X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lnet%2Fklnds%2Fsocklnd%2Fsocklnd.c;h=4d3b7a4c4b08ed5d5d6f66402b238990a1c3e1ba;hp=ff6be40b368dc0344369e32abf0dd76c52b7835b;hb=fdce7be68eb7ab2fc7945e81d2dc49f3595ec790;hpb=35a1096a4741b24dbcd674bda5daf7b8a13cfe28 diff --git a/lnet/klnds/socklnd/socklnd.c b/lnet/klnds/socklnd/socklnd.c index ff6be40..4d3b7a4 100644 --- a/lnet/klnds/socklnd/socklnd.c +++ b/lnet/klnds/socklnd/socklnd.c @@ -15,11 +15,7 @@ * * You should have received a copy of the GNU General Public License * version 2 along with this program; If not, see - * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf - * - * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara, - * CA 95054 USA or visit www.sun.com if you need additional information or - * have any questions. + * http://www.gnu.org/licenses/gpl-2.0.html * * GPL HEADER END */ @@ -27,7 +23,7 @@ * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. * - * Copyright (c) 2011, 2012, Intel Corporation. + * Copyright (c) 2011, 2017, Intel Corporation. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -41,57 +37,117 @@ * Author: Eric Barton */ +#include #include "socklnd.h" +#include -lnd_t the_ksocklnd; -ksock_nal_data_t ksocknal_data; +static const struct lnet_lnd the_ksocklnd; +struct ksock_nal_data ksocknal_data; -ksock_interface_t * -ksocknal_ip2iface(lnet_ni_t *ni, __u32 ip) +static struct ksock_interface * +ksocknal_ip2iface(struct lnet_ni *ni, struct sockaddr *addr) { - ksock_net_t *net = ni->ni_data; - int i; - ksock_interface_t *iface; + struct ksock_net *net = ni->ni_data; + int i; + struct ksock_interface *iface; - for (i = 0; i < net->ksnn_ninterfaces; i++) { - LASSERT(i < LNET_MAX_INTERFACES); - iface = &net->ksnn_interfaces[i]; + for (i = 0; i < net->ksnn_ninterfaces; i++) { + LASSERT(i < LNET_INTERFACES_NUM); + iface = &net->ksnn_interfaces[i]; - if (iface->ksni_ipaddr == ip) - return (iface); - } + if (rpc_cmp_addr((struct sockaddr *)&iface->ksni_addr, addr)) + return iface; + } + + return NULL; +} + +static struct ksock_interface * +ksocknal_index2iface(struct lnet_ni *ni, int index) +{ + struct ksock_net *net = ni->ni_data; + int i; + struct ksock_interface *iface; + + for (i = 0; i < net->ksnn_ninterfaces; i++) { + LASSERT(i < LNET_INTERFACES_NUM); + iface = &net->ksnn_interfaces[i]; + + if (iface->ksni_index == index) + return iface; + } + + return NULL; +} + +static int ksocknal_ip2index(struct sockaddr *addr, struct lnet_ni *ni) +{ + struct net_device *dev; + int ret = -1; + DECLARE_CONST_IN_IFADDR(ifa); - return (NULL); + if (addr->sa_family != AF_INET) + /* No IPv6 support yet */ + return ret; + + rcu_read_lock(); + for_each_netdev(ni->ni_net_ns, dev) { + int flags = dev_get_flags(dev); + struct in_device *in_dev; + + if (flags & IFF_LOOPBACK) /* skip the loopback IF */ + continue; + + if (!(flags & IFF_UP)) + continue; + + in_dev = __in_dev_get_rcu(dev); + if (!in_dev) + continue; + + in_dev_for_each_ifa_rcu(ifa, in_dev) { + if (ifa->ifa_local == + ((struct sockaddr_in *)addr)->sin_addr.s_addr) + ret = dev->ifindex; + } + endfor_ifa(in_dev); + if (ret >= 0) + break; + } + rcu_read_unlock(); + + return ret; } -ksock_route_t * -ksocknal_create_route (__u32 ipaddr, int port) +static struct ksock_route * +ksocknal_create_route(struct sockaddr *addr) { - ksock_route_t *route; + struct ksock_route *route; LIBCFS_ALLOC (route, sizeof (*route)); if (route == NULL) return (NULL); - atomic_set (&route->ksnr_refcount, 1); + refcount_set(&route->ksnr_refcount, 1); route->ksnr_peer = NULL; route->ksnr_retry_interval = 0; /* OK to connect at any time */ - route->ksnr_ipaddr = ipaddr; - route->ksnr_port = port; - route->ksnr_scheduled = 0; - route->ksnr_connecting = 0; - route->ksnr_connected = 0; - route->ksnr_deleted = 0; - route->ksnr_conn_count = 0; - route->ksnr_share_count = 0; - - return (route); + rpc_copy_addr((struct sockaddr *)&route->ksnr_addr, addr); + rpc_set_port((struct sockaddr *)&route->ksnr_addr, rpc_get_port(addr)); + route->ksnr_myiface = -1; + route->ksnr_scheduled = 0; + route->ksnr_connecting = 0; + route->ksnr_connected = 0; + route->ksnr_deleted = 0; + route->ksnr_conn_count = 0; + route->ksnr_share_count = 0; + + return route; } void -ksocknal_destroy_route (ksock_route_t *route) +ksocknal_destroy_route(struct ksock_route *route) { - LASSERT (atomic_read(&route->ksnr_refcount) == 0); + LASSERT(refcount_read(&route->ksnr_refcount) == 0); if (route->ksnr_peer != NULL) ksocknal_peer_decref(route->ksnr_peer); @@ -99,667 +155,675 @@ ksocknal_destroy_route (ksock_route_t *route) LIBCFS_FREE (route, sizeof (*route)); } -int -ksocknal_create_peer (ksock_peer_t **peerp, lnet_ni_t *ni, lnet_process_id_t id) +static struct ksock_peer_ni * +ksocknal_create_peer(struct lnet_ni *ni, struct lnet_process_id id) { - ksock_net_t *net = ni->ni_data; - ksock_peer_t *peer; - - LASSERT (id.nid != LNET_NID_ANY); - LASSERT (id.pid != LNET_PID_ANY); - LASSERT (!in_interrupt()); - - LIBCFS_ALLOC (peer, sizeof (*peer)); - if (peer == NULL) - return -ENOMEM; - - memset (peer, 0, sizeof (*peer)); /* NULL pointers/clear flags etc */ + int cpt = lnet_cpt_of_nid(id.nid, ni); + struct ksock_net *net = ni->ni_data; + struct ksock_peer_ni *peer_ni; - peer->ksnp_ni = ni; - peer->ksnp_id = id; - atomic_set (&peer->ksnp_refcount, 1); /* 1 ref for caller */ - peer->ksnp_closing = 0; - peer->ksnp_accepting = 0; - peer->ksnp_proto = NULL; - peer->ksnp_last_alive = 0; - peer->ksnp_zc_next_cookie = SOCKNAL_KEEPALIVE_PING + 1; + LASSERT(id.nid != LNET_NID_ANY); + LASSERT(id.pid != LNET_PID_ANY); + LASSERT(!in_interrupt()); - CFS_INIT_LIST_HEAD (&peer->ksnp_conns); - CFS_INIT_LIST_HEAD (&peer->ksnp_routes); - CFS_INIT_LIST_HEAD (&peer->ksnp_tx_queue); - CFS_INIT_LIST_HEAD (&peer->ksnp_zc_req_list); - spin_lock_init(&peer->ksnp_lock); - - spin_lock_bh(&net->ksnn_lock); - - if (net->ksnn_shutdown) { - spin_unlock_bh(&net->ksnn_lock); - - LIBCFS_FREE(peer, sizeof(*peer)); - CERROR("Can't create peer: network shutdown\n"); - return -ESHUTDOWN; + if (!atomic_inc_unless_negative(&net->ksnn_npeers)) { + CERROR("Can't create peer_ni: network shutdown\n"); + return ERR_PTR(-ESHUTDOWN); } - net->ksnn_npeers++; - - spin_unlock_bh(&net->ksnn_lock); + LIBCFS_CPT_ALLOC(peer_ni, lnet_cpt_table(), cpt, sizeof(*peer_ni)); + if (!peer_ni) { + atomic_dec(&net->ksnn_npeers); + return ERR_PTR(-ENOMEM); + } - *peerp = peer; - return 0; + peer_ni->ksnp_ni = ni; + peer_ni->ksnp_id = id; + refcount_set(&peer_ni->ksnp_refcount, 1); /* 1 ref for caller */ + peer_ni->ksnp_closing = 0; + peer_ni->ksnp_accepting = 0; + peer_ni->ksnp_proto = NULL; + peer_ni->ksnp_last_alive = 0; + peer_ni->ksnp_zc_next_cookie = SOCKNAL_KEEPALIVE_PING + 1; + + INIT_LIST_HEAD(&peer_ni->ksnp_conns); + INIT_LIST_HEAD(&peer_ni->ksnp_routes); + INIT_LIST_HEAD(&peer_ni->ksnp_tx_queue); + INIT_LIST_HEAD(&peer_ni->ksnp_zc_req_list); + spin_lock_init(&peer_ni->ksnp_lock); + + return peer_ni; } void -ksocknal_destroy_peer (ksock_peer_t *peer) +ksocknal_destroy_peer(struct ksock_peer_ni *peer_ni) { - ksock_net_t *net = peer->ksnp_ni->ni_data; - - CDEBUG (D_NET, "peer %s %p deleted\n", - libcfs_id2str(peer->ksnp_id), peer); - - LASSERT (atomic_read (&peer->ksnp_refcount) == 0); - LASSERT (peer->ksnp_accepting == 0); - LASSERT (cfs_list_empty (&peer->ksnp_conns)); - LASSERT (cfs_list_empty (&peer->ksnp_routes)); - LASSERT (cfs_list_empty (&peer->ksnp_tx_queue)); - LASSERT (cfs_list_empty (&peer->ksnp_zc_req_list)); - - LIBCFS_FREE (peer, sizeof (*peer)); - - /* NB a peer's connections and routes 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. */ - spin_lock_bh(&net->ksnn_lock); - net->ksnn_npeers--; - spin_unlock_bh(&net->ksnn_lock); -} + struct ksock_net *net = peer_ni->ksnp_ni->ni_data; -ksock_peer_t * -ksocknal_find_peer_locked (lnet_ni_t *ni, lnet_process_id_t id) -{ - cfs_list_t *peer_list = ksocknal_nid2peerlist(id.nid); - cfs_list_t *tmp; - ksock_peer_t *peer; + CDEBUG (D_NET, "peer_ni %s %p deleted\n", + libcfs_id2str(peer_ni->ksnp_id), peer_ni); - cfs_list_for_each (tmp, peer_list) { + LASSERT(refcount_read(&peer_ni->ksnp_refcount) == 0); + LASSERT(peer_ni->ksnp_accepting == 0); + LASSERT(list_empty(&peer_ni->ksnp_conns)); + LASSERT(list_empty(&peer_ni->ksnp_routes)); + LASSERT(list_empty(&peer_ni->ksnp_tx_queue)); + LASSERT(list_empty(&peer_ni->ksnp_zc_req_list)); - peer = cfs_list_entry (tmp, ksock_peer_t, ksnp_list); + LIBCFS_FREE(peer_ni, sizeof(*peer_ni)); - LASSERT (!peer->ksnp_closing); + /* NB a peer_ni's connections and routes keep a reference on their + * peer_ni until they are destroyed, so we can be assured that _all_ + * state to do with this peer_ni has been cleaned up when its refcount + * drops to zero. + */ + if (atomic_dec_and_test(&net->ksnn_npeers)) + wake_up_var(&net->ksnn_npeers); +} + +struct ksock_peer_ni * +ksocknal_find_peer_locked(struct lnet_ni *ni, struct lnet_process_id id) +{ + struct ksock_peer_ni *peer_ni; - if (peer->ksnp_ni != ni) - continue; + hash_for_each_possible(ksocknal_data.ksnd_peers, peer_ni, + ksnp_list, id.nid) { + LASSERT(!peer_ni->ksnp_closing); - if (peer->ksnp_id.nid != id.nid || - peer->ksnp_id.pid != id.pid) - continue; + if (peer_ni->ksnp_ni != ni) + continue; + + if (peer_ni->ksnp_id.nid != id.nid || + peer_ni->ksnp_id.pid != id.pid) + continue; - CDEBUG(D_NET, "got peer [%p] -> %s (%d)\n", - peer, libcfs_id2str(id), - atomic_read(&peer->ksnp_refcount)); - return (peer); + CDEBUG(D_NET, "got peer_ni [%p] -> %s (%d)\n", + peer_ni, libcfs_id2str(id), + refcount_read(&peer_ni->ksnp_refcount)); + return peer_ni; } - return (NULL); + return NULL; } -ksock_peer_t * -ksocknal_find_peer (lnet_ni_t *ni, lnet_process_id_t id) +struct ksock_peer_ni * +ksocknal_find_peer(struct lnet_ni *ni, struct lnet_process_id id) { - ksock_peer_t *peer; + struct ksock_peer_ni *peer_ni; read_lock(&ksocknal_data.ksnd_global_lock); - peer = ksocknal_find_peer_locked(ni, id); - if (peer != NULL) /* +1 ref for caller? */ - ksocknal_peer_addref(peer); + peer_ni = ksocknal_find_peer_locked(ni, id); + if (peer_ni != NULL) /* +1 ref for caller? */ + ksocknal_peer_addref(peer_ni); read_unlock(&ksocknal_data.ksnd_global_lock); - return (peer); + return (peer_ni); } -void -ksocknal_unlink_peer_locked (ksock_peer_t *peer) +static void +ksocknal_unlink_peer_locked(struct ksock_peer_ni *peer_ni) { - int i; - __u32 ip; - ksock_interface_t *iface; - - for (i = 0; i < peer->ksnp_n_passive_ips; i++) { - LASSERT (i < LNET_MAX_INTERFACES); - ip = peer->ksnp_passive_ips[i]; - - iface = ksocknal_ip2iface(peer->ksnp_ni, ip); - /* All IPs in peer->ksnp_passive_ips[] come from the - * interface list, therefore the call must succeed. */ - LASSERT (iface != NULL); - - CDEBUG(D_NET, "peer=%p iface=%p ksni_nroutes=%d\n", - peer, iface, iface->ksni_nroutes); - iface->ksni_npeers--; - } + int i; + struct ksock_interface *iface; + + for (i = 0; i < peer_ni->ksnp_n_passive_ips; i++) { + struct sockaddr_in sa = { .sin_family = AF_INET }; + LASSERT(i < LNET_INTERFACES_NUM); + sa.sin_addr.s_addr = htonl(peer_ni->ksnp_passive_ips[i]); + + iface = ksocknal_ip2iface(peer_ni->ksnp_ni, + (struct sockaddr *)&sa); + /* + * All IPs in peer_ni->ksnp_passive_ips[] come from the + * interface list, therefore the call must succeed. + */ + LASSERT(iface != NULL); + + CDEBUG(D_NET, "peer_ni=%p iface=%p ksni_nroutes=%d\n", + peer_ni, iface, iface->ksni_nroutes); + iface->ksni_npeers--; + } - LASSERT (cfs_list_empty(&peer->ksnp_conns)); - LASSERT (cfs_list_empty(&peer->ksnp_routes)); - LASSERT (!peer->ksnp_closing); - peer->ksnp_closing = 1; - cfs_list_del (&peer->ksnp_list); - /* lose peerlist's ref */ - ksocknal_peer_decref(peer); + LASSERT(list_empty(&peer_ni->ksnp_conns)); + LASSERT(list_empty(&peer_ni->ksnp_routes)); + LASSERT(!peer_ni->ksnp_closing); + peer_ni->ksnp_closing = 1; + hlist_del(&peer_ni->ksnp_list); + /* lose peerlist's ref */ + ksocknal_peer_decref(peer_ni); } -int -ksocknal_get_peer_info (lnet_ni_t *ni, int index, - lnet_process_id_t *id, __u32 *myip, __u32 *peer_ip, - int *port, int *conn_count, int *share_count) +static int +ksocknal_get_peer_info(struct lnet_ni *ni, int index, + struct lnet_process_id *id, __u32 *myip, __u32 *peer_ip, + int *port, int *conn_count, int *share_count) { - ksock_peer_t *peer; - cfs_list_t *ptmp; - ksock_route_t *route; - cfs_list_t *rtmp; - int i; - int j; - int rc = -ENOENT; + struct ksock_peer_ni *peer_ni; + struct ksock_route *route; + struct list_head *rtmp; + int i; + int j; + int rc = -ENOENT; read_lock(&ksocknal_data.ksnd_global_lock); - for (i = 0; i < ksocknal_data.ksnd_peer_hash_size; i++) { - - cfs_list_for_each (ptmp, &ksocknal_data.ksnd_peers[i]) { - peer = cfs_list_entry (ptmp, ksock_peer_t, ksnp_list); + hash_for_each(ksocknal_data.ksnd_peers, i, peer_ni, ksnp_list) { - if (peer->ksnp_ni != ni) - continue; - - if (peer->ksnp_n_passive_ips == 0 && - cfs_list_empty(&peer->ksnp_routes)) { - if (index-- > 0) - continue; - - *id = peer->ksnp_id; - *myip = 0; - *peer_ip = 0; - *port = 0; - *conn_count = 0; - *share_count = 0; - rc = 0; - goto out; - } - - for (j = 0; j < peer->ksnp_n_passive_ips; j++) { - if (index-- > 0) - continue; + if (peer_ni->ksnp_ni != ni) + continue; - *id = peer->ksnp_id; - *myip = peer->ksnp_passive_ips[j]; - *peer_ip = 0; - *port = 0; - *conn_count = 0; - *share_count = 0; - rc = 0; - goto out; - } + if (peer_ni->ksnp_n_passive_ips == 0 && + list_empty(&peer_ni->ksnp_routes)) { + if (index-- > 0) + continue; + + *id = peer_ni->ksnp_id; + *myip = 0; + *peer_ip = 0; + *port = 0; + *conn_count = 0; + *share_count = 0; + rc = 0; + goto out; + } - cfs_list_for_each (rtmp, &peer->ksnp_routes) { - if (index-- > 0) - continue; + for (j = 0; j < peer_ni->ksnp_n_passive_ips; j++) { + if (index-- > 0) + continue; + + *id = peer_ni->ksnp_id; + *myip = peer_ni->ksnp_passive_ips[j]; + *peer_ip = 0; + *port = 0; + *conn_count = 0; + *share_count = 0; + rc = 0; + goto out; + } - route = cfs_list_entry(rtmp, ksock_route_t, - ksnr_list); - - *id = peer->ksnp_id; - *myip = route->ksnr_myipaddr; - *peer_ip = route->ksnr_ipaddr; - *port = route->ksnr_port; - *conn_count = route->ksnr_conn_count; - *share_count = route->ksnr_share_count; - rc = 0; - goto out; - } - } - } - out: + list_for_each(rtmp, &peer_ni->ksnp_routes) { + if (index-- > 0) + continue; + + route = list_entry(rtmp, struct ksock_route, + ksnr_list); + + *id = peer_ni->ksnp_id; + if (route->ksnr_addr.ss_family == AF_INET) { + struct sockaddr_in *sa = + (void *)&route->ksnr_addr; + rc = choose_ipv4_src( + myip, + route->ksnr_myiface, + ntohl(sa->sin_addr.s_addr), + ni->ni_net_ns); + *peer_ip = ntohl(sa->sin_addr.s_addr); + *port = ntohs(sa->sin_port); + } else { + *myip = 0xFFFFFFFF; + *peer_ip = 0xFFFFFFFF; + *port = 0; + rc = -ENOTSUPP; + } + *conn_count = route->ksnr_conn_count; + *share_count = route->ksnr_share_count; + goto out; + } + } +out: read_unlock(&ksocknal_data.ksnd_global_lock); - return (rc); + return rc; } -void -ksocknal_associate_route_conn_locked(ksock_route_t *route, ksock_conn_t *conn) +static void +ksocknal_associate_route_conn_locked(struct ksock_route *route, + struct ksock_conn *conn) { - ksock_peer_t *peer = route->ksnr_peer; - int type = conn->ksnc_type; - ksock_interface_t *iface; - - conn->ksnc_route = route; - ksocknal_route_addref(route); - - if (route->ksnr_myipaddr != conn->ksnc_myipaddr) { - if (route->ksnr_myipaddr == 0) { - /* route wasn't bound locally yet (the initial route) */ - CDEBUG(D_NET, "Binding %s %u.