/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*- * vim:expandtab:shiftwidth=8:tabstop=8: * * Copyright (C) 2004 Cluster File Systems, Inc. * Author: Eric Barton * * This file is part of Lustre, http://www.lustre.org. * * Lustre is free software; you can redistribute it and/or * modify it under the terms of version 2 of the GNU General Public * License as published by the Free Software Foundation. * * Lustre is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with Lustre; if not, write to the Free Software * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. * */ #ifndef EXPORT_SYMTAB # define EXPORT_SYMTAB #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define DEBUG_SUBSYSTEM S_NAL #include #include #include #include #include #include #include #if CONFIG_SMP # define IBNAL_N_SCHED num_online_cpus() /* # schedulers */ #else # define IBNAL_N_SCHED 1 /* # schedulers */ #endif #define IBNAL_N_CONND 4 /* # connection daemons */ #define IBNAL_MIN_RECONNECT_INTERVAL HZ /* first failed connection retry... */ #define IBNAL_MAX_RECONNECT_INTERVAL (60*HZ) /* ...exponentially increasing to this */ #define IBNAL_MSG_SIZE (4<<10) /* max size of queued messages (inc hdr) */ #define IBNAL_MSG_QUEUE_SIZE 8 /* # messages/RDMAs in-flight */ #define IBNAL_CREDIT_HIGHWATER 6 /* when to eagerly return credits */ #define IBNAL_RETRY 7 /* # times to retry */ #define IBNAL_RNR_RETRY 7 /* */ #define IBNAL_CM_RETRY 7 /* # times to retry connection */ #define IBNAL_FLOW_CONTROL 1 #define IBNAL_RESPONDER_RESOURCES 8 #define IBNAL_NTX 64 /* # tx descs */ #define IBNAL_NTX_NBLK 256 /* # reserved tx descs */ #define IBNAL_PEER_HASH_SIZE 101 /* # peer lists */ #define IBNAL_RESCHED 100 /* # scheduler loops before reschedule */ #define IBNAL_CONCURRENT_PEERS 1000 /* # nodes all talking at once to me */ /* default vals for runtime tunables */ #define IBNAL_IO_TIMEOUT 50 /* default comms timeout (seconds) */ #define IBNAL_LISTENER_TIMEOUT 5 /* default listener timeout (seconds) */ #define IBNAL_BACKLOG 127 /* default listener backlog */ #define IBNAL_PORT 988 /* default listener port */ /************************/ /* derived constants... */ /* TX messages (shared by all connections) */ #define IBNAL_TX_MSGS (IBNAL_NTX + IBNAL_NTX_NBLK) #define IBNAL_TX_MSG_BYTES (IBNAL_TX_MSGS * IBNAL_MSG_SIZE) #define IBNAL_TX_MSG_PAGES ((IBNAL_TX_MSG_BYTES + PAGE_SIZE - 1)/PAGE_SIZE) /* RX messages (per connection) */ #define IBNAL_RX_MSGS IBNAL_MSG_QUEUE_SIZE #define IBNAL_RX_MSG_BYTES (IBNAL_RX_MSGS * IBNAL_MSG_SIZE) #define IBNAL_RX_MSG_PAGES ((IBNAL_RX_MSG_BYTES + PAGE_SIZE - 1)/PAGE_SIZE) /* we may have up to 2 completions per transmit + 1 completion per receive, per connection */ #define IBNAL_CQ_ENTRIES ((2*IBNAL_TX_MSGS) + \ (IBNAL_RX_MSGS * IBNAL_CONCURRENT_PEERS)) #define IBNAL_RDMA_BASE 0x0eeb0000 #define IBNAL_FMR 1 #define IBNAL_CKSUM 1 //#define IBNAL_CALLBACK_CTXT IB_CQ_CALLBACK_PROCESS #define IBNAL_CALLBACK_CTXT IB_CQ_CALLBACK_INTERRUPT typedef struct { int kib_io_timeout; /* comms timeout (seconds) */ int kib_listener_timeout; /* listener's timeout */ int kib_backlog; /* listenter's accept backlog */ int kib_port; /* where the listener listens */ struct ctl_table_header *kib_sysctl; /* sysctl interface */ } kib_tunables_t; typedef struct { int ibp_npages; /* # pages */ int ibp_mapped; /* mapped? */ __u64 ibp_vaddr; /* mapped region vaddr */ __u32 ibp_lkey; /* mapped region lkey */ __u32 ibp_rkey; /* mapped region rkey */ struct ib_mr *ibp_handle; /* mapped region handle */ struct page *ibp_pages[0]; } kib_pages_t; typedef struct { int kib_init; /* initialisation state */ __u64 kib_incarnation; /* which one am I */ int kib_shutdown; /* shut down? */ atomic_t kib_nthreads; /* # live threads */ __u64 kib_svc_id; /* service number I listen on */ tTS_IB_GID kib_svc_gid; /* device/port GID */ __u16 kib_svc_pkey; /* device/port pkey */ ptl_nid_t kib_nid; /* my NID */ struct semaphore kib_nid_mutex; /* serialise NID ops */ struct semaphore kib_listener_signal; /* signal IP listener completion */ struct socket *kib_listener_sock; /* IP listener's socket */ int kib_listener_shutdown; /* ask IP listener to close */ void *kib_listen_handle; /* IB listen handle */ rwlock_t kib_global_lock; /* stabilize peer/conn ops */ struct list_head *kib_peers; /* hash table of all my known peers */ int kib_peer_hash_size; /* size of kib_peers */ atomic_t kib_npeers; /* # peers extant */ atomic_t kib_nconns; /* # connections extant */ struct list_head kib_reaper_conns; /* connections to reap */ wait_queue_head_t kib_reaper_waitq; /* reaper sleeps here */ unsigned long kib_reaper_waketime; /* when reaper will wake */ spinlock_t kib_reaper_lock; /* serialise */ struct list_head kib_connd_peers; /* peers waiting for a connection */ struct list_head kib_connd_acceptq; /* accepted sockets to handle */ wait_queue_head_t kib_connd_waitq; /* connection daemons sleep here */ spinlock_t kib_connd_lock; /* serialise */ wait_queue_head_t kib_sched_waitq; /* schedulers sleep here */ struct list_head kib_sched_txq; /* tx requiring attention */ struct list_head kib_sched_rxq; /* rx requiring attention */ spinlock_t kib_sched_lock; /* serialise */ struct kib_tx *kib_tx_descs; /* all the tx descriptors */ kib_pages_t *kib_tx_pages; /* premapped tx msg pages */ struct list_head kib_idle_txs; /* idle tx descriptors */ struct list_head kib_idle_nblk_txs; /* idle reserved tx descriptors */ wait_queue_head_t kib_idle_tx_waitq; /* block here for tx descriptor */ __u64 kib_next_tx_cookie; /* RDMA completion cookie */ spinlock_t kib_tx_lock; /* serialise */ struct ib_device *kib_device; /* "the" device */ struct ib_device_properties kib_device_props; /* its properties */ int kib_port; /* port on the device */ struct ib_port_properties kib_port_props; /* its properties */ struct ib_pd *kib_pd; /* protection domain */ #if IBNAL_FMR struct ib_fmr_pool *kib_fmr_pool; /* fast memory region pool */ #endif struct ib_cq *kib_cq; /* completion queue */ } kib_data_t; #define IBNAL_INIT_NOTHING 0 #define IBNAL_INIT_DATA 1 #define IBNAL_INIT_LIB 2 #define IBNAL_INIT_PD 3 #define IBNAL_INIT_FMR 4 #define IBNAL_INIT_TXD 5 #define IBNAL_INIT_CQ 6 #define IBNAL_INIT_ALL 7 typedef struct kib_acceptsock /* accepted socket queued for connd */ { struct list_head ibas_list; /* queue for attention */ struct socket *ibas_sock; /* the accepted socket */ } kib_acceptsock_t; /************************************************************************ * IB Wire message format. * These are sent in sender's byte order (i.e. receiver flips). * They may be sent via TCP/IP (service ID,GID,PKEY query/response), * as private data in the connection request/response, or "normally". */ typedef struct kib_svcrsp /* service response */ { __u64 ibsr_svc_id; /* service's id */ __u8 ibsr_svc_gid[16]; /* service's gid */ __u16 ibsr_svc_pkey; /* service's pkey */ } kib_svcrsp_t; typedef struct kib_connparams { __u32 ibcp_queue_depth; } kib_connparams_t; typedef