From d635711daa98be86d4c7fd01499c34f566b54ccb Mon Sep 17 00:00:00 2001 From: André Fabian Silva Delgado Date: Fri, 10 Jun 2016 05:30:17 -0300 Subject: Linux-libre 4.6.2-gnu --- .../staging/lustre/lnet/klnds/o2iblnd/o2iblnd.c | 705 ++++++------- .../staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h | 159 +-- .../staging/lustre/lnet/klnds/o2iblnd/o2iblnd_cb.c | 1048 ++++++++++++-------- .../lustre/lnet/klnds/o2iblnd/o2iblnd_modparams.c | 8 +- 4 files changed, 1072 insertions(+), 848 deletions(-) (limited to 'drivers/staging/lustre/lnet/klnds/o2iblnd') diff --git a/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.c b/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.c index cb74ae731..0d32e6541 100644 --- a/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.c +++ b/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.c @@ -42,15 +42,7 @@ #include #include "o2iblnd.h" -static lnd_t the_o2iblnd = { - .lnd_type = O2IBLND, - .lnd_startup = kiblnd_startup, - .lnd_shutdown = kiblnd_shutdown, - .lnd_ctl = kiblnd_ctl, - .lnd_query = kiblnd_query, - .lnd_send = kiblnd_send, - .lnd_recv = kiblnd_recv, -}; +static lnd_t the_o2iblnd; kib_data_t kiblnd_data; @@ -63,7 +55,7 @@ static __u32 kiblnd_cksum(void *ptr, int nob) sum = ((sum << 1) | (sum >> 31)) + *c++; /* ensure I don't return 0 (== no checksum) */ - return (sum == 0) ? 1 : sum; + return !sum ? 1 : sum; } static char *kiblnd_msgtype2str(int type) @@ -145,7 +137,7 @@ static int kiblnd_unpack_rd(kib_msg_t *msg, int flip) int i; LASSERT(msg->ibm_type == IBLND_MSG_GET_REQ || - msg->ibm_type == IBLND_MSG_PUT_ACK); + msg->ibm_type == IBLND_MSG_PUT_ACK); rd = msg->ibm_type == IBLND_MSG_GET_REQ ? &msg->ibm_u.get.ibgm_rd : @@ -189,8 +181,10 @@ void kiblnd_pack_msg(lnet_ni_t *ni, kib_msg_t *msg, int version, { kib_net_t *net = ni->ni_data; - /* CAVEAT EMPTOR! all message fields not set here should have been - * initialised previously. */ + /* + * CAVEAT EMPTOR! all message fields not set here should have been + * initialised previously. + */ msg->ibm_magic = IBLND_MSG_MAGIC; msg->ibm_version = version; /* ibm_type */ @@ -249,11 +243,13 @@ int kiblnd_unpack_msg(kib_msg_t *msg, int nob) return -EPROTO; } - /* checksum must be computed with ibm_cksum zero and BEFORE anything - * gets flipped */ + /* + * checksum must be computed with ibm_cksum zero and BEFORE anything + * gets flipped + */ msg_cksum = flip ? __swab32(msg->ibm_cksum) : msg->ibm_cksum; msg->ibm_cksum = 0; - if (msg_cksum != 0 && + if (msg_cksum && msg_cksum != kiblnd_cksum(msg, msg_nob)) { CERROR("Bad checksum\n"); return -EPROTO; @@ -326,21 +322,21 @@ int kiblnd_create_peer(lnet_ni_t *ni, kib_peer_t **peerp, lnet_nid_t nid) int cpt = lnet_cpt_of_nid(nid); unsigned long flags; - LASSERT(net != NULL); + LASSERT(net); LASSERT(nid != LNET_NID_ANY); LIBCFS_CPT_ALLOC(peer, lnet_cpt_table(), cpt, sizeof(*peer)); - if (peer == NULL) { + if (!peer) { CERROR("Cannot allocate peer\n"); return -ENOMEM; } - memset(peer, 0, sizeof(*peer)); /* zero flags etc */ - peer->ibp_ni = ni; peer->ibp_nid = nid; peer->ibp_error = 0; peer->ibp_last_alive = 0; + peer->ibp_max_frags = IBLND_CFG_RDMA_FRAGS; + peer->ibp_queue_depth = *kiblnd_tunables.kib_peertxcredits; atomic_set(&peer->ibp_refcount, 1); /* 1 ref for caller */ INIT_LIST_HEAD(&peer->ibp_list); /* not in the peer table yet */ @@ -350,7 +346,7 @@ int kiblnd_create_peer(lnet_ni_t *ni, kib_peer_t **peerp, lnet_nid_t nid) write_lock_irqsave(&kiblnd_data.kib_global_lock, flags); /* always called with a ref on ni, which prevents ni being shutdown */ - LASSERT(net->ibn_shutdown == 0); + LASSERT(!net->ibn_shutdown); /* npeers only grows with the global lock held */ atomic_inc(&net->ibn_npeers); @@ -365,38 +361,36 @@ void kiblnd_destroy_peer(kib_peer_t *peer) { kib_net_t *net = peer->ibp_ni->ni_data; - LASSERT(net != NULL); - LASSERT(atomic_read(&peer->ibp_refcount) == 0); + LASSERT(net); + LASSERT(!atomic_read(&peer->ibp_refcount)); LASSERT(!kiblnd_peer_active(peer)); - LASSERT(peer->ibp_connecting == 0); - LASSERT(peer->ibp_accepting == 0); - LASSERT(list_empty(&peer->ibp_conns)); + LASSERT(kiblnd_peer_idle(peer)); LASSERT(list_empty(&peer->ibp_tx_queue)); LIBCFS_FREE(peer, sizeof(*peer)); - /* NB a peer's connections keep a reference on their peer until + /* + * NB a peer's connections keep a reference on their peer until * they are destroyed, so we can be assured that _all_ state to do * with this peer has been cleaned up when its refcount drops to - * zero. */ + * zero. + */ atomic_dec(&net->ibn_npeers); } kib_peer_t *kiblnd_find_peer_locked(lnet_nid_t nid) { - /* the caller is responsible for accounting the additional reference - * that this creates */ + /* + * the caller is responsible for accounting the additional reference + * that this creates + */ struct list_head *peer_list = kiblnd_nid2peerlist(nid); struct list_head *tmp; kib_peer_t *peer; list_for_each(tmp, peer_list) { - peer = list_entry(tmp, kib_peer_t, ibp_list); - - LASSERT(peer->ibp_connecting > 0 || /* creating conns */ - peer->ibp_accepting > 0 || - !list_empty(&peer->ibp_conns)); /* active conn */ + LASSERT(!kiblnd_peer_idle(peer)); if (peer->ibp_nid != nid) continue; @@ -431,13 +425,9 @@ static int kiblnd_get_peer_info(lnet_ni_t *ni, int index, read_lock_irqsave(&kiblnd_data.kib_global_lock, flags); for (i = 0; i < kiblnd_data.kib_peer_hash_size; i++) { - list_for_each(ptmp, &kiblnd_data.kib_peers[i]) { - peer = list_entry(ptmp, kib_peer_t, ibp_list); - LASSERT(peer->ibp_connecting > 0 || - peer->ibp_accepting > 0 || - !list_empty(&peer->ibp_conns)); + LASSERT(!kiblnd_peer_idle(peer)); if (peer->ibp_ni != ni) continue; @@ -474,8 +464,10 @@ static void kiblnd_del_peer_locked(kib_peer_t *peer) } /* NB closing peer's last conn unlinked it. */ } - /* NB peer now unlinked; might even be freed if the peer table had the - * last ref on it. */ + /* + * NB peer now unlinked; might even be freed if the peer table had the + * last ref on it. + */ } static int kiblnd_del_peer(lnet_ni_t *ni, lnet_nid_t nid) @@ -493,7 +485,8 @@ static int kiblnd_del_peer(lnet_ni_t *ni, lnet_nid_t nid) write_lock_irqsave(&kiblnd_data.kib_global_lock, flags); if (nid != LNET_NID_ANY) { - lo = hi = kiblnd_nid2peerlist(nid) - kiblnd_data.kib_peers; + lo = kiblnd_nid2peerlist(nid) - kiblnd_data.kib_peers; + hi = kiblnd_nid2peerlist(nid) - kiblnd_data.kib_peers; } else { lo = 0; hi = kiblnd_data.kib_peer_hash_size - 1; @@ -502,9 +495,7 @@ static int kiblnd_del_peer(lnet_ni_t *ni, lnet_nid_t nid) for (i = lo; i <= hi; i++) { list_for_each_safe(ptmp, pnxt, &kiblnd_data.kib_peers[i]) { peer = list_entry(ptmp, kib_peer_t, ibp_list); - LASSERT(peer->ibp_connecting > 0 || - peer->ibp_accepting > 0 || - !list_empty(&peer->ibp_conns)); + LASSERT(!kiblnd_peer_idle(peer)); if (peer->ibp_ni != ni) continue; @@ -516,7 +507,7 @@ static int kiblnd_del_peer(lnet_ni_t *ni, lnet_nid_t nid) LASSERT(list_empty(&peer->ibp_conns)); list_splice_init(&peer->ibp_tx_queue, - &zombies); + &zombies); } kiblnd_del_peer_locked(peer); @@ -544,11 +535,8 @@ static kib_conn_t *kiblnd_get_conn_by_idx(lnet_ni_t *ni, int index) for (i = 0; i < kiblnd_data.kib_peer_hash_size; i++) { list_for_each(ptmp, &kiblnd_data.kib_peers[i]) { - peer = list_entry(ptmp, kib_peer_t, ibp_list); - LASSERT(peer->ibp_connecting > 0 || - peer->ibp_accepting > 0 || - !list_empty(&peer->ibp_conns)); + LASSERT(!kiblnd_peer_idle(peer)); if (peer->ibp_ni != ni) continue; @@ -558,7 +546,7 @@ static kib_conn_t *kiblnd_get_conn_by_idx(lnet_ni_t *ni, int index) continue; conn = list_entry(ctmp, kib_conn_t, - ibc_list); + ibc_list); kiblnd_conn_addref(conn); read_unlock_irqrestore( &kiblnd_data.kib_global_lock, @@ -597,12 +585,12 @@ static void kiblnd_setup_mtu_locked(struct rdma_cm_id *cmid) int mtu; /* XXX There is no path record for iWARP, set by netdev->change_mtu? */ - if (cmid->route.path_rec == NULL) + if (!cmid->route.path_rec) return; mtu = kiblnd_translate_mtu(*kiblnd_tunables.kib_ib_mtu); LASSERT(mtu >= 0); - if (mtu != 0) + if (mtu) cmid->route.path_rec->mtu = mtu; } @@ -619,13 +607,13 @@ static int kiblnd_get_completion_vector(kib_conn_t *conn, int cpt) return 0; mask = cfs_cpt_cpumask(lnet_cpt_table(), cpt); - if (mask == NULL) + if (!mask) return 0; /* hash NID to CPU id in this partition... */ off = do_div(nid, cpumask_weight(mask)); for_each_cpu(i, mask) { - if (off-- == 0) + if (!off--) return i % vectors; } @@ -634,15 +622,17 @@ static int kiblnd_get_completion_vector(kib_conn_t *conn, int cpt) } kib_conn_t *kiblnd_create_conn(kib_peer_t *peer, struct rdma_cm_id *cmid, - int state, int version) + int state, int version) { - /* CAVEAT EMPTOR: + /* + * CAVEAT EMPTOR: * If the new conn is created successfully it takes over the caller's * ref on 'peer'. It also "owns" 'cmid' and destroys it when it itself * is destroyed. On failure, the caller's ref on 'peer' remains and * she must dispose of 'cmid'. (Actually I'd block forever if I tried * to destroy 'cmid' here since I'm called from the CM which still has - * its ref on 'cmid'). */ + * its ref on 'cmid'). + */ rwlock_t *glock = &kiblnd_data.kib_global_lock; kib_net_t *net = peer->ibp_ni->ni_data; kib_dev_t *dev; @@ -656,7 +646,7 @@ kib_conn_t *kiblnd_create_conn(kib_peer_t *peer, struct rdma_cm_id *cmid, int rc; int i; - LASSERT(net != NULL); + LASSERT(net); LASSERT(!in_interrupt()); dev = net->ibn_dev; @@ -668,14 +658,14 @@ kib_conn_t *kiblnd_create_conn(kib_peer_t *peer, struct rdma_cm_id *cmid, LIBCFS_CPT_ALLOC(init_qp_attr, lnet_cpt_table(), cpt, sizeof(*init_qp_attr)); - if (init_qp_attr == NULL) { + if (!init_qp_attr) { CERROR("Can't allocate qp_attr for %s\n", libcfs_nid2str(peer->ibp_nid)); goto failed_0; } LIBCFS_CPT_ALLOC(conn, lnet_cpt_table(), cpt, sizeof(*conn)); - if (conn == NULL) { + if (!conn) { CERROR("Can't allocate connection for %s\n", libcfs_nid2str(peer->ibp_nid)); goto failed_1; @@ -686,6 +676,8 @@ kib_conn_t *kiblnd_create_conn(kib_peer_t *peer, struct rdma_cm_id *cmid, conn->ibc_peer = peer; /* I take the caller's ref */ cmid->context = conn; /* for future CM callbacks */ conn->ibc_cmid = cmid; + conn->ibc_max_frags = peer->ibp_max_frags; + conn->ibc_queue_depth = peer->ibp_queue_depth; INIT_LIST_HEAD(&conn->ibc_early_rxs); INIT_LIST_HEAD(&conn->ibc_tx_noops); @@ -697,7 +689,7 @@ kib_conn_t *kiblnd_create_conn(kib_peer_t *peer, struct rdma_cm_id *cmid, LIBCFS_CPT_ALLOC(conn->ibc_connvars, lnet_cpt_table(), cpt, sizeof(*conn->ibc_connvars)); - if (conn->ibc_connvars == NULL) { + if (!conn->ibc_connvars) { CERROR("Can't allocate in-progress connection state\n"); goto failed_2; } @@ -731,42 +723,42 @@ kib_conn_t *kiblnd_create_conn(kib_peer_t *peer, struct rdma_cm_id *cmid, write_unlock_irqrestore(glock, flags); LIBCFS_CPT_ALLOC(conn->ibc_rxs, lnet_cpt_table(), cpt, - IBLND_RX_MSGS(version) * sizeof(kib_rx_t)); - if (conn->ibc_rxs == NULL) { + IBLND_RX_MSGS(conn) * sizeof(kib_rx_t)); + if (!conn->ibc_rxs) { CERROR("Cannot allocate RX buffers\n"); goto failed_2; } rc = kiblnd_alloc_pages(&conn->ibc_rx_pages, cpt, - IBLND_RX_MSG_PAGES(version)); - if (rc != 0) + IBLND_RX_MSG_PAGES(conn)); + if (rc) goto failed_2; kiblnd_map_rx_descs(conn); - cq_attr.cqe = IBLND_CQ_ENTRIES(version); + cq_attr.cqe = IBLND_CQ_ENTRIES(conn); cq_attr.comp_vector = kiblnd_get_completion_vector(conn, cpt); cq = ib_create_cq(cmid->device, kiblnd_cq_completion, kiblnd_cq_event, conn, &cq_attr); if (IS_ERR(cq)) { - CERROR("Can't create CQ: %ld, cqe: %d\n", - PTR_ERR(cq), IBLND_CQ_ENTRIES(version)); + CERROR("Failed to create CQ with %d CQEs: %ld\n", + IBLND_CQ_ENTRIES(conn), PTR_ERR(cq)); goto failed_2; } conn->ibc_cq = cq; rc = ib_req_notify_cq(cq, IB_CQ_NEXT_COMP); - if (rc != 0) { - CERROR("Can't request completion notificiation: %d\n", rc); + if (rc) { + CERROR("Can't request completion notification: %d\n", rc); goto failed_2; } init_qp_attr->event_handler = kiblnd_qp_event; init_qp_attr->qp_context = conn; - init_qp_attr->cap.max_send_wr = IBLND_SEND_WRS(version); - init_qp_attr->cap.max_recv_wr = IBLND_RECV_WRS(version); + init_qp_attr->cap.max_send_wr = IBLND_SEND_WRS(conn); + init_qp_attr->cap.max_recv_wr = IBLND_RECV_WRS(conn); init_qp_attr->cap.max_send_sge = 1; init_qp_attr->cap.max_recv_sge = 1; init_qp_attr->sq_sig_type = IB_SIGNAL_REQ_WR; @@ -777,7 +769,7 @@ kib_conn_t *kiblnd_create_conn(kib_peer_t *peer, struct rdma_cm_id *cmid, conn->ibc_sched = sched; rc = rdma_create_qp(cmid, conn->ibc_hdev->ibh_pd, init_qp_attr); - if (rc != 0) { + if (rc) { CERROR("Can't create QP: %d, send_wr: %d, recv_wr: %d\n", rc, init_qp_attr->cap.max_send_wr, init_qp_attr->cap.max_recv_wr); @@ -787,33 +779,37 @@ kib_conn_t *kiblnd_create_conn(kib_peer_t *peer, struct rdma_cm_id *cmid, LIBCFS_FREE(init_qp_attr, sizeof(*init_qp_attr)); /* 1 ref for caller and each rxmsg */ - atomic_set(&conn->ibc_refcount, 1 + IBLND_RX_MSGS(version)); - conn->ibc_nrx = IBLND_RX_MSGS(version); + atomic_set(&conn->ibc_refcount, 1 + IBLND_RX_MSGS(conn)); + conn->ibc_nrx = IBLND_RX_MSGS(conn); /* post receives */ - for (i = 0; i < IBLND_RX_MSGS(version); i++) { + for (i = 0; i < IBLND_RX_MSGS(conn); i++) { rc = kiblnd_post_rx(&conn->ibc_rxs[i], IBLND_POSTRX_NO_CREDIT); - if (rc != 0) { + if (rc) { CERROR("Can't post rxmsg: %d\n", rc); /* Make posted receives complete */ kiblnd_abort_receives(conn); - /* correct # of posted buffers - * NB locking needed now I'm racing with completion */ + /* + * correct # of posted buffers + * NB locking needed now I'm racing with completion + */ spin_lock_irqsave(&sched->ibs_lock, flags); - conn->ibc_nrx -= IBLND_RX_MSGS(version) - i; + conn->ibc_nrx -= IBLND_RX_MSGS(conn) - i; spin_unlock_irqrestore(&sched->ibs_lock, flags); - /* cmid will be destroyed by CM(ofed) after cm_callback + /* + * cmid will be destroyed by CM(ofed) after cm_callback * returned, so we can't refer it anymore - * (by kiblnd_connd()->kiblnd_destroy_conn) */ + * (by kiblnd_connd()->kiblnd_destroy_conn) + */ rdma_destroy_qp(conn->ibc_cmid); conn->ibc_cmid = NULL; /* Drop my own and unused rxbuffer refcounts */ - while (i++ <= IBLND_RX_MSGS(version)) + while (i++ <= IBLND_RX_MSGS(conn)) kiblnd_conn_decref(conn); return NULL; @@ -822,7 +818,7 @@ kib_conn_t *kiblnd_create_conn(kib_peer_t *peer, struct rdma_cm_id *cmid, /* Init successful! */ LASSERT(state == IBLND_CONN_ACTIVE_CONNECT || - state == IBLND_CONN_PASSIVE_WAIT); + state == IBLND_CONN_PASSIVE_WAIT); conn->ibc_state = state; /* 1 more conn */ @@ -830,29 +826,29 @@ kib_conn_t *kiblnd_create_conn(kib_peer_t *peer, struct rdma_cm_id *cmid, return conn; failed_2: - kiblnd_destroy_conn(conn); + kiblnd_destroy_conn(conn, true); failed_1: LIBCFS_FREE(init_qp_attr, sizeof(*init_qp_attr)); failed_0: return NULL; } -void kiblnd_destroy_conn(kib_conn_t *conn) +void kiblnd_destroy_conn(kib_conn_t *conn, bool free_conn) { struct rdma_cm_id *cmid = conn->ibc_cmid; kib_peer_t *peer = conn->ibc_peer; int rc; LASSERT(!in_interrupt()); - LASSERT(atomic_read(&conn->ibc_refcount) == 0); + LASSERT(!atomic_read(&conn->ibc_refcount)); LASSERT(list_empty(&conn->ibc_early_rxs)); LASSERT(list_empty(&conn->ibc_tx_noops)); LASSERT(list_empty(&conn->ibc_tx_queue)); LASSERT(list_empty(&conn->ibc_tx_queue_rsrvd)); LASSERT(list_empty(&conn->ibc_tx_queue_nocred)); LASSERT(list_empty(&conn->ibc_active_txs)); - LASSERT(conn->ibc_noops_posted == 0); - LASSERT(conn->ibc_nsends_posted == 0); + LASSERT(!conn->ibc_noops_posted); + LASSERT(!conn->ibc_nsends_posted); switch (conn->ibc_state) { default: @@ -861,7 +857,7 @@ void kiblnd_destroy_conn(kib_conn_t *conn) case IBLND_CONN_DISCONNECTED: /* connvars should have been freed already */ - LASSERT(conn->ibc_connvars == NULL); + LASSERT(!conn->ibc_connvars); break; case IBLND_CONN_INIT: @@ -869,28 +865,27 @@ void kiblnd_destroy_conn(kib_conn_t *conn) } /* conn->ibc_cmid might be destroyed by CM already */ - if (cmid != NULL && cmid->qp != NULL) + if (cmid && cmid->qp) rdma_destroy_qp(cmid); - if (conn->ibc_cq != NULL) { + if (conn->ibc_cq) { rc = ib_destroy_cq(conn->ibc_cq); - if (rc != 0) + if (rc) CWARN("Error destroying CQ: %d\n", rc); } - if (conn->ibc_rx_pages != NULL) + if (conn->ibc_rx_pages) kiblnd_unmap_rx_descs(conn); - if (conn->ibc_rxs != NULL) { + if (conn->ibc_rxs) { LIBCFS_FREE(conn->ibc_rxs, - IBLND_RX_MSGS(conn->ibc_version) - * sizeof(kib_rx_t)); + IBLND_RX_MSGS(conn) * sizeof(kib_rx_t)); } - if (conn->ibc_connvars != NULL) + if (conn->ibc_connvars) LIBCFS_FREE(conn->ibc_connvars, sizeof(*conn->ibc_connvars)); - if (conn->ibc_hdev != NULL) + if (conn->ibc_hdev) kiblnd_hdev_decref(conn->ibc_hdev); /* See CAVEAT EMPTOR above in kiblnd_create_conn */ @@ -927,7 +922,7 @@ int kiblnd_close_peer_conns_locked(kib_peer_t *peer, int why) } int kiblnd_close_stale_conns_locked(kib_peer_t *peer, - int version, __u64 incarnation) + int version, __u64 incarnation) { kib_conn_t *conn; struct list_head *ctmp; @@ -967,20 +962,18 @@ static int kiblnd_close_matching_conns(lnet_ni_t *ni, lnet_nid_t nid) write_lock_irqsave(&kiblnd_data.kib_global_lock, flags); - if (nid != LNET_NID_ANY) - lo = hi = kiblnd_nid2peerlist(nid) - kiblnd_data.kib_peers; - else { + if (nid != LNET_NID_ANY) { + lo = kiblnd_nid2peerlist(nid) - kiblnd_data.kib_peers; + hi = kiblnd_nid2peerlist(nid) - kiblnd_data.kib_peers; + } else { lo = 0; hi = kiblnd_data.kib_peer_hash_size - 1; } for (i = lo; i <= hi; i++) { list_for_each_safe(ptmp, pnxt, &kiblnd_data.kib_peers[i]) { - peer = list_entry(ptmp, kib_peer_t, ibp_list); - LASSERT(peer->ibp_connecting > 0 || - peer->ibp_accepting > 0 || - !list_empty(&peer->ibp_conns)); + LASSERT(!kiblnd_peer_idle(peer)); if (peer->ibp_ni != ni) continue; @@ -998,10 +991,10 @@ static int kiblnd_close_matching_conns(lnet_ni_t *ni, lnet_nid_t nid) if (nid == LNET_NID_ANY) return 0; - return (count == 0) ? -ENOENT : 0; + return !count ? -ENOENT : 0; } -int kiblnd_ctl(lnet_ni_t *ni, unsigned int cmd, void *arg) +static int kiblnd_ctl(lnet_ni_t *ni, unsigned int cmd, void *arg) { struct libcfs_ioctl_data *data = arg; int rc = -EINVAL; @@ -1027,14 +1020,14 @@ int kiblnd_ctl(lnet_ni_t *ni, unsigned int cmd, void *arg) rc = 0; conn = kiblnd_get_conn_by_idx(ni, data->ioc_count); - if (conn == NULL) { + if (!conn) { rc = -ENOENT; break; } - LASSERT(conn->ibc_cmid != NULL); + LASSERT(conn->ibc_cmid); data->ioc_nid = conn->ibc_peer->ibp_nid; - if (conn->ibc_cmid->route.path_rec == NULL) + if (!conn->ibc_cmid->route.path_rec) data->ioc_u32[0] = 0; /* iWarp has no