%u.%u.%u to %u.%u.%u.%u\n", - libcfs_id2str(peer->ksnp_id), - HIPQUAD(route->ksnr_ipaddr), - HIPQUAD(conn->ksnc_myipaddr)); - } else { - CDEBUG(D_NET, "Rebinding %s %u.%u.%u.%u from " - "%u.%u.%u.%u to %u.%u.%u.%u\n", - libcfs_id2str(peer->ksnp_id), - HIPQUAD(route->ksnr_ipaddr), - HIPQUAD(route->ksnr_myipaddr), - HIPQUAD(conn->ksnc_myipaddr)); - - iface = ksocknal_ip2iface(route->ksnr_peer->ksnp_ni, - route->ksnr_myipaddr); - if (iface != NULL) - iface->ksni_nroutes--; - } - route->ksnr_myipaddr = conn->ksnc_myipaddr; - iface = ksocknal_ip2iface(route->ksnr_peer->ksnp_ni, - route->ksnr_myipaddr); - if (iface != NULL) - iface->ksni_nroutes++; - } + struct ksock_peer_ni *peer_ni = route->ksnr_peer; + int type = conn->ksnc_type; + struct ksock_interface *iface; + int conn_iface = + ksocknal_ip2index((struct sockaddr *)&conn->ksnc_myaddr, + route->ksnr_peer->ksnp_ni); + + conn->ksnc_route = route; + ksocknal_route_addref(route); + + if (route->ksnr_myiface != conn_iface) { + if (route->ksnr_myiface < 0) { + /* route wasn't bound locally yet (the initial route) */ + CDEBUG(D_NET, "Binding %s %pIS to interface %d\n", + libcfs_id2str(peer_ni->ksnp_id), + &route->ksnr_addr, + conn_iface); + } else { + CDEBUG(D_NET, + "Rebinding %s %pIS from interface %d to %d\n", + libcfs_id2str(peer_ni->ksnp_id), + &route->ksnr_addr, + route->ksnr_myiface, + conn_iface); + + iface = ksocknal_index2iface(route->ksnr_peer->ksnp_ni, + route->ksnr_myiface); + if (iface) + iface->ksni_nroutes--; + } + route->ksnr_myiface = conn_iface; + iface = ksocknal_index2iface(route->ksnr_peer->ksnp_ni, + route->ksnr_myiface); + if (iface) + iface->ksni_nroutes++; + } - route->ksnr_connected |= (1<ksnr_conn_count++; + route->ksnr_connected |= (1<ksnr_conn_count++; - /* Successful connection => further attempts can - * proceed immediately */ - route->ksnr_retry_interval = 0; + /* Successful connection => further attempts can + * proceed immediately + */ + route->ksnr_retry_interval = 0; } -void -ksocknal_add_route_locked (ksock_peer_t *peer, ksock_route_t *route) +static void +ksocknal_add_route_locked(struct ksock_peer_ni *peer_ni, struct ksock_route *route) { - cfs_list_t *tmp; - ksock_conn_t *conn; - ksock_route_t *route2; - - LASSERT (!peer->ksnp_closing); - LASSERT (route->ksnr_peer == NULL); - LASSERT (!route->ksnr_scheduled); - LASSERT (!route->ksnr_connecting); - LASSERT (route->ksnr_connected == 0); - - /* LASSERT(unique) */ - cfs_list_for_each(tmp, &peer->ksnp_routes) { - route2 = cfs_list_entry(tmp, ksock_route_t, ksnr_list); - - if (route2->ksnr_ipaddr == route->ksnr_ipaddr) { - CERROR ("Duplicate route %s %u.%u.%u.%u\n", - libcfs_id2str(peer->ksnp_id), - HIPQUAD(route->ksnr_ipaddr)); - LBUG(); - } - } + struct list_head *tmp; + struct ksock_conn *conn; + struct ksock_route *route2; + struct ksock_net *net = peer_ni->ksnp_ni->ni_data; + + LASSERT(!peer_ni->ksnp_closing); + LASSERT(route->ksnr_peer == NULL); + LASSERT(!route->ksnr_scheduled); + LASSERT(!route->ksnr_connecting); + LASSERT(route->ksnr_connected == 0); + LASSERT(net->ksnn_ninterfaces > 0); + + /* LASSERT(unique) */ + list_for_each(tmp, &peer_ni->ksnp_routes) { + route2 = list_entry(tmp, struct ksock_route, ksnr_list); + + if (rpc_cmp_addr((struct sockaddr *)&route2->ksnr_addr, + (struct sockaddr *)&route->ksnr_addr)) { + CERROR("Duplicate route %s %pI4h\n", + libcfs_id2str(peer_ni->ksnp_id), + &route->ksnr_addr); + LBUG(); + } + } - route->ksnr_peer = peer; - ksocknal_peer_addref(peer); - /* peer's routelist takes over my ref on 'route' */ - cfs_list_add_tail(&route->ksnr_list, &peer->ksnp_routes); + route->ksnr_peer = peer_ni; + ksocknal_peer_addref(peer_ni); - cfs_list_for_each(tmp, &peer->ksnp_conns) { - conn = cfs_list_entry(tmp, ksock_conn_t, ksnc_list); + /* set the route's interface to the current net's interface */ + route->ksnr_myiface = net->ksnn_interfaces[0].ksni_index; + net->ksnn_interfaces[0].ksni_nroutes++; - if (conn->ksnc_ipaddr != route->ksnr_ipaddr) - continue; + /* peer_ni's routelist takes over my ref on 'route' */ + list_add_tail(&route->ksnr_list, &peer_ni->ksnp_routes); - ksocknal_associate_route_conn_locked(route, conn); - /* keep going (typed routes) */ - } + list_for_each(tmp, &peer_ni->ksnp_conns) { + conn = list_entry(tmp, struct ksock_conn, ksnc_list); + + if (!rpc_cmp_addr((struct sockaddr *)&conn->ksnc_peeraddr, + (struct sockaddr *)&route->ksnr_addr)) + continue; + + ksocknal_associate_route_conn_locked(route, conn); + /* keep going (typed routes) */ + } } -void -ksocknal_del_route_locked (ksock_route_t *route) +static void +ksocknal_del_route_locked(struct ksock_route *route) { - ksock_peer_t *peer = route->ksnr_peer; - ksock_interface_t *iface; - ksock_conn_t *conn; - cfs_list_t *ctmp; - cfs_list_t *cnxt; + struct ksock_peer_ni *peer_ni = route->ksnr_peer; + struct ksock_interface *iface; + struct ksock_conn *conn; + struct ksock_conn *cnxt; - LASSERT (!route->ksnr_deleted); + LASSERT(!route->ksnr_deleted); - /* Close associated conns */ - cfs_list_for_each_safe (ctmp, cnxt, &peer->ksnp_conns) { - conn = cfs_list_entry(ctmp, ksock_conn_t, ksnc_list); - - if (conn->ksnc_route != route) - continue; + /* Close associated conns */ + list_for_each_entry_safe(conn, cnxt, &peer_ni->ksnp_conns, ksnc_list) { + if (conn->ksnc_route != route) + continue; - ksocknal_close_conn_locked (conn, 0); - } + ksocknal_close_conn_locked(conn, 0); + } - if (route->ksnr_myipaddr != 0) { - iface = ksocknal_ip2iface(route->ksnr_peer->ksnp_ni, - route->ksnr_myipaddr); - if (iface != NULL) - iface->ksni_nroutes--; - } + if (route->ksnr_myiface >= 0) { + iface = ksocknal_index2iface(route->ksnr_peer->ksnp_ni, + route->ksnr_myiface); + if (iface) + iface->ksni_nroutes--; + } - route->ksnr_deleted = 1; - cfs_list_del (&route->ksnr_list); - ksocknal_route_decref(route); /* drop peer's ref */ + route->ksnr_deleted = 1; + list_del(&route->ksnr_list); + ksocknal_route_decref(route); /* drop peer_ni's ref */ - if (cfs_list_empty (&peer->ksnp_routes) && - cfs_list_empty (&peer->ksnp_conns)) { - /* I've just removed the last route to a peer with no active - * connections */ - ksocknal_unlink_peer_locked (peer); - } + if (list_empty(&peer_ni->ksnp_routes) && + list_empty(&peer_ni->ksnp_conns)) { + /* I've just removed the last route to a peer_ni with no active + * connections */ + ksocknal_unlink_peer_locked(peer_ni); + } } int -ksocknal_add_peer (lnet_ni_t *ni, lnet_process_id_t id, __u32 ipaddr, int port) +ksocknal_add_peer(struct lnet_ni *ni, struct lnet_process_id id, + struct sockaddr *addr) { - cfs_list_t *tmp; - ksock_peer_t *peer; - ksock_peer_t *peer2; - ksock_route_t *route; - ksock_route_t *route2; - int rc; - - if (id.nid == LNET_NID_ANY || - id.pid == LNET_PID_ANY) - return (-EINVAL); - - /* Have a brand new peer ready... */ - rc = ksocknal_create_peer(&peer, ni, id); - if (rc != 0) - return rc; - - route = ksocknal_create_route (ipaddr, port); - if (route == NULL) { - ksocknal_peer_decref(peer); - return (-ENOMEM); - } + struct list_head *tmp; + struct ksock_peer_ni *peer_ni; + struct ksock_peer_ni *peer2; + struct ksock_route *route; + struct ksock_route *route2; + + if (id.nid == LNET_NID_ANY || + id.pid == LNET_PID_ANY) + return (-EINVAL); + + /* Have a brand new peer_ni ready... */ + peer_ni = ksocknal_create_peer(ni, id); + if (IS_ERR(peer_ni)) + return PTR_ERR(peer_ni); + + route = ksocknal_create_route(addr); + if (route == NULL) { + ksocknal_peer_decref(peer_ni); + return (-ENOMEM); + } write_lock_bh(&ksocknal_data.ksnd_global_lock); - /* always called with a ref on ni, so shutdown can't have started */ - LASSERT (((ksock_net_t *) ni->ni_data)->ksnn_shutdown == 0); + /* always called with a ref on ni, so shutdown can't have started */ + LASSERT(atomic_read(&((struct ksock_net *)ni->ni_data)->ksnn_npeers) + >= 0); - peer2 = ksocknal_find_peer_locked (ni, id); - if (peer2 != NULL) { - ksocknal_peer_decref(peer); - peer = peer2; - } else { - /* peer table takes my ref on peer */ - cfs_list_add_tail (&peer->ksnp_list, - ksocknal_nid2peerlist (id.nid)); - } + peer2 = ksocknal_find_peer_locked(ni, id); + if (peer2 != NULL) { + ksocknal_peer_decref(peer_ni); + peer_ni = peer2; + } else { + /* peer_ni table takes my ref on peer_ni */ + hash_add(ksocknal_data.ksnd_peers, &peer_ni->ksnp_list, id.nid); + } - route2 = NULL; - cfs_list_for_each (tmp, &peer->ksnp_routes) { - route2 = cfs_list_entry(tmp, ksock_route_t, ksnr_list); + route2 = NULL; + list_for_each(tmp, &peer_ni->ksnp_routes) { + route2 = list_entry(tmp, struct ksock_route, ksnr_list); - if (route2->ksnr_ipaddr == ipaddr) - break; + if (rpc_cmp_addr(addr, (struct sockaddr *)&route2->ksnr_addr)) + break; - route2 = NULL; - } - if (route2 == NULL) { - ksocknal_add_route_locked(peer, route); - route->ksnr_share_count++; - } else { - ksocknal_route_decref(route); - route2->ksnr_share_count++; - } + route2 = NULL; + } + if (route2 == NULL) { + ksocknal_add_route_locked(peer_ni, route); + route->ksnr_share_count++; + } else { + ksocknal_route_decref(route); + route2->ksnr_share_count++; + } write_unlock_bh(&ksocknal_data.ksnd_global_lock); - return (0); + return 0; } -void -ksocknal_del_peer_locked (ksock_peer_t *peer, __u32 ip) +static void +ksocknal_del_peer_locked(struct ksock_peer_ni *peer_ni, __u32 ip) { - ksock_conn_t *conn; - ksock_route_t *route; - cfs_list_t *tmp; - cfs_list_t *nxt; - int nshared; - - LASSERT (!peer->ksnp_closing); - - /* Extra ref prevents peer disappearing until I'm done with it */ - ksocknal_peer_addref(peer); - - cfs_list_for_each_safe (tmp, nxt, &peer->ksnp_routes) { - route = cfs_list_entry(tmp, ksock_route_t, ksnr_list); - - /* no match */ - if (!(ip == 0 || route->ksnr_ipaddr == ip)) - continue; - - route->ksnr_share_count = 0; - /* This deletes associated conns too */ - ksocknal_del_route_locked (route); - } - - nshared = 0; - cfs_list_for_each_safe (tmp, nxt, &peer->ksnp_routes) { - route = cfs_list_entry(tmp, ksock_route_t, ksnr_list); - nshared += route->ksnr_share_count; - } - - if (nshared == 0) { - /* remove everything else if there are no explicit entries - * left */ - - cfs_list_for_each_safe (tmp, nxt, &peer->ksnp_routes) { - route = cfs_list_entry(tmp, ksock_route_t, ksnr_list); + struct ksock_conn *conn; + struct ksock_conn *cnxt; + struct ksock_route *route; + struct ksock_route *rnxt; + int nshared; + + LASSERT(!peer_ni->ksnp_closing); + + /* Extra ref prevents peer_ni disappearing until I'm done with it */ + ksocknal_peer_addref(peer_ni); + + list_for_each_entry_safe(route, rnxt, &peer_ni->ksnp_routes, + ksnr_list) { + /* no match */ + if (ip) { + if (route->ksnr_addr.ss_family != AF_INET) + continue; + if (((struct sockaddr_in *)&route->ksnr_addr) + ->sin_addr.s_addr != htonl(ip)) + continue; + } - /* we should only be removing auto-entries */ - LASSERT(route->ksnr_share_count == 0); - ksocknal_del_route_locked (route); - } + route->ksnr_share_count = 0; + /* This deletes associated conns too */ + ksocknal_del_route_locked(route); + } - cfs_list_for_each_safe (tmp, nxt, &peer->ksnp_conns) { - conn = cfs_list_entry(tmp, ksock_conn_t, ksnc_list); + nshared = 0; + list_for_each_entry(route, &peer_ni->ksnp_routes, ksnr_list) + nshared += route->ksnr_share_count; + + if (nshared == 0) { + /* remove everything else if there are no explicit entries + * left + */ + list_for_each_entry_safe(route, rnxt, &peer_ni->ksnp_routes, + ksnr_list) { + /* we should only be removing auto-entries */ + LASSERT(route->ksnr_share_count == 0); + ksocknal_del_route_locked(route); + } - ksocknal_close_conn_locked(conn, 0); - } - } + list_for_each_entry_safe(conn, cnxt, &peer_ni->ksnp_conns, + ksnc_list) + ksocknal_close_conn_locked(conn, 0); + } - ksocknal_peer_decref(peer); - /* NB peer unlinks itself when last conn/route is removed */ + ksocknal_peer_decref(peer_ni); + /* NB peer_ni unlinks itself when last conn/route is removed */ } -int -ksocknal_del_peer (lnet_ni_t *ni, lnet_process_id_t id, __u32 ip) +static int +ksocknal_del_peer(struct lnet_ni *ni, struct lnet_process_id id, __u32 ip) { - CFS_LIST_HEAD (zombies); - cfs_list_t *ptmp; - cfs_list_t *pnxt; - ksock_peer_t *peer; - int lo; - int hi; - int i; - int rc = -ENOENT; + LIST_HEAD(zombies); + struct hlist_node *pnxt; + struct ksock_peer_ni *peer_ni; + int lo; + int hi; + int i; + int rc = -ENOENT; write_lock_bh(&ksocknal_data.ksnd_global_lock); - if (id.nid != LNET_NID_ANY) - lo = hi = (int)(ksocknal_nid2peerlist(id.nid) - ksocknal_data.ksnd_peers); - else { - lo = 0; - hi = ksocknal_data.ksnd_peer_hash_size - 1; - } - - for (i = lo; i <= hi; i++) { - cfs_list_for_each_safe (ptmp, pnxt, - &ksocknal_data.ksnd_peers[i]) { - peer = cfs_list_entry (ptmp, ksock_peer_t, ksnp_list); + if (id.nid != LNET_NID_ANY) { + lo = hash_min(id.nid, HASH_BITS(ksocknal_data.ksnd_peers)); + hi = lo; + } else { + lo = 0; + hi = HASH_SIZE(ksocknal_data.ksnd_peers) - 1; + } - if (peer->ksnp_ni != ni) - continue; + for (i = lo; i <= hi; i++) { + hlist_for_each_entry_safe(peer_ni, pnxt, + &ksocknal_data.ksnd_peers[i], + ksnp_list) { + if (peer_ni->ksnp_ni != ni) + continue; - if (!