struct { union { struct ib_mr *mr; struct ib_fmr *fmr; } md_handle; __u32 md_lkey; __u32 md_rkey; __u64 md_addr; } kib_md_t; typedef struct { __u32 rd_key; /* remote key */ __u32 rd_nob; /* # of bytes */ __u64 rd_addr; /* remote io vaddr */ } kib_rdma_desc_t; typedef struct { ptl_hdr_t ibim_hdr; /* portals header */ char ibim_payload[0]; /* piggy-backed payload */ } kib_immediate_msg_t; typedef struct { ptl_hdr_t ibrm_hdr; /* portals header */ __u64 ibrm_cookie; /* opaque completion cookie */ kib_rdma_desc_t ibrm_desc; /* where to suck/blow */ } kib_rdma_msg_t; typedef struct { __u64 ibcm_cookie; /* opaque completion cookie */ __u32 ibcm_status; /* completion status */ } kib_completion_msg_t; typedef struct { /* First 2 fields fixed FOR ALL TIME */ __u32 ibm_magic; /* I'm an openibnal message */ __u16 ibm_version; /* this is my version number */ __u8 ibm_type; /* msg type */ __u8 ibm_credits; /* returned credits */ __u32 ibm_nob; /* # bytes in whole message */ __u32 ibm_cksum; /* checksum (0 == no checksum) */ __u64 ibm_srcnid; /* sender's NID */ __u64 ibm_srcstamp; /* sender's incarnation */ __u64 ibm_dstnid; /* destination's NID */ __u64 ibm_dststamp; /* destination's incarnation */ union { kib_svcrsp_t svcrsp; kib_connparams_t connparams; kib_immediate_msg_t immediate; kib_rdma_msg_t rdma; kib_completion_msg_t completion; } ibm_u; } kib_msg_t; #define IBNAL_MSG_MAGIC 0x0be91b91 /* unique magic */ #define IBNAL_MSG_VERSION 2 /* current protocol version */ #define IBNAL_MSG_SVCQRY 0xb0 /* service query */ #define IBNAL_MSG_SVCRSP 0xb1 /* service response */ #define IBNAL_MSG_CONNREQ 0xc0 /* connection request */ #define IBNAL_MSG_CONNACK 0xc1 /* connection acknowledge */ #define IBNAL_MSG_NOOP 0xd0 /* nothing (just credits) */ #define IBNAL_MSG_IMMEDIATE 0xd1 /* portals hdr + payload */ #define IBNAL_MSG_PUT_RDMA 0xd2 /* portals PUT hdr + source rdma desc */ #define IBNAL_MSG_PUT_DONE 0xd3 /* signal PUT rdma completion */ #define IBNAL_MSG_GET_RDMA 0xd4 /* portals GET hdr + sink rdma desc */ #define IBNAL_MSG_GET_DONE 0xd5 /* signal GET rdma completion */ /***********************************************************************/ typedef struct kib_rx /* receive message */ { struct list_head rx_list; /* queue for attention */ struct kib_conn *rx_conn; /* owning conn */ int rx_rdma; /* RDMA completion posted? */ int rx_posted; /* posted? */ __u64 rx_vaddr; /* pre-mapped buffer (hca vaddr) */ kib_msg_t *rx_msg; /* pre-mapped buffer (host vaddr) */ struct ib_receive_param rx_sp; /* receive work item */ struct ib_gather_scatter rx_gl; /* and it's memory */ } kib_rx_t; typedef struct kib_tx /* transmit message */ { struct list_head tx_list; /* queue on idle_txs ibc_tx_queue etc. */ int tx_isnblk; /* I'm reserved for non-blocking sends */ struct kib_conn *tx_conn; /* owning conn */ int tx_mapped; /* mapped for RDMA? */ int tx_sending; /* # tx callbacks outstanding */ int tx_status; /* completion status */ unsigned long tx_deadline; /* completion deadline */ int tx_passive_rdma; /* peer sucks/blows */ int tx_passive_rdma_wait; /* waiting for peer to complete */ __u64 tx_passive_rdma_cookie; /* completion cookie */ lib_msg_t *tx_libmsg[2]; /* lib msgs to finalize on completion */ kib_md_t tx_md; /* RDMA mapping (active/passive) */ __u64 tx_vaddr; /* pre-mapped buffer (hca vaddr) */ kib_msg_t *tx_msg; /* pre-mapped buffer (host vaddr) */ int tx_nsp; /* # send work items */ struct ib_send_param tx_sp[2]; /* send work items... */ struct