path MTU */ else data->ioc_u32[0] = @@ -1054,7 +1047,7 @@ int kiblnd_ctl(lnet_ni_t *ni, unsigned int cmd, void *arg) return rc; } -void kiblnd_query(lnet_ni_t *ni, lnet_nid_t nid, unsigned long *when) +static void kiblnd_query(lnet_ni_t *ni, lnet_nid_t nid, unsigned long *when) { unsigned long last_alive = 0; unsigned long now = cfs_time_current(); @@ -1065,21 +1058,19 @@ void kiblnd_query(lnet_ni_t *ni, lnet_nid_t nid, unsigned long *when) read_lock_irqsave(glock, flags); peer = kiblnd_find_peer_locked(nid); - if (peer != NULL) { - LASSERT(peer->ibp_connecting > 0 || /* creating conns */ - peer->ibp_accepting > 0 || - !list_empty(&peer->ibp_conns)); /* active conn */ + if (peer) last_alive = peer->ibp_last_alive; - } read_unlock_irqrestore(glock, flags); - if (last_alive != 0) + if (last_alive) *when = last_alive; - /* peer is not persistent in hash, trigger peer creation - * and connection establishment with a NULL tx */ - if (peer == NULL) + /* + * peer is not persistent in hash, trigger peer creation + * and connection establishment with a NULL tx + */ + if (!peer) kiblnd_launch_tx(ni, NULL, nid); CDEBUG(D_NET, "Peer %s %p, alive %ld secs ago\n", @@ -1087,13 +1078,13 @@ void kiblnd_query(lnet_ni_t *ni, lnet_nid_t nid, unsigned long *when) last_alive ? cfs_duration_sec(now - last_alive) : -1); } -void kiblnd_free_pages(kib_pages_t *p) +static void kiblnd_free_pages(kib_pages_t *p) { int npages = p->ibp_npages; int i; for (i = 0; i < npages; i++) { - if (p->ibp_pages[i] != NULL) + if (p->ibp_pages[i]) __free_page(p->ibp_pages[i]); } @@ -1107,7 +1098,7 @@ int kiblnd_alloc_pages(kib_pages_t **pp, int cpt, int npages) LIBCFS_CPT_ALLOC(p, lnet_cpt_table(), cpt, offsetof(kib_pages_t, ibp_pages[npages])); - if (p == NULL) { + if (!p) { CERROR("Can't allocate descriptor for %d pages\n", npages); return -ENOMEM; } @@ -1119,7 +1110,7 @@ int kiblnd_alloc_pages(kib_pages_t **pp, int cpt, int npages) p->ibp_pages[i] = alloc_pages_node( cfs_cpt_spread_node(lnet_cpt_table(), cpt), GFP_NOFS, 0); - if (p->ibp_pages[i] == NULL) { + if (!p->ibp_pages[i]) { CERROR("Can't allocate page %d of %d\n", i, npages); kiblnd_free_pages(p); return -ENOMEM; @@ -1135,10 +1126,10 @@ void kiblnd_unmap_rx_descs(kib_conn_t *conn) kib_rx_t *rx; int i; - LASSERT(conn->ibc_rxs != NULL); - LASSERT(conn->ibc_hdev != NULL); + LASSERT(conn->ibc_rxs); + LASSERT(conn->ibc_hdev); - for (i = 0; i < IBLND_RX_MSGS(conn->ibc_version); i++) { + for (i = 0; i < IBLND_RX_MSGS(conn); i++) { rx = &conn->ibc_rxs[i]; LASSERT(rx->rx_nob >= 0); /* not posted */ @@ -1162,7 +1153,7 @@ void kiblnd_map_rx_descs(kib_conn_t *conn) int ipg; int i; - for (pg_off = ipg = i = 0; i < IBLND_RX_MSGS(conn->ibc_version); i++) { + for (pg_off = ipg = i = 0; i < IBLND_RX_MSGS(conn); i++) { pg = conn->ibc_rx_pages->ibp_pages[ipg]; rx = &conn->ibc_rxs[i]; @@ -1174,7 +1165,7 @@ void kiblnd_map_rx_descs(kib_conn_t *conn) IBLND_MSG_SIZE, DMA_FROM_DEVICE); LASSERT(!kiblnd_dma_mapping_error(conn->ibc_hdev->ibh_ibdev, - rx->rx_msgaddr)); + rx->rx_msgaddr)); KIBLND_UNMAP_ADDR_SET(rx, rx_msgunmap, rx->rx_msgaddr); CDEBUG(D_NET, "rx %d: %p %#llx(%#llx)\n", @@ -1187,7 +1178,7 @@ void kiblnd_map_rx_descs(kib_conn_t *conn) if (pg_off == PAGE_SIZE) { pg_off = 0; ipg++; - LASSERT(ipg <= IBLND_RX_MSG_PAGES(conn->ibc_version)); + LASSERT(ipg <= IBLND_RX_MSG_PAGES(conn)); } } } @@ -1198,9 +1189,9 @@ static void kiblnd_unmap_tx_pool(kib_tx_pool_t *tpo) kib_tx_t *tx; int i; - LASSERT(tpo->tpo_pool.po_allocated == 0); + LASSERT(!tpo->tpo_pool.po_allocated); - if (hdev == NULL) + if (!hdev) return; for (i = 0; i < tpo->tpo_pool.po_size; i++) { @@ -1224,9 +1215,10 @@ static kib_hca_dev_t *kiblnd_current_hdev(kib_dev_t *dev) read_lock_irqsave(&kiblnd_data.kib_global_lock, flags); while (dev->ibd_failover) { read_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags); - if (i++ % 50 == 0) + if (!(i++ % 50)) CDEBUG(D_NET, "%s: Wait for failover\n", dev->ibd_ifname); + set_current_state(TASK_INTERRUPTIBLE); schedule_timeout(cfs_time_seconds(1) / 100); read_lock_irqsave(&kiblnd_data.kib_global_lock, flags); @@ -1252,7 +1244,7 @@ static void kiblnd_map_tx_pool(kib_tx_pool_t *tpo) int ipage; int i; - LASSERT(net != NULL); + LASSERT(net); dev = net->ibn_dev; @@ -1260,7 +1252,7 @@ static void kiblnd_map_tx_pool(kib_tx_pool_t *tpo) CLASSERT(IBLND_MSG_SIZE <= PAGE_SIZE); /* No fancy arithmetic when we do the buffer calculations */ - CLASSERT(PAGE_SIZE % IBLND_MSG_SIZE == 0); + CLASSERT(!(PAGE_SIZE % IBLND_MSG_SIZE)); tpo->tpo_hdev = kiblnd_current_hdev(dev); @@ -1275,7 +1267,7 @@ static void kiblnd_map_tx_pool(kib_tx_pool_t *tpo) tpo->tpo_hdev->ibh_ibdev, tx->tx_msg, IBLND_MSG_SIZE, DMA_TO_DEVICE); LASSERT(!kiblnd_dma_mapping_error(tpo->tpo_hdev->ibh_ibdev, - tx->tx_msgaddr)); + tx->tx_msgaddr)); KIBLND_UNMAP_ADDR_SET(tx, tx_msgunmap, tx->tx_msgaddr); list_add(&tx->tx_list, &pool->po_free_list); @@ -1291,68 +1283,32 @@ static void kiblnd_map_tx_pool(kib_tx_pool_t *tpo) } } -struct ib_mr *kiblnd_find_dma_mr(kib_hca_dev_t *hdev, __u64 addr, __u64 size) +struct ib_mr *kiblnd_find_rd_dma_mr(kib_hca_dev_t *hdev, kib_rdma_desc_t *rd, + int negotiated_nfrags) { - __u64 index; - - LASSERT(hdev->ibh_mrs[0] != NULL); - - if (hdev->ibh_nmrs == 1) - return hdev->ibh_mrs[0]; - - index = addr >> hdev->ibh_mr_shift; + __u16 nfrags = (negotiated_nfrags != -1) ? + negotiated_nfrags : *kiblnd_tunables.kib_map_on_demand; - if (index < hdev->ibh_nmrs && - index == ((addr + size - 1) >> hdev->ibh_mr_shift)) - return hdev->ibh_mrs[index]; - - return NULL; -} - -struct ib_mr *kiblnd_find_rd_dma_mr(kib_hca_dev_t *hdev, kib_rdma_desc_t *rd) -{ - struct ib_mr *prev_mr; - struct ib_mr *mr; - int i; - - LASSERT(hdev->ibh_mrs[0] != NULL); + LASSERT(hdev->ibh_mrs); if (*kiblnd_tunables.kib_map_on_demand > 0 && - *kiblnd_tunables.kib_map_on_demand <= rd->rd_nfrags) + nfrags <= rd->rd_nfrags) return NULL; - if (hdev->ibh_nmrs == 1) - return hdev->ibh_mrs[0]; - - for (i = 0, mr = prev_mr = NULL; - i < rd->rd_nfrags; i++) { - mr = kiblnd_find_dma_mr(hdev, - rd->rd_frags[i].rf_addr, - rd->rd_frags[i].rf_nob); - if (prev_mr == NULL) - prev_mr = mr; - - if (mr == NULL || prev_mr != mr) { - /* Can't covered by one single MR */ - mr = NULL; - break; - } - } - - return mr; + return hdev->ibh_mrs; } static void kiblnd_destroy_fmr_pool(kib_fmr_pool_t *pool) { - LASSERT(pool->fpo_map_count == 0); + LASSERT(!pool->fpo_map_count); - if (pool->fpo_fmr_pool != NULL) + if (pool->fpo_fmr_pool) ib_destroy_fmr_pool(pool->fpo_fmr_pool); - if (pool->fpo_hdev != NULL) + if (pool->fpo_hdev) kiblnd_hdev_decref(pool->fpo_hdev); - LIBCFS_FREE(pool, sizeof(kib_fmr_pool_t)); + LIBCFS_FREE(pool, sizeof(*pool)); } static void kiblnd_destroy_fmr_pool_list(struct list_head *head) @@ -1387,7 +1343,7 @@ static int kiblnd_create_fmr_pool(kib_fmr_poolset_t *fps, kib_dev_t *dev = fps->fps_net->ibn_dev; kib_fmr_pool_t *fpo; struct ib_fmr_pool_param param = { - .max_pages_per_fmr = LNET_MAX_PAYLOAD/PAGE_SIZE, + .max_pages_per_fmr = LNET_MAX_PAYLOAD / PAGE_SIZE, .page_shift = PAGE_SHIFT, .access = (IB_ACCESS_LOCAL_WRITE | IB_ACCESS_REMOTE_WRITE), @@ -1399,7 +1355,7 @@ static int kiblnd_create_fmr_pool(kib_fmr_poolset_t *fps, int rc; LIBCFS_CPT_ALLOC(fpo, lnet_cpt_table(), fps->fps_cpt, sizeof(*fpo)); - if (fpo == NULL) + if (!fpo) return -ENOMEM; fpo->fpo_hdev = kiblnd_current_hdev(dev); @@ -1410,7 +1366,7 @@ static int kiblnd_create_fmr_pool(kib_fmr_poolset_t *fps, CERROR("Failed to create FMR pool: %d\n", rc); kiblnd_hdev_decref(fpo->fpo_hdev); - LIBCFS_FREE(fpo, sizeof(kib_fmr_pool_t)); + LIBCFS_FREE(fpo, sizeof(*fpo)); return rc; } @@ -1424,7 +1380,7 @@ static int kiblnd_create_fmr_pool(kib_fmr_poolset_t *fps, static void kiblnd_fail_fmr_poolset(kib_fmr_poolset_t *fps, struct list_head *zombies) { - if (fps->fps_net == NULL) /* intialized? */ + if (!fps->fps_net) /* intialized? */ return; spin_lock(&fps->fps_lock); @@ -1434,7 +1390,7 @@ static void kiblnd_fail_fmr_poolset(kib_fmr_poolset_t *fps, kib_fmr_pool_t, fpo_list); fpo->fpo_failed = 1; list_del(&fpo->fpo_list); - if (fpo->fpo_map_count == 0) + if (!fpo->fpo_map_count) list_add(&fpo->fpo_list, zombies); else list_add(&fpo->fpo_list, &fps->fps_failed_pool_list); @@ -1445,7 +1401,7 @@ static void kiblnd_fail_fmr_poolset(kib_fmr_poolset_t *fps, static void kiblnd_fini_fmr_poolset(kib_fmr_poolset_t *fps) { - if (fps->fps_net != NULL) { /* initialized? */ + if (fps->fps_net) { /* initialized? */ kiblnd_destroy_fmr_pool_list(&fps->fps_failed_pool_list); kiblnd_destroy_fmr_pool_list(&fps->fps_pool_list); } @@ -1458,7 +1414,7 @@ static int kiblnd_init_fmr_poolset(kib_fmr_poolset_t *fps, int cpt, kib_fmr_pool_t *fpo; int rc; - memset(fps, 0, sizeof(kib_fmr_poolset_t)); + memset(fps, 0, sizeof(*fps)); fps->fps_net = net; fps->fps_cpt = cpt; @@ -1469,7 +1425,7 @@ static int kiblnd_init_fmr_poolset(kib_fmr_poolset_t *fps, int cpt, INIT_LIST_HEAD(&fps->fps_failed_pool_list); rc = kiblnd_create_fmr_pool(fps, &fpo); - if (rc == 0) + if (!rc) list_add_tail(&fpo->fpo_list, &fps->fps_pool_list); return rc; @@ -1477,7 +1433,7 @@ static int kiblnd_init_fmr_poolset(kib_fmr_poolset_t *fps, int cpt, static int kiblnd_fmr_pool_is_idle(kib_fmr_pool_t *fpo, unsigned long now) { - if (fpo->fpo_map_count != 0) /* still in use */ + if (fpo->fpo_map_count) /* still in use */ return 0; if (fpo->fpo_failed) return 1; @@ -1494,11 +1450,11 @@ void kiblnd_fmr_pool_unmap(kib_fmr_t *fmr, int status) int rc; rc = ib_fmr_pool_unmap(fmr->fmr_pfmr); - LASSERT(rc == 0); + LASSERT(!rc); - if (status != 0) { + if (status) { rc = ib_flush_fmr_pool(fpo->fpo_fmr_pool); - LASSERT(rc == 0); + LASSERT(!rc); } fmr->fmr_pool = NULL; @@ -1563,11 +1519,9 @@ int kiblnd_fmr_pool_map(kib_fmr_poolset_t *fps, __u64 *pages, int npages, if (fps->fps_increasing) { spin_unlock(&fps->fps_lock); - CDEBUG(D_NET, - "Another thread is allocating new FMR pool, waiting for her to complete\n"); + CDEBUG(D_NET, "Another thread is allocating new FMR pool, waiting for her to complete\n"); schedule(); goto again; - } if (time_before(cfs_time_current(), fps->fps_next_retry)) { @@ -1583,7 +1537,7 @@ int kiblnd_fmr_pool_map(kib_fmr_poolset_t *fps, __u64 *pages, int npages, rc = kiblnd_create_fmr_pool(fps, &fpo); spin_lock(&fps->fps_lock); fps->fps_increasing = 0; - if (rc == 0) { + if (!rc) { fps->fps_version++; list_add_tail(&fpo->fpo_list, &fps->fps_pool_list); } else { @@ -1597,7 +1551,7 @@ int kiblnd_fmr_pool_map(kib_fmr_poolset_t *fps, __u64 *pages, int npages, static void kiblnd_fini_pool(kib_pool_t *pool) { LASSERT(list_empty(&pool->po_free_list)); - LASSERT(pool->po_allocated == 0); + LASSERT(!pool->po_allocated); CDEBUG(D_NET, "Finalize %s pool\n", pool->po_owner->ps_name); } @@ -1606,7 +1560,7 @@ static void kiblnd_init_pool(kib_poolset_t *ps, kib_pool_t *pool, int size) { CDEBUG(D_NET, "Initialize %s pool\n", ps->ps_name); - memset(pool, 0, sizeof(kib_pool_t)); + memset(pool, 0, sizeof(*pool)); INIT_LIST_HEAD(&pool->po_free_list); pool->po_deadline = cfs_time_shift(IBLND_POOL_DEADLINE); pool->po_owner = ps; @@ -1621,14 +1575,14 @@ static void kiblnd_destroy_pool_list(struct list_head *head) pool = list_entry(head->next, kib_pool_t, po_list); list_del(&pool->po_list); - LASSERT(pool->po_owner != NULL); + LASSERT(pool->po_owner); pool->po_owner->ps_pool_destroy(pool); } } static void kiblnd_fail_poolset(kib_poolset_t *ps, struct list_head *zombies) { - if (ps->ps_net == NULL) /* intialized? */ + if (!ps->ps_net) /* intialized? */ return; spin_lock(&ps->ps_lock); @@ -1637,7 +1591,7 @@ static void kiblnd_fail_poolset(kib_poolset_t *ps, struct list_head *zombies) kib_pool_t, po_list); po->po_failed = 1; list_del(&po->po_list); - if (po->po_allocated == 0) + if (!po->po_allocated) list_add(&po->po_list, zombies); else list_add(&po->po_list, &ps->ps_failed_pool_list); @@ -1647,7 +1601,7 @@ static void kiblnd_fail_poolset(kib_poolset_t *ps, struct list_head *zombies) static void kiblnd_fini_poolset(kib_poolset_t *ps) { - if (ps->ps_net != NULL) { /* initialized? */ + if (ps->ps_net) { /* initialized? */ kiblnd_destroy_pool_list(&ps->ps_failed_pool_list); kiblnd_destroy_pool_list(&ps->ps_pool_list); } @@ -1663,7 +1617,7 @@ static int kiblnd_init_poolset(kib_poolset_t *ps, int cpt, kib_pool_t *pool; int rc; - memset(ps, 0, sizeof(kib_poolset_t)); + memset(ps, 0, sizeof(*ps)); ps->ps_cpt = cpt; ps->ps_net = net; @@ -1680,7 +1634,7 @@ static int kiblnd_init_poolset(kib_poolset_t *ps, int cpt, INIT_LIST_HEAD(&ps->ps_failed_pool_list); rc = ps->ps_pool_create(ps, size, &pool); - if (rc == 0) + if (!rc) list_add(&pool->po_list, &ps->ps_pool_list); else CERROR("Failed to create the first pool for %s\n", ps->ps_name); @@ -1690,7 +1644,7 @@ static int kiblnd_init_poolset(kib_poolset_t *ps, int cpt, static int kiblnd_pool_is_idle(kib_pool_t *pool, unsigned long now) { - if (pool->po_allocated != 0) /* still in use */ + if (pool->po_allocated) /* still in use */ return 0; if (pool->po_failed) return 1; @@ -1706,7 +1660,7 @@ void kiblnd_pool_free_node(kib_pool_t *pool, struct list_head *node) spin_lock(&ps->ps_lock); - if (ps->ps_node_fini != NULL) + if (ps->ps_node_fini) ps->ps_node_fini(pool, node); LASSERT(pool->po_allocated > 0); @@ -1731,6 +1685,9 @@ struct list_head *kiblnd_pool_alloc_node(kib_poolset_t *ps) { struct list_head *node; kib_pool_t *pool; + unsigned int interval = 1; + unsigned long time_before; + unsigned int trips = 0; int rc; again: @@ -1744,7 +1701,7 @@ struct list_head *kiblnd_pool_alloc_node(kib_poolset_t *ps) node = pool->po_free_list.next; list_del(node); - if (ps->ps_node_init != NULL) { + if (ps->ps_node_init) { /* still hold the lock */ ps->ps_node_init(pool, node); } @@ -1756,9 +1713,15 @@ struct list_head *kiblnd_pool_alloc_node(kib_poolset_t *ps) if (ps->ps_increasing) { /* another thread is allocating a new pool */ spin_unlock(&ps->ps_lock); - CDEBUG(D_NET, "Another thread is allocating new %s pool, waiting for her to complete\n", - ps->ps_name); - schedule(); + trips++; + CDEBUG(D_NET, "Another thread is allocating new %s pool, waiting %d HZs for her to complete. trips = %d\n", + ps->ps_name, interval, trips); + + set_current_state(TASK_INTERRUPTIBLE); + schedule_timeout(interval); + if (interval < cfs_time_seconds(1)) + interval *= 2; + goto again; } @@ -1772,12 +1735,14 @@ struct list_head *kiblnd_pool_alloc_node(kib_poolset_t *ps) spin_unlock(&ps->ps_lock); CDEBUG(D_NET, "%s pool exhausted, allocate new pool\n", ps->ps_name); - + time_before = cfs_time_current(); rc = ps->ps_pool_create(ps, ps->ps_pool_size, &pool); + CDEBUG(D_NET, "ps_pool_create took %lu HZ to complete", + cfs_time_current() - time_before); spin_lock(&ps->ps_lock); ps->ps_increasing = 0; - if (rc == 0) { + if (!rc) { list_add_tail(&pool->po_list, &ps->ps_pool_list); } else { ps->ps_next_retry = cfs_time_shift(IBLND_POOL_RETRY); @@ -1794,37 +1759,37 @@ static void kiblnd_destroy_tx_pool(kib_pool_t *pool) kib_tx_pool_t *tpo = container_of(pool, kib_tx_pool_t, tpo_pool); int i; - LASSERT(pool->po_allocated == 0); + LASSERT(!pool->po_allocated); - if (tpo->tpo_tx_pages != NULL) { + if (tpo->tpo_tx_pages) { kiblnd_unmap_tx_pool(tpo); kiblnd_free_pages(tpo->tpo_tx_pages); } - if (tpo->tpo_tx_descs == NULL) + if (!tpo->tpo_tx_descs) goto out; for (i = 0; i < pool->po_size; i++) { kib_tx_t *tx = &tpo->tpo_tx_descs[i]; list_del(&tx->tx_list); - if (tx->tx_pages != NULL) + if (tx->tx_pages) LIBCFS_FREE(tx->tx_pages, LNET_MAX_IOV * sizeof(*tx->tx_pages)); - if (tx->tx_frags != NULL) + if (tx->tx_frags) LIBCFS_FREE(tx->tx_frags, IBLND_MAX_RDMA_FRAGS * sizeof(*tx->tx_frags)); - if (tx->tx_wrq != NULL) + if (tx->tx_wrq) LIBCFS_FREE(tx->tx_wrq, (1 + IBLND_MAX_RDMA_FRAGS) * sizeof(*tx->tx_wrq)); - if (tx->tx_sge != NULL) + if (tx->tx_sge) LIBCFS_FREE(tx->tx_sge, (1 + IBLND_MAX_RDMA_FRAGS) * sizeof(*tx->tx_sge)); - if (tx->tx_rd != NULL) + if (tx->tx_rd) LIBCFS_FREE(tx->tx_rd, offsetof(kib_rdma_desc_t, rd_frags[IBLND_MAX_RDMA_FRAGS])); @@ -1834,7 +1799,7 @@ static void kiblnd_destroy_tx_pool(kib_pool_t *pool) pool->po_size * sizeof(kib_tx_t)); out: kiblnd_fini_pool(pool); - LIBCFS_FREE(tpo, sizeof(kib_tx_pool_t)); + LIBCFS_FREE(tpo, sizeof(*tpo)); } static int kiblnd_tx_pool_size(int ncpts) @@ -1853,7 +1818,7 @@ static int kiblnd_create_tx_pool(kib_poolset_t *ps, int size, kib_tx_pool_t *tpo; LIBCFS_CPT_ALLOC(tpo, lnet_cpt_table(), ps->ps_cpt, sizeof(*tpo)); - if (tpo == NULL) { + if (!tpo) { CERROR("Failed to allocate TX pool\n"); return -ENOMEM; } @@ -1864,15 +1829,15 @@ static int kiblnd_create_tx_pool(kib_poolset_t *ps, int size, tpo->tpo_tx_pages = NULL; npg = (size * IBLND_MSG_SIZE + PAGE_SIZE - 1) / PAGE_SIZE; - if (kiblnd_alloc_pages(&tpo->tpo_tx_pages, ps->ps_cpt, npg) != 0) { + if (kiblnd_alloc_pages(&tpo->tpo_tx_pages, ps->ps_cpt, npg)) { CERROR("Can't allocate tx pages: %d\n", npg); - LIBCFS_FREE(tpo, sizeof(kib_tx_pool_t)); + LIBCFS_FREE(tpo, sizeof(*tpo)); return -ENOMEM; } LIBCFS_CPT_ALLOC(tpo->tpo_tx_descs, lnet_cpt_table(), ps->ps_cpt, size * sizeof(kib_tx_t)); - if (tpo->tpo_tx_descs == NULL) { + if (!tpo->tpo_tx_descs) { CERROR("Can't allocate %d tx descriptors\n", size); ps->ps_pool_destroy(pool); return -ENOMEM; @@ -1884,17 +1849,17 @@ static int kiblnd_create_tx_pool(kib_poolset_t *ps, int size, kib_tx_t *tx = &tpo->tpo_tx_descs[i]; tx->tx_pool = tpo; - if (ps->ps_net->ibn_fmr_ps != NULL) { + if (ps->ps_net->ibn_fmr_ps) { LIBCFS_CPT_ALLOC(tx->tx_pages, lnet_cpt_table(), ps->ps_cpt, LNET_MAX_IOV * sizeof(*tx->tx_pages)); - if (tx->tx_pages == NULL) + if (!tx->tx_pages) break; } LIBCFS_CPT_ALLOC(tx->tx_frags, lnet_cpt_table(), ps->ps_cpt, IBLND_MAX_RDMA_FRAGS * sizeof(*tx->tx_frags)); - if (tx->tx_frags == NULL) + if (!tx->tx_frags) break; sg_init_table(tx->tx_frags, IBLND_MAX_RDMA_FRAGS); @@ -1902,19 +1867,19 @@ static int kiblnd_create_tx_pool(kib_poolset_t *ps, int size, LIBCFS_CPT_ALLOC(tx->tx_wrq, lnet_cpt_table(), ps->ps_cpt, (1 + IBLND_MAX_RDMA_FRAGS) * sizeof(*tx->tx_wrq)); - if (tx->tx_wrq == NULL) + if (!tx->tx_wrq) break; LIBCFS_CPT_ALLOC(tx->tx_sge, lnet_cpt_table(), ps->ps_cpt, (1 + IBLND_MAX_RDMA_FRAGS) * sizeof(*tx->tx_sge)); - if (tx->tx_sge == NULL) + if (!tx->tx_sge) break; LIBCFS_CPT_ALLOC(tx->tx_rd, lnet_cpt_table(), ps->ps_cpt, offsetof(kib_rdma_desc_t, rd_frags[IBLND_MAX_RDMA_FRAGS])); - if (tx->tx_rd == NULL) + if (!tx->tx_rd) break; } @@ -1945,23 +1910,23 @@ static void kiblnd_net_fini_pools(kib_net_t *net) kib_tx_poolset_t *tps; kib_fmr_poolset_t *fps; - if (net->ibn_tx_ps != NULL) { + if (net->ibn_tx_ps) { tps = net->ibn_tx_ps[i]; kiblnd_fini_poolset(&tps->tps_poolset); } - if (net->ibn_fmr_ps != NULL) { + if (net->ibn_fmr_ps) { fps = net->ibn_fmr_ps[i]; kiblnd_fini_fmr_poolset(fps); } } - if (net->ibn_tx_ps != NULL) { + if (net->ibn_tx_ps) { cfs_percpt_free(net->ibn_tx_ps); net->ibn_tx_ps = NULL; } - if (net->ibn_fmr_ps != NULL) { + if (net->ibn_fmr_ps) { cfs_percpt_free(net->ibn_fmr_ps); net->ibn_fmr_ps = NULL; } @@ -1975,8 +1940,7 @@ static int kiblnd_net_init_pools(kib_net_t *net, __u32 *cpts, int ncpts) int i; read_lock_irqsave(&kiblnd_data.kib_global_lock, flags); - if (*kiblnd_tunables.kib_map_on_demand == 0 && - net->ibn_dev->ibd_hdev->ibh_nmrs == 1) { + if (!