((id.nid == LNET_NID_ANY || peer->ksnp_id.nid == id.nid) && - (id.pid == LNET_PID_ANY || peer->ksnp_id.pid == id.pid))) - continue; + if (!((id.nid == LNET_NID_ANY || + peer_ni->ksnp_id.nid == id.nid) && + (id.pid == LNET_PID_ANY || + peer_ni->ksnp_id.pid == id.pid))) + continue; - ksocknal_peer_addref(peer); /* a ref for me... */ + ksocknal_peer_addref(peer_ni); /* a ref for me... */ - ksocknal_del_peer_locked (peer, ip); + ksocknal_del_peer_locked(peer_ni, ip); - if (peer->ksnp_closing && - !cfs_list_empty(&peer->ksnp_tx_queue)) { - LASSERT (cfs_list_empty(&peer->ksnp_conns)); - LASSERT (cfs_list_empty(&peer->ksnp_routes)); + if (peer_ni->ksnp_closing && + !list_empty(&peer_ni->ksnp_tx_queue)) { + LASSERT(list_empty(&peer_ni->ksnp_conns)); + LASSERT(list_empty(&peer_ni->ksnp_routes)); - cfs_list_splice_init(&peer->ksnp_tx_queue, - &zombies); - } + list_splice_init(&peer_ni->ksnp_tx_queue, + &zombies); + } - ksocknal_peer_decref(peer); /* ...till here */ + ksocknal_peer_decref(peer_ni); /* ...till here */ - rc = 0; /* matched! */ - } - } + rc = 0; /* matched! */ + } + } write_unlock_bh(&ksocknal_data.ksnd_global_lock); - ksocknal_txlist_done(ni, &zombies, 1); + ksocknal_txlist_done(ni, &zombies, -ENETDOWN); - return (rc); + return rc; } -ksock_conn_t * -ksocknal_get_conn_by_idx (lnet_ni_t *ni, int index) +static struct ksock_conn * +ksocknal_get_conn_by_idx(struct lnet_ni *ni, int index) { - ksock_peer_t *peer; - cfs_list_t *ptmp; - ksock_conn_t *conn; - cfs_list_t *ctmp; - int i; + struct ksock_peer_ni *peer_ni; + struct ksock_conn *conn; + struct list_head *ctmp; + int i; read_lock(&ksocknal_data.ksnd_global_lock); - for (i = 0; i < ksocknal_data.ksnd_peer_hash_size; i++) { - cfs_list_for_each (ptmp, &ksocknal_data.ksnd_peers[i]) { - peer = cfs_list_entry (ptmp, ksock_peer_t, ksnp_list); - - LASSERT (!peer->ksnp_closing); + hash_for_each(ksocknal_data.ksnd_peers, i, peer_ni, ksnp_list) { + LASSERT(!peer_ni->ksnp_closing); - if (peer->ksnp_ni != ni) - continue; + if (peer_ni->ksnp_ni != ni) + continue; - cfs_list_for_each (ctmp, &peer->ksnp_conns) { - if (index-- > 0) - continue; + list_for_each(ctmp, &peer_ni->ksnp_conns) { + if (index-- > 0) + continue; - conn = cfs_list_entry (ctmp, ksock_conn_t, - ksnc_list); - ksocknal_conn_addref(conn); - read_unlock(&ksocknal_data. \ - ksnd_global_lock); - return (conn); - } - } - } + conn = list_entry(ctmp, struct ksock_conn, + ksnc_list); + ksocknal_conn_addref(conn); + read_unlock(&ksocknal_data.ksnd_global_lock); + return conn; + } + } read_unlock(&ksocknal_data.ksnd_global_lock); - return (NULL); + return NULL; } -ksock_sched_t * +static struct ksock_sched * ksocknal_choose_scheduler_locked(unsigned int cpt) { - struct ksock_sched_info *info = ksocknal_data.ksnd_sched_info[cpt]; - ksock_sched_t *sched; - int i; - - LASSERT(info->ksi_nthreads > 0); - - sched = &info->ksi_scheds[0]; - /* - * NB: it's safe so far, but info->ksi_nthreads could be changed - * at runtime when we have dynamic LNet configuration, then we - * need to take care of this. - */ - for (i = 1; i < info->ksi_nthreads; i++) { - if (sched->kss_nconns > info->ksi_scheds[i].kss_nconns) - sched = &info->ksi_scheds[i]; + struct ksock_sched *sched = ksocknal_data.ksnd_schedulers[cpt]; + int i; + + if (sched->kss_nthreads == 0) { + cfs_percpt_for_each(sched, i, ksocknal_data.ksnd_schedulers) { + if (sched->kss_nthreads > 0) { + CDEBUG(D_NET, "scheduler[%d] has no threads. selected scheduler[%d]\n", + cpt, sched->kss_cpt); + return sched; + } + } + return NULL; } return sched; } -int -ksocknal_local_ipvec (lnet_ni_t *ni, __u32 *ipaddrs) +static int +ksocknal_local_ipvec(struct lnet_ni *ni, __u32 *ipaddrs) { - ksock_net_t *net = ni->ni_data; - int i; - int nip; + struct ksock_net *net = ni->ni_data; + int i, j; + int nip; read_lock(&ksocknal_data.ksnd_global_lock); - nip = net->ksnn_ninterfaces; - LASSERT (nip <= LNET_MAX_INTERFACES); + nip = net->ksnn_ninterfaces; + LASSERT(nip <= LNET_INTERFACES_NUM); - /* Only offer interfaces for additional connections if I have - * more than one. */ - if (nip < 2) { - read_unlock(&ksocknal_data.ksnd_global_lock); - return 0; - } + for (i = 0, j = 0; i < nip; i++) + if (net->ksnn_interfaces[i].ksni_addr.ss_family == AF_INET) { + struct sockaddr_in *sa = + (void *)&net->ksnn_interfaces[i].ksni_addr; - for (i = 0; i < nip; i++) { - ipaddrs[i] = net->ksnn_interfaces[i].ksni_ipaddr; - LASSERT (ipaddrs[i] != 0); - } + ipaddrs[j] = ntohl(sa->sin_addr.s_addr); + LASSERT(ipaddrs[j] != 0); + j += 1; + } + nip = j; read_unlock(&ksocknal_data.ksnd_global_lock); - return (nip); + /* + * Only offer interfaces for additional connections if I have + * more than one. + */ + return nip < 2 ? 0 : nip; } -int -ksocknal_match_peerip (ksock_interface_t *iface, __u32 *ips, int nips) +static int +ksocknal_match_peerip(struct ksock_interface *iface, __u32 *ips, int nips) { - int best_netmatch = 0; - int best_xor = 0; - int best = -1; - int this_xor; - int this_netmatch; - int i; - - for (i = 0; i < nips; i++) { - if (ips[i] == 0) - continue; - - this_xor = (ips[i] ^ iface->ksni_ipaddr); - this_netmatch = ((this_xor & iface->ksni_netmask) == 0) ? 1 : 0; - - if (!(best < 0 || - best_netmatch < this_netmatch || - (best_netmatch == this_netmatch && - best_xor > this_xor))) - continue; - - best = i; - best_netmatch = this_netmatch; - best_xor = this_xor; - } + int best_netmatch = 0; + int best_xor = 0; + int best = -1; + int this_xor; + int this_netmatch; + int i; + struct sockaddr_in *sa; + __u32 ip; + + sa = (struct sockaddr_in *)&iface->ksni_addr; + LASSERT(sa->sin_family == AF_INET); + ip = ntohl(sa->sin_addr.s_addr); + + for (i = 0; i < nips; i++) { + if (ips[i] == 0) + continue; - LASSERT (best >= 0); - return (best); + this_xor = ips[i] ^ ip; + this_netmatch = ((this_xor & iface->ksni_netmask) == 0) ? 1 : 0; + + if (!(best < 0 || + best_netmatch < this_netmatch || + (best_netmatch == this_netmatch && + best_xor > this_xor))) + continue; + + best = i; + best_netmatch = this_netmatch; + best_xor = this_xor; + } + + LASSERT(best >= 0); + return best; } -int -ksocknal_select_ips(ksock_peer_t *peer, __u32 *peerips, int n_peerips) +static int +ksocknal_select_ips(struct ksock_peer_ni *peer_ni, __u32 *peerips, int n_peerips) { - rwlock_t *global_lock = &ksocknal_data.ksnd_global_lock; - ksock_net_t *net = peer->ksnp_ni->ni_data; - ksock_interface_t *iface; - ksock_interface_t *best_iface; - int n_ips; - int i; - int j; - int k; - __u32 ip; - __u32 xor; - int this_netmatch; - int best_netmatch; - int best_npeers; + rwlock_t *global_lock = &ksocknal_data.ksnd_global_lock; + struct ksock_net *net = peer_ni->ksnp_ni->ni_data; + struct ksock_interface *iface; + struct ksock_interface *best_iface; + int n_ips; + int i; + int j; + int k; + u32 ip; + u32 xor; + int this_netmatch; + int best_netmatch; + int best_npeers; /* CAVEAT EMPTOR: We do all our interface matching with an * exclusive hold of global lock at IRQ priority. We're only @@ -771,47 +835,57 @@ ksocknal_select_ips(ksock_peer_t *peer, __u32 *peerips, int n_peerips) write_lock_bh(global_lock); - LASSERT (n_peerips <= LNET_MAX_INTERFACES); - LASSERT (net->ksnn_ninterfaces <= LNET_MAX_INTERFACES); + LASSERT(n_peerips <= LNET_INTERFACES_NUM); + LASSERT(net->ksnn_ninterfaces <= LNET_INTERFACES_NUM); - /* Only match interfaces for additional connections - * if I have > 1 interface */ - n_ips = (net->ksnn_ninterfaces < 2) ? 0 : - MIN(n_peerips, net->ksnn_ninterfaces); + /* Only match interfaces for additional connections + * if I have > 1 interface + */ + n_ips = (net->ksnn_ninterfaces < 2) ? 0 : + min(n_peerips, net->ksnn_ninterfaces); - for (i = 0; peer->ksnp_n_passive_ips < n_ips; i++) { + for (i = 0; peer_ni->ksnp_n_passive_ips < n_ips; i++) { /* ^ yes really... */ /* If we have any new interfaces, first tick off all the - * peer IPs that match old interfaces, then choose new - * interfaces to match the remaining peer IPS. + * peer_ni IPs that match old interfaces, then choose new + * interfaces to match the remaining peer_ni IPS. * We don't forget interfaces we've stopped using; we might * start using them again... */ - if (i < peer->ksnp_n_passive_ips) { - /* Old interface. */ - ip = peer->ksnp_passive_ips[i]; - best_iface = ksocknal_ip2iface(peer->ksnp_ni, ip); + if (i < peer_ni->ksnp_n_passive_ips) { + /* Old interface. */ + struct sockaddr_in sa = { .sin_family = AF_INET}; - /* peer passive ips are kept up to date */ - LASSERT(best_iface != NULL); + sa.sin_addr.s_addr = + htonl(peer_ni->ksnp_passive_ips[i]); + best_iface = ksocknal_ip2iface(peer_ni->ksnp_ni, + (struct sockaddr *)&sa); + + /* peer_ni passive ips are kept up to date */ + LASSERT(best_iface != NULL); } else { /* choose a new interface */ - LASSERT (i == peer->ksnp_n_passive_ips); + struct sockaddr_in *sa; + + LASSERT (i == peer_ni->ksnp_n_passive_ips); best_iface = NULL; best_netmatch = 0; best_npeers = 0; - for (j = 0; j < net->ksnn_ninterfaces; j++) { - iface = &net->ksnn_interfaces[j]; - ip = iface->ksni_ipaddr; + for (j = 0; j < net->ksnn_ninterfaces; j++) { + iface = &net->ksnn_interfaces[j]; + sa = (void *)&iface->ksni_addr; + if (sa->sin_family != AF_INET) + continue; + ip = ntohl(sa->sin_addr.s_addr); - for (k = 0; k < peer->ksnp_n_passive_ips; k++) - if (peer->ksnp_passive_ips[k] == ip) + for (k = 0; k < peer_ni->ksnp_n_passive_ips; k++) + if (peer_ni->ksnp_passive_ips[k] == ip) break; - if (k < peer->ksnp_n_passive_ips) /* using it already */ + if (k < peer_ni->ksnp_n_passive_ips) /* using it already */ continue; k = ksocknal_match_peerip(iface, peerips, n_peerips); @@ -829,44 +903,45 @@ ksocknal_select_ips(ksock_peer_t *peer, __u32 *peerips, int n_peerips) best_npeers = iface->ksni_npeers; } - best_iface->ksni_npeers++; - ip = best_iface->ksni_ipaddr; - peer->ksnp_passive_ips[i] = ip; - peer->ksnp_n_passive_ips = i+1; - } + LASSERT(best_iface != NULL); - LASSERT (best_iface != NULL); + best_iface->ksni_npeers++; + sa = (void *)&best_iface->ksni_addr; + ip = ntohl(sa->sin_addr.s_addr); + peer_ni->ksnp_passive_ips[i] = ip; + peer_ni->ksnp_n_passive_ips = i+1; + } - /* mark the best matching peer IP used */ + /* mark the best matching peer_ni IP used */ j = ksocknal_match_peerip(best_iface, peerips, n_peerips); peerips[j] = 0; } - /* Overwrite input peer IP addresses */ - memcpy(peerips, peer->ksnp_passive_ips, n_ips * sizeof(*peerips)); + /* Overwrite input peer_ni IP addresses */ + memcpy(peerips, peer_ni->ksnp_passive_ips, n_ips * sizeof(*peerips)); write_unlock_bh(global_lock); return (n_ips); } -void -ksocknal_create_routes(ksock_peer_t *peer, int port, +static void +ksocknal_create_routes(struct ksock_peer_ni *peer_ni, int port, __u32 *peer_ipaddrs, int npeer_ipaddrs) { - ksock_route_t *newroute = NULL; + struct ksock_route *newroute = NULL; rwlock_t *global_lock = &ksocknal_data.ksnd_global_lock; - lnet_ni_t *ni = peer->ksnp_ni; - ksock_net_t *net = ni->ni_data; - cfs_list_t *rtmp; - ksock_route_t *route; - ksock_interface_t *iface; - ksock_interface_t *best_iface; - int best_netmatch; - int this_netmatch; - int best_nroutes; - int i; - int j; + struct lnet_ni *ni = peer_ni->ksnp_ni; + struct ksock_net *net = ni->ni_data; + struct list_head *rtmp; + struct ksock_route *route; + struct ksock_interface *iface; + struct ksock_interface *best_iface; + int best_netmatch; + int this_netmatch; + int best_nroutes; + int i; + int j; /* CAVEAT EMPTOR: We do all our interface matching with an * exclusive hold of global lock at IRQ priority. We're only @@ -876,165 +951,186 @@ ksocknal_create_routes(ksock_peer_t *peer, int port, write_lock_bh(global_lock); if (net->ksnn_ninterfaces < 2) { - /* Only create additional connections + /* Only create additional connections * if I have > 1 interface */ write_unlock_bh(global_lock); return; } - LASSERT (npeer_ipaddrs <= LNET_MAX_INTERFACES); + LASSERT(npeer_ipaddrs <= LNET_INTERFACES_NUM); + + for (i = 0; i < npeer_ipaddrs; i++) { + if (newroute) { + struct sockaddr_in *sa = (void *)&newroute->ksnr_addr; + + memset(sa, 0, sizeof(*sa)); + sa->sin_family = AF_INET; + sa->sin_addr.s_addr = htonl(peer_ipaddrs[i]); + } else { + struct sockaddr_in sa = {.sin_family = AF_INET}; - for (i = 0; i < npeer_ipaddrs; i++) { - if (newroute != NULL) { - newroute->ksnr_ipaddr = peer_ipaddrs[i]; - } else { write_unlock_bh(global_lock); - newroute = ksocknal_create_route(peer_ipaddrs[i], port); - if (newroute == NULL) - return; + sa.sin_addr.s_addr = htonl(peer_ipaddrs[i]); + sa.sin_port = htons(port); + newroute = + ksocknal_create_route((struct sockaddr *)&sa); + if (!newroute) + return; write_lock_bh(global_lock); - } + } - if (peer->ksnp_closing) { - /* peer got closed under me */ - break; - } + if (peer_ni->ksnp_closing) { + /* peer_ni got closed under me */ + break; + } - /* Already got a route? */ - route = NULL; - cfs_list_for_each(rtmp, &peer->ksnp_routes) { - route = cfs_list_entry(rtmp, ksock_route_t, ksnr_list); + /* Already got a route? */ + route = NULL; + list_for_each(rtmp, &peer_ni->ksnp_routes) { + route = list_entry(rtmp, struct ksock_route, ksnr_list); - if (route->ksnr_ipaddr == newroute->ksnr_ipaddr) - break; + if (rpc_cmp_addr( + (struct sockaddr *)&route->ksnr_addr, + (struct sockaddr *)&newroute->ksnr_addr)) + break; - route = NULL; - } - if (route != NULL) - continue; + route = NULL; + } + if (route != NULL) + continue; - best_iface = NULL; - best_nroutes = 0; - best_netmatch = 0; + best_iface = NULL; + best_nroutes = 0; + best_netmatch = 0; - LASSERT (net->ksnn_ninterfaces <= LNET_MAX_INTERFACES); + LASSERT(net->ksnn_ninterfaces <= LNET_INTERFACES_NUM); - /* Select interface to connect from */ - for (j = 0; j < net->ksnn_ninterfaces; j++) { - iface = &net->ksnn_interfaces[j]; + /* Select interface to connect from */ + for (j = 0; j < net->ksnn_ninterfaces; j++) { + __u32 iface_ip, route_ip; - /* Using this interface already? */ - cfs_list_for_each(rtmp, &peer->ksnp_routes) { - route = cfs_list_entry(rtmp, ksock_route_t, - ksnr_list); + iface = &net->ksnn_interfaces[j]; - if (route->ksnr_myipaddr == iface->ksni_ipaddr) - break; + /* Using this interface already? */ + list_for_each(rtmp, &peer_ni->ksnp_routes) { + route = list_entry(rtmp, struct ksock_route, + ksnr_list); - route = NULL; - } - if (route != NULL) - continue; - - this_netmatch = (((iface->ksni_ipaddr ^ - newroute->ksnr_ipaddr) & - iface->ksni_netmask) == 0) ? 1 : 0; - - if (!(best_iface == NULL || - best_netmatch < this_netmatch || - (best_netmatch == this_netmatch && - best_nroutes > iface->ksni_nroutes))) - continue; - - best_iface = iface; - best_netmatch = this_netmatch; - best_nroutes = iface->ksni_nroutes; - } + if (route->ksnr_myiface == iface->ksni_index) + break; + + route = NULL; + } + if (route != NULL) + continue; + if (iface->ksni_addr.ss_family != AF_INET) + continue; + if (newroute->ksnr_addr.ss_family != AF_INET) + continue; + + iface_ip = + ntohl(((struct sockaddr_in *) + &iface->ksni_addr)->sin_addr.s_addr); + route_ip = + ntohl(((struct sockaddr_in *) + &newroute->ksnr_addr)->sin_addr.s_addr); + + this_netmatch = (((iface_ip ^ route_ip) & + iface->ksni_netmask) == 0) ? 1 : 0; + + if (!(best_iface == NULL || + best_netmatch < this_netmatch || + (best_netmatch == this_netmatch && + best_nroutes > iface->ksni_nroutes))) + continue; + + best_iface = iface; + best_netmatch = this_netmatch; + best_nroutes = iface->ksni_nroutes; + } - if (best_iface == NULL) - continue; + if (best_iface == NULL) + continue; - newroute->ksnr_myipaddr = best_iface->ksni_ipaddr; - best_iface->ksni_nroutes++; + newroute->ksnr_myiface = best_iface->ksni_index; + best_iface->ksni_nroutes++; - ksocknal_add_route_locked(peer, newroute); - newroute = NULL; - } + ksocknal_add_route_locked(peer_ni, newroute); + newroute = NULL; + } write_unlock_bh(global_lock); - if (newroute != NULL) - ksocknal_route_decref(newroute); + if (newroute != NULL) + ksocknal_route_decref(newroute); } int -ksocknal_accept (lnet_ni_t *ni, cfs_socket_t *sock) +ksocknal_accept(struct lnet_ni *ni, struct socket *sock) { - ksock_connreq_t *cr; - int rc; - __u32 peer_ip; - int peer_port; - - rc = libcfs_sock_getaddr(sock, 1, &peer_ip, &peer_port); - LASSERT (rc == 0); /* we succeeded before */ - - LIBCFS_ALLOC(cr, sizeof(*cr)); - if (cr == NULL) { - LCONSOLE_ERROR_MSG(0x12f, "Dropping connection request from " - "%u.%u.%u.%u: memory exhausted\n", - HIPQUAD(peer_ip)); - return -ENOMEM; - } + struct ksock_connreq *cr; + int rc; + struct sockaddr_storage peer; + + rc = lnet_sock_getaddr(sock, true, &peer); + LASSERT(rc == 0); /* we succeeded before */ + + LIBCFS_ALLOC(cr, sizeof(*cr)); + if (cr == NULL) { + LCONSOLE_ERROR_MSG(0x12f, + "Dropping connection request from %pIS: memory exhausted\n", + &peer); + return -ENOMEM; + } - lnet_ni_addref(ni); - cr->ksncr_ni = ni; - cr->ksncr_sock = sock; + lnet_ni_addref(ni); + cr->ksncr_ni = ni; + cr->ksncr_sock = sock; spin_lock_bh(&ksocknal_data.ksnd_connd_lock); - cfs_list_add_tail(&cr->ksncr_list, &ksocknal_data.ksnd_connd_connreqs); + list_add_tail(&cr->ksncr_list, &ksocknal_data.ksnd_connd_connreqs); wake_up(&ksocknal_data.ksnd_connd_waitq); spin_unlock_bh(&ksocknal_data.ksnd_connd_lock); - return 0; + return 0; } -int -ksocknal_connecting (ksock_peer_t *peer, __u32 ipaddr) +static int +ksocknal_connecting(struct ksock_peer_ni *peer_ni, struct sockaddr *sa) { - ksock_route_t *route; - - cfs_list_for_each_entry_typed (route, &peer->ksnp_routes, - ksock_route_t, ksnr_list) { + struct ksock_route *route; - if (route->ksnr_ipaddr == ipaddr) - return route->ksnr_connecting; - } - return 0; + list_for_each_entry(route, &peer_ni->ksnp_routes, ksnr_list) { + if (rpc_cmp_addr((struct sockaddr *)&route->ksnr_addr, sa)) + return route->ksnr_connecting; + } + return 0; } int -ksocknal_create_conn (lnet_ni_t *ni, ksock_route_t *route, - cfs_socket_t *sock, int type) +ksocknal_create_conn(struct lnet_ni *ni, struct ksock_route *route, + struct socket *sock, int type) { - rwlock_t *global_lock = &ksocknal_data.ksnd_global_lock; - CFS_LIST_HEAD (zombies); - lnet_process_id_t peerid; - cfs_list_t *tmp; - __u64 incarnation; - ksock_conn_t *conn; - ksock_conn_t *conn2; - ksock_peer_t *peer = NULL; - ksock_peer_t *peer2; - ksock_sched_t *sched; - ksock_hello_msg_t *hello; - int cpt; - ksock_tx_t *tx; - ksock_tx_t *txtmp; - int rc; - int active; - char *warn = NULL; + rwlock_t *global_lock = &ksocknal_data.ksnd_global_lock; + LIST_HEAD(zombies); + struct lnet_process_id peerid; + struct list_head *tmp; + u64 incarnation; + struct ksock_conn *conn; + struct ksock_conn *conn2; + struct ksock_peer_ni *peer_ni = NULL; + struct ksock_peer_ni *peer2; + struct ksock_sched *sched; + struct ksock_hello_msg *hello; + int cpt; + struct ksock_tx *tx; + struct ksock_tx *txtmp; + int rc; + int rc2; + int active; + char *warn = NULL; active = (route != NULL); @@ -1046,29 +1142,27 @@ ksocknal_create_conn (lnet_ni_t *ni, ksock_route_t *route, goto failed_0; } - memset (conn, 0, sizeof (*conn)); - conn->ksnc_peer = NULL; conn->ksnc_route = NULL; conn->ksnc_sock = sock; /* 2 ref, 1 for conn, another extra ref prevents socket * being closed before establishment of connection */ - atomic_set (&conn->ksnc_sock_refcount, 2); + refcount_set(&conn->ksnc_sock_refcount, 2); conn->ksnc_type = type; ksocknal_lib_save_callback(sock, conn); - atomic_set (&conn->ksnc_conn_refcount, 1); /* 1 ref for me */ + refcount_set(&conn->ksnc_conn_refcount, 1); /* 1 ref for me */ conn->ksnc_rx_ready = 0; conn->ksnc_rx_scheduled = 0; - CFS_INIT_LIST_HEAD (&conn->ksnc_tx_queue); + INIT_LIST_HEAD(&conn->ksnc_tx_queue); conn->ksnc_tx_ready = 0; conn->ksnc_tx_scheduled = 0; conn->ksnc_tx_carrier = NULL; atomic_set (&conn->ksnc_tx_nob, 0); - LIBCFS_ALLOC(hello, offsetof(ksock_hello_msg_t, - kshm_ips[LNET_MAX_INTERFACES])); + LIBCFS_ALLOC(hello, offsetof(struct ksock_hello_msg, + kshm_ips[LNET_INTERFACES_NUM])); if (hello == NULL) { rc = -ENOMEM; goto failed_1; @@ -1079,21 +1173,21 @@ ksocknal_create_conn (lnet_ni_t *ni, ksock_route_t *route, if (rc != 0) goto failed_1; - /* Find out/confirm peer's NID and connection type and get the + /* Find out/confirm peer_ni's NID and connection type and get the * vector of interfaces she's willing to let me connect to. - * Passive connections use the listener timeout since the peer sends + * Passive connections use the listener timeout since the peer_ni sends * eagerly */ if (active) { - peer = route->ksnr_peer; - LASSERT(ni == peer->ksnp_ni); + peer_ni = route->ksnr_peer; + LASSERT(ni == peer_ni->ksnp_ni); /* Active connection sends HELLO eagerly */ hello->kshm_nips = ksocknal_local_ipvec(ni, hello->kshm_ips); - peerid = peer->ksnp_id; + peerid = peer_ni->ksnp_id; write_lock_bh(global_lock); - conn->ksnc_proto = peer->ksnp_proto; + conn->ksnc_proto = peer_ni->ksnp_proto; write_unlock_bh(global_lock); if (conn->ksnc_proto == NULL) { @@ -1113,7 +1207,7 @@ ksocknal_create_conn (lnet_ni_t *ni, ksock_route_t *route, peerid.nid = LNET_NID_ANY; peerid.pid = LNET_PID_ANY; - /* Passive, get protocol from peer */ + /* Passive, get protocol from peer_ni */ conn->ksnc_proto = NULL; } @@ -1125,77 +1219,81 @@ ksocknal_create_conn (lnet_ni_t *ni, ksock_route_t *route, LASSERT (conn->ksnc_proto != NULL); LASSERT (peerid.nid != LNET_NID_ANY); - cpt = lnet_cpt_of_nid(peerid.nid); + cpt = lnet_cpt_of_nid(peerid.nid, ni); - if (active) { - ksocknal_peer_addref(peer); + if (active) { + ksocknal_peer_addref(peer_ni); write_lock_bh(global_lock); - } else { - rc = ksocknal_create_peer(&peer, ni, peerid); - if (rc != 0) - goto failed_1; + } else { + peer_ni = ksocknal_create_peer(ni, peerid); + if (IS_ERR(peer_ni)) { + rc = PTR_ERR(peer_ni); + goto failed_1; + } write_lock_bh(global_lock); - /* called with a ref on ni, so shutdown can't have started */ - LASSERT (((ksock_net_t *) ni->ni_data)->ksnn_shutdown == 0); + /* called with a ref on ni, so shutdown can't have started */ + LASSERT(atomic_read(&((struct ksock_net *)ni->ni_data)->ksnn_npeers) >= 0); - peer2 = ksocknal_find_peer_locked(ni, peerid); - if (peer2 == NULL) { - /* NB this puts an "empty" peer in the peer - * table (which takes my ref) */ - cfs_list_add_tail(&peer->ksnp_list, - ksocknal_nid2peerlist(peerid.nid)); - } else { - ksocknal_peer_decref(peer); - peer = peer2; - } + peer2 = ksocknal_find_peer_locked(ni, peerid); + if (peer2 == NULL) { + /* NB this puts an "empty" peer_ni in the peer_ni + * table (which takes my ref) */ + hash_add(ksocknal_data.ksnd_peers, + &peer_ni->ksnp_list, peerid.nid); + } else { + ksocknal_peer_decref(peer_ni); + peer_ni = peer2; + } /* +1 ref for me */ - ksocknal_peer_addref(peer); - peer->ksnp_accepting++; - - /* Am I already connecting to this guy? Resolve in - * favour of higher NID... */ - if (peerid.nid < ni->ni_nid && - ksocknal_connecting(peer, conn->ksnc_ipaddr)) { - rc = EALREADY; - warn = "connection race resolution"; - goto failed_2; - } + ksocknal_peer_addref(peer_ni); + peer_ni->ksnp_accepting++; + + /* Am I already connecting to this guy? Resolve in + * favour of higher NID... + */ + if (peerid.nid < ni->ni_nid && + ksocknal_connecting(peer_ni, ((struct sockaddr *) + &conn->ksnc_peeraddr))) { + rc = EALREADY; + warn = "connection race resolution"; + goto failed_2; + } } - if (peer->ksnp_closing || + if (peer_ni->ksnp_closing || (active && route->ksnr_deleted)) { - /* peer/route got closed under me */ + /* peer_ni/route got closed under me */ rc = -ESTALE; - warn = "peer/route removed"; + warn = "peer_ni/route removed"; goto failed_2; } - if (peer->ksnp_proto == NULL) { - /* Never connected before. - * NB recv_hello may have returned EPROTO to signal my peer - * wants a different protocol than the one I asked for. - */ - LASSERT (cfs_list_empty(&peer->ksnp_conns)); - - peer->ksnp_proto = conn->ksnc_proto; - peer->ksnp_incarnation = incarnation; - } + if (peer_ni->ksnp_proto == NULL) { + /* Never connected before. + * NB recv_hello may have returned EPROTO to signal my peer_ni + * wants a different protocol than the one I asked for. + */ + LASSERT(list_empty(&peer_ni->ksnp_conns)); - if (peer->ksnp_proto != conn->ksnc_proto || - peer->ksnp_incarnation != incarnation) { - /* Peer rebooted or I've got the wrong protocol version */ - ksocknal_close_peer_conns_locked(peer, 0, 0); + peer_ni->ksnp_proto = conn->ksnc_proto; + peer_ni->ksnp_incarnation = incarnation; + } - peer->ksnp_proto = NULL; - rc = ESTALE; - warn = peer->ksnp_incarnation != incarnation ? - "peer rebooted" : - "wrong proto version"; - goto failed_2; - } + if (peer_ni->ksnp_proto != conn->ksnc_proto || + peer_ni->ksnp_incarnation != incarnation) { + /* peer_ni rebooted or I've got the wrong protocol version */ + ksocknal_close_peer_conns_locked(peer_ni, NULL, 0); + + peer_ni->ksnp_proto = NULL; + rc = ESTALE; + warn = peer_ni->ksnp_incarnation != incarnation ? + "peer_ni rebooted" : + "wrong proto version"; + goto failed_2; + } switch (rc) { default: @@ -1210,18 +1308,23 @@ ksocknal_create_conn (lnet_ni_t *ni, ksock_route_t *route, goto failed_2; } - /* Refuse to duplicate an existing connection, unless this is a - * loopback connection */ - if (conn->ksnc_ipaddr != conn->ksnc_myipaddr) { - cfs_list_for_each(tmp, &peer->ksnp_conns) { - conn2 = cfs_list_entry(tmp, ksock_conn_t, ksnc_list); + /* Refuse to duplicate an existing connection, unless this is a + * loopback connection */ + if (!rpc_cmp_addr((struct sockaddr *)&conn->ksnc_peeraddr, + (struct sockaddr *)&conn->ksnc_myaddr)) { + list_for_each(tmp, &peer_ni->ksnp_conns) { + conn2 = list_entry(tmp, struct ksock_conn, ksnc_list); - if (conn2->ksnc_ipaddr != conn->ksnc_ipaddr || - conn2->ksnc_myipaddr != conn->ksnc_myipaddr || - conn2->ksnc_type != conn->ksnc_type) - continue; + if (!rpc_cmp_addr( + (struct sockaddr *)&conn2->ksnc_peeraddr, + (struct sockaddr *)&conn->ksnc_peeraddr) || + !rpc_cmp_addr( + (struct sockaddr *)&conn2->ksnc_myaddr, + (struct sockaddr *)&conn->ksnc_myaddr) || + conn2->ksnc_type != conn->ksnc_type) + continue; - /* Reply on a passive connection attempt so the peer + /* Reply on a passive connection attempt so the peer_ni * realises we're connected. */ LASSERT (rc == 0); if (!active) @@ -1236,87 +1339,101 @@ ksocknal_create_conn (lnet_ni_t *ni, ksock_route_t *route, * address the route connected to, the connection/route matching * code below probably isn't going to work. */ if (active && - route->ksnr_ipaddr != conn->ksnc_ipaddr) { - CERROR("Route %s %u.