ib_gather_scatter tx_gl[2]; /* ...and their memory */ } kib_tx_t; #define KIB_TX_UNMAPPED 0 #define KIB_TX_MAPPED 1 #define KIB_TX_MAPPED_FMR 2 typedef struct kib_connreq { /* active connection-in-progress state */ struct kib_conn *cr_conn; kib_msg_t cr_msg; __u64 cr_tid; tTS_IB_GID cr_gid; kib_svcrsp_t cr_svcrsp; struct ib_path_record cr_path; struct ib_cm_active_param cr_connparam; } kib_connreq_t; typedef struct kib_conn { struct kib_peer *ibc_peer; /* owning peer */ struct list_head ibc_list; /* stash on peer's conn list */ __u64 ibc_incarnation; /* which instance of the peer */ atomic_t ibc_refcount; /* # users */ int ibc_state; /* what's happening */ atomic_t ibc_nob; /* # bytes buffered */ int ibc_nsends_posted; /* # uncompleted sends */ int ibc_credits; /* # credits I have */ int ibc_outstanding_credits; /* # credits to return */ struct list_head ibc_tx_queue; /* send queue */ struct list_head ibc_active_txs; /* active tx awaiting completion */ spinlock_t ibc_lock; /* serialise */ kib_rx_t *ibc_rxs; /* the rx descs */ kib_pages_t *ibc_rx_pages; /* premapped rx msg pages */ struct ib_qp *ibc_qp; /* queue pair */ __u32 ibc_qpn; /* queue pair number */ tTS_IB_CM_COMM_ID ibc_comm_id; /* connection ID? */ kib_connreq_t *ibc_connreq; /* connection request state */ } kib_conn_t; #define IBNAL_CONN_INIT_NOTHING 0 /* initial state */ #define IBNAL_CONN_INIT_QP 1 /* ibc_qp set up */ #define IBNAL_CONN_CONNECTING 2 /* started to connect */ #define IBNAL_CONN_ESTABLISHED 3 /* connection established */ #define IBNAL_CONN_DEATHROW 4 /* waiting to be closed */ #define IBNAL_CONN_ZOMBIE 5 /* waiting to be freed */ typedef struct kib_peer { struct list_head ibp_list; /* stash on global peer list */ struct list_head ibp_connd_list; /* schedule on kib_connd_peers */ ptl_nid_t ibp_nid; /* who's on the other end(s) */ __u32 ibp_ip; /* IP to query for peer conn params */ int ibp_port; /* port to qery for peer conn params */ __u64 ibp_incarnation; /* peer's incarnation */ atomic_t ibp_refcount; /* # users */ int ibp_persistence; /* "known" peer refs */ struct list_head ibp_conns; /* all active connections */ struct list_head ibp_tx_queue; /* msgs waiting for a conn */ int ibp_connecting; /* connecting+accepting */ unsigned long ibp_reconnect_time; /* when reconnect may be attempted */ unsigned long ibp_reconnect_interval; /* exponential backoff */ } kib_peer_t; extern lib_nal_t kibnal_lib; extern kib_data_t kibnal_data; extern kib_tunables_t kibnal_tunables; static inline struct list_head * kibnal_nid2peerlist (ptl_nid_t nid) { unsigned int hash = ((unsigned int)nid) % kibnal_data.kib_peer_hash_size; return (&kibnal_data.kib_peers [hash]); } static inline int kibnal_peer_active(kib_peer_t *peer) { /* Am I in the peer hash table? */ return (!list_empty(&peer->ibp_list)); } static inline void kibnal_queue_tx_locked (kib_tx_t *tx, kib_conn_t *conn) { /* CAVEAT EMPTOR: tx takes caller's ref on conn */ LASSERT (tx->tx_nsp > 0); /* work items set up */ LASSERT (tx->tx_conn == NULL); /* only set here */ tx->tx_conn = conn; tx->tx_deadline = jiffies + kibnal_tunables.kib_io_timeout * HZ; list_add_tail(&tx->tx_list, &conn->ibc_tx_queue); } #if 0 static inline void kibnal_show_rdma_attr (kib_conn_t *conn) { struct ib_qp_attribute qp_attr; int rc; memset (&qp_attr, 0, sizeof(qp_attr)); rc = ib_qp_query(conn->ibc_qp, &qp_attr); if (rc != 0) { CERROR ("Can't get qp attrs: %d\n", rc); return; } CWARN ("RDMA CAPABILITY: write %s read %s\n", (qp_attr.valid_fields & TS_IB_QP_ATTRIBUTE_RDMA_ATOMIC_ENABLE) ? (qp_attr.enable_rdma_write ? "enabled" : "disabled") : "invalid", (qp_attr.valid_fields & TS_IB_QP_ATTRIBUTE_RDMA_ATOMIC_ENABLE) ? (qp_attr.enable_rdma_read ? "enabled" : "disabled") : "invalid"); } #endif static inline __u64 kibnal_page2phys (struct page *p) { return page_to_phys(p); } /* CAVEAT EMPTOR: * We rely on tx/rx descriptor alignment to allow us to use the lowest bit * of the work request id as a flag to determine if the completion is for a * transmit or a receive. It seems that that the CQ entry's 'op' field * isn't always set correctly on completions that occur after QP teardown. */ static inline __u64 kibnal_ptr2wreqid (void *ptr, int isrx) { unsigned long lptr = (unsigned long)ptr; LASSERT ((lptr & 1) == 0); return (__u64)(lptr | (isrx ? 1 : 0)); } static inline void * kibnal_wreqid2ptr (__u64 wreqid) { return (void *)(((unsigned long)wreqid) & ~1UL); } static inline int kibnal_wreqid_is_rx (__u64 wreqid) { return (wreqid & 1) != 0; } #if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0)) # define sk_allocation allocation # define sk_data_ready data_ready # define sk_write_space write_space # define sk_user_data user_data # define sk_prot prot # define sk_sndbuf sndbuf # define sk_socket socket # define sk_wmem_queued wmem_queued # define sk_err err # define sk_sleep sleep #endif extern void kibnal_init_msg(kib_msg_t *msg, int type, int body_nob); extern void kibnal_pack_msg(kib_msg_t *msg, int credits, ptl_nid_t dstnid, __u64 dststamp); extern int kibnal_unpack_msg(kib_msg_t *msg, int nob); extern void kibnal_handle_svcqry (struct socket *sock); extern int kibnal_make_svcqry (kib_conn_t *conn); extern void kibnal_free_acceptsock (kib_acceptsock_t *as); extern int kibnal_listener_procint(ctl_table *table, int write, struct file *filp, void *buffer, size_t *lenp); extern kib_peer_t *kibnal_create_peer (ptl_nid_t nid); extern void kibnal_put_peer (kib_peer_t *peer); extern int kibnal_del_peer (ptl_nid_t nid, int single_share); extern kib_peer_t *kibnal_find_peer_locked (ptl_nid_t nid); extern void kibnal_unlink_peer_locked (kib_peer_t *peer); extern int kibnal_close_stale_conns_locked (kib_peer_t *peer, __u64 incarnation); extern kib_conn_t *kibnal_create_conn (void); extern void kibnal_put_conn (kib_conn_t *conn); extern void kibnal_destroy_conn (kib_conn_t *conn); extern int kibnal_alloc_pages (kib_pages_t **pp, int npages, int access); extern void kibnal_free_pages (kib_pages_t *p); extern void kibnal_check_sends (kib_conn_t *conn); extern tTS_IB_CM_CALLBACK_RETURN kibnal_conn_callback (tTS_IB_CM_EVENT event, tTS_IB_CM_COMM_ID cid, void *param, void *arg); extern tTS_IB_CM_CALLBACK_RETURN kibnal_passive_conn_callback (tTS_IB_CM_EVENT event, tTS_IB_CM_COMM_ID cid, void *param, void *arg); extern void kibnal_close_conn_locked (kib_conn_t *conn, int error); extern void kibnal_destroy_conn (kib_conn_t *conn); extern int kibnal_thread_start (int (*fn)(void *arg), void *arg); extern int kibnal_scheduler(void *arg); extern int kibnal_connd (void *arg); extern int kibnal_reaper (void *arg); extern void kibnal_callback (struct ib_cq *cq, struct ib_cq_entry *e, void *arg); extern void kibnal_init_tx_msg (kib_tx_t *tx, int type, int body_nob); extern int kibnal_close_conn (kib_conn_t *conn, int why); extern void kibnal_start_active_rdma (int type, int status, kib_rx_t *rx, lib_msg_t *libmsg, unsigned int niov, struct iovec *iov, ptl_kiov_t *kiov, int offset, int nob);