*kiblnd_tunables.kib_map_on_demand) { read_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags); goto create_tx_pool; } @@ -1996,7 +1960,7 @@ static int kiblnd_net_init_pools(kib_net_t *net, __u32 *cpts, int ncpts) * TX pool must be created later than FMR, see LU-2268 * for details */ - LASSERT(net->ibn_tx_ps == NULL); + LASSERT(!net->ibn_tx_ps); /* * premapping can fail if ibd_nmr > 1, so we always create @@ -2005,56 +1969,45 @@ static int kiblnd_net_init_pools(kib_net_t *net, __u32 *cpts, int ncpts) net->ibn_fmr_ps = cfs_percpt_alloc(lnet_cpt_table(), sizeof(kib_fmr_poolset_t)); - if (net->ibn_fmr_ps == NULL) { + if (!net->ibn_fmr_ps) { CERROR("Failed to allocate FMR pool array\n"); rc = -ENOMEM; goto failed; } for (i = 0; i < ncpts; i++) { - cpt = (cpts == NULL) ? i : cpts[i]; + cpt = !cpts ? i : cpts[i]; rc = kiblnd_init_fmr_poolset(net->ibn_fmr_ps[cpt], cpt, net, kiblnd_fmr_pool_size(ncpts), kiblnd_fmr_flush_trigger(ncpts)); - if (rc == -ENOSYS && i == 0) /* no FMR */ - break; - - if (rc != 0) { /* a real error */ + if (rc) { CERROR("Can't initialize FMR pool for CPT %d: %d\n", cpt, rc); goto failed; } } - if (i > 0) { + if (i > 0) LASSERT(i == ncpts); - goto create_tx_pool; - } - - cfs_percpt_free(net->ibn_fmr_ps); - net->ibn_fmr_ps = NULL; - - CWARN("Device does not support FMR\n"); - goto failed; create_tx_pool: net->ibn_tx_ps = cfs_percpt_alloc(lnet_cpt_table(), sizeof(kib_tx_poolset_t)); - if (net->ibn_tx_ps == NULL) { + if (!net->ibn_tx_ps) { CERROR("Failed to allocate tx pool array\n"); rc = -ENOMEM; goto failed; } for (i = 0; i < ncpts; i++) { - cpt = (cpts == NULL) ? i : cpts[i]; + cpt = !cpts ? i : cpts[i]; rc = kiblnd_init_poolset(&net->ibn_tx_ps[cpt]->tps_poolset, cpt, net, "TX", kiblnd_tx_pool_size(ncpts), kiblnd_create_tx_pool, kiblnd_destroy_tx_pool, kiblnd_tx_init, NULL); - if (rc != 0) { + if (rc) { CERROR("Can't initialize TX pool for CPT %d: %d\n", cpt, rc); goto failed; @@ -2064,14 +2017,16 @@ static int kiblnd_net_init_pools(kib_net_t *net, __u32 *cpts, int ncpts) return 0; failed: kiblnd_net_fini_pools(net); - LASSERT(rc != 0); + LASSERT(rc); return rc; } static int kiblnd_hdev_get_attr(kib_hca_dev_t *hdev) { - /* It's safe to assume a HCA can handle a page size - * matching that of the native system */ + /* + * It's safe to assume a HCA can handle a page size + * matching that of the native system + */ hdev->ibh_page_shift = PAGE_SHIFT; hdev->ibh_page_size = 1 << PAGE_SHIFT; hdev->ibh_page_mask = ~((__u64)hdev->ibh_page_size - 1); @@ -2082,44 +2037,28 @@ static int kiblnd_hdev_get_attr(kib_hca_dev_t *hdev) return 0; } - for (hdev->ibh_mr_shift = 0; - hdev->ibh_mr_shift < 64; hdev->ibh_mr_shift++) { - if (hdev->ibh_mr_size == (1ULL << hdev->ibh_mr_shift) || - hdev->ibh_mr_size == (1ULL << hdev->ibh_mr_shift) - 1) - return 0; - } - CERROR("Invalid mr size: %#llx\n", hdev->ibh_mr_size); return -EINVAL; } static void kiblnd_hdev_cleanup_mrs(kib_hca_dev_t *hdev) { - int i; - - if (hdev->ibh_nmrs == 0 || hdev->ibh_mrs == NULL) + if (!hdev->ibh_mrs) return; - for (i = 0; i < hdev->ibh_nmrs; i++) { - if (hdev->ibh_mrs[i] == NULL) - break; + ib_dereg_mr(hdev->ibh_mrs); - ib_dereg_mr(hdev->ibh_mrs[i]); - } - - LIBCFS_FREE(hdev->ibh_mrs, sizeof(*hdev->ibh_mrs) * hdev->ibh_nmrs); - hdev->ibh_mrs = NULL; - hdev->ibh_nmrs = 0; + hdev->ibh_mrs = NULL; } void kiblnd_hdev_destroy(kib_hca_dev_t *hdev) { kiblnd_hdev_cleanup_mrs(hdev); - if (hdev->ibh_pd != NULL) + if (hdev->ibh_pd) ib_dealloc_pd(hdev->ibh_pd); - if (hdev->ibh_cmid != NULL) + if (hdev->ibh_cmid) rdma_destroy_id(hdev->ibh_cmid); LIBCFS_FREE(hdev, sizeof(*hdev)); @@ -2132,18 +2071,9 @@ static int kiblnd_hdev_setup_mrs(kib_hca_dev_t *hdev) int acflags = IB_ACCESS_LOCAL_WRITE | IB_ACCESS_REMOTE_WRITE; rc = kiblnd_hdev_get_attr(hdev); - if (rc != 0) + if (rc) return rc; - LIBCFS_ALLOC(hdev->ibh_mrs, 1 * sizeof(*hdev->ibh_mrs)); - if (hdev->ibh_mrs == NULL) { - CERROR("Failed to allocate MRs table\n"); - return -ENOMEM; - } - - hdev->ibh_mrs[0] = NULL; - hdev->ibh_nmrs = 1; - mr = ib_get_dma_mr(hdev->ibh_pd, acflags); if (IS_ERR(mr)) { CERROR("Failed ib_get_dma_mr : %ld\n", PTR_ERR(mr)); @@ -2151,7 +2081,7 @@ static int kiblnd_hdev_setup_mrs(kib_hca_dev_t *hdev) return PTR_ERR(mr); } - hdev->ibh_mrs[0] = mr; + hdev->ibh_mrs = mr; return 0; } @@ -2170,12 +2100,13 @@ static int kiblnd_dev_need_failover(kib_dev_t *dev) struct sockaddr_in dstaddr; int rc; - if (dev->ibd_hdev == NULL || /* initializing */ - dev->ibd_hdev->ibh_cmid == NULL || /* listener is dead */ + if (!dev->ibd_hdev || /* initializing */ + !dev->ibd_hdev->ibh_cmid || /* listener is dead */ *kiblnd_tunables.kib_dev_failover > 1) /* debugging */ return 1; - /* XXX: it's UGLY, but I don't have better way to find + /* + * XXX: it's UGLY, but I don't have better way to find * ib-bonding HCA failover because: * * a. no reliable CM event for HCA failover... @@ -2184,7 +2115,8 @@ static int kiblnd_dev_need_failover(kib_dev_t *dev) * We have only two choices at this point: * * a. rdma_bind_addr(), it will conflict with listener cmid - * b. rdma_resolve_addr() to zero addr */ + * b. rdma_resolve_addr() to zero addr + */ cmid = kiblnd_rdma_create_id(kiblnd_dummy_callback, dev, RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(cmid)) { @@ -2201,7 +2133,7 @@ static int kiblnd_dev_need_failover(kib_dev_t *dev) dstaddr.sin_family = AF_INET; rc = rdma_resolve_addr(cmid, (struct sockaddr *)&srcaddr, (struct sockaddr *)&dstaddr, 1); - if (rc != 0 || cmid->device == NULL) { + if (rc || !cmid->device) { CERROR("Failed to bind %s:%pI4h to device(%p): %d\n", dev->ibd_ifname, &dev->ibd_ifip, cmid->device, rc); @@ -2230,24 +2162,27 @@ int kiblnd_dev_failover(kib_dev_t *dev) int i; LASSERT(*kiblnd_tunables.kib_dev_failover > 1 || - dev->ibd_can_failover || - dev->ibd_hdev == NULL); + dev->ibd_can_failover || !dev->ibd_hdev); rc = kiblnd_dev_need_failover(dev); if (rc <= 0) goto out; - if (dev->ibd_hdev != NULL && - dev->ibd_hdev->ibh_cmid != NULL) { - /* XXX it's not good to close old listener at here, + if (dev->ibd_hdev && + dev->ibd_hdev->ibh_cmid) { + /* + * XXX it's not good to close old listener at here, * because we can fail to create new listener. * But we have to close it now, otherwise rdma_bind_addr - * will return EADDRINUSE... How crap! */ + * will return EADDRINUSE... How crap! + */ write_lock_irqsave(&kiblnd_data.kib_global_lock, flags); cmid = dev->ibd_hdev->ibh_cmid; - /* make next schedule of kiblnd_dev_need_failover() - * return 1 for me */ + /* + * make next schedule of kiblnd_dev_need_failover() + * return 1 for me + */ dev->ibd_hdev->ibh_cmid = NULL; write_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags); @@ -2269,7 +2204,7 @@ int kiblnd_dev_failover(kib_dev_t *dev) /* Bind to failover device or port */ rc = rdma_bind_addr(cmid, (struct sockaddr *)&addr); - if (rc != 0 || cmid->device == NULL) { + if (rc || !cmid->device) { CERROR("Failed to bind %s:%pI4h to device(%p): %d\n", dev->ibd_ifname, &dev->ibd_ifip, cmid->device, rc); @@ -2278,7 +2213,7 @@ int kiblnd_dev_failover(kib_dev_t *dev) } LIBCFS_ALLOC(hdev, sizeof(*hdev)); - if (hdev == NULL) { + if (!hdev) { CERROR("Failed to allocate kib_hca_dev\n"); rdma_destroy_id(cmid); rc = -ENOMEM; @@ -2300,13 +2235,13 @@ int kiblnd_dev_failover(kib_dev_t *dev) hdev->ibh_pd = pd; rc = rdma_listen(cmid, 0); - if (rc != 0) { + if (rc) { CERROR("Can't start new listener: %d\n", rc); goto out; } rc = kiblnd_hdev_setup_mrs(hdev); - if (rc != 0) { + if (rc) { CERROR("Can't setup device: %d\n", rc); goto out; } @@ -2334,10 +2269,10 @@ int kiblnd_dev_failover(kib_dev_t *dev) kiblnd_destroy_pool_list(&zombie_ppo); if (!list_empty(&zombie_fpo)) kiblnd_destroy_fmr_pool_list(&zombie_fpo); - if (hdev != NULL) + if (hdev) kiblnd_hdev_decref(hdev); - if (rc != 0) + if (rc) dev->ibd_failed_failover++; else dev->ibd_failed_failover = 0; @@ -2347,13 +2282,13 @@ int kiblnd_dev_failover(kib_dev_t *dev) void kiblnd_destroy_dev(kib_dev_t *dev) { - LASSERT(dev->ibd_nnets == 0); + LASSERT(!dev->ibd_nnets); LASSERT(list_empty(&dev->ibd_nets)); list_del(&dev->ibd_fail_list); list_del(&dev->ibd_list); - if (dev->ibd_hdev != NULL) + if (dev->ibd_hdev) kiblnd_hdev_decref(dev->ibd_hdev); LIBCFS_FREE(dev, sizeof(*dev)); @@ -2369,7 +2304,7 @@ static kib_dev_t *kiblnd_create_dev(char *ifname) int rc; rc = lnet_ipif_query(ifname, &up, &ip, &netmask); - if (rc != 0) { + if (rc) { CERROR("Can't query IPoIB interface %s: %d\n", ifname, rc); return NULL; @@ -2381,11 +2316,11 @@ static kib_dev_t *kiblnd_create_dev(char *ifname) } LIBCFS_ALLOC(dev, sizeof(*dev)); - if (dev == NULL) + if (!dev) return NULL; netdev = dev_get_by_name(&init_net, ifname); - if (netdev == NULL) { + if (!netdev) { dev->ibd_can_failover = 0; } else { dev->ibd_can_failover = !!(netdev->flags & IFF_MASTER); @@ -2400,14 +2335,13 @@ static kib_dev_t *kiblnd_create_dev(char *ifname) /* initialize the device */ rc = kiblnd_dev_failover(dev); - if (rc != 0) { + if (rc) { CERROR("Can't initialize device: %d\n", rc); LIBCFS_FREE(dev, sizeof(*dev)); return NULL; } - list_add_tail(&dev->ibd_list, - &kiblnd_data.kib_devs); + list_add_tail(&dev->ibd_list, &kiblnd_data.kib_devs); return dev; } @@ -2424,18 +2358,22 @@ static void kiblnd_base_shutdown(void) case IBLND_INIT_ALL: case IBLND_INIT_DATA: - LASSERT(kiblnd_data.kib_peers != NULL); + LASSERT(kiblnd_data.kib_peers); for (i = 0; i < kiblnd_data.kib_peer_hash_size; i++) LASSERT(list_empty(&kiblnd_data.kib_peers[i])); LASSERT(list_empty(&kiblnd_data.kib_connd_zombies)); LASSERT(list_empty(&kiblnd_data.kib_connd_conns)); + LASSERT(list_empty(&kiblnd_data.kib_reconn_list)); + LASSERT(list_empty(&kiblnd_data.kib_reconn_wait)); /* flag threads to terminate; wake and wait for them to die */ kiblnd_data.kib_shutdown = 1; - /* NB: we really want to stop scheduler threads net by net + /* + * NB: we really want to stop scheduler threads net by net * instead of the whole module, this should be improved - * with dynamic configuration LNet */ + * with dynamic configuration LNet + */ cfs_percpt_for_each(sched, i, kiblnd_data.kib_scheds) wake_up_all(&sched->ibs_waitq); @@ -2443,7 +2381,7 @@ static void kiblnd_base_shutdown(void) wake_up_all(&kiblnd_data.kib_failover_waitq); i = 2; - while (atomic_read(&kiblnd_data.kib_nthreads) != 0) { + while (atomic_read(&kiblnd_data.kib_nthreads)) { i++; /* power of 2 ? */ CDEBUG(((i & (-i)) == i) ? D_WARNING : D_NET, @@ -2459,20 +2397,20 @@ static void kiblnd_base_shutdown(void) break; } - if (kiblnd_data.kib_peers != NULL) { + if (kiblnd_data.kib_peers) { LIBCFS_FREE(kiblnd_data.kib_peers, sizeof(struct list_head) * kiblnd_data.kib_peer_hash_size); } - if (kiblnd_data.kib_scheds != NULL) + if (kiblnd_data.kib_scheds) cfs_percpt_free(kiblnd_data.kib_scheds); kiblnd_data.kib_init = IBLND_INIT_NOTHING; module_put(THIS_MODULE); } -void kiblnd_shutdown(lnet_ni_t *ni) +static void kiblnd_shutdown(lnet_ni_t *ni) { kib_net_t *net = ni->ni_data; rwlock_t *g_lock = &kiblnd_data.kib_global_lock; @@ -2481,7 +2419,7 @@ void kiblnd_shutdown(lnet_ni_t *ni) LASSERT(kiblnd_data.kib_init == IBLND_INIT_ALL); - if (net == NULL) + if (!net) goto out; write_lock_irqsave(g_lock, flags); @@ -2498,7 +2436,7 @@ void kiblnd_shutdown(lnet_ni_t *ni) /* Wait for all peer state to clean up */ i = 2; - while (atomic_read(&net->ibn_npeers) != 0) { + while (atomic_read(&net->ibn_npeers)) { i++; CDEBUG(((i & (-i)) == i) ? D_WARNING : D_NET, /* 2**n? */ "%s: waiting for %d peers to disconnect\n", @@ -2519,10 +2457,9 @@ void kiblnd_shutdown(lnet_ni_t *ni) /* fall through */ case IBLND_INIT_NOTHING: - LASSERT(atomic_read(&net->ibn_nconns) == 0); + LASSERT(!atomic_read(&net->ibn_nconns)); - if (net->ibn_dev != NULL && - net->ibn_dev->ibd_nnets == 0) + if (net->ibn_dev && !net->ibn_dev->ibd_nnets) kiblnd_destroy_dev(net->ibn_dev); break; @@ -2558,7 +2495,7 @@ static int kiblnd_base_startup(void) kiblnd_data.kib_peer_hash_size = IBLND_PEER_HASH_SIZE; LIBCFS_ALLOC(kiblnd_data.kib_peers, sizeof(struct list_head) * kiblnd_data.kib_peer_hash_size); - if (kiblnd_data.kib_peers == NULL) + if (!kiblnd_data.kib_peers) goto failed; for (i = 0; i < kiblnd_data.kib_peer_hash_size; i++) INIT_LIST_HEAD(&kiblnd_data.kib_peers[i]); @@ -2566,12 +2503,15 @@ static int kiblnd_base_startup(void) spin_lock_init(&kiblnd_data.kib_connd_lock); INIT_LIST_HEAD(&kiblnd_data.kib_connd_conns); INIT_LIST_HEAD(&kiblnd_data.kib_connd_zombies); + INIT_LIST_HEAD(&kiblnd_data.kib_reconn_list); + INIT_LIST_HEAD(&kiblnd_data.kib_reconn_wait); + init_waitqueue_head(&kiblnd_data.kib_connd_waitq); init_waitqueue_head(&kiblnd_data.kib_failover_waitq); kiblnd_data.kib_scheds = cfs_percpt_alloc(lnet_cpt_table(), sizeof(*sched)); - if (kiblnd_data.kib_scheds == NULL) + if (!kiblnd_data.kib_scheds) goto failed; cfs_percpt_for_each(sched, i, kiblnd_data.kib_scheds) { @@ -2585,8 +2525,10 @@ static int kiblnd_base_startup(void) if (*kiblnd_tunables.kib_nscheds > 0) { nthrs = min(nthrs, *kiblnd_tunables.kib_nscheds); } else { - /* max to half of CPUs, another half is reserved for - * upper layer modules */ + /* + * max to half of CPUs, another half is reserved for + * upper layer modules + */ nthrs = min(max(IBLND_N_SCHED, nthrs >> 1), nthrs); } @@ -2601,16 +2543,16 @@ static int kiblnd_base_startup(void) /*****************************************************/ rc = kiblnd_thread_start(kiblnd_connd, NULL, "kiblnd_connd"); - if (rc != 0) { + if (rc) { CERROR("Can't spawn o2iblnd connd: %d\n", rc); goto failed; } - if (*kiblnd_tunables.kib_dev_failover != 0) + if (*kiblnd_tunables.kib_dev_failover) rc = kiblnd_thread_start(kiblnd_failover_thread, NULL, "kiblnd_failover"); - if (rc != 0) { + if (rc) { CERROR("Can't spawn o2iblnd failover thread: %d\n", rc); goto failed; } @@ -2632,7 +2574,7 @@ static int kiblnd_start_schedulers(struct kib_sched_info *sched) int nthrs; int i; - if (sched->ibs_nthreads == 0) { + if (!sched->ibs_nthreads) { if (*kiblnd_tunables.kib_nscheds > 0) { nthrs = sched->ibs_nthreads_max; } else { @@ -2655,7 +2597,7 @@ static int kiblnd_start_schedulers(struct kib_sched_info *sched) snprintf(name, sizeof(name), "kiblnd_sd_%02ld_%02ld", KIB_THREAD_CPT(id), KIB_THREAD_TID(id)); rc = kiblnd_thread_start(kiblnd_scheduler, (void *)id, name); - if (rc == 0) + if (!rc) continue; CERROR("Can't spawn thread %d for scheduler[%d]: %d\n", @@ -2677,14 +2619,14 @@ static int kiblnd_dev_start_threads(kib_dev_t *dev, int newdev, __u32 *cpts, for (i = 0; i < ncpts; i++) { struct kib_sched_info *sched; - cpt = (cpts == NULL) ? i : cpts[i]; + cpt = !cpts ? i : cpts[i]; sched = kiblnd_data.kib_scheds[cpt]; if (!newdev && sched->ibs_nthreads > 0) continue; rc = kiblnd_start_schedulers(kiblnd_data.kib_scheds[cpt]); - if (rc != 0) { + if (rc) { CERROR("Failed to start scheduler threads for %s\n", dev->ibd_ifname); return rc; @@ -2702,30 +2644,30 @@ static kib_dev_t *kiblnd_dev_search(char *ifname) colon = strchr(ifname, ':'); list_for_each_entry(dev, &kiblnd_data.kib_devs, ibd_list) { - if (strcmp(&dev->ibd_ifname[0], ifname) == 0) + if (!strcmp(&dev->ibd_ifname[0], ifname)) return dev; - if (alias != NULL) + if (alias) continue; colon2 = strchr(dev->ibd_ifname, ':'); - if (colon != NULL) + if (colon) *colon = 0; - if (colon2 != NULL) + if (colon2) *colon2 = 0; - if (strcmp(&dev->ibd_ifname[0], ifname) == 0) + if (!strcmp(&dev->ibd_ifname[0], ifname)) alias = dev; - if (colon != NULL) + if (colon) *colon = ':'; - if (colon2 != NULL) + if (colon2) *colon2 = ':'; } return alias; } -int kiblnd_startup(lnet_ni_t *ni) +static int kiblnd_startup(lnet_ni_t *ni) { char *ifname; kib_dev_t *ibdev = NULL; @@ -2739,13 +2681,13 @@ int kiblnd_startup(lnet_ni_t *ni) if (kiblnd_data.kib_init == IBLND_INIT_NOTHING) { rc = kiblnd_base_startup(); - if (rc != 0) + if (rc) return rc; } LIBCFS_ALLOC(net, sizeof(*net)); ni->ni_data = net; - if (net == NULL) + if (!net) goto net_failed; ktime_get_real_ts64(&tv); @@ -2757,11 +2699,11 @@ int kiblnd_startup(lnet_ni_t *ni) ni->ni_peertxcredits = *kiblnd_tunables.kib_peertxcredits; ni->ni_peerrtrcredits = *kiblnd_tunables.kib_peerrtrcredits; - if (ni->ni_interfaces[0] != NULL) { + if (ni->ni_interfaces[0]) { /* Use the IPoIB interface specified in 'networks=' */ CLASSERT(LNET_MAX_INTERFACES > 1); - if (ni->ni_interfaces[1] != NULL) { + if (ni->ni_interfaces[1]) { CERROR("Multiple interfaces not supported\n"); goto failed; } @@ -2778,12 +2720,12 @@ int kiblnd_startup(lnet_ni_t *ni) ibdev = kiblnd_dev_search(ifname); - newdev = ibdev == NULL; + newdev = !ibdev; /* hmm...create kib_dev even for alias */ - if (ibdev == NULL || strcmp(&ibdev->ibd_ifname[0], ifname) != 0) + if (!ibdev || strcmp(&ibdev->ibd_ifname[0], ifname)) ibdev = kiblnd_create_dev(ifname); - if (ibdev == NULL) + if (!ibdev) goto failed; net->ibn_dev = ibdev; @@ -2791,11 +2733,11 @@ int kiblnd_startup(lnet_ni_t *ni) rc = kiblnd_dev_start_threads(ibdev, newdev, ni->ni_cpts, ni->ni_ncpts); - if (rc != 0) + if (rc) goto failed; rc = kiblnd_net_init_pools(net, ni->ni_cpts, ni->ni_ncpts); - if (rc != 0) { + if (rc) { CERROR("Failed to initialize NI pools: %d\n", rc); goto failed; } @@ -2810,7 +2752,7 @@ int kiblnd_startup(lnet_ni_t *ni) return 0; failed: - if (net->ibn_dev == NULL && ibdev != NULL) + if (!net->ibn_dev && ibdev) kiblnd_destroy_dev(ibdev); net_failed: @@ -2820,25 +2762,35 @@ net_failed: return -ENETDOWN; } -static void __exit kiblnd_module_fini(void) +static lnd_t the_o2iblnd = { + .lnd_type = O2IBLND, + .lnd_startup = kiblnd_startup, + .lnd_shutdown = kiblnd_shutdown, + .lnd_ctl = kiblnd_ctl, + .lnd_query = kiblnd_query, + .lnd_send = kiblnd_send, + .lnd_recv = kiblnd_recv, +}; + +static void __exit ko2iblnd_exit(void) { lnet_unregister_lnd(&the_o2iblnd); } -static int __init kiblnd_module_init(void) +static int __init ko2iblnd_init(void) { int rc; CLASSERT(sizeof(kib_msg_t) <= IBLND_MSG_SIZE); CLASSERT(offsetof(kib_msg_t, - ibm_u.get.ibgm_rd.rd_frags[IBLND_MAX_RDMA_FRAGS]) - <= IBLND_MSG_SIZE); + ibm_u.get.ibgm_rd.rd_frags[IBLND_MAX_RDMA_FRAGS]) + <= IBLND_MSG_SIZE); CLASSERT(offsetof(kib_msg_t, - ibm_u.putack.ibpam_rd.rd_frags[IBLND_MAX_RDMA_FRAGS]) - <= IBLND_MSG_SIZE); + ibm_u.putack.ibpam_rd.rd_frags[IBLND_MAX_RDMA_FRAGS]) + <= IBLND_MSG_SIZE); rc = kiblnd_tunables_init(); - if (rc != 0) + if (rc) return rc; lnet_register_lnd(&the_o2iblnd); @@ -2847,8 +2799,9 @@ static int __init kiblnd_module_init(void) } MODULE_AUTHOR("OpenSFS, Inc. "); -MODULE_DESCRIPTION("Kernel OpenIB gen2 LND v2.00"); +MODULE_DESCRIPTION("OpenIB gen2 LNet Network Driver"); +MODULE_VERSION("2.7.0"); MODULE_LICENSE("GPL"); -module_init(kiblnd_module_init); -module_exit(kiblnd_module_fini); +module_init(ko2iblnd_init); +module_exit(ko2iblnd_exit); diff --git a/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h b/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h index 025faa9f8..bfcbdd167 100644 --- a/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h +++ b/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h @@ -60,17 +60,17 @@ #include #include +#include +#include +#include +#include + #define DEBUG_SUBSYSTEM S_LND #include "../../../include/linux/libcfs/libcfs.h" #include "../../../include/linux/lnet/lnet.h" #include "../../../include/linux/lnet/lib-lnet.h" -#include -#include -#include -#include - #define IBLND_PEER_HASH_SIZE 101 /* # peer lists */ /* # scheduler loops before reschedule */ #define IBLND_RESCHED 100 @@ -146,9 +146,9 @@ kiblnd_concurrent_sends_v1(void) #define IBLND_OOB_CAPABLE(v) ((v) != IBLND_MSG_VERSION_1) #define IBLND_OOB_MSGS(v) (IBLND_OOB_CAPABLE(v) ? 