%u.%u.%u connected to %u.%u.%u.%u\n", - libcfs_id2str(peer->ksnp_id), - HIPQUAD(route->ksnr_ipaddr), - HIPQUAD(conn->ksnc_ipaddr)); - } - - /* Search for a route corresponding to the new connection and - * create an association. This allows incoming connections created - * by routes in my peer to match my own route entries so I don't - * continually create duplicate routes. */ - cfs_list_for_each (tmp, &peer->ksnp_routes) { - route = cfs_list_entry(tmp, ksock_route_t, ksnr_list); - - if (route->ksnr_ipaddr != conn->ksnc_ipaddr) - continue; + !rpc_cmp_addr((struct sockaddr *)&route->ksnr_addr, + (struct sockaddr *)&conn->ksnc_peeraddr)) { + CERROR("Route %s %pIS connected to %pIS\n", + libcfs_id2str(peer_ni->ksnp_id), + &route->ksnr_addr, + &conn->ksnc_peeraddr); + } + + /* Search for a route corresponding to the new connection and + * create an association. This allows incoming connections created + * by routes in my peer_ni to match my own route entries so I don't + * continually create duplicate routes. */ + list_for_each(tmp, &peer_ni->ksnp_routes) { + route = list_entry(tmp, struct ksock_route, ksnr_list); + + if (!rpc_cmp_addr((struct sockaddr *)&route->ksnr_addr, + (struct sockaddr *)&conn->ksnc_peeraddr)) + continue; - ksocknal_associate_route_conn_locked(route, conn); - break; - } + ksocknal_associate_route_conn_locked(route, conn); + break; + } - conn->ksnc_peer = peer; /* conn takes my ref on peer */ - peer->ksnp_last_alive = cfs_time_current(); - peer->ksnp_send_keepalive = 0; - peer->ksnp_error = 0; + conn->ksnc_peer = peer_ni; /* conn takes my ref on peer_ni */ + peer_ni->ksnp_last_alive = ktime_get_seconds(); + peer_ni->ksnp_send_keepalive = 0; + peer_ni->ksnp_error = 0; sched = ksocknal_choose_scheduler_locked(cpt); + if (!sched) { + CERROR("no schedulers available. node is unhealthy\n"); + goto failed_2; + } + /* + * The cpt might have changed if we ended up selecting a non cpt + * native scheduler. So use the scheduler's cpt instead. + */ + cpt = sched->kss_cpt; sched->kss_nconns++; conn->ksnc_scheduler = sched; - conn->ksnc_tx_last_post = cfs_time_current(); + conn->ksnc_tx_last_post = ktime_get_seconds(); /* Set the deadline for the outgoing HELLO to drain */ - conn->ksnc_tx_bufnob = libcfs_sock_wmem_queued(sock); - conn->ksnc_tx_deadline = cfs_time_shift(*ksocknal_tunables.ksnd_timeout); - smp_mb(); /* order with adding to peer's conn list */ + conn->ksnc_tx_bufnob = sock->sk->sk_wmem_queued; + conn->ksnc_tx_deadline = ktime_get_seconds() + + ksocknal_timeout(); + smp_mb(); /* order with adding to peer_ni's conn list */ - cfs_list_add (&conn->ksnc_list, &peer->ksnp_conns); + list_add(&conn->ksnc_list, &peer_ni->ksnp_conns); ksocknal_conn_addref(conn); ksocknal_new_packet(conn, 0); conn->ksnc_zc_capable = ksocknal_lib_zc_capable(conn); - /* Take packets blocking for this connection. */ - cfs_list_for_each_entry_safe(tx, txtmp, &peer->ksnp_tx_queue, tx_list) { - if (conn->ksnc_proto->pro_match_tx(conn, tx, tx->tx_nonblk) == SOCKNAL_MATCH_NO) - continue; + /* Take packets blocking for this connection. */ + list_for_each_entry_safe(tx, txtmp, &peer_ni->ksnp_tx_queue, tx_list) { + if (conn->ksnc_proto->pro_match_tx(conn, tx, tx->tx_nonblk) == + SOCKNAL_MATCH_NO) + continue; - cfs_list_del (&tx->tx_list); - ksocknal_queue_tx_locked (tx, conn); - } + list_del(&tx->tx_list); + ksocknal_queue_tx_locked(tx, conn); + } write_unlock_bh(global_lock); /* We've now got a new connection. Any errors from here on are just * like "normal" comms errors and we close the connection normally. * NB (a) we still have to send the reply HELLO for passive - * connections, + * connections, * (b) normal I/O on the conn is blocked until I setup and call the * socket callbacks. */ - CDEBUG(D_NET, "New conn %s p %d.x %u.%u.%u.%u -> %u.%u.%u.%u/%d" - " incarnation:"LPD64" sched[%d:%d]\n", + CDEBUG(D_NET, "New conn %s p %d.x %pIS -> %pISp" + " incarnation:%lld sched[%d]\n", libcfs_id2str(peerid), conn->ksnc_proto->pro_version, - HIPQUAD(conn->ksnc_myipaddr), HIPQUAD(conn->ksnc_ipaddr), - conn->ksnc_port, incarnation, cpt, - (int)(sched - &sched->kss_info->ksi_scheds[0])); - - if (active) { - /* additional routes after interface exchange? */ - ksocknal_create_routes(peer, conn->ksnc_port, - hello->kshm_ips, hello->kshm_nips); + &conn->ksnc_myaddr, &conn->ksnc_peeraddr, + incarnation, cpt); + + if (active) { + /* additional routes after interface exchange? */ + ksocknal_create_routes( + peer_ni, + rpc_get_port((struct sockaddr *)&conn->ksnc_peeraddr), + hello->kshm_ips, hello->kshm_nips); } else { - hello->kshm_nips = ksocknal_select_ips(peer, hello->kshm_ips, + hello->kshm_nips = ksocknal_select_ips(peer_ni, hello->kshm_ips, hello->kshm_nips); rc = ksocknal_send_hello(ni, conn, peerid.nid, hello); } - LIBCFS_FREE(hello, offsetof(ksock_hello_msg_t, - kshm_ips[LNET_MAX_INTERFACES])); + LIBCFS_FREE(hello, offsetof(struct ksock_hello_msg, + kshm_ips[LNET_INTERFACES_NUM])); /* setup the socket AFTER I've received hello (it disables * SO_LINGER). I might call back to the acceptor who may want @@ -1332,7 +1449,7 @@ ksocknal_create_conn (lnet_ni_t *ni, ksock_route_t *route, ksocknal_lib_set_callback(sock, conn); if (!active) - peer->ksnp_accepting--; + peer_ni->ksnp_accepting--; write_unlock_bh(global_lock); @@ -1354,14 +1471,13 @@ ksocknal_create_conn (lnet_ni_t *ni, ksock_route_t *route, ksocknal_conn_decref(conn); return rc; - failed_2: - if (!peer->ksnp_closing && - cfs_list_empty (&peer->ksnp_conns) && - cfs_list_empty (&peer->ksnp_routes)) { - cfs_list_add(&zombies, &peer->ksnp_tx_queue); - cfs_list_del_init(&peer->ksnp_tx_queue); - ksocknal_unlink_peer_locked(peer); - } +failed_2: + if (!peer_ni->ksnp_closing && + list_empty(&peer_ni->ksnp_conns) && + list_empty(&peer_ni->ksnp_routes)) { + list_splice_init(&peer_ni->ksnp_tx_queue, &zombies); + ksocknal_unlink_peer_locked(peer_ni); + } write_unlock_bh(global_lock); @@ -1376,7 +1492,7 @@ ksocknal_create_conn (lnet_ni_t *ni, ksock_route_t *route, if (!active) { if (rc > 0) { - /* Request retry by replying with CONN_NONE + /* Request retry by replying with CONN_NONE * ksnc_proto has been set already */ conn->ksnc_type = SOCKLND_CONN_NONE; hello->kshm_nips = 0; @@ -1384,200 +1500,201 @@ ksocknal_create_conn (lnet_ni_t *ni, ksock_route_t *route, } write_lock_bh(global_lock); - peer->ksnp_accepting--; + peer_ni->ksnp_accepting--; write_unlock_bh(global_lock); } - ksocknal_txlist_done(ni, &zombies, 1); - ksocknal_peer_decref(peer); + /* + * If we get here without an error code, just use -EALREADY. + * Depending on how we got here, the error may be positive + * or negative. Normalize the value for ksocknal_txlist_done(). + */ + rc2 = (rc == 0 ? -EALREADY : (rc > 0 ? -rc : rc)); + ksocknal_txlist_done(ni, &zombies, rc2); + ksocknal_peer_decref(peer_ni); - failed_1: - if (hello != NULL) - LIBCFS_FREE(hello, offsetof(ksock_hello_msg_t, - kshm_ips[LNET_MAX_INTERFACES])); +failed_1: + if (hello != NULL) + LIBCFS_FREE(hello, offsetof(struct ksock_hello_msg, + kshm_ips[LNET_INTERFACES_NUM])); - LIBCFS_FREE (conn, sizeof(*conn)); + LIBCFS_FREE(conn, sizeof(*conn)); - failed_0: - libcfs_sock_release(sock); - return rc; +failed_0: + sock_release(sock); + return rc; } void -ksocknal_close_conn_locked (ksock_conn_t *conn, int error) +ksocknal_close_conn_locked(struct ksock_conn *conn, int error) { /* This just does the immmediate housekeeping, and queues the * connection for the reaper to terminate. * Caller holds ksnd_global_lock exclusively in irq context */ - ksock_peer_t *peer = conn->ksnc_peer; - ksock_route_t *route; - ksock_conn_t *conn2; - cfs_list_t *tmp; - - LASSERT (peer->ksnp_error == 0); - LASSERT (!conn->ksnc_closing); - conn->ksnc_closing = 1; - - /* ksnd_deathrow_conns takes over peer's ref */ - cfs_list_del (&conn->ksnc_list); - - route = conn->ksnc_route; - if (route != NULL) { - /* dissociate conn from route... */ - LASSERT (!route->ksnr_deleted); - LASSERT ((route->ksnr_connected & (1 << conn->ksnc_type)) != 0); - - conn2 = NULL; - cfs_list_for_each(tmp, &peer->ksnp_conns) { - conn2 = cfs_list_entry(tmp, ksock_conn_t, ksnc_list); - - if (conn2->ksnc_route == route && - conn2->ksnc_type == conn->ksnc_type) - break; + struct ksock_peer_ni *peer_ni = conn->ksnc_peer; + struct ksock_route *route; + struct ksock_conn *conn2; + struct list_head *tmp; + + LASSERT(peer_ni->ksnp_error == 0); + LASSERT(!conn->ksnc_closing); + conn->ksnc_closing = 1; + + /* ksnd_deathrow_conns takes over peer_ni's ref */ + list_del(&conn->ksnc_list); + + route = conn->ksnc_route; + if (route != NULL) { + /* dissociate conn from route... */ + LASSERT(!route->ksnr_deleted); + LASSERT((route->ksnr_connected & BIT(conn->ksnc_type)) != 0); + + conn2 = NULL; + list_for_each(tmp, &peer_ni->ksnp_conns) { + conn2 = list_entry(tmp, struct ksock_conn, ksnc_list); + + if (conn2->ksnc_route == route && + conn2->ksnc_type == conn->ksnc_type) + break; - conn2 = NULL; - } - if (conn2 == NULL) - route->ksnr_connected &= ~(1 << conn->ksnc_type); + conn2 = NULL; + } + if (conn2 == NULL) + route->ksnr_connected &= ~BIT(conn->ksnc_type); - conn->ksnc_route = NULL; + conn->ksnc_route = NULL; -#if 0 /* irrelevent with only eager routes */ - /* make route least favourite */ - cfs_list_del (&route->ksnr_list); - cfs_list_add_tail (&route->ksnr_list, &peer->ksnp_routes); -#endif - ksocknal_route_decref(route); /* drop conn's ref on route */ - } + ksocknal_route_decref(route); /* drop conn's ref on route */ + } - if (cfs_list_empty (&peer->ksnp_conns)) { - /* No more connections to this peer */ + if (list_empty(&peer_ni->ksnp_conns)) { + /* No more connections to this peer_ni */ - if (!cfs_list_empty(&peer->ksnp_tx_queue)) { - ksock_tx_t *tx; + if (!list_empty(&peer_ni->ksnp_tx_queue)) { + struct ksock_tx *tx; - LASSERT (conn->ksnc_proto == &ksocknal_protocol_v3x); + LASSERT(conn->ksnc_proto == &ksocknal_protocol_v3x); - /* throw them to the last connection..., - * these TXs will be send to /dev/null by scheduler */ - cfs_list_for_each_entry(tx, &peer->ksnp_tx_queue, - tx_list) - ksocknal_tx_prep(conn, tx); + /* throw them to the last connection..., + * these TXs will be send to /dev/null by scheduler */ + list_for_each_entry(tx, &peer_ni->ksnp_tx_queue, + tx_list) + ksocknal_tx_prep(conn, tx); spin_lock_bh(&conn->ksnc_scheduler->kss_lock); - cfs_list_splice_init(&peer->ksnp_tx_queue, - &conn->ksnc_tx_queue); + list_splice_init(&peer_ni->ksnp_tx_queue, + &conn->ksnc_tx_queue); spin_unlock_bh(&conn->ksnc_scheduler->kss_lock); - } + } - peer->ksnp_proto = NULL; /* renegotiate protocol version */ - peer->ksnp_error = error; /* stash last conn close reason */ + /* renegotiate protocol version */ + peer_ni->ksnp_proto = NULL; + /* stash last conn close reason */ + peer_ni->ksnp_error = error; - if (cfs_list_empty (&peer->ksnp_routes)) { - /* I've just closed last conn belonging to a - * peer with no routes to it */ - ksocknal_unlink_peer_locked (peer); - } - } + if (list_empty(&peer_ni->ksnp_routes)) { + /* I've just closed last conn belonging to a + * peer_ni with no routes to it */ + ksocknal_unlink_peer_locked(peer_ni); + } + } spin_lock_bh(&ksocknal_data.ksnd_reaper_lock); - cfs_list_add_tail(&conn->ksnc_list, - &ksocknal_data.ksnd_deathrow_conns); + list_add_tail(&conn->ksnc_list, &ksocknal_data.ksnd_deathrow_conns); wake_up(&ksocknal_data.ksnd_reaper_waitq); spin_unlock_bh(&ksocknal_data.ksnd_reaper_lock); } void -ksocknal_peer_failed (ksock_peer_t *peer) +ksocknal_peer_failed(struct ksock_peer_ni *peer_ni) { - int notify = 0; - cfs_time_t last_alive = 0; + bool notify = false; + time64_t last_alive = 0; - /* There has been a connection failure or comms error; but I'll only - * tell LNET I think the peer is dead if it's to another kernel and - * there are no connections or connection attempts in existance. */ + /* There has been a connection failure or comms error; but I'll only + * tell LNET I think the peer_ni is dead if it's to another kernel and + * there are no connections or connection attempts in existence. */ read_lock(&ksocknal_data.ksnd_global_lock); - if ((peer->ksnp_id.pid & LNET_PID_USERFLAG) == 0 && - cfs_list_empty(&peer->ksnp_conns) && - peer->ksnp_accepting == 0 && - ksocknal_find_connecting_route_locked(peer) == NULL) { - notify = 1; - last_alive = peer->ksnp_last_alive; - } + if ((peer_ni->ksnp_id.pid & LNET_PID_USERFLAG) == 0 && + list_empty(&peer_ni->ksnp_conns) && + peer_ni->ksnp_accepting == 0 && + ksocknal_find_connecting_route_locked(peer_ni) == NULL) { + notify = true; + last_alive = peer_ni->ksnp_last_alive; + } read_unlock(&ksocknal_data.ksnd_global_lock); - if (notify) - lnet_notify (peer->ksnp_ni, peer->ksnp_id.nid, 0, - last_alive); + if (notify) + lnet_notify(peer_ni->ksnp_ni, peer_ni->ksnp_id.nid, + false, false, last_alive); } void -ksocknal_finalize_zcreq(ksock_conn_t *conn) +ksocknal_finalize_zcreq(struct ksock_conn *conn) { - ksock_peer_t *peer = conn->ksnc_peer; - ksock_tx_t *tx; - ksock_tx_t *tmp; - CFS_LIST_HEAD (zlist); + struct ksock_peer_ni *peer_ni = conn->ksnc_peer; + struct ksock_tx *tx; + struct ksock_tx *tmp; + LIST_HEAD(zlist); - /* NB safe to finalize TXs because closing of socket will - * abort all buffered data */ - LASSERT (conn->ksnc_sock == NULL); + /* NB safe to finalize TXs because closing of socket will + * abort all buffered data */ + LASSERT(conn->ksnc_sock == NULL); - spin_lock(&peer->ksnp_lock); + spin_lock(&peer_ni->ksnp_lock); - cfs_list_for_each_entry_safe_typed(tx, tmp, &peer->ksnp_zc_req_list, - ksock_tx_t, tx_zc_list) { - if (tx->tx_conn != conn) - continue; + list_for_each_entry_safe(tx, tmp, &peer_ni->ksnp_zc_req_list, tx_zc_list) { + if (tx->tx_conn != conn) + continue; - LASSERT (tx->tx_msg.ksm_zc_cookies[0] != 0); + LASSERT(tx->tx_msg.ksm_zc_cookies[0] != 0); - tx->tx_msg.ksm_zc_cookies[0] = 0; - tx->tx_zc_aborted = 1; /* mark it as not-acked */ - cfs_list_del(&tx->tx_zc_list); - cfs_list_add(&tx->tx_zc_list, &zlist); - } + tx->tx_msg.ksm_zc_cookies[0] = 0; + tx->tx_zc_aborted = 1; /* mark it as not-acked */ + list_move(&tx->tx_zc_list, &zlist); + } - spin_unlock(&peer->ksnp_lock); + spin_unlock(&peer_ni->ksnp_lock); - while (!cfs_list_empty(&zlist)) { - tx = cfs_list_entry(zlist.next, ksock_tx_t, tx_zc_list); + while (!list_empty(&zlist)) { + tx = list_entry(zlist.next, struct ksock_tx, tx_zc_list); - cfs_list_del(&tx->tx_zc_list); - ksocknal_tx_decref(tx); - } + list_del(&tx->tx_zc_list); + ksocknal_tx_decref(tx); + } } void -ksocknal_terminate_conn (ksock_conn_t *conn) +ksocknal_terminate_conn(struct ksock_conn *conn) { - /* This gets called by the reaper (guaranteed thread context) to - * disengage the socket from its callbacks and close it. - * ksnc_refcount will eventually hit zero, and then the reaper will - * destroy it. */ - ksock_peer_t *peer = conn->ksnc_peer; - ksock_sched_t *sched = conn->ksnc_scheduler; - int failed = 0; + /* This gets called by the reaper (guaranteed thread context) to + * disengage the socket from its callbacks and close it. + * ksnc_refcount will eventually hit zero, and then the reaper will + * destroy it. + */ + struct ksock_peer_ni *peer_ni = conn->ksnc_peer; + struct ksock_sched *sched = conn->ksnc_scheduler; + bool failed = false; - LASSERT(conn->ksnc_closing); + LASSERT(conn->ksnc_closing); - /* wake up the scheduler to "send" all remaining packets to /dev/null */ + /* wake up the scheduler to "send" all remaining packets to /dev/null */ spin_lock_bh(&sched->kss_lock); - /* a closing conn is always ready to tx */ - conn->ksnc_tx_ready = 1; + /* a closing conn is always ready to tx */ + conn->ksnc_tx_ready = 1; - if (!conn->ksnc_tx_scheduled && - !cfs_list_empty(&conn->ksnc_tx_queue)){ - cfs_list_add_tail (&conn->ksnc_tx_list, - &sched->kss_tx_conns); - conn->ksnc_tx_scheduled = 1; - /* extra ref for scheduler */ + if (!conn->ksnc_tx_scheduled && + !list_empty(&conn->ksnc_tx_queue)) { + list_add_tail(&conn->ksnc_tx_list, + &sched->kss_tx_conns); + conn->ksnc_tx_scheduled = 1; + /* extra ref for scheduler */ ksocknal_conn_addref(conn); wake_up (&sched->kss_waitq); @@ -1588,100 +1705,97 @@ ksocknal_terminate_conn (ksock_conn_t *conn) /* serialise with callbacks */ write_lock_bh(&ksocknal_data.ksnd_global_lock); - ksocknal_lib_reset_callback(conn->ksnc_sock, conn); + ksocknal_lib_reset_callback(conn->ksnc_sock, conn); - /* OK, so this conn may not be completely disengaged from its - * scheduler yet, but it _has_ committed to terminate... */ - conn->ksnc_scheduler->kss_nconns--; + /* OK, so this conn may not be completely disengaged from its + * scheduler yet, but it _has_ committed to terminate... + */ + conn->ksnc_scheduler->kss_nconns--; - if (peer->ksnp_error != 0) { - /* peer's last conn closed in error */ - LASSERT (cfs_list_empty (&peer->ksnp_conns)); - failed = 1; - peer->ksnp_error = 0; /* avoid multiple notifications */ - } + if (peer_ni->ksnp_error != 0) { + /* peer_ni's last conn closed in error */ + LASSERT(list_empty(&peer_ni->ksnp_conns)); + failed = true; + peer_ni->ksnp_error = 0; /* avoid multiple notifications */ + } write_unlock_bh(&ksocknal_data.ksnd_global_lock); - if (failed) - ksocknal_peer_failed(peer); + if (failed) + ksocknal_peer_failed(peer_ni); - /* The socket is closed on the final put; either here, or in - * ksocknal_{send,recv}msg(). Since we set up the linger2 option - * when the connection was established, this will close the socket - * immediately, aborting anything buffered in it. Any hung - * zero-copy transmits will therefore complete in finite time. */ - ksocknal_connsock_decref(conn); + /* The socket is closed on the final put; either here, or in + * ksocknal_{send,recv}msg(). Since we set up the linger2 option + * when the connection was established, this will close the socket + * immediately, aborting anything buffered in it. Any hung + * zero-copy transmits will therefore complete in finite time. + */ + ksocknal_connsock_decref(conn); } void -ksocknal_queue_zombie_conn (ksock_conn_t *conn) +ksocknal_queue_zombie_conn(struct ksock_conn *conn) { /* Queue the conn for the reaper to destroy */ - - LASSERT(atomic_read(&conn->ksnc_conn_refcount) == 0); + LASSERT(refcount_read(&conn->ksnc_conn_refcount) == 0); spin_lock_bh(&ksocknal_data.ksnd_reaper_lock); - cfs_list_add_tail(&conn->ksnc_list, &ksocknal_data.ksnd_zombie_conns); + list_add_tail(&conn->ksnc_list, &ksocknal_data.ksnd_zombie_conns); wake_up(&ksocknal_data.ksnd_reaper_waitq); spin_unlock_bh(&ksocknal_data.ksnd_reaper_lock); } void -ksocknal_destroy_conn (ksock_conn_t *conn) +ksocknal_destroy_conn(struct ksock_conn *conn) { - cfs_time_t last_rcv; + time64_t last_rcv; /* Final coup-de-grace of the reaper */ CDEBUG (D_NET, "connection %p\n", conn); - LASSERT (atomic_read (&conn->ksnc_conn_refcount) == 0); - LASSERT (atomic_read (&conn->ksnc_sock_refcount) == 0); + LASSERT(refcount_read(&conn->ksnc_conn_refcount) == 0); + LASSERT(refcount_read(&conn->ksnc_sock_refcount) == 0); LASSERT (conn->ksnc_sock == NULL); LASSERT (conn->ksnc_route == NULL); LASSERT (!conn->ksnc_tx_scheduled); LASSERT (!conn->ksnc_rx_scheduled); - LASSERT (cfs_list_empty(&conn->ksnc_tx_queue)); + LASSERT(list_empty(&conn->ksnc_tx_queue)); /* complete current receive if any */ switch (conn->ksnc_rx_state) { case SOCKNAL_RX_LNET_PAYLOAD: last_rcv = conn->ksnc_rx_deadline - - cfs_time_seconds(*ksocknal_tunables.ksnd_timeout); - CERROR("Completing partial receive from %s[%d]" - ", ip %d.%d.%d.%d:%d, with error, wanted: %d, left: %d, " - "last alive is %ld secs ago\n", + ksocknal_timeout(); + CERROR("Completing partial receive from %s[%d], ip %pISp, with error, wanted: %d, left: %d, last alive is %lld secs ago\n", libcfs_id2str(conn->ksnc_peer->ksnp_id), conn->ksnc_type, - HIPQUAD(conn->ksnc_ipaddr), conn->ksnc_port, + &conn->ksnc_peeraddr, conn->ksnc_rx_nob_wanted, conn->ksnc_rx_nob_left, - cfs_duration_sec(cfs_time_sub(cfs_time_current(), - last_rcv))); - lnet_finalize (conn->ksnc_peer->ksnp_ni, - conn->ksnc_cookie, -EIO); - break; - case SOCKNAL_RX_LNET_HEADER: - if (conn->ksnc_rx_started) - CERROR("Incomplete receive of lnet header from %s" - ", ip %d.%d.%d.%d:%d, with error, protocol: %d.x.\n", - libcfs_id2str(conn->ksnc_peer->ksnp_id), - HIPQUAD(conn->ksnc_ipaddr), conn->ksnc_port, - conn->ksnc_proto->pro_version); - break; + ktime_get_seconds() - last_rcv); + if (conn->ksnc_lnet_msg) + conn->ksnc_lnet_msg->msg_health_status = + LNET_MSG_STATUS_REMOTE_ERROR; + lnet_finalize(conn->ksnc_lnet_msg, -EIO); + break; + case SOCKNAL_RX_LNET_HEADER: + if (conn->ksnc_rx_started) + CERROR("Incomplete receive of lnet header from %s, ip %pISp, with error, protocol: %d.x.\n", + libcfs_id2str(conn->ksnc_peer->ksnp_id), + &conn->ksnc_peeraddr, + conn->ksnc_proto->pro_version); + break; case SOCKNAL_RX_KSM_HEADER: if (conn->ksnc_rx_started) - CERROR("Incomplete receive of ksock message from %s" - ", ip %d.%d.%d.%d:%d, with error, protocol: %d.x.\n", - libcfs_id2str(conn->ksnc_peer->ksnp_id), - HIPQUAD(conn->ksnc_ipaddr), conn->ksnc_port, - conn->ksnc_proto->pro_version); + CERROR("Incomplete receive of ksock message from %s, ip %pISp, with error, protocol: %d.x.\n", + libcfs_id2str(conn->ksnc_peer->ksnp_id), + &conn->ksnc_peeraddr, + conn->ksnc_proto->pro_version); break; case SOCKNAL_RX_SLOP: if (conn->ksnc_rx_started) - CERROR("Incomplete receive of slops from %s" - ", ip %d.%d.%d.%d:%d, with error\n", - libcfs_id2str(conn->ksnc_peer->ksnp_id), - HIPQUAD(conn->ksnc_ipaddr), conn->ksnc_port); + CERROR("Incomplete receive of slops from %s, ip %pISp, with error\n", + libcfs_id2str(conn->ksnc_peer->ksnp_id), + &conn->ksnc_peeraddr); break; default: LBUG (); @@ -1694,176 +1808,117 @@ ksocknal_destroy_conn (ksock_conn_t *conn) } int -ksocknal_close_peer_conns_locked (ksock_peer_t *peer, __u32 ipaddr, int why) +ksocknal_close_peer_conns_locked(struct ksock_peer_ni *peer_ni, + struct sockaddr *addr, int why) { - ksock_conn_t *conn; - cfs_list_t *ctmp; - cfs_list_t *cnxt; - int count = 0; - - cfs_list_for_each_safe (ctmp, cnxt, &peer->ksnp_conns) { - conn = cfs_list_entry (ctmp, ksock_conn_t, ksnc_list); - - if (ipaddr == 0 || - conn->ksnc_ipaddr == ipaddr) { - count++; - ksocknal_close_conn_locked (conn, why); - } - } + struct ksock_conn *conn; + struct ksock_conn *cnxt; + int count = 0; + + list_for_each_entry_safe(conn, cnxt, &peer_ni->ksnp_conns, ksnc_list) { + if (!addr || + rpc_cmp_addr(addr, + (struct sockaddr *)&conn->ksnc_peeraddr)) { + count++; + ksocknal_close_conn_locked(conn, why); + } + } - return (count); + return count; } int -ksocknal_close_conn_and_siblings (ksock_conn_t *conn, int why) +ksocknal_close_conn_and_siblings(struct ksock_conn *conn, int why) { - ksock_peer_t *peer = conn->ksnc_peer; - __u32 ipaddr = conn->ksnc_ipaddr; - int count; + struct ksock_peer_ni *peer_ni = conn->ksnc_peer; + int count; write_lock_bh(&ksocknal_data.ksnd_global_lock); - count = ksocknal_close_peer_conns_locked (peer, ipaddr, why); + count = ksocknal_close_peer_conns_locked( + peer_ni, (struct sockaddr *)&conn->ksnc_peeraddr, why); write_unlock_bh(&ksocknal_data.ksnd_global_lock); - return (count); + return count; } int -ksocknal_close_matching_conns (lnet_process_id_t id, __u32 ipaddr) +ksocknal_close_matching_conns(struct lnet_process_id id, __u32 ipaddr) { - ksock_peer_t *peer; - cfs_list_t *ptmp; - cfs_list_t *pnxt; - int lo; - int hi; - int i; - int count = 0; + struct ksock_peer_ni *peer_ni; + struct hlist_node *pnxt; + int lo; + int hi; + int i; + int count = 0; + struct sockaddr_in sa = {.sin_family = AF_INET}; write_lock_bh(&ksocknal_data.ksnd_global_lock); - if (id.nid != LNET_NID_ANY) - lo = hi = (int)(ksocknal_nid2peerlist(id.nid) - ksocknal_data.ksnd_peers); - else { - lo = 0; - hi = ksocknal_data.ksnd_peer_hash_size - 1; - } - - for (i = lo; i <= hi; i++) { - cfs_list_for_each_safe (ptmp, pnxt, - &ksocknal_data.ksnd_peers[i]) { - - peer = cfs_list_entry (ptmp, ksock_peer_t, ksnp_list); - - if (!((id.nid == LNET_NID_ANY || id.nid == peer->ksnp_id.nid) && - (id.pid == LNET_PID_ANY || id.pid == peer->ksnp_id.pid))) - continue; + if (id.nid != LNET_NID_ANY) { + lo = hash_min(id.nid, HASH_BITS(ksocknal_data.ksnd_peers)); + hi = lo; + } else { + lo = 0; + hi = HASH_SIZE(ksocknal_data.ksnd_peers) - 1; + } - count += ksocknal_close_peer_conns_locked (peer, ipaddr, 0); - } - } + sa.sin_addr.s_addr = htonl(ipaddr); + for (i = lo; i <= hi; i++) { + hlist_for_each_entry_safe(peer_ni, pnxt, + &ksocknal_data.ksnd_peers[i], + ksnp_list) { + + if (!((id.nid == LNET_NID_ANY || + id.nid == peer_ni->ksnp_id.nid) && + (id.pid == LNET_PID_ANY || + id.pid == peer_ni->ksnp_id.pid))) + continue; + + count += ksocknal_close_peer_conns_locked( + peer_ni, + ipaddr ? (struct sockaddr *)&sa : NULL, 0); + } + } write_unlock_bh(&ksocknal_data.ksnd_global_lock); - /* wildcards always succeed */ - if (id.nid == LNET_NID_ANY || id.pid == LNET_PID_ANY || ipaddr == 0) - return (0); + /* wildcards always succeed */ + if (id.nid == LNET_NID_ANY || id.pid == LNET_PID_ANY || ipaddr == 0) + return 0; - return (count == 0 ? -ENOENT : 0); + return (count == 0 ? -ENOENT : 0); } void -ksocknal_notify (lnet_ni_t *ni, lnet_nid_t gw_nid, int alive) +ksocknal_notify_gw_down(lnet_nid_t gw_nid) { - /* The router is telling me she's been notified of a change in - * gateway state.... */ - lnet_process_id_t id = {0}; - - id.nid = gw_nid; - id.pid = LNET_PID_ANY; - - CDEBUG (D_NET, "gw %s %s\n", libcfs_nid2str(gw_nid), - alive ? "up" : "down"); - - if (!alive) { - /* If the gateway crashed, close all open connections... */ - ksocknal_close_matching_conns (id, 0); - return; - } - - /* ...otherwise do nothing. We can only establish new connections - * if we have autroutes, and these connect on demand. */ -} - -void -ksocknal_query (lnet_ni_t *ni, lnet_nid_t nid, cfs_time_t *when) -{ - int connect = 1; - cfs_time_t last_alive = 0; - cfs_time_t now = cfs_time_current(); - ksock_peer_t *peer = NULL; - rwlock_t *glock = &ksocknal_data.ksnd_global_lock; - lnet_process_id_t id = {.nid = nid, .pid = LUSTRE_SRV_LNET_PID}; - - read_lock(glock); - - peer = ksocknal_find_peer_locked(ni, id); - if (peer != NULL) { - cfs_list_t *tmp; - ksock_conn_t *conn; - int bufnob; - - cfs_list_for_each (tmp, &peer->ksnp_conns) { - conn = cfs_list_entry(tmp, ksock_conn_t, ksnc_list); - bufnob = libcfs_sock_wmem_queued(conn->ksnc_sock); - - if (bufnob < conn->ksnc_tx_bufnob) { - /* something got ACKed */ - conn->ksnc_tx_deadline = - cfs_time_shift(*ksocknal_tunables.ksnd_timeout); - peer->ksnp_last_alive = now; - conn->ksnc_tx_bufnob = bufnob; - } - } - - last_alive = peer->ksnp_last_alive; - if (ksocknal_find_connectable_route_locked(peer) == NULL) - connect = 0; - } - - read_unlock(glock); - - if (last_alive != 0) - *when = last_alive; - - CDEBUG(D_NET, "Peer %s %p, alive %ld secs ago, connect %d\n", - libcfs_nid2str(nid), peer, - last_alive ? cfs_duration_sec(now - last_alive) : -1, - connect); - - if (!connect) - return; - - ksocknal_add_peer(ni, id, LNET_NIDADDR(nid), lnet_acceptor_port()); + /* The router is telling me she's been notified of a change in + * gateway state.... + */ + struct lnet_process_id id = { + .nid = gw_nid, + .pid = LNET_PID_ANY, + }; - write_lock_bh(glock); + CDEBUG(D_NET, "gw %s down\n", libcfs_nid2str(gw_nid)); - peer = ksocknal_find_peer_locked(ni, id); - if (peer != NULL) - ksocknal_launch_all_connections_locked(peer); + /* If the gateway crashed, close all open connections... */ + ksocknal_close_matching_conns(id, 0); + return; - write_unlock_bh(glock); - return; + /* We can only establish new connections + * if we have autroutes, and these connect on demand. */ } -void -ksocknal_push_peer (ksock_peer_t *peer) +static void +ksocknal_push_peer(struct ksock_peer_ni *peer_ni) { - int index; - int i; - cfs_list_t *tmp; - ksock_conn_t *conn; + int index; + int i; + struct list_head *tmp; + struct ksock_conn *conn; for (index = 0; ; index++) { read_lock(&ksocknal_data.ksnd_global_lock); @@ -1871,10 +1926,10 @@ ksocknal_push_peer (ksock_peer_t *peer) i = 0; conn = NULL; - cfs_list_for_each (tmp, &peer->ksnp_conns) { + list_for_each(tmp, &peer_ni->ksnp_conns) { if (i++ == index) { - conn = cfs_list_entry (tmp, ksock_conn_t, - ksnc_list); + conn = list_entry(tmp, struct ksock_conn, + ksnc_list); ksocknal_conn_addref(conn); break; } @@ -1890,217 +1945,224 @@ ksocknal_push_peer (ksock_peer_t *peer) } } -int -ksocknal_push (lnet_ni_t *ni, lnet_process_id_t id) +static int +ksocknal_push(struct lnet_ni *ni, struct lnet_process_id id) { - ksock_peer_t *peer; - cfs_list_t *tmp; - int index; - int i; - int j; - int rc = -ENOENT; - - for (i = 0; i < ksocknal_data.ksnd_peer_hash_size; i++) { - for (j = 0; ; j++) { - read_lock(&ksocknal_data.ksnd_global_lock); - - index = 0; - peer = NULL; + int lo; + int hi; + int bkt; + int rc = -ENOENT; + + if (id.nid != LNET_NID_ANY) { + lo = hash_min(id.nid, HASH_BITS(ksocknal_data.ksnd_peers)); + hi = lo; + } else { + lo = 0; + hi = HASH_SIZE(ksocknal_data.ksnd_peers) - 1; + } - cfs_list_for_each (tmp, &ksocknal_data.ksnd_peers[i]) { - peer = cfs_list_entry(tmp, ksock_peer_t, - ksnp_list); + for (bkt = lo; bkt <= hi; bkt++) { + int peer_off; /* searching offset in peer_ni hash table */ - if (!((id.nid == LNET_NID_ANY || - id.nid == peer->ksnp_id.nid) && - (id.pid == LNET_PID_ANY || - id.pid == peer->ksnp_id.pid))) { - peer = NULL; - continue; - } + for (peer_off = 0; ; peer_off++) { + struct ksock_peer_ni *peer_ni; + int i = 0; - if (index++ == j) { - ksocknal_peer_addref(peer); - break; - } - } + read_lock(&ksocknal_data.ksnd_global_lock); + hlist_for_each_entry(peer_ni, + &ksocknal_data.ksnd_peers[bkt], + ksnp_list) { + if (!((id.nid == LNET_NID_ANY || + id.nid == peer_ni->ksnp_id.nid) && + (id.pid == LNET_PID_ANY || + id.pid == peer_ni->ksnp_id.pid))) + continue; + if (i++ == peer_off) { + ksocknal_peer_addref(peer_ni); + break; + } + } read_unlock(&ksocknal_data.ksnd_global_lock); - if (peer != NULL) { - rc = 0; - ksocknal_push_peer (peer); - ksocknal_peer_decref(peer); - } - } - - } + if (i <= peer_off) /* no match */ + break; - return (rc); + rc = 0; + ksocknal_push_peer(peer_ni); + ksocknal_peer_decref(peer_ni); + } + } + return rc; } -int -ksocknal_add_interface(lnet_ni_t *ni, __u32 ipaddress, __u32 netmask) +static int +ksocknal_add_interface(struct lnet_ni *ni, __u32 ipaddress, __u32 netmask) { - ksock_net_t *net = ni->ni_data; - ksock_interface_t *iface; - int rc; - int i; - int j; - cfs_list_t *ptmp; - ksock_peer_t *peer; - cfs_list_t *rtmp; - ksock_route_t *route; - - if (ipaddress == 0 || - netmask == 0) - return (-EINVAL); + struct ksock_net *net = ni->ni_data; + struct ksock_interface *iface; + struct sockaddr_in sa = { .sin_family = AF_INET }; + int rc; + int i; + int j; + struct ksock_peer_ni *peer_ni; + struct list_head *rtmp; + struct ksock_route *route; + + if (ipaddress == 0 || + netmask == 0) + return -EINVAL; write_lock_bh(&ksocknal_data.ksnd_global_lock); - iface = ksocknal_ip2iface(ni, ipaddress); - if (iface != NULL) { - /* silently ignore dups */ - rc = 0; - } else if (net->ksnn_ninterfaces == LNET_MAX_INTERFACES) { - rc = -ENOSPC; - } else { - iface = &net->ksnn_interfaces[net->ksnn_ninterfaces++]; - - iface->ksni_ipaddr = ipaddress; - iface->ksni_netmask = netmask; - iface->ksni_nroutes = 0; - iface->ksni_npeers = 0; - - for (i = 0; i < ksocknal_data.ksnd_peer_hash_size; i++) { - cfs_list_for_each(ptmp, &ksocknal_data.ksnd_peers[i]) { - peer = cfs_list_entry(ptmp, ksock_peer_t, - ksnp_list); - - for (j = 0; j < peer->ksnp_n_passive_ips; j++) - if (peer->ksnp_passive_ips[j] == ipaddress) - iface->ksni_npeers++; - - cfs_list_for_each(rtmp, &peer->ksnp_routes) { - route = cfs_list_entry(rtmp, - ksock_route_t, - ksnr_list); - - if (route->ksnr_myipaddr == ipaddress) - iface->ksni_nroutes++; - } - } - } + sa.sin_addr.s_addr = htonl(ipaddress); + iface = ksocknal_ip2iface(ni, (struct sockaddr *)&sa); + if (iface != NULL) { + /* silently ignore dups */ + rc = 0; + } else if (net->ksnn_ninterfaces == LNET_INTERFACES_NUM) { + rc = -ENOSPC; + } else { + iface = &net->ksnn_interfaces[net->ksnn_ninterfaces++]; + + iface->ksni_index = ksocknal_ip2index((struct sockaddr *)&sa, + ni); + rpc_copy_addr((struct sockaddr *)&iface->ksni_addr, + (struct sockaddr *)&sa); + iface->ksni_netmask = netmask; + iface->ksni_nroutes = 0; + iface->ksni_npeers = 0; + + hash_for_each(ksocknal_data.ksnd_peers, i, peer_ni, ksnp_list) { + for (j = 0; j < peer_ni->ksnp_n_passive_ips; j++) + if (peer_ni->ksnp_passive_ips[j] == ipaddress) + iface->ksni_npeers++; + + list_for_each(rtmp, &peer_ni->ksnp_routes) { + route = list_entry(rtmp, + struct ksock_route, + ksnr_list); + + if (route->ksnr_myiface == + iface->ksni_index) + iface->ksni_nroutes++; + } + } - rc = 0; - /* NB only new connections will pay attention to the new interface! */ - } + rc = 0; + /* NB only new connections will pay attention to the new + * interface! + */ + } write_unlock_bh(&ksocknal_data.ksnd_global_lock); - return (rc); + return rc; } -void -ksocknal_peer_del_interface_locked(ksock_peer_t *peer, __u32 ipaddr) +static void +ksocknal_peer_del_interface_locked(struct ksock_peer_ni *peer_ni, + __u32 ipaddr, int index) { - cfs_list_t *tmp; - cfs_list_t *nxt; - ksock_route_t *route; - ksock_conn_t *conn; - int i; - int j; - - for (i = 0; i < peer->ksnp_n_passive_ips; i++) - if (peer->ksnp_passive_ips[i] == ipaddr) { - for (j = i+1; j < peer->ksnp_n_passive_ips; j++) - peer->ksnp_passive_ips[j-1] = - peer->ksnp_passive_ips[j]; - peer->ksnp_n_passive_ips--; + struct ksock_route *route; + struct ksock_route *rnxt; + struct ksock_conn *conn; + struct ksock_conn *cnxt; + int i; + int j; + + for (i = 0; i < peer_ni->ksnp_n_passive_ips; i++) + if (peer_ni->ksnp_passive_ips[i] == ipaddr) { + for (j = i+1; j < peer_ni->ksnp_n_passive_ips; j++) + peer_ni->ksnp_passive_ips[j-1] = + peer_ni->ksnp_passive_ips[j]; + peer_ni->ksnp_n_passive_ips--; break; } - cfs_list_for_each_safe(tmp, nxt, &peer->ksnp_routes) { - route = cfs_list_entry (tmp, ksock_route_t, ksnr_list); - - if (route->ksnr_myipaddr != ipaddr) - continue; - - if (route->ksnr_share_count != 0) { - /* Manually created; keep, but unbind */ - route->ksnr_myipaddr = 0; - } else { - ksocknal_del_route_locked(route); - } - } + list_for_each_entry_safe(route, rnxt, &peer_ni->ksnp_routes, + ksnr_list) { + if (route->ksnr_myiface != index) + continue; - cfs_list_for_each_safe(tmp, nxt, &peer->ksnp_conns) { - conn = cfs_list_entry(tmp, ksock_conn_t, ksnc_list); + if (route->ksnr_share_count != 0) { + /* Manually created; keep, but unbind */ + route->ksnr_myiface = -1; + } else { + ksocknal_del_route_locked(route); + } + } - if (conn->ksnc_myipaddr == ipaddr) - ksocknal_close_conn_locked (conn, 0); - } + list_for_each_entry_safe(conn, cnxt, &peer_ni->ksnp_conns, ksnc_list) + if (conn->ksnc_route->ksnr_myiface == index) + ksocknal_close_conn_locked (conn, 0); } -int -ksocknal_del_interface(lnet_ni_t *ni, __u32 ipaddress) +static int +ksocknal_del_interface(struct lnet_ni *ni, __u32 ipaddress) { - ksock_net_t *net = ni->ni_data; - int rc = -ENOENT; - cfs_list_t *tmp; - cfs_list_t *nxt; - ksock_peer_t *peer; - __u32 this_ip; - int i; - int j; + struct ksock_net *net = ni->ni_data; + int rc = -ENOENT; + struct hlist_node *nxt; + struct ksock_peer_ni *peer_ni; + u32 this_ip; + struct sockaddr_in sa = {.sin_family = AF_INET }; + int index; + int i; + int j; + + sa.sin_addr.s_addr = htonl(ipaddress); + index = ksocknal_ip2index((struct sockaddr *)&sa, ni); write_lock_bh(&ksocknal_data.ksnd_global_lock); - for (i = 0; i < net->ksnn_ninterfaces; i++) { - this_ip = net->ksnn_interfaces[i].ksni_ipaddr; + for (i = 0; i < net->ksnn_ninterfaces; i++) { + struct sockaddr_in *sa = + (void *)&net->ksnn_interfaces[i].ksni_addr; - if (!(ipaddress == 0 || - ipaddress == this_ip)) - continue; + if (sa->sin_family != AF_INET) + continue; + this_ip = ntohl(sa->sin_addr.s_addr); - rc = 0; + if (!(ipaddress == 0 || + ipaddress == this_ip)) + continue; - for (j = i+1; j < net->ksnn_ninterfaces; j++) - net->ksnn_interfaces[j-1] = - net->ksnn_interfaces[j]; + rc = 0; - net->ksnn_ninterfaces--; + for (j = i+1; j < net->ksnn_ninterfaces; j++) + net->ksnn_interfaces[j-1] = + net->ksnn_interfaces[j]; - for (j = 0; j < ksocknal_data.ksnd_peer_hash_size; j++) { - cfs_list_for_each_safe(tmp, nxt, - &ksocknal_data.ksnd_peers[j]) { - peer = cfs_list_entry(tmp, ksock_peer_t, - ksnp_list); + net->ksnn_ninterfaces--; - if (peer->ksnp_ni != ni) - continue; + hash_for_each_safe(ksocknal_data.ksnd_peers, j, + nxt, peer_ni, ksnp_list) { + if (peer_ni->ksnp_ni != ni) + continue; - ksocknal_peer_del_interface_locked(peer, this_ip); - } - } - } + ksocknal_peer_del_interface_locked(peer_ni, + this_ip, index); + } + } write_unlock_bh(&ksocknal_data.ksnd_global_lock); - return (rc); + return rc; } int -ksocknal_ctl(lnet_ni_t *ni, unsigned int cmd, void *arg) +ksocknal_ctl(struct lnet_ni *ni, unsigned int cmd, void *arg) { - lnet_process_id_t id = {0}; + struct lnet_process_id id = {0}; struct libcfs_ioctl_data *data = arg; int rc; switch(cmd) { case IOC_LIBCFS_GET_INTERFACE: { - ksock_net_t *net = ni->ni_data; - ksock_interface_t *iface; + struct ksock_net *net = ni->ni_data; + struct ksock_interface *iface; + struct sockaddr_in *sa; read_lock(&ksocknal_data.ksnd_global_lock); @@ -2110,11 +2172,15 @@ ksocknal_ctl(lnet_ni_t *ni, unsigned int cmd, void *arg) rc = 0; iface = &net->ksnn_interfaces[data->ioc_count]; - data->ioc_u32[0] = iface->ksni_ipaddr; - data->ioc_u32[1] = iface->ksni_netmask; - data->ioc_u32[2] = iface->ksni_npeers; - data->ioc_u32[3] = iface->ksni_nroutes; - } + sa = (void *)&iface->ksni_addr; + if (sa->sin_family == AF_INET) + data->ioc_u32[0] = ntohl(sa->sin_addr.s_addr); + else + data->ioc_u32[0] = 0xFFFFFFFF; + data->ioc_u32[1] = iface->ksni_netmask; + data->ioc_u32[2] = iface->ksni_npeers; + data->ioc_u32[3] = iface->ksni_nroutes; + } read_unlock(&ksocknal_data.ksnd_global_lock); return rc; @@ -2152,13 +2218,15 @@ ksocknal_ctl(lnet_ni_t *ni, unsigned int cmd, void *arg) return 0; } - case IOC_LIBCFS_ADD_PEER: - id.nid = data->ioc_nid; - id.pid = LUSTRE_SRV_LNET_PID; - return ksocknal_add_peer (ni, id, - data->ioc_u32[0], /* IP */ - data->ioc_u32[1]); /* port */ + case IOC_LIBCFS_ADD_PEER: { + struct sockaddr_in sa = {.sin_family = AF_INET}; + id.nid = data->ioc_nid; + id.pid = LNET_PID_LUSTRE; + sa.sin_addr.s_addr = htonl(data->ioc_u32[0]); + sa.sin_port = htons(data->ioc_u32[1]); + return ksocknal_add_peer(ni, id, (struct sockaddr *)&sa); + } case IOC_LIBCFS_DEL_PEER: id.nid = data->ioc_nid; id.pid = LNET_PID_ANY; @@ -2169,7 +2237,9 @@ ksocknal_ctl(lnet_ni_t *ni, unsigned int cmd, void *arg) int txmem; int rxmem; int nagle; - ksock_conn_t *conn = ksocknal_get_conn_by_idx (ni, data->ioc_count); + struct ksock_conn *conn = ksocknal_get_conn_by_idx(ni, data->ioc_count); + struct sockaddr_in *psa = (void *)&conn->ksnc_peeraddr; + struct sockaddr_in *mysa = (void *)&conn->ksnc_myaddr; if (conn == NULL) return -ENOENT; @@ -2179,11 +2249,18 @@ ksocknal_ctl(lnet_ni_t *ni, unsigned int cmd, void *arg) data->ioc_count = txmem; data->ioc_nid = conn->ksnc_peer->ksnp_id.nid; data->ioc_flags = nagle; - data->ioc_u32[0] = conn->ksnc_ipaddr; - data->ioc_u32[1] = conn->ksnc_port; - data->ioc_u32[2] = conn->ksnc_myipaddr; - data->ioc_u32[3] = conn->ksnc_type; - data->ioc_u32[4] = conn->ksnc_scheduler->kss_info->ksi_cpt; + if (psa->sin_family == AF_INET) + data->ioc_u32[0] = ntohl(psa->sin_addr.s_addr); + else + data->ioc_u32[0] = 0xFFFFFFFF; + data->ioc_u32[1] = rpc_get_port((struct sockaddr *) + &conn->ksnc_peeraddr); + if (mysa->sin_family == AF_INET) + data->ioc_u32[2] = ntohl(mysa->sin_addr.s_addr); + else + data->ioc_u32[2] = 0xFFFFFFFF; + data->ioc_u32[3] = conn->ksnc_type; + data->ioc_u32[4] = conn->ksnc_scheduler->kss_cpt; data->ioc_u32[5] = rxmem; data->ioc_u32[6] = conn->ksnc_peer->ksnp_id.pid; ksocknal_conn_decref(conn); @@ -2217,42 +2294,26 @@ ksocknal_ctl(lnet_ni_t *ni, unsigned int cmd, void *arg) /* not reached */ } -void +static void ksocknal_free_buffers (void) { LASSERT (atomic_read(&ksocknal_data.ksnd_nactive_txs) == 0); - if (ksocknal_data.ksnd_sched_info != NULL) { - struct ksock_sched_info *info; - int