : 
      58              :     struct CclBufferInfo {
      59              :         uint64_t addr{0};
      60              :         uint32_t size{0};
      61              :         uint32_t tokenId{0};
      62              :         uint32_t tokenValue{0};
      63              : 
      64              :         void Pack(BinaryStream& binaryStream) const
      65              :         {
      66              :             binaryStream << addr << size << tokenId << tokenValue;
      67              :             HCCL_INFO("Pack Ccl Buffer Info: addr[%llu] size[%u]", addr, size);
      68              :         }
      69              : 
      70              :         void Unpack(BinaryStream& binaryStream)
      71              :         {
      72              :             binaryStream >> addr >> size >> tokenId >> tokenValue;
      73              :             HCCL_INFO("Unpack Ccl Buffer Info: addr[%llu] size[%u]", addr, size);
      74              :         }
      75              :     };
      76              : 
      77              : private:
      78              :     UBTransportStatus ubStatus{UBTransportStatus::INIT};
      79              :     vector<char> rmtHandshakeMsg{0}; // 远端握手消息
      80              :     vector<char> localHandshakeMsg{0};
      81              :     vector<char> recvDataMsg{0};
      82              :     AcceleratorState rmtOpAcceState{AcceleratorState::AIV};
      83              :     AcceleratorState locOpAcceState{AcceleratorState::AIV};
      84              : 
      85              :     CclBufferInfo locCclBufInfo;
      86              :     CclBufferInfo rmtCclBufInfo;
      87              :     uint32_t exchangeDataSize{0};
      88              : 
      89              :     //  新增
      90              :     std::shared_ptr<Buffer> cclBuffer;
      91              :     std::shared_ptr<Buffer> aivTagBuffer;
      92              :     std::shared_ptr<Buffer> aivOffloadTagBuffer;
      93              :     Socket* socket{};
      94              :     int32_t deviceLogicId{0};
      95              : 
      96              :     std::vector<std::unique_ptr<RemoteIpcRmaBuffer>> rmtBufferVec;
      97              :     std::vector<std::unique_ptr<LocalIpcRmaBuffer>> localBufferVec;
      98              :     std::vector<RemoteRmaBuffer*> rmtRmaBufferVec;
      99              : 
     100              :     void HandshakeMsgPack(BinaryStream& binaryStream);
     101              :     void HandshakeMsgUnpack(BinaryStream& binaryStream);
     102              :     UBTransportStatus StateMachine();
     103              :     void ReleaseRes();
     104              :     void SendMemInfo();
     105              :     void RecvMemInfo();
     106              :     void RecvMemProcess();
     107              :     void BufferPack(BinaryStream& binaryStream);
     108              :     void RmtBufferUnpackProc(BinaryStream& binaryStream);
     109              :     void SendName();
     110              :     void RecvName();
     111              : };
     112              : } // namespace Hccl
     113              : 
     114              : #endif