2 : 0) -#define IBLND_MSG_SIZE (4<<10) /* max size of queued messages (inc hdr) */ +#define IBLND_MSG_SIZE (4 << 10) /* max size of queued messages (inc hdr) */ #define IBLND_MAX_RDMA_FRAGS LNET_MAX_IOV /* max # of fragments supported */ -#define IBLND_CFG_RDMA_FRAGS (*kiblnd_tunables.kib_map_on_demand != 0 ? \ +#define IBLND_CFG_RDMA_FRAGS (*kiblnd_tunables.kib_map_on_demand ? \ *kiblnd_tunables.kib_map_on_demand : \ IBLND_MAX_RDMA_FRAGS) /* max # of fragments configured by user */ #define IBLND_RDMA_FRAGS(v) ((v) == IBLND_MSG_VERSION_1 ? \ @@ -162,18 +162,17 @@ kiblnd_concurrent_sends_v1(void) #define IBLND_FMR_POOL 256 #define IBLND_FMR_POOL_FLUSH 192 -/* TX messages (shared by all connections) */ -#define IBLND_TX_MSGS() (*kiblnd_tunables.kib_ntx) - -/* RX messages (per connection) */ -#define IBLND_RX_MSGS(v) (IBLND_MSG_QUEUE_SIZE(v) * 2 + IBLND_OOB_MSGS(v)) -#define IBLND_RX_MSG_BYTES(v) (IBLND_RX_MSGS(v) * IBLND_MSG_SIZE) -#define IBLND_RX_MSG_PAGES(v) ((IBLND_RX_MSG_BYTES(v) + PAGE_SIZE - 1) / PAGE_SIZE) +#define IBLND_RX_MSGS(c) \ + ((c->ibc_queue_depth) * 2 + IBLND_OOB_MSGS(c->ibc_version)) +#define IBLND_RX_MSG_BYTES(c) (IBLND_RX_MSGS(c) * IBLND_MSG_SIZE) +#define IBLND_RX_MSG_PAGES(c) \ + ((IBLND_RX_MSG_BYTES(c) + PAGE_SIZE - 1) / PAGE_SIZE) /* WRs and CQEs (per connection) */ -#define IBLND_RECV_WRS(v) IBLND_RX_MSGS(v) -#define IBLND_SEND_WRS(v) ((IBLND_RDMA_FRAGS(v) + 1) * IBLND_CONCURRENT_SENDS(v)) -#define IBLND_CQ_ENTRIES(v) (IBLND_RECV_WRS(v) + IBLND_SEND_WRS(v)) +#define IBLND_RECV_WRS(c) IBLND_RX_MSGS(c) +#define IBLND_SEND_WRS(c) \ + ((c->ibc_max_frags + 1) * IBLND_CONCURRENT_SENDS(c->ibc_version)) +#define IBLND_CQ_ENTRIES(c) (IBLND_RECV_WRS(c) + IBLND_SEND_WRS(c)) struct kib_hca_dev; @@ -209,8 +208,7 @@ typedef struct kib_hca_dev { __u64 ibh_page_mask; /* page mask of current HCA */ int ibh_mr_shift; /* bits shift of max MR size */ __u64 ibh_mr_size; /* size of MR */ - int ibh_nmrs; /* # of global MRs */ - struct ib_mr **ibh_mrs; /* global MR */ + struct ib_mr *ibh_mrs; /* global MR */ struct ib_pd *ibh_pd; /* PD */ kib_dev_t *ibh_dev; /* owner */ atomic_t ibh_ref; /* refcount */ @@ -350,6 +348,16 @@ typedef struct { void *kib_connd; /* the connd task (serialisation assertions) */ struct list_head kib_connd_conns; /* connections to setup/teardown */ struct list_head kib_connd_zombies; /* connections with zero refcount */ + /* connections to reconnect */ + struct list_head kib_reconn_list; + /* peers wait for reconnection */ + struct list_head kib_reconn_wait; + /** + * The second that peers are pulled out from \a kib_reconn_wait + * for reconnection. + */ + time64_t kib_reconn_sec; + wait_queue_head_t kib_connd_waitq; /* connection daemon sleeps here */ spinlock_t kib_connd_lock; /* serialise */ struct ib_qp_attr kib_error_qpa; /* QP->ERROR */ @@ -465,10 +473,10 @@ typedef struct { #define IBLND_REJECT_FATAL 3 /* Anything else */ #define IBLND_REJECT_CONN_UNCOMPAT 4 /* incompatible version peer */ #define IBLND_REJECT_CONN_STALE 5 /* stale peer */ -#define IBLND_REJECT_RDMA_FRAGS 6 /* Fatal: peer's rdma frags can't match */ - /* mine */ -#define IBLND_REJECT_MSG_QUEUE_SIZE 7 /* Fatal: peer's msg queue size can't */ - /* match mine */ +/* peer's rdma frags doesn't match mine */ +#define IBLND_REJECT_RDMA_FRAGS 6 +/* peer's msg queue size doesn't match mine */ +#define IBLND_REJECT_MSG_QUEUE_SIZE 7 /***********************************************************************/ @@ -527,6 +535,8 @@ typedef struct kib_conn { struct list_head ibc_list; /* stash on peer's conn list */ struct list_head ibc_sched_list; /* schedule for attention */ __u16 ibc_version; /* version of connection */ + /* reconnect later */ + __u16 ibc_reconnect:1; __u64 ibc_incarnation; /* which instance of the peer */ atomic_t ibc_refcount; /* # users */ int ibc_state; /* what's happening */ @@ -536,6 +546,10 @@ typedef struct kib_conn { int ibc_outstanding_credits; /* # credits to return */ int ibc_reserved_credits; /* # ACK/DONE msg credits */ int ibc_comms_error; /* set on comms error */ + /* connections queue depth */ + __u16 ibc_queue_depth; + /* connections max frags */ + __u16 ibc_max_frags; unsigned int ibc_nrx:16; /* receive buffers owned */ unsigned int ibc_scheduled:1; /* scheduled for attention */ unsigned int ibc_ready:1; /* CQ callback fired */ @@ -572,18 +586,29 @@ typedef struct kib_peer { struct list_head ibp_list; /* stash on global peer list */ lnet_nid_t ibp_nid; /* who's on the other end(s) */ lnet_ni_t *ibp_ni; /* LNet interface */ - atomic_t ibp_refcount; /* # users */ struct list_head ibp_conns; /* all active connections */ struct list_head ibp_tx_queue; /* msgs waiting for a conn */ - __u16 ibp_version; /* version of peer */ __u64 ibp_incarnation; /* incarnation of peer */ - int ibp_connecting; /* current active connection attempts - */ - int ibp_accepting; /* current passive connection attempts - */ - int ibp_error; /* errno on closing this peer */ - unsigned long ibp_last_alive; /* when (in jiffies) I was last alive - */ + /* when (in jiffies) I was last alive */ + unsigned long ibp_last_alive; + /* # users */ + atomic_t ibp_refcount; + /* version of peer */ + __u16 ibp_version; + /* current passive connection attempts */ + unsigned short ibp_accepting; + /* current active connection attempts */ + unsigned short ibp_connecting; + /* reconnect this peer later */ + unsigned short ibp_reconnecting:1; + /* # consecutive reconnection attempts to this peer */ + unsigned int ibp_reconnected; + /* errno on closing this peer */ + int ibp_error; + /* max map_on_demand */ + __u16 ibp_max_frags; + /* max_peer_credits */ + __u16 ibp_queue_depth; } kib_peer_t; extern kib_data_t kiblnd_data; @@ -611,7 +636,7 @@ kiblnd_dev_can_failover(kib_dev_t *dev) if (!list_empty(&dev->ibd_fail_list)) /* already scheduled */ return 0; - if (*kiblnd_tunables.kib_dev_failover == 0) /* disabled */ + if (!*kiblnd_tunables.kib_dev_failover) /* disabled */ return 0; if (*kiblnd_tunables.kib_dev_failover > 1) /* force failover */ @@ -661,6 +686,20 @@ do { \ kiblnd_destroy_peer(peer); \ } while (0) +static inline bool +kiblnd_peer_connecting(kib_peer_t *peer) +{ + return peer->ibp_connecting || + peer->ibp_reconnecting || + peer->ibp_accepting; +} + +static inline bool +kiblnd_peer_idle(kib_peer_t *peer) +{ + return !kiblnd_peer_connecting(peer) && list_empty(&peer->ibp_conns); +} + static inline struct list_head * kiblnd_nid2peerlist(lnet_nid_t nid) { @@ -691,7 +730,8 @@ kiblnd_send_keepalive(kib_conn_t *conn) { return (*kiblnd_tunables.kib_keepalive > 0) && cfs_time_after(jiffies, conn->ibc_last_send + - *kiblnd_tunables.kib_keepalive*HZ); + msecs_to_jiffies(*kiblnd_tunables.kib_keepalive * + MSEC_PER_SEC)); } static inline int @@ -710,16 +750,16 @@ kiblnd_need_noop(kib_conn_t *conn) /* No tx to piggyback NOOP onto or no credit to send a tx */ return (list_empty(&conn->ibc_tx_queue) || - conn->ibc_credits == 0); + !conn->ibc_credits); } if (!list_empty(&conn->ibc_tx_noops) || /* NOOP already queued */ !list_empty(&conn->ibc_tx_queue_nocred) || /* piggyback NOOP */ - conn->ibc_credits == 0) /* no credit */ + !conn->ibc_credits) /* no credit */ return 0; if (conn->ibc_credits == 1 && /* last credit reserved for */ - conn->ibc_outstanding_credits == 0) /* giving back credits */ + !conn->ibc_outstanding_credits) /* giving back credits */ return 0; /* No tx to piggyback NOOP onto or no credit to send a tx */ @@ -755,18 +795,19 @@ kiblnd_queue2str(kib_conn_t *conn, struct list_head *q) /* CAVEAT EMPTOR: We rely on descriptor alignment to allow us to use the */ /* lowest bits of the work request id to stash the work item type. */ -#define IBLND_WID_TX 0 -#define IBLND_WID_RDMA 1 -#define IBLND_WID_RX 2 -#define IBLND_WID_MASK 3UL +#define IBLND_WID_INVAL 0 +#define IBLND_WID_TX 1 +#define IBLND_WID_RX 2 +#define IBLND_WID_RDMA 3 +#define IBLND_WID_MASK 3UL static inline __u64 kiblnd_ptr2wreqid(void *ptr, int type) { unsigned long lptr = (unsigned long)ptr; - LASSERT((lptr & IBLND_WID_MASK) == 0); - LASSERT((type & ~IBLND_WID_MASK) == 0); + LASSERT(!(lptr & IBLND_WID_MASK)); + LASSERT(!(type & ~IBLND_WID_MASK)); return (__u64)(lptr | type); } @@ -907,9 +948,8 @@ static inline unsigned int kiblnd_sg_dma_len(struct ib_device *dev, #define KIBLND_CONN_PARAM_LEN(e) ((e)->param.conn.private_data_len) struct ib_mr *kiblnd_find_rd_dma_mr(kib_hca_dev_t *hdev, - kib_rdma_desc_t *rd); -struct ib_mr *kiblnd_find_dma_mr(kib_hca_dev_t *hdev, - __u64 addr, __u64 size); + kib_rdma_desc_t *rd, + int negotiated_nfrags); void kiblnd_map_rx_descs(kib_conn_t *conn); void kiblnd_unmap_rx_descs(kib_conn_t *conn); void kiblnd_pool_free_node(kib_pool_t *pool, struct list_head *node); @@ -919,11 +959,6 @@ int kiblnd_fmr_pool_map(kib_fmr_poolset_t *fps, __u64 *pages, int npages, __u64 iov, kib_fmr_t *fmr); void kiblnd_fmr_pool_unmap(kib_fmr_t *fmr, int status); -int kiblnd_startup(lnet_ni_t *ni); -void kiblnd_shutdown(lnet_ni_t *ni); -int kiblnd_ctl(lnet_ni_t *ni, unsigned int cmd, void *arg); -void kiblnd_query(struct lnet_ni *ni, lnet_nid_t nid, unsigned long *when); - int kiblnd_tunables_init(void); void kiblnd_tunables_fini(void); @@ -933,7 +968,6 @@ int kiblnd_thread_start(int (*fn)(void *arg), void *arg, char *name); int kiblnd_failover_thread(void *arg); int kiblnd_alloc_pages(kib_pages_t **pp, int cpt, int npages); -void kiblnd_free_pages(kib_pages_t *p); int kiblnd_cm_callback(struct rdma_cm_id *cmid, struct rdma_cm_event *event); @@ -942,39 +976,30 @@ int kiblnd_translate_mtu(int value); int kiblnd_dev_failover(kib_dev_t *dev); int kiblnd_create_peer(lnet_ni_t *ni, kib_peer_t **peerp, lnet_nid_t nid); void kiblnd_destroy_peer(kib_peer_t *peer); +bool kiblnd_reconnect_peer(kib_peer_t *peer); void kiblnd_destroy_dev(kib_dev_t *dev); void kiblnd_unlink_peer_locked(kib_peer_t *peer); -void kiblnd_peer_alive(kib_peer_t *peer); kib_peer_t *kiblnd_find_peer_locked(lnet_nid_t nid); -void kiblnd_peer_connect_failed(kib_peer_t *peer, int active, int error); int kiblnd_close_stale_conns_locked(kib_peer_t *peer, - int version, __u64 incarnation); + int version, __u64 incarnation); int kiblnd_close_peer_conns_locked(kib_peer_t *peer, int why); -void kiblnd_connreq_done(kib_conn_t *conn, int status); kib_conn_t *kiblnd_create_conn(kib_peer_t *peer, struct rdma_cm_id *cmid, - int state, int version); -void kiblnd_destroy_conn(kib_conn_t *conn); + int state, int version); +void kiblnd_destroy_conn(kib_conn_t *conn, bool free_conn); void kiblnd_close_conn(kib_conn_t *conn, int error); void kiblnd_close_conn_locked(kib_conn_t *conn, int error); -int kiblnd_init_rdma(kib_conn_t *conn, kib_tx_t *tx, int type, - int nob, kib_rdma_desc_t *dstrd, __u64 dstcookie); - void kiblnd_launch_tx(lnet_ni_t *ni, kib_tx_t *tx, lnet_nid_t nid); -void kiblnd_queue_tx_locked(kib_tx_t *tx, kib_conn_t *conn); -void kiblnd_queue_tx(kib_tx_t *tx, kib_conn_t *conn); -void kiblnd_init_tx_msg(lnet_ni_t *ni, kib_tx_t *tx, int type, int body_nob); void kiblnd_txlist_done(lnet_ni_t *ni, struct list_head *txlist, - int status); -void kiblnd_check_sends (kib_conn_t *conn); + int status); void kiblnd_qp_event(struct ib_event *event, void *arg); void kiblnd_cq_event(struct ib_event *event, void *arg); void kiblnd_cq_completion(struct ib_cq *cq, void *arg); void kiblnd_pack_msg(lnet_ni_t *ni, kib_msg_t *msg, int version, - int credits, lnet_nid_t dstnid, __u64 dststamp); + int credits, lnet_nid_t dstnid, __u64 dststamp); int kiblnd_unpack_msg(kib_msg_t *msg, int nob); int kiblnd_post_rx(kib_rx_t *rx, int credit); diff --git a/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd_cb.c b/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd_cb.c index c7b9ccb13..2323e8d3a 100644 --- a/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd_cb.c +++ b/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd_cb.c @@ -40,6 +40,15 @@ #include "o2iblnd.h" +static void kiblnd_peer_alive(kib_peer_t *peer); +static void kiblnd_peer_connect_failed(kib_peer_t *peer, int active, int error); +static void kiblnd_check_sends(kib_conn_t *conn); +static void kiblnd_init_tx_msg(lnet_ni_t *ni, kib_tx_t *tx, + int type, int body_nob); +static int kiblnd_init_rdma(kib_conn_t *conn, kib_tx_t *tx, int type, + int resid, kib_rdma_desc_t *dstrd, __u64 dstcookie); +static void kiblnd_queue_tx_locked(kib_tx_t *tx, kib_conn_t *conn); +static void kiblnd_queue_tx(kib_tx_t *tx, kib_conn_t *conn); static void kiblnd_unmap_tx(lnet_ni_t *ni, kib_tx_t *tx); static void @@ -50,12 +59,12 @@ kiblnd_tx_done(lnet_ni_t *ni, kib_tx_t *tx) int rc; int i; - LASSERT(net != NULL); + LASSERT(net); LASSERT(!in_interrupt()); LASSERT(!tx->tx_queued); /* mustn't be queued for sending */ - LASSERT(tx->tx_sending == 0); /* mustn't be awaiting sent callback */ + LASSERT(!tx->tx_sending); /* mustn't be awaiting sent callback */ LASSERT(!tx->tx_waiting); /* mustn't be awaiting peer response */ - LASSERT(tx->tx_pool != NULL); + LASSERT(tx->tx_pool); kiblnd_unmap_tx(ni, tx); @@ -64,7 +73,7 @@ kiblnd_tx_done(lnet_ni_t *ni, kib_tx_t *tx) lntmsg[1] = tx->tx_lntmsg[1]; tx->tx_lntmsg[1] = NULL; rc = tx->tx_status; - if (tx->tx_conn != NULL) { + if (tx->tx_conn) { LASSERT(ni == tx->tx_conn->ibc_peer->ibp_ni); kiblnd_conn_decref(tx->tx_conn); @@ -78,7 +87,7 @@ kiblnd_tx_done(lnet_ni_t *ni, kib_tx_t *tx) /* delay finalize until my descs have been freed */ for (i = 0; i < 2; i++) { - if (lntmsg[i] == NULL) + if (!lntmsg[i]) continue; lnet_finalize(ni, lntmsg[i], rc); @@ -111,19 +120,19 @@ kiblnd_get_idle_tx(lnet_ni_t *ni, lnet_nid_t target) tps = net->ibn_tx_ps[lnet_cpt_of_nid(target)]; node = kiblnd_pool_alloc_node(&tps->tps_poolset); - if (node == NULL) + if (!node) return NULL; - tx = container_of(node, kib_tx_t, tx_list); + tx = list_entry(node, kib_tx_t, tx_list); - LASSERT(tx->tx_nwrq == 0); + LASSERT(!tx->tx_nwrq); LASSERT(!tx->tx_queued); - LASSERT(tx->tx_sending == 0); + LASSERT(!tx->tx_sending); LASSERT(!tx->tx_waiting); - LASSERT(tx->tx_status == 0); - LASSERT(tx->tx_conn == NULL); - LASSERT(tx->tx_lntmsg[0] == NULL); - LASSERT(tx->tx_lntmsg[1] == NULL); - LASSERT(tx->tx_nfrags == 0); + LASSERT(!tx->tx_status); + LASSERT(!tx->tx_conn); + LASSERT(!tx->tx_lntmsg[0]); + LASSERT(!tx->tx_lntmsg[1]); + LASSERT(!tx->tx_nfrags); return tx; } @@ -149,17 +158,15 @@ kiblnd_post_rx(kib_rx_t *rx, int credit) kib_conn_t *conn = rx->rx_conn; kib_net_t *net = conn->ibc_peer->ibp_ni->ni_data; struct ib_recv_wr *bad_wrq = NULL; - struct ib_mr *mr; + struct ib_mr *mr = conn->ibc_hdev->ibh_mrs; int rc; - LASSERT(net != NULL); + LASSERT(net); LASSERT(!in_interrupt()); LASSERT(credit == IBLND_POSTRX_NO_CREDIT || credit == IBLND_POSTRX_PEER_CREDIT || credit == IBLND_POSTRX_RSRVD_CREDIT); - - mr = kiblnd_find_dma_mr(conn->ibc_hdev, rx->rx_msgaddr, IBLND_MSG_SIZE); - LASSERT(mr != NULL); + LASSERT(mr); rx->rx_sge.lkey = mr->lkey; rx->rx_sge.addr = rx->rx_msgaddr; @@ -185,7 +192,7 @@ kiblnd_post_rx(kib_rx_t *rx, int credit) */ kiblnd_conn_addref(conn); rc = ib_post_recv(conn->ibc_cmid->qp, &rx->rx_wrq, &bad_wrq); - if (unlikely(rc != 0)) { + if (unlikely(rc)) { CERROR("Can't post rx for %s: %d, bad_wrq: %p\n", libcfs_nid2str(conn->ibc_peer->ibp_nid), rc, bad_wrq); rx->rx_nob = 0; @@ -194,7 +201,7 @@ kiblnd_post_rx(kib_rx_t *rx, int credit) if (conn->ibc_state < IBLND_CONN_ESTABLISHED) /* Initial post */ goto out; - if (unlikely(rc != 0)) { + if (unlikely(rc)) { kiblnd_close_conn(conn, rc); kiblnd_drop_rx(rx); /* No more posts for this rx */ goto out; @@ -225,7 +232,7 @@ kiblnd_find_waiting_tx_locked(kib_conn_t *conn, int txtype, __u64 cookie) kib_tx_t *tx = list_entry(tmp, kib_tx_t, tx_list); LASSERT(!tx->tx_queued); - LASSERT(tx->tx_sending != 0 || tx->tx_waiting); + LASSERT(tx->tx_sending || tx->tx_waiting); if (tx->tx_cookie != cookie) continue; @@ -251,7 +258,7 @@ kiblnd_handle_completion(kib_conn_t *conn, int txtype, int status, __u64 cookie) spin_lock(&conn->ibc_lock); tx = kiblnd_find_waiting_tx_locked(conn, txtype, cookie); - if (tx == NULL) { + if (!tx) { spin_unlock(&conn->ibc_lock); CWARN("Unmatched completion type %x cookie %#llx from %s\n", @@ -260,7 +267,7 @@ kiblnd_handle_completion(kib_conn_t *conn, int txtype, int status, __u64 cookie) return; } - if (tx->tx_status == 0) { /* success so far */ + if (!tx->tx_status) { /* success so far */ if (status < 0) /* failed? */ tx->tx_status = status; else if (txtype == IBLND_MSG_GET_REQ) @@ -269,7 +276,7 @@ kiblnd_handle_completion(kib_conn_t *conn, int txtype, int status, __u64 cookie) tx->tx_waiting = 0; - idle = !tx->tx_queued && (tx->tx_sending == 0); + idle = !tx->tx_queued && !tx->tx_sending; if (idle) list_del(&tx->tx_list); @@ -285,7 +292,7 @@ kiblnd_send_completion(kib_conn_t *conn, int type, int status, __u64 cookie) lnet_ni_t *ni = conn->ibc_peer->ibp_ni; kib_tx_t *tx = kiblnd_get_idle_tx(ni, conn->ibc_peer->ibp_nid); - if (tx == NULL) { + if (!tx) { CERROR("Can't get tx for completion %x for %s\n", type, libcfs_nid2str(conn->ibc_peer->ibp_nid)); return; @@ -316,19 +323,18 @@ kiblnd_handle_rx(kib_rx_t *rx) msg->ibm_type, credits, libcfs_nid2str(conn->ibc_peer->ibp_nid)); - if (credits != 0) { + if (credits) { /* Have I received credits that will let me send? */ spin_lock(&conn->ibc_lock); if (conn->ibc_credits + credits > - IBLND_MSG_QUEUE_SIZE(conn->ibc_version)) { + conn->ibc_queue_depth) { rc2 = conn->ibc_credits; spin_unlock(&conn->ibc_lock); CERROR("Bad credits from %s: %d + %d > %d\n", libcfs_nid2str(conn->ibc_peer->ibp_nid), - rc2, credits, - IBLND_MSG_QUEUE_SIZE(conn->ibc_version)); + rc2, credits, conn->ibc_queue_depth); kiblnd_close_conn(conn, -EPROTO); kiblnd_post_rx(rx, IBLND_POSTRX_NO_CREDIT); @@ -360,7 +366,7 @@ kiblnd_handle_rx(kib_rx_t *rx) break; } - if (credits != 0) /* credit already posted */ + if (credits) /* credit already posted */ post_credit = IBLND_POSTRX_NO_CREDIT; else /* a keepalive NOOP */ post_credit = IBLND_POSTRX_PEER_CREDIT; @@ -396,12 +402,12 @@ kiblnd_handle_rx(kib_rx_t *rx) spin_lock(&conn->ibc_lock); tx = kiblnd_find_waiting_tx_locked(conn, IBLND_MSG_PUT_REQ, - msg->ibm_u.putack.ibpam_src_cookie); - if (tx != NULL) + msg->ibm_u.putack.ibpam_src_cookie); + if (tx) list_del(&tx->tx_list); spin_unlock(&conn->ibc_lock); - if (tx == NULL) { + if (!tx) { CERROR("Unmatched PUT_ACK from %s\n", libcfs_nid2str(conn->ibc_peer->ibp_nid)); rc = -EPROTO; @@ -409,10 +415,11 @@ kiblnd_handle_rx(kib_rx_t *rx) } LASSERT(tx->tx_waiting); - /* CAVEAT EMPTOR: I could be racing with tx_complete, but... + /* + * CAVEAT EMPTOR: I could be racing with tx_complete, but... * (a) I can overwrite tx_msg since my peer has received it! - * (b) tx_waiting set tells tx_complete() it's not done. */ - + * (b) tx_waiting set tells tx_complete() it's not done. + */ tx->tx_nwrq = 0; /* overwrite PUT_REQ */ rc2 = kiblnd_init_rdma(conn, tx, IBLND_MSG_PUT_DONE, @@ -469,7 +476,7 @@ kiblnd_rx_complete(kib_rx_t *rx, int status, int nob) int rc; int err = -EIO; - LASSERT(net != NULL); + LASSERT(net); LASSERT(rx->rx_nob < 0); /* was posted */ rx->rx_nob = 0; /* isn't now */ @@ -486,9 +493,9 @@ kiblnd_rx_complete(kib_rx_t *rx, int status, int nob) rx->rx_nob = nob; rc = kiblnd_unpack_msg(msg, rx->rx_nob); - if (rc != 0) { + if (rc) { CERROR("Error %d unpacking rx from %s\n", - rc, libcfs_nid2str(conn->ibc_peer->ibp_nid)); + rc, libcfs_nid2str(conn->ibc_peer->ibp_nid)); goto failed; } @@ -497,7 +504,7 @@ kiblnd_rx_complete(kib_rx_t *rx, int status, int nob) msg->ibm_srcstamp != conn->ibc_incarnation || msg->ibm_dststamp != net->ibn_incarnation) { CERROR("Stale rx from %s\n", - libcfs_nid2str(conn->ibc_peer->ibp_nid)); + libcfs_nid2str(conn->ibc_peer->ibp_nid)); err = -ESTALE; goto failed; } @@ -537,7 +544,7 @@ kiblnd_kvaddr_to_page(unsigned long vaddr) if (is_vmalloc_addr((void *)vaddr)) { page = vmalloc_to_page((void *)vaddr); - LASSERT(page != NULL); + LASSERT(page); return page; } #ifdef CONFIG_HIGHMEM @@ -549,7 +556,7 @@ kiblnd_kvaddr_to_page(unsigned long vaddr) } #endif page = virt_to_page(vaddr); - LASSERT(page != NULL); + LASSERT(page); return page; } @@ -565,8 +572,8 @@ kiblnd_fmr_map_tx(kib_net_t *net, kib_tx_t *tx, kib_rdma_desc_t *rd, int nob) int rc; int i; - LASSERT(tx->tx_pool != NULL); - LASSERT(tx->tx_pool->tpo_pool.po_owner != NULL); + LASSERT(tx->tx_pool); + LASSERT(tx->tx_pool->tpo_pool.po_owner); hdev = tx->tx_pool->tpo_hdev; @@ -582,13 +589,15 @@ kiblnd_fmr_map_tx(kib_net_t *net, kib_tx_t *tx, kib_rdma_desc_t *rd, int nob) fps = net->ibn_fmr_ps[cpt]; rc = kiblnd_fmr_pool_map(fps, pages, npages, 0, &tx->fmr); - if (rc != 0) { + if (rc) { CERROR("Can't map %d pages: %d\n", npages, rc); return rc; } - /* If rd is not tx_rd, it's going to get sent to a peer, who will need - * the rkey */ + /* + * If rd is not tx_rd, it's going to get sent to a peer, who will need + * the rkey + */ rd->rd_key = (rd != tx->tx_rd) ? tx->fmr.fmr_pfmr->fmr->rkey : tx->fmr.fmr_pfmr->fmr->lkey; rd->rd_frags[0].rf_addr &= ~hdev->ibh_page_mask; @@ -602,14 +611,14 @@ static void kiblnd_unmap_tx(lnet_ni_t *ni, kib_tx_t *tx) { kib_net_t *net = ni->ni_data; - LASSERT(net != NULL); + LASSERT(net); if (net->ibn_fmr_ps && tx->fmr.fmr_pfmr) { kiblnd_fmr_pool_unmap(&tx->fmr, tx->tx_status); tx->fmr.fmr_pfmr = NULL; } - if (tx->tx_nfrags != 0) { + if (tx->tx_nfrags) { kiblnd_dma_unmap_sg(tx->tx_pool->tpo_hdev->ibh_ibdev, tx->tx_frags, tx->tx_nfrags, tx->tx_dmadir); tx->tx_nfrags = 0; @@ -625,8 +634,10 @@ static int kiblnd_map_tx(lnet_ni_t *ni, kib_tx_t *tx, kib_rdma_desc_t *rd, __u32 nob; int i; - /* If rd is not tx_rd, it's going to get sent to a peer and I'm the - * RDMA sink */ + /* + * If rd is not tx_rd, it's going to get sent to a peer and I'm the + * RDMA sink + */ tx->tx_dmadir = (rd != tx->tx_rd) ? DMA_FROM_DEVICE : DMA_TO_DEVICE; tx->tx_nfrags = nfrags; @@ -641,15 +652,15 @@ static int kiblnd_map_tx(lnet_ni_t *ni, kib_tx_t *tx, kib_rdma_desc_t *rd, nob += rd->rd_frags[i].rf_nob; } - /* looking for pre-mapping MR */ - mr = kiblnd_find_rd_dma_mr(hdev, rd); - if (mr != NULL) { + mr = kiblnd_find_rd_dma_mr(hdev, rd, tx->tx_conn ? + tx->tx_conn->ibc_max_frags : -1); + if (mr) { /* found pre-mapping MR */ rd->rd_key = (rd != tx->tx_rd) ? mr->rkey : mr->lkey; return 0; } - if (net->ibn_fmr_ps != NULL) + if (net->ibn_fmr_ps) return kiblnd_fmr_map_tx(net, tx, rd, nob); return -EINVAL; @@ -668,7 +679,7 @@ kiblnd_setup_rd_iov(lnet_ni_t *ni, kib_tx_t *tx, kib_rdma_desc_t *rd, LASSERT(nob > 0); LASSERT(niov > 0); - LASSERT(net != NULL); + LASSERT(net); while (offset >= iov->iov_len) { offset -= iov->iov_len; @@ -684,7 +695,7 @@ kiblnd_setup_rd_iov(lnet_ni_t *ni, kib_tx_t *tx, kib_rdma_desc_t *rd, vaddr = ((unsigned long)iov->iov_base) + offset; page_offset = vaddr & (PAGE_SIZE - 1); page = kiblnd_kvaddr_to_page(vaddr); - if (page == NULL) { + if (!page) { CERROR("Can't find page\n"); return -EFAULT; } @@ -710,7 +721,7 @@ kiblnd_setup_rd_iov(lnet_ni_t *ni, kib_tx_t *tx, kib_rdma_desc_t *rd, static int kiblnd_setup_rd_kiov(lnet_ni_t *ni, kib_tx_t *tx, kib_rdma_desc_t *rd, - int nkiov, lnet_kiov_t *kiov, int offset, int nob) + int nkiov, lnet_kiov_t *kiov, int offset, int nob) { kib_net_t *net = ni->ni_data; struct scatterlist *sg; @@ -720,7 +731,7 @@ kiblnd_setup_rd_kiov(lnet_ni_t *ni, kib_tx_t *tx, kib_rdma_desc_t *rd, LASSERT(nob > 0); LASSERT(nkiov > 0); - LASSERT(net != NULL); + LASSERT(net); while (offset >= kiov->kiov_len) { offset -= kiov->kiov_len; @@ -750,26 +761,24 @@ kiblnd_setup_rd_kiov(lnet_ni_t *ni, kib_tx_t *tx, kib_rdma_desc_t *rd, static int kiblnd_post_tx_locked(kib_conn_t *conn, kib_tx_t *tx, int credit) - __releases(conn->ibc_lock) - __acquires(conn->ibc_lock) + __must_hold(&conn->ibc_lock) { kib_msg_t *msg = tx->tx_msg; kib_peer_t *peer = conn->ibc_peer; int ver = conn->ibc_version; int rc; int done; - struct ib_send_wr *bad_wrq; LASSERT(tx->tx_queued); /* We rely on this for QP sizing */ LASSERT(tx->tx_nwrq > 0); - LASSERT(tx->tx_nwrq <= 1 + IBLND_RDMA_FRAGS(ver)); + LASSERT(tx->tx_nwrq <= 1 + conn->ibc_max_frags); - LASSERT(credit == 0 || credit == 1); + LASSERT(!credit || credit == 1); LASSERT(conn->ibc_outstanding_credits >= 0); - LASSERT(conn->ibc_outstanding_credits <= IBLND_MSG_QUEUE_SIZE(ver)); + LASSERT(conn->ibc_outstanding_credits <= conn->ibc_queue_depth); LASSERT(conn->ibc_credits >= 0); - LASSERT(conn->ibc_credits <= IBLND_MSG_QUEUE_SIZE(ver)); + LASSERT(conn->ibc_credits <= conn->ibc_queue_depth); if (conn->ibc_nsends_posted == IBLND_CONCURRENT_SENDS(ver)) { /* tx completions outstanding... */ @@ -778,13 +787,13 @@ kiblnd_post_tx_locked(kib_conn_t *conn, kib_tx_t *tx, int credit) return -EAGAIN; } - if (credit != 0 && conn->ibc_credits == 0) { /* no credits */ + if (credit && !conn->ibc_credits) { /* no credits */ CDEBUG(D_NET, "%s: no credits\n", libcfs_nid2str(peer->ibp_nid)); return -EAGAIN; } - if (credit != 0 && !IBLND_OOB_CAPABLE(ver) && + if (credit && !IBLND_OOB_CAPABLE(ver) && conn->ibc_credits == 1 && /* last credit reserved */ msg->ibm_type != IBLND_MSG_NOOP) { /* for NOOP */ CDEBUG(D_NET, "%s: not using last credit\n", @@ -800,9 +809,11 @@ kiblnd_post_tx_locked(kib_conn_t *conn, kib_tx_t *tx, int credit) (!kiblnd_need_noop(conn) || /* redundant NOOP */ (IBLND_OOB_CAPABLE(ver) && /* posted enough NOOP */ conn->ibc_noops_posted == IBLND_OOB_MSGS(ver)))) { - /* OK to drop when posted enough NOOPs, since + /* + * OK to drop when posted enough NOOPs, since * kiblnd_check_sends will queue NOOP again when - * posted NOOPs complete */ + * posted NOOPs complete + */ spin_unlock(&conn->ibc_lock); kiblnd_tx_done(peer->ibp_ni, tx); spin_lock(&conn->ibc_lock); @@ -821,12 +832,14 @@ kiblnd_post_tx_locked(kib_conn_t *conn, kib_tx_t *tx, int credit) if (msg->ibm_type == IBLND_MSG_NOOP) conn->ibc_noops_posted++; - /* CAVEAT EMPTOR! This tx could be the PUT_DONE of an RDMA + /* + * CAVEAT EMPTOR! This tx could be the PUT_DONE of an RDMA * PUT. If so, it was first queued here as a PUT_REQ, sent and * stashed on ibc_active_txs, matched by an incoming PUT_ACK, * and then re-queued here. It's (just) possible that * tx_sending is non-zero if we've not done the tx_complete() - * from the first send; hence the ++ rather than = below. */ + * from the first send; hence the ++ rather than = below. + */ tx->tx_sending++; list_add(&tx->tx_list, &conn->ibc_active_txs); @@ -838,16 +851,25 @@ kiblnd_post_tx_locked(kib_conn_t *conn, kib_tx_t *tx, int credit) /* close_conn will launch failover */ rc = -ENETDOWN; } else { - rc = ib_post_send(conn->ibc_cmid->qp, &tx->tx_wrq->wr, &bad_wrq); + struct ib_send_wr *wrq = &tx->tx_wrq[tx->tx_nwrq - 1].wr; + + LASSERTF(wrq->wr_id == kiblnd_ptr2wreqid(tx, IBLND_WID_TX), + "bad wr_id %llx, opc %d, flags %d, peer: %s\n", + wrq->wr_id, wrq->opcode, wrq->send_flags, + libcfs_nid2str(conn->ibc_peer->ibp_nid)); + wrq = NULL; + rc = ib_post_send(conn->ibc_cmid->qp, &tx->tx_wrq->wr, &wrq); } conn->ibc_last_send = jiffies; - if (rc == 0) + if (!rc) return 0; - /* NB credits are transferred in the actual - * message, which can only be the last work item */ + /* + * NB credits are transferred in the actual + * message, which can only be the last work item + */ conn->ibc_credits += credit; conn->ibc_outstanding_credits += msg->ibm_credits; conn->ibc_nsends_posted--; @@ -858,7 +880,7 @@ kiblnd_post_tx_locked(kib_conn_t *conn, kib_tx_t *tx, int credit) tx->tx_waiting = 0; tx->tx_sending--; - done = (tx->tx_sending == 0); + done = !tx->tx_sending; if (done) list_del(&tx->tx_list); @@ -881,7 +903,7 @@ kiblnd_post_tx_locked(kib_conn_t *conn, kib_tx_t *tx, int credit) return -EIO; } -void +static void kiblnd_check_sends(kib_conn_t *conn) { int ver = conn->ibc_version; @@ -899,13 +921,13 @@ kiblnd_check_sends(kib_conn_t *conn) LASSERT(conn->ibc_nsends_posted <= IBLND_CONCURRENT_SENDS(ver)); LASSERT(!IBLND_OOB_CAPABLE(ver) || - conn->ibc_noops_posted <= IBLND_OOB_MSGS(ver)); + conn->ibc_noops_posted <= IBLND_OOB_MSGS(ver)); LASSERT(conn->ibc_reserved_credits >= 0); while (conn->ibc_reserved_credits > 0 && !list_empty(&conn->ibc_tx_queue_rsrvd)) { tx = list_entry(conn->ibc_tx_queue_rsrvd.next, - kib_tx_t, tx_list); + kib_tx_t, tx_list); list_del(&tx->tx_list); list_add_tail(&tx->tx_list, &conn->ibc_tx_queue); conn->ibc_reserved_credits--; @@ -915,23 +937,21 @@ kiblnd_check_sends(kib_conn_t *conn) spin_unlock(&conn->ibc_lock); tx = kiblnd_get_idle_tx(ni, conn->ibc_peer->ibp_nid); - if (tx != NULL) + if (tx) kiblnd_init_tx_msg(ni, tx, IBLND_MSG_NOOP, 0); spin_lock(&conn->ibc_lock); - if (tx != NULL) + if (tx) kiblnd_queue_tx_locked(tx, conn); } - kiblnd_conn_addref(conn); /* 1 ref for me.... (see b21911) */ - for (;;) { int credit; if (!list_empty(&conn->ibc_tx_queue_nocred)) { credit = 0; tx = list_entry(conn->ibc_tx_queue_nocred.next, - kib_tx_t, tx_list); + kib_tx_t, tx_list); } else if (!list_empty(&conn->ibc_tx_noops)) { LASSERT(!IBLND_OOB_CAPABLE(ver)); credit = 1; @@ -940,17 +960,16 @@ kiblnd_check_sends(kib_conn_t *conn) } else if (!list_empty(&conn->ibc_tx_queue)) { credit = 1; tx = list_entry(conn->ibc_tx_queue.next, - kib_tx_t, tx_list); - } else + kib_tx_t, tx_list); + } else { break; + } - if (kiblnd_post_tx_locked(conn, tx, credit) != 0) + if (kiblnd_post_tx_locked(conn, tx, credit)) break; } spin_unlock(&conn->ibc_lock); - - kiblnd_conn_decref(conn); /* ...until here */ } static void @@ -976,9 +995,10 @@ kiblnd_tx_complete(kib_tx_t *tx, int status) spin_lock(&conn->ibc_lock); - /* I could be racing with rdma completion. Whoever makes 'tx' idle - * gets to free it, which also drops its ref on 'conn'. */ - + /* + * I could be racing with rdma completion. Whoever makes 'tx' idle + * gets to free it, which also drops its ref on 'conn'. + */ tx->tx_sending--; conn->ibc_nsends_posted--; if (tx->tx_msg->ibm_type == IBLND_MSG_NOOP) @@ -989,7 +1009,7 @@ kiblnd_tx_complete(kib_tx_t *tx, int status) tx->tx_status = -EIO; } - idle = (tx->tx_sending == 0) && /* This is the final callback */ + idle = !tx->tx_sending && /* This is the final callback */ !tx->tx_waiting && /* Not waiting for peer */ !tx->tx_queued; /* Not re-queued (PUT_DONE) */ if (idle) @@ -1007,24 +1027,22 @@ kiblnd_tx_complete(kib_tx_t *tx, int status) kiblnd_conn_decref(conn); /* ...until here */ } -void +static void kiblnd_init_tx_msg(lnet_ni_t *ni, kib_tx_t *tx, int type, int body_nob) { kib_hca_dev_t *hdev = tx->tx_pool->tpo_hdev; struct ib_sge *sge = &tx->tx_sge[tx->tx_nwrq]; struct ib_rdma_wr *wrq = &tx->tx_wrq[tx->tx_nwrq]; int nob = offsetof(kib_msg_t, ibm_u) + body_nob; - struct ib_mr *mr; + struct ib_mr *mr = hdev->ibh_mrs; LASSERT(tx->tx_nwrq >= 0); LASSERT(tx->tx_nwrq < IBLND_MAX_RDMA_FRAGS + 1); LASSERT(nob <= IBLND_MSG_SIZE); + LASSERT(mr); kiblnd_init_msg(tx->tx_msg, type, body_nob); - mr = kiblnd_find_dma_mr(hdev, tx->tx_msgaddr, nob); - LASSERT(mr != NULL); - sge->lkey = mr->lkey; sge->addr = tx->tx_msgaddr; sge->length = nob; @@ -1041,25 +1059,23 @@ kiblnd_init_tx_msg(lnet_ni_t *ni, kib_tx_t *tx, int type, int body_nob) tx->tx_nwrq++; } -int +static int kiblnd_init_rdma(kib_conn_t *conn, kib_tx_t *tx, int type, - int resid, kib_rdma_desc_t *dstrd, __u64 dstcookie) + int resid, kib_rdma_desc_t *dstrd, __u64 dstcookie) { kib_msg_t *ibmsg = tx->tx_msg; kib_rdma_desc_t *srcrd = tx->tx_rd; struct ib_sge *sge = &tx->tx_sge[0]; struct ib_rdma_wr *wrq = &tx->tx_wrq[0], *next; int rc = resid; - int srcidx; - int dstidx; + int srcidx = 0; + int dstidx = 0; int wrknob; LASSERT(!in_interrupt()); - LASSERT(tx->tx_nwrq == 0); + LASSERT(!tx->tx_nwrq); LASSERT(type == IBLND_MSG_GET_DONE || - type == IBLND_MSG_PUT_DONE); - - srcidx = dstidx = 0; + type == IBLND_MSG_PUT_DONE); while (resid > 0) { if (srcidx >= srcrd->rd_nfrags) { @@ -1074,10 +1090,10 @@ kiblnd_init_rdma(kib_conn_t *conn, kib_tx_t *tx, int type, break; } - if (tx->tx_nwrq == IBLND_RDMA_FRAGS(conn->ibc_version)) { - CERROR("RDMA too fragmented for %s (%d): %d/%d src %d/%d dst frags\n", + if (tx->tx_nwrq >= conn->ibc_max_frags) { + CERROR("RDMA has too many fragments for peer %s (%d), src idx/frags: %d/%d dst idx/frags: %d/%d\n", libcfs_nid2str(conn->ibc_peer->ibp_nid), - IBLND_RDMA_FRAGS(conn->ibc_version), + conn->ibc_max_frags, srcidx, srcrd->rd_nfrags, dstidx, dstrd->rd_nfrags); rc = -EMSGSIZE; @@ -1127,7 +1143,7 @@ kiblnd_init_rdma(kib_conn_t *conn, kib_tx_t *tx, int type, return rc; } -void +static void kiblnd_queue_tx_locked(kib_tx_t *tx, kib_conn_t *conn) { struct list_head *q; @@ -1137,9 +1153,11 @@ kiblnd_queue_tx_locked(kib_tx_t *tx, kib_conn_t *conn) LASSERT(conn->ibc_state >= IBLND_CONN_ESTABLISHED); tx->tx_queued = 1; - tx->tx_deadline = jiffies + (*kiblnd_tunables.kib_timeout * HZ); + tx->tx_deadline = jiffies + + msecs_to_jiffies(*kiblnd_tunables.kib_timeout * + MSEC_PER_SEC); - if (tx->tx_conn == NULL) { + if (!tx->tx_conn) { kiblnd_conn_addref(conn); tx->tx_conn = conn; LASSERT(tx->tx_msg->ibm_type != IBLND_MSG_PUT_DONE); @@ -1180,7 +1198,7 @@ kiblnd_queue_tx_locked(kib_tx_t *tx, kib_conn_t *conn) list_add_tail(&tx->tx_list, q); } -void +static void kiblnd_queue_tx(kib_tx_t *tx, kib_conn_t *conn) { spin_lock(&conn->ibc_lock); @@ -1200,19 +1218,19 @@ static int kiblnd_resolve_addr(struct rdma_cm_id *cmid, /* allow the port to be reused */ rc = rdma_set_reuseaddr(cmid, 1); - if (rc != 0) { + if (rc) { CERROR("Unable to set reuse on cmid: %d\n", rc); return rc; } /* look for a free privileged port */ - for (port = PROT_SOCK-1; port > 0; port--) { + for (port = PROT_SOCK - 1; port > 0; port--) { srcaddr->sin_port = htons(port); rc = rdma_resolve_addr(cmid, (struct sockaddr *)srcaddr, (struct sockaddr *)dstaddr, timeout_ms); - if (rc == 0) { + if (!rc) { CDEBUG(D_NET, "bound to port %hu\n", port); return 0; } else if (rc == -EADDRINUSE || rc == -EADDRNOTAVAIL) { @@ -1237,8 +1255,9 @@ kiblnd_connect_peer(kib_peer_t *peer) struct sockaddr_in dstaddr; int rc; - LASSERT(net != NULL); + LASSERT(net); LASSERT(peer->ibp_connecting > 0); + LASSERT(!peer->ibp_reconnecting); cmid = kiblnd_rdma_create_id(kiblnd_cm_callback, peer, RDMA_PS_TCP, IB_QPT_RC); @@ -1271,14 +1290,14 @@ kiblnd_connect_peer(kib_peer_t *peer) (struct sockaddr *)&dstaddr, *kiblnd_tunables.kib_timeout * 1000); } - if (rc != 0) { + if (rc) { /* Can't initiate address resolution: */ CERROR("Can't resolve addr for %s: %d\n", libcfs_nid2str(peer->ibp_nid), rc); goto failed2; } - LASSERT(cmid->device != NULL); + LASSERT(cmid->device); CDEBUG(D_NET, "%s: connection bound to %s:%pI4h:%s\n", libcfs_nid2str(peer->ibp_nid), dev->ibd_ifname, &dev->ibd_ifip, cmid->device->name); @@ -1286,12 +1305,64 @@ kiblnd_connect_peer(kib_peer_t *peer) return; failed2: + kiblnd_peer_connect_failed(peer, 1, rc); kiblnd_peer_decref(peer); /* cmid's ref */ rdma_destroy_id(cmid); + return; failed: kiblnd_peer_connect_failed(peer, 1, rc); } +bool +kiblnd_reconnect_peer(kib_peer_t *peer) +{ + rwlock_t *glock = &kiblnd_data.kib_global_lock; + char *reason = NULL; + struct list_head txs; + unsigned long flags; + + INIT_LIST_HEAD(&txs); + + write_lock_irqsave(glock, flags); + if (!peer->ibp_reconnecting) { + if (peer->ibp_accepting) + reason = "accepting"; + else if (peer->ibp_connecting) + reason = "connecting"; + else if (!list_empty(&peer->ibp_conns)) + reason = "connected"; + else /* connected then closed */ + reason = "closed"; + + goto no_reconnect; + } + + LASSERT(!peer->ibp_accepting && !peer->ibp_connecting && + list_empty(&peer->ibp_conns)); + peer->ibp_reconnecting = 0; + + if (!kiblnd_peer_active(peer)) { + list_splice_init(&peer->ibp_tx_queue, &txs); + reason = "unlinked"; + goto no_reconnect; + } + + peer->ibp_connecting++; + peer->ibp_reconnected++; + write_unlock_irqrestore(glock, flags); + + kiblnd_connect_peer(peer); + return true; + +no_reconnect: + write_unlock_irqrestore(glock, flags); + + CWARN("Abort reconnection of %s: %s\n", + libcfs_nid2str(peer->ibp_nid), reason); + kiblnd_txlist_done(peer->ibp_ni, &txs, -ECONNABORTED); + return false; +} + void kiblnd_launch_tx(lnet_ni_t *ni, kib_tx_t *tx, lnet_nid_t nid) { @@ -1302,25 +1373,28 @@ kiblnd_launch_tx(lnet_ni_t *ni, kib_tx_t *tx, lnet_nid_t nid) unsigned long flags; int rc; - /* If I get here, I've committed to send, so I complete the tx with - * failure on any problems */ - - LASSERT(tx == NULL || tx->tx_conn == NULL); /* only set when assigned a conn */ - LASSERT(tx == NULL || tx->tx_nwrq > 0); /* work items have been set up */ + /* + * If I get here, I've committed to send, so I complete the tx with + * failure on any problems + */ + LASSERT(!tx || !tx->tx_conn); /* only set when assigned a conn */ + LASSERT(!tx || tx->tx_nwrq > 0); /* work items have been set up */ - /* First time, just use a read lock since I expect to find my peer - * connected */ + /* + * First time, just use a read lock since I expect to find my peer + * connected + */ read_lock_irqsave(g_lock, flags); peer = kiblnd_find_peer_locked(nid); - if (peer != NULL && !list_empty(&peer->ibp_conns)) { + if (peer && !list_empty(&peer->ibp_conns)) { /* Found a peer with an established connection */ conn = kiblnd_get_conn_locked(peer); kiblnd_conn_addref(conn); /* 1 ref for me... */ read_unlock_irqrestore(g_lock, flags); - if (tx != NULL) + if (tx) kiblnd_queue_tx(tx, conn); kiblnd_conn_decref(conn); /* ...to here */ return; @@ -1331,14 +1405,13 @@ kiblnd_launch_tx(lnet_ni_t *ni, kib_tx_t *tx, lnet_nid_t nid) write_lock(g_lock); peer = kiblnd_find_peer_locked(nid); - if (peer != NULL) { + if (peer) { if (list_empty(&peer->ibp_conns)) { /* found a peer, but it's still connecting... */ - LASSERT(peer->ibp_connecting != 0 || - peer->ibp_accepting != 0); - if (tx != NULL) + LASSERT(kiblnd_peer_connecting(peer)); + if (tx) list_add_tail(&tx->tx_list, - &peer->ibp_tx_queue); + &peer->ibp_tx_queue); write_unlock_irqrestore(g_lock, flags); } else { conn = kiblnd_get_conn_locked(peer); @@ -1346,7 +1419,7 @@ kiblnd_launch_tx(lnet_ni_t *ni, kib_tx_t *tx, lnet_nid_t nid) write_unlock_irqrestore(g_lock, flags); - if (tx != NULL) + if (tx) kiblnd_queue_tx(tx, conn); kiblnd_conn_decref(conn); /* ...to here */ } @@ -1357,9 +1430,9 @@ kiblnd_launch_tx(lnet_ni_t *ni, kib_tx_t *tx, lnet_nid_t nid) /* Allocate a peer ready to add to the peer table and retry */ rc = kiblnd_create_peer(ni, &peer, nid); - if (rc != 0) { + if (rc) { CERROR("Can't create peer %s\n", libcfs_nid2str(nid)); - if (tx != NULL) { + if (tx) { tx->tx_status = -EHOSTUNREACH; tx->tx_waiting = 0; kiblnd_tx_done(ni, tx); @@ -1370,14 +1443,13 @@ kiblnd_launch_tx(lnet_ni_t *ni, kib_tx_t *tx, lnet_nid_t nid) write_lock_irqsave(g_lock, flags); peer2 = kiblnd_find_peer_locked(nid); - if (peer2 != NULL) { + if (peer2) { if (list_empty(&peer2->ibp_conns)) { /* found a peer, but it's still connecting... */ - LASSERT(peer2->ibp_connecting != 0 || - peer2->ibp_accepting != 0); - if (tx != NULL) + LASSERT(kiblnd_peer_connecting(peer2)); + if (tx) list_add_tail(&tx->tx_list, - &peer2->ibp_tx_queue); + &peer2->ibp_tx_queue); write_unlock_irqrestore(g_lock, flags); } else { conn = kiblnd_get_conn_locked(peer2); @@ -1385,7 +1457,7 @@ kiblnd_launch_tx(lnet_ni_t *ni, kib_tx_t *tx, lnet_nid_t nid) write_unlock_irqrestore(g_lock, flags); - if (tx != NULL) + if (tx) kiblnd_queue_tx(tx, conn); kiblnd_conn_decref(conn); /* ...to here */ } @@ -1395,13 +1467,13 @@ kiblnd_launch_tx(lnet_ni_t *ni, kib_tx_t *tx, lnet_nid_t nid) } /* Brand new peer */ - LASSERT(peer->ibp_connecting == 0); + LASSERT(!peer->ibp_connecting); peer->ibp_connecting = 1; /* always called with a ref on ni, which prevents ni being shutdown */ - LASSERT(((kib_net_t *)ni->ni_data)->ibn_shutdown == 0); + LASSERT(!((kib_net_t *)ni->ni_data)->ibn_shutdown); - if (tx != NULL) + if (tx) list_add_tail(&tx->tx_list, &peer->ibp_tx_queue); kiblnd_peer_addref(peer); @@ -1437,13 +1509,13 @@ kiblnd_send(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg) CDEBUG(D_NET, "sending %d bytes in %d frags to %s\n", payload_nob, payload_niov, libcfs_id2str(target)); - LASSERT(payload_nob == 0 || payload_niov > 0); + LASSERT(!payload_nob || payload_niov > 0); LASSERT(payload_niov <= LNET_MAX_IOV); /* Thread context */ LASSERT(!in_interrupt()); /* payload is either all vaddrs or all pages */ - LASSERT(!(payload_kiov != NULL && payload_iov != NULL)); + LASSERT(!(payload_kiov && payload_iov)); switch (type) { default: @@ -1451,7 +1523,7 @@ kiblnd_send(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg) return -EIO; case LNET_MSG_ACK: - LASSERT(payload_nob == 0); + LASSERT(!payload_nob); break; case LNET_MSG_GET: @@ -1464,7 +1536,7 @@ kiblnd_send(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg) break; /* send IMMEDIATE */ tx = kiblnd_get_idle_tx(ni, target.nid); - if (tx == NULL) { + if (!tx) { CERROR("Can't allocate txd for GET to %s\n", libcfs_nid2str(target.nid)); return -ENOMEM; @@ -1472,7 +1544,7 @@ kiblnd_send(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg) ibmsg = tx->tx_msg; rd = &ibmsg->ibm_u.get.ibgm_rd; - if ((lntmsg->msg_md->md_options & LNET_MD_KIOV) == 0) + if (!(lntmsg->msg_md->md_options & LNET_MD_KIOV)) rc = kiblnd_setup_rd_iov(ni, tx, rd, lntmsg->msg_md->md_niov, lntmsg->msg_md->md_iov.iov, @@ -1482,7 +1554,7 @@ kiblnd_send(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg) lntmsg->msg_md->md_niov, lntmsg->msg_md->md_iov.kiov, 0, lntmsg->msg_md->md_length); - if (rc != 0) { + if (rc) { CERROR("Can't setup GET sink for %s: %d\n", libcfs_nid2str(target.nid), rc); kiblnd_tx_done(ni, tx); @@ -1496,7 +1568,7 @@ kiblnd_send(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg) kiblnd_init_tx_msg(ni, tx, IBLND_MSG_GET_REQ, nob); tx->tx_lntmsg[1] = lnet_create_reply_msg(ni, lntmsg); - if (tx->tx_lntmsg[1] == NULL) { + if (!tx->tx_lntmsg[1]) { CERROR("Can't create reply for GET -> %s\n", libcfs_nid2str(target.nid)); kiblnd_tx_done(ni, tx); @@ -1516,14 +1588,14 @@ kiblnd_send(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg) break; /* send IMMEDIATE */ tx = kiblnd_get_idle_tx(ni, target.nid); - if (tx == NULL) { + if (!tx) { CERROR("Can't allocate %s txd for %s\n", type == LNET_MSG_PUT ? "PUT" : "REPLY", libcfs_nid2str(target.nid)); return -ENOMEM; } - if (payload_kiov == NULL) + if (!payload_kiov) rc = kiblnd_setup_rd_iov(ni, tx, tx->tx_rd, payload_niov, payload_iov, payload_offset, payload_nob); @@ -1531,7 +1603,7 @@ kiblnd_send(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg) rc = kiblnd_setup_rd_kiov(ni, tx, tx->tx_rd, payload_niov, payload_kiov, payload_offset, payload_nob); - if (rc != 0) { + if (rc) { CERROR("Can't setup PUT src for %s: %d\n", libcfs_nid2str(target.nid), rc); kiblnd_tx_done(ni, tx); @@ -1555,16 +1627,16 @@ kiblnd_send(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg) <= IBLND_MSG_SIZE); tx = kiblnd_get_idle_tx(ni, target.nid); - if (tx == NULL) { + if (!tx) { CERROR("Can't send %d to %s: tx descs exhausted\n", - type, libcfs_nid2str(target.nid)); + type, libcfs_nid2str(target.nid)); return -ENOMEM; } ibmsg = tx->tx_msg; ibmsg->ibm_u.immediate.ibim_hdr = *hdr; - if (payload_kiov != NULL) + if (payload_kiov) lnet_copy_kiov2flat(IBLND_MSG_SIZE, ibmsg, offsetof(kib_msg_t, ibm_u.immediate.ibim_payload), payload_niov, payload_kiov, @@ -1596,22 +1668,22 @@ kiblnd_reply(lnet_ni_t *ni, kib_rx_t *rx, lnet_msg_t *lntmsg) int rc; tx = kiblnd_get_idle_tx(ni, rx->rx_conn->ibc_peer->ibp_nid); - if (tx == NULL) { + if (!tx) { CERROR("Can't get tx for REPLY to %s\n", libcfs_nid2str(target.nid)); goto failed_0; } - if (nob == 0) + if (!nob) rc = 0; - else if (kiov == NULL) + else if (!kiov) rc = kiblnd_setup_rd_iov(ni, tx, tx->tx_rd, niov, iov, offset, nob); else rc = kiblnd_setup_rd_kiov(ni, tx, tx->tx_rd, niov, kiov, offset, nob); - if (rc != 0) { + if (rc) { CERROR("Can't setup GET src for %s: %d\n", libcfs_nid2str(target.nid), rc); goto failed_1; @@ -1627,12 +1699,11 @@ kiblnd_reply(lnet_ni_t *ni, kib_rx_t *rx, lnet_msg_t *lntmsg) goto failed_1; } - if (nob == 0) { + if (!nob) { /* No RDMA: local completion may happen now! */ lnet_finalize(ni, lntmsg, 0); } else { - /* RDMA: lnet_finalize(lntmsg) when it - * completes */ + /* RDMA: lnet_finalize(lntmsg) when it completes */ tx->tx_lntmsg[0] = lntmsg; } @@ -1647,8 +1718,8 @@ kiblnd_reply(lnet_ni_t *ni, kib_rx_t *rx, lnet_msg_t *lntmsg) int kiblnd_recv(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg, int delayed, - unsigned int niov, struct kvec *iov, lnet_kiov_t *kiov, - unsigned int offset, unsigned int mlen, unsigned int rlen) + unsigned int niov, struct kvec *iov, lnet_kiov_t *kiov, + unsigned int offset, unsigned int mlen, unsigned int rlen) { kib_rx_t *rx = private; kib_msg_t *rxmsg = rx->rx_msg; @@ -1661,7 +1732,7 @@ kiblnd_recv(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg, int delayed, LASSERT(mlen <= rlen); LASSERT(!in_interrupt()); /* Either all pages or all vaddrs */ - LASSERT(!(kiov != NULL && iov != NULL)); + LASSERT(!(kiov && iov)); switch (rxmsg->ibm_type) { default: @@ -1671,13 +1742,13 @@ kiblnd_recv(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg, int delayed, nob = offsetof(kib_msg_t, ibm_u.immediate.ibim_payload[rlen]); if (nob > rx->rx_nob) { CERROR("Immediate message from %s too big: %d(%d)\n", - libcfs_nid2str(rxmsg->ibm_u.immediate.ibim_hdr.src_nid), - nob, rx->rx_nob); + libcfs_nid2str(rxmsg->ibm_u.immediate.ibim_hdr.src_nid), + nob, rx->rx_nob); rc = -EPROTO; break; } - if (kiov != NULL) + if (kiov) lnet_copy_flat2kiov(niov, kiov, offset, IBLND_MSG_SIZE, rxmsg, offsetof(kib_msg_t, ibm_u.immediate.ibim_payload), @@ -1694,7 +1765,7 @@ kiblnd_recv(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg, int delayed, kib_msg_t *txmsg; kib_rdma_desc_t *rd; - if (mlen == 0) { + if (!mlen) { lnet_finalize(ni, lntmsg, 0); kiblnd_send_completion(rx->rx_conn, IBLND_MSG_PUT_NAK, 0, rxmsg->ibm_u.putreq.ibprm_cookie); @@ -1702,7 +1773,7 @@ kiblnd_recv(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg, int delayed, } tx = kiblnd_get_idle_tx(ni, conn->ibc_peer->ibp_nid); - if (tx == NULL) { + if (!tx) { CERROR("Can't allocate tx for %s\n", libcfs_nid2str(conn->ibc_peer->ibp_nid)); /* Not replying will break the connection */ @@ -1712,13 +1783,13 @@ kiblnd_recv(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg, int delayed, txmsg = tx->tx_msg; rd = &txmsg->ibm_u.putack.ibpam_rd; - if (kiov == NULL) + if (!kiov) rc = kiblnd_setup_rd_iov(ni, tx, rd, niov, iov, offset, mlen); else rc = kiblnd_setup_rd_kiov(ni, tx, rd, niov, kiov, offset, mlen); - if (rc != 0) { + if (rc) { CERROR("Can't setup PUT sink for %s: %d\n", libcfs_nid2str(conn->ibc_peer->ibp_nid), rc); kiblnd_tx_done(ni, tx); @@ -1744,7 +1815,7 @@ kiblnd_recv(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg, int delayed, } case IBLND_MSG_GET_REQ: - if (lntmsg != NULL) { + if (lntmsg) { /* Optimized GET; RDMA lntmsg's payload */ kiblnd_reply(ni, rx, lntmsg); } else { @@ -1778,7 +1849,7 @@ kiblnd_thread_fini(void) atomic_dec(&kiblnd_data.kib_nthreads); } -void +static void kiblnd_peer_alive(kib_peer_t *peer) { /* This is racy, but everyone's only writing cfs_time_current() */ @@ -1795,10 +1866,7 @@ kiblnd_peer_notify(kib_peer_t *peer) read_lock_irqsave(&kiblnd_data.kib_global_lock, flags); - if (list_empty(&peer->ibp_conns) && - peer->ibp_accepting == 0 && - peer->ibp_connecting == 0 && - peer->ibp_error != 0) { + if (kiblnd_peer_idle(peer) && peer->ibp_error) { error = peer->ibp_error; peer->ibp_error = 0; @@ -1807,7 +1875,7 @@ kiblnd_peer_notify(kib_peer_t *peer) read_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags); - if (error != 0) + if (error) lnet_notify(peer->ibp_ni, peer->ibp_nid, 0, last_alive); } @@ -1815,25 +1883,27 @@ kiblnd_peer_notify(kib_peer_t *peer) void kiblnd_close_conn_locked(kib_conn_t *conn, int error) { - /* This just does the immediate housekeeping. 'error' is zero for a + /* + * This just does the immediate housekeeping. 'error' is zero for a * normal shutdown which can happen only after the connection has been * established. If the connection is established, schedule the - * connection to be finished off by the connd. Otherwise the connd is + * connection to be finished off by the connd. Otherwise the connd is * already dealing with it (either to set it up or tear it down). - * Caller holds kib_global_lock exclusively in irq context */ + * Caller holds kib_global_lock exclusively in irq context + */ kib_peer_t *peer = conn->ibc_peer; kib_dev_t *dev; unsigned long flags; - LASSERT(error != 0 || conn->ibc_state >= IBLND_CONN_ESTABLISHED); + LASSERT(error || conn->ibc_state >= IBLND_CONN_ESTABLISHED); - if (error != 0 && conn->ibc_comms_error == 0) + if (error && !conn->ibc_comms_error) conn->ibc_comms_error = error; if (conn->ibc_state != IBLND_CONN_ESTABLISHED) return; /* already being handled */ - if (error == 0 && + if (!error && list_empty(&conn->ibc_tx_noops) && list_empty(&conn->ibc_tx_queue) && list_empty(&conn->ibc_tx_queue_rsrvd) && @@ -1843,12 +1913,12 @@ kiblnd_close_conn_locked(kib_conn_t *conn, int error) libcfs_nid2str(peer->ibp_nid)); } else { CNETERR("Closing conn to %s: error %d%s%s%s%s%s\n", - libcfs_nid2str(peer->ibp_nid), error, - list_empty(&conn->ibc_tx_queue) ? "" : "(sending)", - list_empty(&conn->ibc_tx_noops) ? "" : "(sending_noops)", - list_empty(&conn->ibc_tx_queue_rsrvd) ? "" : "(sending_rsrvd)", - list_empty(&conn->ibc_tx_queue_nocred) ? "" : "(sending_nocred)", - list_empty(&conn->ibc_active_txs) ? "" : "(waiting)"); + libcfs_nid2str(peer->ibp_nid), error, + list_empty(&conn->ibc_tx_queue) ? "" : "(sending)", + list_empty(&conn->ibc_tx_noops) ? "" : "(sending_noops)", + list_empty(&conn->ibc_tx_queue_rsrvd) ? "" : "(sending_rsrvd)", + list_empty(&conn->ibc_tx_queue_nocred) ? "" : "(sending_nocred)", + list_empty(&conn->ibc_active_txs) ? "" : "(waiting)"); } dev = ((kib_net_t *)peer->ibp_ni->ni_data)->ibn_dev; @@ -1865,7 +1935,7 @@ kiblnd_close_conn_locked(kib_conn_t *conn, int error) kiblnd_set_conn_state(conn, IBLND_CONN_CLOSING); - if (error != 0 && + if (error && kiblnd_dev_can_failover(dev)) { list_add_tail(&dev->ibd_fail_list, &kiblnd_data.kib_failed_devs); @@ -1929,8 +1999,7 @@ kiblnd_abort_txs(kib_conn_t *conn, struct list_head *txs) if (txs == &conn->ibc_active_txs) { LASSERT(!tx->tx_queued); - LASSERT(tx->tx_waiting || - tx->tx_sending != 0); + LASSERT(tx->tx_waiting || tx->tx_sending); } else { LASSERT(tx->tx_queued); } @@ -1938,7 +2007,7 @@ kiblnd_abort_txs(kib_conn_t *conn, struct list_head *txs) tx->tx_status = -ECONNABORTED; tx->tx_waiting = 0; - if (tx->tx_sending == 0) { + if (!tx->tx_sending) { tx->tx_queued = 0; list_del(&tx->tx_list); list_add(&tx->tx_list, &zombies); @@ -1958,14 +2027,17 @@ kiblnd_finalise_conn(kib_conn_t *conn) kiblnd_set_conn_state(conn, IBLND_CONN_DISCONNECTED); - /* abort_receives moves QP state to IB_QPS_ERR. This is only required + /* + * abort_receives moves QP state to IB_QPS_ERR. This is only required * for connections that didn't get as far as being connected, because - * rdma_disconnect() does this for free. */ + * rdma_disconnect() does this for free. + */ kiblnd_abort_receives(conn); - /* Complete all tx descs not waiting for sends to complete. - * NB we should be safe from RDMA now that the QP has changed state */ - + /* + * Complete all tx descs not waiting for sends to complete. + * NB we should be safe from RDMA now that the QP has changed state + */ kiblnd_abort_txs(conn, &conn->ibc_tx_noops); kiblnd_abort_txs(conn, &conn->ibc_tx_queue); kiblnd_abort_txs(conn, &conn->ibc_tx_queue_rsrvd); @@ -1975,13 +2047,13 @@ kiblnd_finalise_conn(kib_conn_t *conn) kiblnd_handle_early_rxs(conn); } -void +static void kiblnd_peer_connect_failed(kib_peer_t *peer, int active, int error) { LIST_HEAD(zombies); unsigned long flags; - LASSERT(error != 0); + LASSERT(error); LASSERT(!in_interrupt()); write_lock_irqsave(&kiblnd_data.kib_global_lock, flags); @@ -1994,14 +2066,14 @@ kiblnd_peer_connect_failed(kib_peer_t *peer, int active, int error) peer->ibp_accepting--; } - if (peer->ibp_connecting != 0 || - peer->ibp_accepting != 0) { + if (kiblnd_peer_connecting(peer)) { /* another connection attempt under way... */ write_unlock_irqrestore(&kiblnd_data.kib_global_lock, - flags); + flags); return; } + peer->ibp_reconnected = 0; if (list_empty(&peer->ibp_conns)) { /* Take peer's blocked transmits to complete with error */ list_add(&zombies, &peer->ibp_tx_queue); @@ -2029,7 +2101,7 @@ kiblnd_peer_connect_failed(kib_peer_t *peer, int active, int error) kiblnd_txlist_done(peer->ibp_ni, &zombies, -EHOSTUNREACH); } -void +static void kiblnd_connreq_done(kib_conn_t *conn, int status) { kib_peer_t *peer = conn->ibc_peer; @@ -2047,14 +2119,14 @@ kiblnd_connreq_done(kib_conn_t *conn, int status) LASSERT(!in_interrupt()); LASSERT((conn->ibc_state == IBLND_CONN_ACTIVE_CONNECT && - peer->ibp_connecting > 0) || + peer->ibp_connecting > 0) || (conn->ibc_state == IBLND_CONN_PASSIVE_WAIT && - peer->ibp_accepting > 0)); + peer->ibp_accepting > 0)); LIBCFS_FREE(conn->ibc_connvars, sizeof(*conn->ibc_connvars)); conn->ibc_connvars = NULL; - if (status != 0) { + if (status) { /* failed to establish connection */ kiblnd_peer_connect_failed(peer, active, status); kiblnd_finalise_conn(conn); @@ -2068,16 +2140,19 @@ kiblnd_connreq_done(kib_conn_t *conn, int status) kiblnd_set_conn_state(conn, IBLND_CONN_ESTABLISHED); kiblnd_peer_alive(peer); - /* Add conn to peer's list and nuke any dangling conns from a different - * peer instance... */ + /* + * Add conn to peer's list and nuke any dangling conns from a different + * peer instance... + */ kiblnd_conn_addref(conn); /* +1 ref for ibc_list */ list_add(&conn->ibc_list, &peer->ibp_conns); + peer->ibp_reconnected = 0; if (active) peer->ibp_connecting--; else peer->ibp_accepting--; - if (peer->ibp_version == 0) { + if (!peer->ibp_version) { peer->ibp_version = conn->ibc_version; peer->ibp_incarnation = conn->ibc_incarnation; } @@ -2095,7 +2170,7 @@ kiblnd_connreq_done(kib_conn_t *conn, int status) list_del_init(&peer->ibp_tx_queue); if (!kiblnd_peer_active(peer) || /* peer has been deleted */ - conn->ibc_comms_error != 0) { /* error has happened already */ + conn->ibc_comms_error) { /* error has happened already */ lnet_ni_t *ni = peer->ibp_ni; /* start to shut down connection */ @@ -2107,6 +2182,16 @@ kiblnd_connreq_done(kib_conn_t *conn, int status) return; } + /** + * refcount taken by cmid is not reliable after I released the glock + * because this connection is visible to other threads now, another + * thread can find and close this connection right after I released + * the glock, if kiblnd_cm_callback for RDMA_CM_EVENT_DISCONNECTED is + * called, it can release the connection refcount taken by cmid. + * It means the connection could be destroyed before I finish my + * operations on it. + */ + kiblnd_conn_addref(conn); write_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags); /* Schedule blocked txs */ @@ -2122,6 +2207,8 @@ kiblnd_connreq_done(kib_conn_t *conn, int status) /* schedule blocked rxs */ kiblnd_handle_early_rxs(conn); + + kiblnd_conn_decref(conn); } static void @@ -2131,7 +2218,7 @@ kiblnd_reject(struct rdma_cm_id *cmid, kib_rej_t *rej) rc = rdma_reject(cmid, rej, sizeof(*rej)); - if (rc != 0) + if (rc) CWARN("Error %d sending reject\n", rc); } @@ -2159,14 +2246,14 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) /* cmid inherits 'context' from the corresponding listener id */ ibdev = (kib_dev_t *)cmid->context; - LASSERT(ibdev != NULL); + LASSERT(ibdev); memset(&rej, 0, sizeof(rej)); rej.ibr_magic = IBLND_MSG_MAGIC; rej.ibr_why = IBLND_REJECT_FATAL; rej.ibr_cp.ibcp_max_msg_size = IBLND_MSG_SIZE; - peer_addr = (struct sockaddr_in *)&(cmid->route.addr.dst_addr); + peer_addr = (struct sockaddr_in *)&cmid->route.addr.dst_addr; if (*kiblnd_tunables.kib_require_priv_port && ntohs(peer_addr->sin_port) >= PROT_SOCK) { __u32 ip = ntohl(peer_addr->sin_addr.s_addr); @@ -2181,12 +2268,14 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) goto failed; } - /* Future protocol version compatibility support! If the + /* + * Future protocol version compatibility support! If the * o2iblnd-specific protocol changes, or when LNET unifies * protocols over all LNDs, the initial connection will * negotiate a protocol version. I trap this here to avoid * console errors; the reject tells the peer which protocol I - * speak. */ + * speak. + */ if (reqmsg->ibm_magic == LNET_PROTO_MAGIC || reqmsg->ibm_magic == __swab32(LNET_PROTO_MAGIC)) goto failed; @@ -2200,7 +2289,7 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) goto failed; rc = kiblnd_unpack_msg(reqmsg, priv_nob); - if (rc != 0) { + if (rc) { CERROR("Can't parse connection request: %d\n", rc); goto failed; } @@ -2208,17 +2297,17 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) nid = reqmsg->ibm_srcnid; ni = lnet_net2ni(LNET_NIDNET(reqmsg->ibm_dstnid)); - if (ni != NULL) { + if (ni) { net = (kib_net_t *)ni->ni_data; rej.ibr_incarnation = net->ibn_incarnation; } - if (ni == NULL || /* no matching net */ + if (!ni || /* no matching net */ ni->ni_nid != reqmsg->ibm_dstnid || /* right NET, wrong NID! */ net->ibn_dev != ibdev) { /* wrong device */ - CERROR("Can't accept %s on %s (%s:%d:%pI4h): bad dst nid %s\n", + CERROR("Can't accept conn from %s on %s (%s:%d:%pI4h): bad dst nid %s\n", libcfs_nid2str(nid), - ni == NULL ? "NA" : libcfs_nid2str(ni->ni_nid), + !ni ? "NA" : libcfs_nid2str(ni->ni_nid), ibdev->ibd_ifname, ibdev->ibd_nnets, &ibdev->ibd_ifip, libcfs_nid2str(reqmsg->ibm_dstnid)); @@ -2227,7 +2316,7 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) } /* check time stamp as soon as possible */ - if (reqmsg->ibm_dststamp != 0 && + if (reqmsg->ibm_dststamp && reqmsg->ibm_dststamp != net->ibn_incarnation) { CWARN("Stale connection request\n"); rej.ibr_why = IBLND_REJECT_CONN_STALE; @@ -2243,10 +2332,11 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) goto failed; } - if (reqmsg->ibm_u.connparams.ibcp_queue_depth != + if (reqmsg->ibm_u.connparams.ibcp_queue_depth > IBLND_MSG_QUEUE_SIZE(version)) { - CERROR("Can't accept %s: incompatible queue depth %d (%d wanted)\n", - libcfs_nid2str(nid), reqmsg->ibm_u.connparams.ibcp_queue_depth, + CERROR("Can't accept conn from %s, queue depth too large: %d (<=%d wanted)\n", + libcfs_nid2str(nid), + reqmsg->ibm_u.connparams.ibcp_queue_depth, IBLND_MSG_QUEUE_SIZE(version)); if (version == IBLND_MSG_VERSION) @@ -2255,18 +2345,28 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) goto failed; } - if (reqmsg->ibm_u.connparams.ibcp_max_frags != + if (reqmsg->ibm_u.connparams.ibcp_max_frags > IBLND_RDMA_FRAGS(version)) { - CERROR("Can't accept %s(version %x): incompatible max_frags %d (%d wanted)\n", - libcfs_nid2str(nid), version, - reqmsg->ibm_u.connparams.ibcp_max_frags, - IBLND_RDMA_FRAGS(version)); + CWARN("Can't accept conn from %s (version %x): max_frags %d too large (%d wanted)\n", + libcfs_nid2str(nid), version, + reqmsg->ibm_u.connparams.ibcp_max_frags, + IBLND_RDMA_FRAGS(version)); - if (version == IBLND_MSG_VERSION) + if (version >= IBLND_MSG_VERSION) rej.ibr_why = IBLND_REJECT_RDMA_FRAGS; goto failed; + } else if (reqmsg->ibm_u.connparams.ibcp_max_frags < + IBLND_RDMA_FRAGS(version) && !net->ibn_fmr_ps) { + CWARN("Can't accept conn from %s (version %x): max_frags %d incompatible without FMR pool (%d wanted)\n", + libcfs_nid2str(nid), version, + reqmsg->ibm_u.connparams.ibcp_max_frags, + IBLND_RDMA_FRAGS(version)); + + if (version >= IBLND_MSG_VERSION) + rej.ibr_why = IBLND_REJECT_RDMA_FRAGS; + goto failed; } if (reqmsg->ibm_u.connparams.ibcp_max_msg_size > IBLND_MSG_SIZE) { @@ -2279,17 +2379,21 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) /* assume 'nid' is a new peer; create */ rc = kiblnd_create_peer(ni, &peer, nid); - if (rc != 0) { + if (rc) { CERROR("Can't create peer for %s\n", libcfs_nid2str(nid)); rej.ibr_why = IBLND_REJECT_NO_RESOURCES; goto failed; } + /* We have validated the peer's parameters so use those */ + peer->ibp_max_frags = reqmsg->ibm_u.connparams.ibcp_max_frags; + peer->ibp_queue_depth = reqmsg->ibm_u.connparams.ibcp_queue_depth; + write_lock_irqsave(g_lock, flags); peer2 = kiblnd_find_peer_locked(nid); - if (peer2 != NULL) { - if (peer2->ibp_version == 0) { + if (peer2) { + if (!peer2->ibp_version) { peer2->ibp_version = version; peer2->ibp_incarnation = reqmsg->ibm_srcstamp; } @@ -2298,10 +2402,16 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) if (peer2->ibp_incarnation != reqmsg->ibm_srcstamp || peer2->ibp_version != version) { kiblnd_close_peer_conns_locked(peer2, -ESTALE); + + if (kiblnd_peer_active(peer2)) { + peer2->ibp_incarnation = reqmsg->ibm_srcstamp; + peer2->ibp_version = version; + } write_unlock_irqrestore(g_lock, flags); - CWARN("Conn stale %s [old ver: %x, new ver: %x]\n", - libcfs_nid2str(nid), peer2->ibp_version, version); + CWARN("Conn stale %s version %x/%x incarnation %llu/%llu\n", + libcfs_nid2str(nid), peer2->ibp_version, version, + peer2->ibp_incarnation, reqmsg->ibm_srcstamp); kiblnd_peer_decref(peer); rej.ibr_why = IBLND_REJECT_CONN_STALE; @@ -2309,7 +2419,7 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) } /* tie-break connection race in favour of the higher NID */ - if (peer2->ibp_connecting != 0 && + if (peer2->ibp_connecting && nid < ni->ni_nid) { write_unlock_irqrestore(g_lock, flags); @@ -2320,24 +2430,37 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) goto failed; } + /** + * passive connection is allowed even this peer is waiting for + * reconnection. + */ + peer2->ibp_reconnecting = 0; peer2->ibp_accepting++; kiblnd_peer_addref(peer2); + /** + * Race with kiblnd_launch_tx (active connect) to create peer + * so copy validated parameters since we now know what the + * peer's limits are + */ + peer2->ibp_max_frags = peer->ibp_max_frags; + peer2->ibp_queue_depth = peer->ibp_queue_depth; + write_unlock_irqrestore(g_lock, flags); kiblnd_peer_decref(peer); peer = peer2; } else { /* Brand new peer */ - LASSERT(peer->ibp_accepting == 0); - LASSERT(peer->ibp_version == 0 && - peer->ibp_incarnation == 0); + LASSERT(!peer->ibp_accepting); + LASSERT(!peer->ibp_version && + !peer->ibp_incarnation); peer->ibp_accepting = 1; peer->ibp_version = version; peer->ibp_incarnation = reqmsg->ibm_srcstamp; /* I have a ref on ni that prevents it being shutdown */ - LASSERT(net->ibn_shutdown == 0); + LASSERT(!net->ibn_shutdown); kiblnd_peer_addref(peer); list_add_tail(&peer->ibp_list, kiblnd_nid2peerlist(nid)); @@ -2345,31 +2468,33 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) write_unlock_irqrestore(g_lock, flags); } - conn = kiblnd_create_conn(peer, cmid, IBLND_CONN_PASSIVE_WAIT, version); - if (conn == NULL) { + conn = kiblnd_create_conn(peer, cmid, IBLND_CONN_PASSIVE_WAIT, + version); + if (!conn) { kiblnd_peer_connect_failed(peer, 0, -ENOMEM); kiblnd_peer_decref(peer); rej.ibr_why = IBLND_REJECT_NO_RESOURCES; goto failed; } - /* conn now "owns" cmid, so I return success from here on to ensure the - * CM callback doesn't destroy cmid. */ - + /* + * conn now "owns" cmid, so I return success from here on to ensure the + * CM callback doesn't destroy cmid. + */ conn->ibc_incarnation = reqmsg->ibm_srcstamp; - conn->ibc_credits = IBLND_MSG_QUEUE_SIZE(version); - conn->ibc_reserved_credits = IBLND_MSG_QUEUE_SIZE(version); - LASSERT(conn->ibc_credits + conn->ibc_reserved_credits + IBLND_OOB_MSGS(version) - <= IBLND_RX_MSGS(version)); + conn->ibc_credits = conn->ibc_queue_depth; + conn->ibc_reserved_credits = conn->ibc_queue_depth; + LASSERT(conn->ibc_credits + conn->ibc_reserved_credits + + IBLND_OOB_MSGS(version) <= IBLND_RX_MSGS(conn)); ackmsg = &conn->ibc_connvars->cv_msg; memset(ackmsg, 0, sizeof(*ackmsg)); kiblnd_init_msg(ackmsg, IBLND_MSG_CONNACK, sizeof(ackmsg->ibm_u.connparams)); - ackmsg->ibm_u.connparams.ibcp_queue_depth = IBLND_MSG_QUEUE_SIZE(version); + ackmsg->ibm_u.connparams.ibcp_queue_depth = conn->ibc_queue_depth; + ackmsg->ibm_u.connparams.ibcp_max_frags = conn->ibc_max_frags; ackmsg->ibm_u.connparams.ibcp_max_msg_size = IBLND_MSG_SIZE; - ackmsg->ibm_u.connparams.ibcp_max_frags = IBLND_RDMA_FRAGS(version); kiblnd_pack_msg(ni, ackmsg, version, 0, nid, reqmsg->ibm_srcstamp); @@ -2385,7 +2510,7 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) CDEBUG(D_NET, "Accept %s\n", libcfs_nid2str(nid)); rc = rdma_accept(cmid, &cp); - if (rc != 0) { + if (rc) { CERROR("Can't accept %s: %d\n", libcfs_nid2str(nid), rc); rej.ibr_version = version; rej.ibr_why = IBLND_REJECT_FATAL; @@ -2399,7 +2524,7 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) return 0; failed: - if (ni != NULL) + if (ni) lnet_ni_decref(ni); rej.ibr_version = version; @@ -2411,45 +2536,82 @@ kiblnd_passive_connect(struct rdma_cm_id *cmid, void *priv, int priv_nob) } static void -kiblnd_reconnect(kib_conn_t *conn, int version, - __u64 incarnation, int why, kib_connparams_t *cp) +kiblnd_check_reconnect(kib_conn_t *conn, int version, + __u64 incarnation, int why, kib_connparams_t *cp) { + rwlock_t *glock = &kiblnd_data.kib_global_lock; kib_peer_t *peer = conn->ibc_peer; char *reason; - int retry = 0; + int msg_size = IBLND_MSG_SIZE; + int frag_num = -1; + int queue_dep = -1; + bool reconnect; unsigned long flags; LASSERT(conn->ibc_state == IBLND_CONN_ACTIVE_CONNECT); LASSERT(peer->ibp_connecting > 0); /* 'conn' at least */ + LASSERT(!peer->ibp_reconnecting); - write_lock_irqsave(&kiblnd_data.kib_global_lock, flags); + if (cp) { + msg_size = cp->ibcp_max_msg_size; + frag_num = cp->ibcp_max_frags; + queue_dep = cp->ibcp_queue_depth; + } - /* retry connection if it's still needed and no other connection + write_lock_irqsave(glock, flags); + /** + * retry connection if it's still needed and no other connection * attempts (active or passive) are in progress * NB: reconnect is still needed even when ibp_tx_queue is * empty if ibp_version != version because reconnect may be - * initiated by kiblnd_query() */ - if ((!list_empty(&peer->ibp_tx_queue) || - peer->ibp_version != version) && - peer->ibp_connecting == 1 && - peer->ibp_accepting == 0) { - retry = 1; - peer->ibp_connecting++; - - peer->ibp_version = version; - peer->ibp_incarnation = incarnation; + * initiated by kiblnd_query() + */ + reconnect = (!list_empty(&peer->ibp_tx_queue) || + peer->ibp_version != version) && + peer->ibp_connecting == 1 && + !peer->ibp_accepting; + if (!reconnect) { + reason = "no need"; + goto out; } - write_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags); - - if (!retry) - return; - switch (why) { default: reason = "Unknown"; break; + case IBLND_REJECT_RDMA_FRAGS: + if (!cp) { + reason = "can't negotiate max frags"; + goto out; + } + if (!*kiblnd_tunables.kib_map_on_demand) { + reason = "map_on_demand must be enabled"; + goto out; + } + if (conn->ibc_max_frags <= frag_num) { + reason = "unsupported max frags"; + goto out; + } + + peer->ibp_max_frags = frag_num; + reason = "rdma fragments"; + break; + + case IBLND_REJECT_MSG_QUEUE_SIZE: + if (!cp) { + reason = "can't negotiate queue depth"; + goto out; + } + if (conn->ibc_queue_depth <= queue_dep) { + reason = "unsupported queue depth"; + goto out; + } + + peer->ibp_queue_depth = queue_dep; + reason = "queue depth"; + break; + case IBLND_REJECT_CONN_STALE: reason = "stale"; break; @@ -2463,14 +2625,24 @@ kiblnd_reconnect(kib_conn_t *conn, int version, break; } - CNETERR("%s: retrying (%s), %x, %x, queue_dep: %d, max_frag: %d, msg_size: %d\n", - libcfs_nid2str(peer->ibp_nid), - reason, IBLND_MSG_VERSION, version, - cp != NULL ? cp->ibcp_queue_depth : IBLND_MSG_QUEUE_SIZE(version), - cp != NULL ? cp->ibcp_max_frags : IBLND_RDMA_FRAGS(version), - cp != NULL ? cp->ibcp_max_msg_size : IBLND_MSG_SIZE); + conn->ibc_reconnect = 1; + peer->ibp_reconnecting = 1; + peer->ibp_version = version; + if (incarnation) + peer->ibp_incarnation = incarnation; +out: + write_unlock_irqrestore(glock, flags); - kiblnd_connect_peer(peer); + CNETERR("%s: %s (%s), %x, %x, msg_size: %d, queue_depth: %d/%d, max_frags: %d/%d\n", + libcfs_nid2str(peer->ibp_nid), + reconnect ? "reconnect" : "don't reconnect", + reason, IBLND_MSG_VERSION, version, msg_size, + conn->ibc_queue_depth, queue_dep, + conn->ibc_max_frags, frag_num); + /** + * if conn::ibc_reconnect is TRUE, connd will reconnect to the peer + * while destroying the zombie + */ } static void @@ -2483,8 +2655,8 @@ kiblnd_rejected(kib_conn_t *conn, int reason, void *priv, int priv_nob) switch (reason) { case IB_CM_REJ_STALE_CONN: - kiblnd_reconnect(conn, IBLND_MSG_VERSION, 0, - IBLND_REJECT_CONN_STALE, NULL); + kiblnd_check_reconnect(conn, IBLND_MSG_VERSION, 0, + IBLND_REJECT_CONN_STALE, NULL); break; case IB_CM_REJ_INVALID_SERVICE_ID: @@ -2521,9 +2693,11 @@ kiblnd_rejected(kib_conn_t *conn, int reason, void *priv, int priv_nob) if (priv_nob >= sizeof(kib_rej_t) && rej->ibr_version > IBLND_MSG_VERSION_1) { - /* priv_nob is always 148 in current version + /* + * priv_nob is always 148 in current version * of OFED, so we still need to check version. - * (define of IB_CM_REJ_PRIVATE_DATA_SIZE) */ + * (define of IB_CM_REJ_PRIVATE_DATA_SIZE) + */ cp = &rej->ibr_cp; if (flip) { @@ -2564,24 +2738,11 @@ kiblnd_rejected(kib_conn_t *conn, int reason, void *priv, int priv_nob) case IBLND_REJECT_CONN_RACE: case IBLND_REJECT_CONN_STALE: case IBLND_REJECT_CONN_UNCOMPAT: - kiblnd_reconnect(conn, rej->ibr_version, - incarnation, rej->ibr_why, cp); - break; - case IBLND_REJECT_MSG_QUEUE_SIZE: - CERROR("%s rejected: incompatible message queue depth %d, %d\n", - libcfs_nid2str(peer->ibp_nid), - cp != NULL ? cp->ibcp_queue_depth : - IBLND_MSG_QUEUE_SIZE(rej->ibr_version), - IBLND_MSG_QUEUE_SIZE(conn->ibc_version)); - break; - case IBLND_REJECT_RDMA_FRAGS: - CERROR("%s rejected: incompatible # of RDMA fragments %d, %d\n", - libcfs_nid2str(peer->ibp_nid), - cp != NULL ? cp->ibcp_max_frags : - IBLND_RDMA_FRAGS(rej->ibr_version), - IBLND_RDMA_FRAGS(conn->ibc_version)); + kiblnd_check_reconnect(conn, rej->ibr_version, + incarnation, + rej->ibr_why, cp); break; case IBLND_REJECT_NO_RESOURCES: @@ -2623,9 +2784,9 @@ kiblnd_check_connreply(kib_conn_t *conn, void *priv, int priv_nob) int rc = kiblnd_unpack_msg(msg, priv_nob); unsigned long flags; - LASSERT(net != NULL); + LASSERT(net); - if (rc != 0) { + if (rc) { CERROR("Can't unpack connack from %s: %d\n", libcfs_nid2str(peer->ibp_nid), rc); goto failed; @@ -2645,22 +2806,22 @@ kiblnd_check_connreply(kib_conn_t *conn, void *priv, int priv_nob) goto failed; } - if (msg->ibm_u.connparams.ibcp_queue_depth != - IBLND_MSG_QUEUE_SIZE(ver)) { - CERROR("%s has incompatible queue depth %d(%d wanted)\n", + if (msg->ibm_u.connparams.ibcp_queue_depth > + conn->ibc_queue_depth) { + CERROR("%s has incompatible queue depth %d (<=%d wanted)\n", libcfs_nid2str(peer->ibp_nid), msg->ibm_u.connparams.ibcp_queue_depth, - IBLND_MSG_QUEUE_SIZE(ver)); + conn->ibc_queue_depth); rc = -EPROTO; goto failed; } - if (msg->ibm_u.connparams.ibcp_max_frags != - IBLND_RDMA_FRAGS(ver)) { - CERROR("%s has incompatible max_frags %d (%d wanted)\n", + if (msg->ibm_u.connparams.ibcp_max_frags > + conn->ibc_max_frags) { + CERROR("%s has incompatible max_frags %d (<=%d wanted)\n", libcfs_nid2str(peer->ibp_nid), msg->ibm_u.connparams.ibcp_max_frags, - IBLND_RDMA_FRAGS(ver)); + conn->ibc_max_frags); rc = -EPROTO; goto failed; } @@ -2682,7 +2843,7 @@ kiblnd_check_connreply(kib_conn_t *conn, void *priv, int priv_nob) rc = -ESTALE; read_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags); - if (rc != 0) { + if (rc) { CERROR("Bad connection reply from %s, rc = %d, version: %x max_frags: %d\n", libcfs_nid2str(peer->ibp_nid), rc, msg->ibm_version, msg->ibm_u.connparams.ibcp_max_frags); @@ -2690,21 +2851,24 @@ kiblnd_check_connreply(kib_conn_t *conn, void *priv, int priv_nob) } conn->ibc_incarnation = msg->ibm_srcstamp; - conn->ibc_credits = - conn->ibc_reserved_credits = IBLND_MSG_QUEUE_SIZE(ver); - LASSERT(conn->ibc_credits + conn->ibc_reserved_credits + IBLND_OOB_MSGS(ver) - <= IBLND_RX_MSGS(ver)); + conn->ibc_credits = msg->ibm_u.connparams.ibcp_queue_depth; + conn->ibc_reserved_credits = msg->ibm_u.connparams.ibcp_queue_depth; + conn->ibc_queue_depth = msg->ibm_u.connparams.ibcp_queue_depth; + conn->ibc_max_frags = msg->ibm_u.connparams.ibcp_max_frags; + LASSERT(conn->ibc_credits + conn->ibc_reserved_credits + + IBLND_OOB_MSGS(ver) <= IBLND_RX_MSGS(conn)); kiblnd_connreq_done(conn, 0); return; failed: - /* NB My QP has already established itself, so I handle anything going + /* + * NB My QP has already established itself, so I handle anything going * wrong here by setting ibc_comms_error. * kiblnd_connreq_done(0) moves the conn state to ESTABLISHED, but then - * immediately tears it down. */ - - LASSERT(rc != 0); + * immediately tears it down. + */ + LASSERT(rc); conn->ibc_comms_error = rc; kiblnd_connreq_done(conn, 0); } @@ -2724,28 +2888,30 @@ kiblnd_active_connect(struct rdma_cm_id *cmid) read_lock_irqsave(&kiblnd_data.kib_global_lock, flags); incarnation = peer->ibp_incarnation; - version = (peer->ibp_version == 0) ? IBLND_MSG_VERSION : - peer->ibp_version; + version = !peer->ibp_version ? IBLND_MSG_VERSION : + peer->ibp_version; read_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags); - conn = kiblnd_create_conn(peer, cmid, IBLND_CONN_ACTIVE_CONNECT, version); - if (conn == NULL) { + conn = kiblnd_create_conn(peer, cmid, IBLND_CONN_ACTIVE_CONNECT, + version); + if (!conn) { kiblnd_peer_connect_failed(peer, 1, -ENOMEM); kiblnd_peer_decref(peer); /* lose cmid's ref */ return -ENOMEM; } - /* conn "owns" cmid now, so I return success from here on to ensure the + /* + * conn "owns" cmid now, so I return success from here on to ensure the * CM callback doesn't destroy cmid. conn also takes over cmid's ref - * on peer */ - + * on peer + */ msg = &conn->ibc_connvars->cv_msg; memset(msg, 0, sizeof(*msg)); kiblnd_init_msg(msg, IBLND_MSG_CONNREQ, sizeof(msg->ibm_u.connparams)); - msg->ibm_u.connparams.ibcp_queue_depth = IBLND_MSG_QUEUE_SIZE(version); - msg->ibm_u.connparams.ibcp_max_frags = IBLND_RDMA_FRAGS(version); + msg->ibm_u.connparams.ibcp_queue_depth = conn->ibc_queue_depth; + msg->ibm_u.connparams.ibcp_max_frags = conn->ibc_max_frags; msg->ibm_u.connparams.ibcp_max_msg_size = IBLND_MSG_SIZE; kiblnd_pack_msg(peer->ibp_ni, msg, version, @@ -2764,7 +2930,7 @@ kiblnd_active_connect(struct rdma_cm_id *cmid) LASSERT(conn->ibc_cmid == cmid); rc = rdma_connect(cmid, &cp); - if (rc != 0) { + if (rc) { CERROR("Can't connect to %s: %d\n", libcfs_nid2str(peer->ibp_nid), rc); kiblnd_connreq_done(conn, rc); @@ -2798,10 +2964,10 @@ kiblnd_cm_callback(struct rdma_cm_id *cmid, struct rdma_cm_event *event) case RDMA_CM_EVENT_ADDR_ERROR: peer = (kib_peer_t *)cmid->context; CNETERR("%s: ADDR ERROR %d\n", - libcfs_nid2str(peer->ibp_nid), event->status); + libcfs_nid2str(peer->ibp_nid), event->status); kiblnd_peer_connect_failed(peer, 1, -EHOSTUNREACH); kiblnd_peer_decref(peer); - return -EHOSTUNREACH; /* rc != 0 destroys cmid */ + return -EHOSTUNREACH; /* rc destroys cmid */ case RDMA_CM_EVENT_ADDR_RESOLVED: peer = (kib_peer_t *)cmid->context; @@ -2809,14 +2975,14 @@ kiblnd_cm_callback(struct rdma_cm_id *cmid, struct rdma_cm_event *event) CDEBUG(D_NET, "%s Addr resolved: %d\n", libcfs_nid2str(peer->ibp_nid), event->status); - if (event->status != 0) { + if (event->status) { CNETERR("Can't resolve address for %s: %d\n", libcfs_nid2str(peer->ibp_nid), event->status); rc = event->status; } else { rc = rdma_resolve_route( cmid, *kiblnd_tunables.kib_timeout * 1000); - if (rc == 0) + if (!rc) return 0; /* Can't initiate route resolution */ CERROR("Can't resolve route for %s: %d\n", @@ -2824,7 +2990,7 @@ kiblnd_cm_callback(struct rdma_cm_id *cmid, struct rdma_cm_event *event) } kiblnd_peer_connect_failed(peer, 1, rc); kiblnd_peer_decref(peer); - return rc; /* rc != 0 destroys cmid */ + return rc; /* rc destroys cmid */ case RDMA_CM_EVENT_ROUTE_ERROR: peer = (kib_peer_t *)cmid->context; @@ -2832,28 +2998,28 @@ kiblnd_cm_callback(struct rdma_cm_id *cmid, struct rdma_cm_event *event) libcfs_nid2str(peer->ibp_nid), event->status); kiblnd_peer_connect_failed(peer, 1, -EHOSTUNREACH); kiblnd_peer_decref(peer); - return -EHOSTUNREACH; /* rc != 0 destroys cmid */ + return -EHOSTUNREACH; /* rc destroys cmid */ case RDMA_CM_EVENT_ROUTE_RESOLVED: peer = (kib_peer_t *)cmid->context; CDEBUG(D_NET, "%s Route resolved: %d\n", libcfs_nid2str(peer->ibp_nid), event->status); - if (event->status == 0) + if (!event->status) return kiblnd_active_connect(cmid); CNETERR("Can't resolve route for %s: %d\n", - libcfs_nid2str(peer->ibp_nid), event->status); + libcfs_nid2str(peer->ibp_nid), event->status); kiblnd_peer_connect_failed(peer, 1, event->status); kiblnd_peer_decref(peer); - return event->status; /* rc != 0 destroys cmid */ + return event->status; /* rc destroys cmid */ case RDMA_CM_EVENT_UNREACHABLE: conn = (kib_conn_t *)cmid->context; LASSERT(conn->ibc_state == IBLND_CONN_ACTIVE_CONNECT || conn->ibc_state == IBLND_CONN_PASSIVE_WAIT); CNETERR("%s: UNREACHABLE %d\n", - libcfs_nid2str(conn->ibc_peer->ibp_nid), event->status); + libcfs_nid2str(conn->ibc_peer->ibp_nid), event->status); kiblnd_connreq_done(conn, -ENETDOWN); kiblnd_conn_decref(conn); return 0; @@ -2876,8 +3042,8 @@ kiblnd_cm_callback(struct rdma_cm_id *cmid, struct rdma_cm_event *event) case IBLND_CONN_PASSIVE_WAIT: CERROR("%s: REJECTED %d\n", - libcfs_nid2str(conn->ibc_peer->ibp_nid), - event->status); + libcfs_nid2str(conn->ibc_peer->ibp_nid), + event->status); kiblnd_connreq_done(conn, -ECONNRESET); break; @@ -2933,8 +3099,10 @@ kiblnd_cm_callback(struct rdma_cm_id *cmid, struct rdma_cm_event *event) LCONSOLE_ERROR_MSG(0x131, "Received notification of device removal\n" "Please shutdown LNET to allow this to proceed\n"); - /* Can't remove network from underneath LNET for now, so I have - * to ignore this */ + /* + * Can't remove network from underneath LNET for now, so I have + * to ignore this + */ return 0; case RDMA_CM_EVENT_ADDR_CHANGE: @@ -2956,7 +3124,7 @@ kiblnd_check_txs_locked(kib_conn_t *conn, struct list_head *txs) LASSERT(tx->tx_queued); } else { LASSERT(!tx->tx_queued); - LASSERT(tx->tx_waiting || tx->tx_sending != 0); + LASSERT(tx->tx_waiting || tx->tx_sending); } if (cfs_time_aftereq(jiffies, tx->tx_deadline)) { @@ -2989,13 +3157,16 @@ kiblnd_check_conns(int idx) struct list_head *ptmp; kib_peer_t *peer; kib_conn_t *conn; + kib_conn_t *temp; kib_conn_t *tmp; struct list_head *ctmp; unsigned long flags; - /* NB. We expect to have a look at all the peers and not find any + /* + * NB. We expect to have a look at all the peers and not find any * RDMAs to time out, so we just use a shared lock while we - * take a look... */ + * take a look... + */ read_lock_irqsave(&kiblnd_data.kib_global_lock, flags); list_for_each(ptmp, peers) { @@ -3028,8 +3199,7 @@ kiblnd_check_conns(int idx) conn->ibc_reserved_credits); list_add(&conn->ibc_connd_list, &closes); } else { - list_add(&conn->ibc_connd_list, - &checksends); + list_add(&conn->ibc_connd_list, &checksends); } /* +ref for 'closes' or 'checksends' */ kiblnd_conn_addref(conn); @@ -3040,21 +3210,23 @@ kiblnd_check_conns(int idx) read_unlock_irqrestore(&kiblnd_data.kib_global_lock, flags); - /* Handle timeout by closing the whole + /* + * Handle timeout by closing the whole * connection. We can only be sure RDMA activity - * has ceased once the QP has been modified. */ + * has ceased once the QP has been modified. + */ list_for_each_entry_safe(conn, tmp, &closes, ibc_connd_list) { list_del(&conn->ibc_connd_list); kiblnd_close_conn(conn, -ETIMEDOUT); kiblnd_conn_decref(conn); } - /* In case we have enough credits to return via a + /* + * In case we have enough credits to return via a * NOOP, but there were no non-blocking tx descs - * free to do it last time... */ - while (!list_empty(&checksends)) { - conn = list_entry(checksends.next, - kib_conn_t, ibc_connd_list); + * free to do it last time... + */ + list_for_each_entry_safe(conn, temp, &checksends, ibc_connd_list) { list_del(&conn->ibc_connd_list); kiblnd_check_sends(conn); kiblnd_conn_decref(conn); @@ -3074,9 +3246,21 @@ kiblnd_disconnect_conn(kib_conn_t *conn) kiblnd_peer_notify(conn->ibc_peer); } +/** + * High-water for reconnection to the same peer, reconnection attempt should + * be delayed after trying more than KIB_RECONN_HIGH_RACE. + */ +#define KIB_RECONN_HIGH_RACE 10 +/** + * Allow connd to take a break and handle other things after consecutive + * reconnection attemps. + */ +#define KIB_RECONN_BREAK 100 + int kiblnd_connd(void *arg) { + spinlock_t *lock= &kiblnd_data.kib_connd_lock; wait_queue_t wait; unsigned long flags; kib_conn_t *conn; @@ -3091,39 +3275,79 @@ kiblnd_connd(void *arg) init_waitqueue_entry(&wait, current); kiblnd_data.kib_connd = current; - spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags); + spin_lock_irqsave(lock, flags); while (!kiblnd_data.kib_shutdown) { + int reconn = 0; dropped_lock = 0; if (!list_empty(&kiblnd_data.kib_connd_zombies)) { + kib_peer_t *peer = NULL; + conn = list_entry(kiblnd_data.kib_connd_zombies.next, - kib_conn_t, ibc_list); + kib_conn_t, ibc_list); list_del(&conn->ibc_list); + if (conn->ibc_reconnect) { + peer = conn->ibc_peer; + kiblnd_peer_addref(peer); + } - spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock, - flags); + spin_unlock_irqrestore(lock, flags); dropped_lock = 1; - kiblnd_destroy_conn(conn); + kiblnd_destroy_conn(conn, !peer); + + spin_lock_irqsave(lock, flags); + if (!peer) + continue; - spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags); + conn->ibc_peer = peer; + if (peer->ibp_reconnected < KIB_RECONN_HIGH_RACE) + list_add_tail(&conn->ibc_list, + &kiblnd_data.kib_reconn_list); + else + list_add_tail(&conn->ibc_list, + &kiblnd_data.kib_reconn_wait); } if (!list_empty(&kiblnd_data.kib_connd_conns)) { conn = list_entry(kiblnd_data.kib_connd_conns.next, - kib_conn_t, ibc_list); + kib_conn_t, ibc_list); list_del(&conn->ibc_list); - spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock, - flags); + spin_unlock_irqrestore(lock, flags); dropped_lock = 1; kiblnd_disconnect_conn(conn); kiblnd_conn_decref(conn); - spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags); + spin_lock_irqsave(lock, flags); + } + + while (reconn < KIB_RECONN_BREAK) { + if (kiblnd_data.kib_reconn_sec != + ktime_get_real_seconds()) { + kiblnd_data.kib_reconn_sec = ktime_get_real_seconds(); + list_splice_init(&kiblnd_data.kib_reconn_wait, + &kiblnd_data.kib_reconn_list); + } + + if (list_empty(&kiblnd_data.kib_reconn_list)) + break; + + conn = list_entry(kiblnd_data.kib_reconn_list.next, + kib_conn_t, ibc_list); + list_del(&conn->ibc_list); + + spin_unlock_irqrestore(lock, flags); + dropped_lock = 1; + + reconn += kiblnd_reconnect_peer(conn->ibc_peer); + kiblnd_peer_decref(conn->ibc_peer); + LIBCFS_FREE(conn, sizeof(*conn)); + + spin_lock_irqsave(lock, flags); } /* careful with the jiffy wrap... */ @@ -3133,21 +3357,22 @@ kiblnd_connd(void *arg) const int p = 1; int chunk = kiblnd_data.kib_peer_hash_size; - spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock, flags); + spin_unlock_irqrestore(lock, flags); dropped_lock = 1; - /* Time to check for RDMA timeouts on a few more + /* + * Time to check for RDMA timeouts on a few more * peers: I do checks every 'p' seconds on a * proportion of the peer table and I need to check * every connection 'n' times within a timeout * interval, to ensure I detect a timeout on any * connection within (n+1)/n times the timeout - * interval. */ - + * interval. + */ if (*kiblnd_tunables.kib_timeout > n * p) chunk = (chunk * n * p) / *kiblnd_tunables.kib_timeout; - if (chunk == 0) + if (!chunk) chunk = 1; for (i = 0; i < chunk; i++) { @@ -3156,8 +3381,8 @@ kiblnd_connd(void *arg) kiblnd_data.kib_peer_hash_size; } - deadline += p * HZ; - spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags); + deadline += msecs_to_jiffies(p * MSEC_PER_SEC); + spin_lock_irqsave(lock, flags); } if (dropped_lock) @@ -3166,15 +3391,15 @@ kiblnd_connd(void *arg) /* Nothing to do for 'timeout' */ set_current_state(TASK_INTERRUPTIBLE); add_wait_queue(&kiblnd_data.kib_connd_waitq, &wait); - spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock, flags); + spin_unlock_irqrestore(lock, flags); schedule_timeout(timeout); remove_wait_queue(&kiblnd_data.kib_connd_waitq, &wait); - spin_lock_irqsave(&kiblnd_data.kib_connd_lock, flags); + spin_lock_irqsave(lock, flags); } - spin_unlock_irqrestore(&kiblnd_data.kib_connd_lock, flags); + spin_unlock_irqrestore(lock, flags); kiblnd_thread_fini(); return 0; @@ -3206,12 +3431,14 @@ kiblnd_complete(struct ib_wc *wc) LBUG(); case IBLND_WID_RDMA: - /* We only get RDMA completion notification if it fails. All + /* + * We only get RDMA completion notification if it fails. All * subsequent work items, including the final SEND will fail * too. However we can't print out any more info about the * failing RDMA because 'tx' might be back on the idle list or * even reused already if we didn't manage to post all our work - * items */ + * items + */ CNETERR("RDMA (tx: %p) failed: %d\n", kiblnd_wreqid2ptr(wc->wr_id), wc->status); return; @@ -3230,11 +3457,13 @@ kiblnd_complete(struct ib_wc *wc) void kiblnd_cq_completion(struct ib_cq *cq, void *arg) { - /* NB I'm not allowed to schedule this conn once its refcount has + /* + * NB I'm not allowed to schedule this conn once its refcount has * reached 0. Since fundamentally I'm racing with scheduler threads * consuming my CQ I could be called after all completions have - * occurred. But in this case, ibc_nrx == 0 && ibc_nsends_posted == 0 - * and this CQ is about to be destroyed so I NOOP. */ + * occurred. But in this case, !ibc_nrx && !ibc_nsends_posted + * and this CQ is about to be destroyed so I NOOP. + */ kib_conn_t *conn = arg; struct kib_sched_info *sched = conn->ibc_sched; unsigned long flags; @@ -3288,7 +3517,7 @@ kiblnd_scheduler(void *arg) sched = kiblnd_data.kib_scheds[KIB_THREAD_CPT(id)]; rc = cfs_cpt_bind(lnet_cpt_table(), sched->ibs_cpt); - if (rc != 0) { + if (rc) { CWARN("Failed to bind on CPT %d, please verify whether all CPUs are healthy and reload modules if necessary, otherwise your system might under risk of low performance\n", sched->ibs_cpt); } @@ -3308,8 +3537,8 @@ kiblnd_scheduler(void *arg) did_something = 0; if (!list_empty(&sched->ibs_conns)) { - conn = list_entry(sched->ibs_conns.next, - kib_conn_t, ibc_sched_list); + conn = list_entry(sched->ibs_conns.next, kib_conn_t, + ibc_sched_list); /* take over kib_sched_conns' ref on conn... */ LASSERT(conn->ibc_scheduled); list_del(&conn->ibc_sched_list); @@ -3317,8 +3546,10 @@ kiblnd_scheduler(void *arg) spin_unlock_irqrestore(&sched->ibs_lock, flags); + wc.wr_id = IBLND_WID_INVAL; + rc = ib_poll_cq(conn->ibc_cq, 1, &wc); - if (rc == 0) { + if (!rc) { rc = ib_req_notify_cq(conn->ibc_cq, IB_CQ_NEXT_COMP); if (rc < 0) { @@ -3327,13 +3558,22 @@ kiblnd_scheduler(void *arg) kiblnd_close_conn(conn, -EIO); kiblnd_conn_decref(conn); spin_lock_irqsave(&sched->ibs_lock, - flags); + flags); continue; } rc = ib_poll_cq(conn->ibc_cq, 1, &wc); } + if (unlikely(rc > 0 && wc.wr_id == IBLND_WID_INVAL)) { + LCONSOLE_ERROR("ib_poll_cq (rc: %d) returned invalid wr_id, opcode %d, status: %d, vendor_err: %d, conn: %s status: %d\nplease upgrade firmware and OFED or contact vendor.\n", + rc, wc.opcode, wc.status, + wc.vendor_err, + libcfs_nid2str(conn->ibc_peer->ibp_nid), + conn->ibc_state); + rc = -EINVAL; + } + if (rc < 0) { CWARN("%s: ib_poll_cq failed: %d, closing connection\n", libcfs_nid2str(conn->ibc_peer->ibp_nid), @@ -3346,21 +3586,23 @@ kiblnd_scheduler(void *arg) spin_lock_irqsave(&sched->ibs_lock, flags); - if (rc != 0 || conn->ibc_ready) { - /* There may be another completion waiting; get + if (rc || conn->ibc_ready) { + /* + * There may be another completion waiting; get * another scheduler to check while I handle - * this one... */ + * this one... + */ /* +1 ref for sched_conns */ kiblnd_conn_addref(conn); list_add_tail(&conn->ibc_sched_list, - &sched->ibs_conns); + &sched->ibs_conns); if (waitqueue_active(&sched->ibs_waitq)) wake_up(&sched->ibs_waitq); } else { conn->ibc_scheduled = 0; } - if (rc != 0) { + if (rc) { spin_unlock_irqrestore(&sched->ibs_lock, flags); kiblnd_complete(&wc); @@ -3400,7 +3642,7 @@ kiblnd_failover_thread(void *arg) unsigned long flags; int rc; - LASSERT(*kiblnd_tunables.kib_dev_failover != 0); + LASSERT(*kiblnd_tunables.kib_dev_failover); cfs_block_allsigs(); @@ -3459,13 +3701,15 @@ kiblnd_failover_thread(void *arg) remove_wait_queue(&kiblnd_data.kib_failover_waitq, &wait); write_lock_irqsave(glock, flags); - if (!long_sleep || rc != 0) + if (!long_sleep || rc) continue; - /* have a long sleep, routine check all active devices, + /* + * have a long sleep, routine check all active devices, * we need checking like this because if there is not active * connection on the dev and no SEND from local, we may listen - * on wrong HCA for ever while there is a bonding failover */ + * on wrong HCA for ever while there is a bonding failover + */ list_for_each_entry(dev, &kiblnd_data.kib_devs, ibd_list) { if (kiblnd_dev_can_failover(dev)) { list_add_tail(&dev->ibd_fail_list, diff --git a/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd_modparams.c b/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd_modparams.c index 1d4e7efb5..b4607dad3 100644 --- a/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd_modparams.c +++ b/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd_modparams.c @@ -52,8 +52,10 @@ static int timeout = 50; module_param(timeout, int, 0644); MODULE_PARM_DESC(timeout, "timeout (seconds)"); -/* Number of threads in each scheduler pool which is percpt, - * we will estimate reasonable value based on CPUs if it's set to zero. */ +/* + * Number of threads in each scheduler pool which is percpt, + * we will estimate reasonable value based on CPUs if it's set to zero. + */ static int nscheds; module_param(nscheds, int, 0444); MODULE_PARM_DESC(nscheds, "number of threads in each scheduler pool"); @@ -200,7 +202,7 @@ kiblnd_tunables_init(void) if (*kiblnd_tunables.kib_map_on_demand == 1) *kiblnd_tunables.kib_map_on_demand = 2; /* don't make sense to create map if only one fragment */ - if (*kiblnd_tunables.kib_concurrent_sends == 0) { + if (!*kiblnd_tunables.kib_concurrent_sends) { if (*kiblnd_tunables.kib_map_on_demand > 0 && *kiblnd_tunables.kib_map_on_demand <= IBLND_MAX_RDMA_FRAGS / 8) *kiblnd_tunables.kib_concurrent_sends = (*kiblnd_tunables.kib_peertxcredits) * 2; -- cgit v1.2.3-54-g00ecf