i; - - cfs_percpt_for_each(info, i, ksocknal_data.ksnd_sched_info) { - if (info->ksi_scheds != NULL) { - LIBCFS_FREE(info->ksi_scheds, - info->ksi_nthreads_max * - sizeof(info->ksi_scheds[0])); - } - } - cfs_percpt_free(ksocknal_data.ksnd_sched_info); - } - - LIBCFS_FREE (ksocknal_data.ksnd_peers, - sizeof (cfs_list_t) * - ksocknal_data.ksnd_peer_hash_size); + if (ksocknal_data.ksnd_schedulers != NULL) + cfs_percpt_free(ksocknal_data.ksnd_schedulers); spin_lock(&ksocknal_data.ksnd_tx_lock); - if (!cfs_list_empty(&ksocknal_data.ksnd_idle_noop_txs)) { - cfs_list_t zlist; - ksock_tx_t *tx; + if (!list_empty(&ksocknal_data.ksnd_idle_noop_txs)) { + LIST_HEAD(zlist); + struct ksock_tx *tx; - cfs_list_add(&zlist, &ksocknal_data.ksnd_idle_noop_txs); - cfs_list_del_init(&ksocknal_data.ksnd_idle_noop_txs); + list_splice_init(&ksocknal_data.ksnd_idle_noop_txs, &zlist); spin_unlock(&ksocknal_data.ksnd_tx_lock); - while (!cfs_list_empty(&zlist)) { - tx = cfs_list_entry(zlist.next, ksock_tx_t, tx_list); - cfs_list_del(&tx->tx_list); + while (!list_empty(&zlist)) { + tx = list_entry(zlist.next, struct ksock_tx, tx_list); + list_del(&tx->tx_list); LIBCFS_FREE(tx, tx->tx_desc_size); } } else { @@ -2260,52 +2321,41 @@ ksocknal_free_buffers (void) } } -void +static void ksocknal_base_shutdown(void) { - struct ksock_sched_info *info; - ksock_sched_t *sched; - int i; - int j; + struct ksock_sched *sched; + struct ksock_peer_ni *peer_ni; + int i; - CDEBUG(D_MALLOC, "before NAL cleanup: kmem %d\n", - atomic_read (&libcfs_kmemory)); + CDEBUG(D_MALLOC, "before NAL cleanup: kmem %lld\n", + libcfs_kmem_read()); LASSERT (ksocknal_data.ksnd_nnets == 0); - switch (ksocknal_data.ksnd_init) { - default: - LASSERT (0); - - case SOCKNAL_INIT_ALL: - case SOCKNAL_INIT_DATA: - LASSERT (ksocknal_data.ksnd_peers != NULL); - for (i = 0; i < ksocknal_data.ksnd_peer_hash_size; i++) { - LASSERT (cfs_list_empty (&ksocknal_data.ksnd_peers[i])); - } - - LASSERT(cfs_list_empty(&ksocknal_data.ksnd_nets)); - LASSERT (cfs_list_empty (&ksocknal_data.ksnd_enomem_conns)); - LASSERT (cfs_list_empty (&ksocknal_data.ksnd_zombie_conns)); - LASSERT (cfs_list_empty (&ksocknal_data.ksnd_connd_connreqs)); - LASSERT (cfs_list_empty (&ksocknal_data.ksnd_connd_routes)); - - if (ksocknal_data.ksnd_sched_info != NULL) { - cfs_percpt_for_each(info, i, - ksocknal_data.ksnd_sched_info) { - if (info->ksi_scheds == NULL) - continue; - - for (j = 0; j < info->ksi_nthreads_max; j++) { - - sched = &info->ksi_scheds[j]; - LASSERT(cfs_list_empty(&sched->\ - kss_tx_conns)); - LASSERT(cfs_list_empty(&sched->\ - kss_rx_conns)); - LASSERT(cfs_list_empty(&sched-> \ - kss_zombie_noop_txs)); - LASSERT(sched->kss_nconns == 0); - } + switch (ksocknal_data.ksnd_init) { + default: + LASSERT(0); + /* fallthrough */ + + case SOCKNAL_INIT_ALL: + case SOCKNAL_INIT_DATA: + hash_for_each(ksocknal_data.ksnd_peers, i, peer_ni, ksnp_list) + LASSERT(0); + + LASSERT(list_empty(&ksocknal_data.ksnd_nets)); + LASSERT(list_empty(&ksocknal_data.ksnd_enomem_conns)); + LASSERT(list_empty(&ksocknal_data.ksnd_zombie_conns)); + LASSERT(list_empty(&ksocknal_data.ksnd_connd_connreqs)); + LASSERT(list_empty(&ksocknal_data.ksnd_connd_routes)); + + if (ksocknal_data.ksnd_schedulers != NULL) { + cfs_percpt_for_each(sched, i, + ksocknal_data.ksnd_schedulers) { + + LASSERT(list_empty(&sched->kss_tx_conns)); + LASSERT(list_empty(&sched->kss_rx_conns)); + LASSERT(list_empty(&sched->kss_zombie_noop_txs)); + LASSERT(sched->kss_nconns == 0); } } @@ -2314,144 +2364,104 @@ ksocknal_base_shutdown(void) wake_up_all(&ksocknal_data.ksnd_connd_waitq); wake_up_all(&ksocknal_data.ksnd_reaper_waitq); - if (ksocknal_data.ksnd_sched_info != NULL) { - cfs_percpt_for_each(info, i, - ksocknal_data.ksnd_sched_info) { - if (info->ksi_scheds == NULL) - continue; - - for (j = 0; j < info->ksi_nthreads_max; j++) { - sched = &info->ksi_scheds[j]; + if (ksocknal_data.ksnd_schedulers != NULL) { + cfs_percpt_for_each(sched, i, + ksocknal_data.ksnd_schedulers) wake_up_all(&sched->kss_waitq); - } - } } - i = 4; - read_lock(&ksocknal_data.ksnd_global_lock); - while (ksocknal_data.ksnd_nthreads != 0) { - i++; - CDEBUG(((i & (-i)) == i) ? D_WARNING : D_NET, /* power of 2? */ - "waiting for %d threads to terminate\n", - ksocknal_data.ksnd_nthreads); - read_unlock(&ksocknal_data.ksnd_global_lock); - cfs_pause(cfs_time_seconds(1)); - read_lock(&ksocknal_data.ksnd_global_lock); - } - read_unlock(&ksocknal_data.ksnd_global_lock); + wait_var_event_warning(&ksocknal_data.ksnd_nthreads, + atomic_read(&ksocknal_data.ksnd_nthreads) == 0, + "waiting for %d threads to terminate\n", + atomic_read(&ksocknal_data.ksnd_nthreads)); - ksocknal_free_buffers(); + ksocknal_free_buffers(); - ksocknal_data.ksnd_init = SOCKNAL_INIT_NOTHING; - break; - } + ksocknal_data.ksnd_init = SOCKNAL_INIT_NOTHING; + break; + } - CDEBUG(D_MALLOC, "after NAL cleanup: kmem %d\n", - atomic_read (&libcfs_kmemory)); + CDEBUG(D_MALLOC, "after NAL cleanup: kmem %lld\n", + libcfs_kmem_read()); module_put(THIS_MODULE); } -__u64 ksocknal_new_incarnation (void) -{ - struct timeval tv; - - /* The incarnation number is the time this module loaded and it - * identifies this particular instance of the socknal. Hopefully - * we won't be able to reboot more frequently than 1MHz for the - * forseeable future :) */ - - do_gettimeofday(&tv); - - return (((__u64)tv.tv_sec) * 1000000) + tv.tv_usec; -} - -int +static int ksocknal_base_startup(void) { - struct ksock_sched_info *info; - int rc; - int i; + struct ksock_sched *sched; + int rc; + int i; - LASSERT (ksocknal_data.ksnd_init == SOCKNAL_INIT_NOTHING); - LASSERT (ksocknal_data.ksnd_nnets == 0); + LASSERT(ksocknal_data.ksnd_init == SOCKNAL_INIT_NOTHING); + LASSERT(ksocknal_data.ksnd_nnets == 0); - memset (&ksocknal_data, 0, sizeof (ksocknal_data)); /* zero pointers */ + memset(&ksocknal_data, 0, sizeof(ksocknal_data)); /* zero pointers */ - ksocknal_data.ksnd_peer_hash_size = SOCKNAL_PEER_HASH_SIZE; - LIBCFS_ALLOC (ksocknal_data.ksnd_peers, - sizeof (cfs_list_t) * - ksocknal_data.ksnd_peer_hash_size); - if (ksocknal_data.ksnd_peers == NULL) - return -ENOMEM; - - for (i = 0; i < ksocknal_data.ksnd_peer_hash_size; i++) - CFS_INIT_LIST_HEAD(&ksocknal_data.ksnd_peers[i]); + hash_init(ksocknal_data.ksnd_peers); rwlock_init(&ksocknal_data.ksnd_global_lock); - CFS_INIT_LIST_HEAD(&ksocknal_data.ksnd_nets); + INIT_LIST_HEAD(&ksocknal_data.ksnd_nets); spin_lock_init(&ksocknal_data.ksnd_reaper_lock); - CFS_INIT_LIST_HEAD (&ksocknal_data.ksnd_enomem_conns); - CFS_INIT_LIST_HEAD (&ksocknal_data.ksnd_zombie_conns); - CFS_INIT_LIST_HEAD (&ksocknal_data.ksnd_deathrow_conns); + INIT_LIST_HEAD(&ksocknal_data.ksnd_enomem_conns); + INIT_LIST_HEAD(&ksocknal_data.ksnd_zombie_conns); + INIT_LIST_HEAD(&ksocknal_data.ksnd_deathrow_conns); init_waitqueue_head(&ksocknal_data.ksnd_reaper_waitq); spin_lock_init(&ksocknal_data.ksnd_connd_lock); - CFS_INIT_LIST_HEAD (&ksocknal_data.ksnd_connd_connreqs); - CFS_INIT_LIST_HEAD (&ksocknal_data.ksnd_connd_routes); + INIT_LIST_HEAD(&ksocknal_data.ksnd_connd_connreqs); + INIT_LIST_HEAD(&ksocknal_data.ksnd_connd_routes); init_waitqueue_head(&ksocknal_data.ksnd_connd_waitq); spin_lock_init(&ksocknal_data.ksnd_tx_lock); - CFS_INIT_LIST_HEAD (&ksocknal_data.ksnd_idle_noop_txs); + INIT_LIST_HEAD(&ksocknal_data.ksnd_idle_noop_txs); /* NB memset above zeros whole of ksocknal_data */ /* flag lists/ptrs/locks initialised */ ksocknal_data.ksnd_init = SOCKNAL_INIT_DATA; - try_module_get(THIS_MODULE); + if (!try_module_get(THIS_MODULE)) + goto failed; - ksocknal_data.ksnd_sched_info = cfs_percpt_alloc(lnet_cpt_table(), - sizeof(*info)); - if (ksocknal_data.ksnd_sched_info == NULL) + /* Create a scheduler block per available CPT */ + ksocknal_data.ksnd_schedulers = cfs_percpt_alloc(lnet_cpt_table(), + sizeof(*sched)); + if (ksocknal_data.ksnd_schedulers == NULL) goto failed; - cfs_percpt_for_each(info, i, ksocknal_data.ksnd_sched_info) { - ksock_sched_t *sched; - int nthrs; + cfs_percpt_for_each(sched, i, ksocknal_data.ksnd_schedulers) { + int nthrs; + /* + * make sure not to allocate more threads than there are + * cores/CPUs in teh CPT + */ nthrs = cfs_cpt_weight(lnet_cpt_table(), i); if (*ksocknal_tunables.ksnd_nscheds > 0) { nthrs = min(nthrs, *ksocknal_tunables.ksnd_nscheds); } else { - /* max to half of CPUs, assume another half should be - * reserved for upper layer modules */ + /* + * max to half of CPUs, assume another half should be + * reserved for upper layer modules + */ nthrs = min(max(SOCKNAL_NSCHEDS, nthrs >> 1), nthrs); } - info->ksi_nthreads_max = nthrs; - info->ksi_cpt = i; + sched->kss_nthreads_max = nthrs; + sched->kss_cpt = i; - LIBCFS_CPT_ALLOC(info->ksi_scheds, lnet_cpt_table(), i, - info->ksi_nthreads_max * sizeof(*sched)); - if (info->ksi_scheds == NULL) - goto failed; - - for (; nthrs > 0; nthrs--) { - sched = &info->ksi_scheds[nthrs - 1]; - - sched->kss_info = info; - spin_lock_init(&sched->kss_lock); - CFS_INIT_LIST_HEAD(&sched->kss_rx_conns); - CFS_INIT_LIST_HEAD(&sched->kss_tx_conns); - CFS_INIT_LIST_HEAD(&sched->kss_zombie_noop_txs); - init_waitqueue_head(&sched->kss_waitq); - } + spin_lock_init(&sched->kss_lock); + INIT_LIST_HEAD(&sched->kss_rx_conns); + INIT_LIST_HEAD(&sched->kss_tx_conns); + INIT_LIST_HEAD(&sched->kss_zombie_noop_txs); + init_waitqueue_head(&sched->kss_waitq); } ksocknal_data.ksnd_connd_starting = 0; ksocknal_data.ksnd_connd_failed_stamp = 0; - ksocknal_data.ksnd_connd_starting_stamp = cfs_time_current_sec(); + ksocknal_data.ksnd_connd_starting_stamp = ktime_get_real_seconds(); /* must have at least 2 connds to remain responsive to accepts while * connecting */ if (*ksocknal_tunables.ksnd_nconnds < SOCKNAL_CONND_RESV + 1) @@ -2472,7 +2482,7 @@ ksocknal_base_startup(void) snprintf(name, sizeof(name), "socknal_cd%02d", i); rc = ksocknal_thread_start(ksocknal_connd, - (void *)((ulong_ptr_t)i), name); + (void *)((uintptr_t)i), name); if (rc != 0) { spin_lock_bh(&ksocknal_data.ksnd_connd_lock); ksocknal_data.ksnd_connd_starting--; @@ -2498,189 +2508,113 @@ ksocknal_base_startup(void) return -ENETDOWN; } -void -ksocknal_debug_peerhash (lnet_ni_t *ni) +static int +ksocknal_debug_peerhash(struct lnet_ni *ni) { - ksock_peer_t *peer = NULL; - cfs_list_t *tmp; - int i; + struct ksock_peer_ni *peer_ni; + int i; read_lock(&ksocknal_data.ksnd_global_lock); - for (i = 0; i < ksocknal_data.ksnd_peer_hash_size; i++) { - cfs_list_for_each (tmp, &ksocknal_data.ksnd_peers[i]) { - peer = cfs_list_entry (tmp, ksock_peer_t, ksnp_list); + hash_for_each(ksocknal_data.ksnd_peers, i, peer_ni, ksnp_list) { + struct ksock_route *route; + struct ksock_conn *conn; - if (peer->ksnp_ni == ni) break; - - peer = NULL; - } - } + if (peer_ni->ksnp_ni != ni) + continue; - if (peer != NULL) { - ksock_route_t *route; - ksock_conn_t *conn; - - CWARN ("Active peer on shutdown: %s, ref %d, scnt %d, " - "closing %d, accepting %d, err %d, zcookie "LPU64", " - "txq %d, zc_req %d\n", libcfs_id2str(peer->ksnp_id), - atomic_read(&peer->ksnp_refcount), - peer->ksnp_sharecount, peer->ksnp_closing, - peer->ksnp_accepting, peer->ksnp_error, - peer->ksnp_zc_next_cookie, - !cfs_list_empty(&peer->ksnp_tx_queue), - !cfs_list_empty(&peer->ksnp_zc_req_list)); - - cfs_list_for_each (tmp, &peer->ksnp_routes) { - route = cfs_list_entry(tmp, ksock_route_t, ksnr_list); - CWARN ("Route: ref %d, schd %d, conn %d, cnted %d, " - "del %d\n", atomic_read(&route->ksnr_refcount), - route->ksnr_scheduled, route->ksnr_connecting, - route->ksnr_connected, route->ksnr_deleted); + CWARN("Active peer_ni on shutdown: %s, ref %d, " + "closing %d, accepting %d, err %d, zcookie %llu, " + "txq %d, zc_req %d\n", libcfs_id2str(peer_ni->ksnp_id), + refcount_read(&peer_ni->ksnp_refcount), + peer_ni->ksnp_closing, + peer_ni->ksnp_accepting, peer_ni->ksnp_error, + peer_ni->ksnp_zc_next_cookie, + !list_empty(&peer_ni->ksnp_tx_queue), + !list_empty(&peer_ni->ksnp_zc_req_list)); + + list_for_each_entry(route, &peer_ni->ksnp_routes, ksnr_list) { + CWARN("Route: ref %d, schd %d, conn %d, cnted %d, del %d\n", + refcount_read(&route->ksnr_refcount), + route->ksnr_scheduled, route->ksnr_connecting, + route->ksnr_connected, route->ksnr_deleted); } - cfs_list_for_each (tmp, &peer->ksnp_conns) { - conn = cfs_list_entry(tmp, ksock_conn_t, ksnc_list); - CWARN ("Conn: ref %d, sref %d, t %d, c %d\n", - atomic_read(&conn->ksnc_conn_refcount), - atomic_read(&conn->ksnc_sock_refcount), - conn->ksnc_type, conn->ksnc_closing); + list_for_each_entry(conn, &peer_ni->ksnp_conns, ksnc_list) { + CWARN("Conn: ref %d, sref %d, t %d, c %d\n", + refcount_read(&conn->ksnc_conn_refcount), + refcount_read(&conn->ksnc_sock_refcount), + conn->ksnc_type, conn->ksnc_closing); } + break; } read_unlock(&ksocknal_data.ksnd_global_lock); - return; + return 0; } void -ksocknal_shutdown (lnet_ni_t *ni) +ksocknal_shutdown(struct lnet_ni *ni) { - ksock_net_t *net = ni->ni_data; - int i; - lnet_process_id_t anyid = {0}; + struct ksock_net *net = ni->ni_data; + struct lnet_process_id anyid = { + .nid = LNET_NID_ANY, + .pid = LNET_PID_ANY, + }; + int i; - anyid.nid = LNET_NID_ANY; - anyid.pid = LNET_PID_ANY; + LASSERT(ksocknal_data.ksnd_init == SOCKNAL_INIT_ALL); + LASSERT(ksocknal_data.ksnd_nnets > 0); - LASSERT(ksocknal_data.ksnd_init == SOCKNAL_INIT_ALL); - LASSERT(ksocknal_data.ksnd_nnets > 0); - - spin_lock_bh(&net->ksnn_lock); - net->ksnn_shutdown = 1; /* prevent new peers */ - spin_unlock_bh(&net->ksnn_lock); + /* prevent new peers */ + atomic_add(SOCKNAL_SHUTDOWN_BIAS, &net->ksnn_npeers); /* Delete all peers */ ksocknal_del_peer(ni, anyid, 0); - /* Wait for all peer state to clean up */ - i = 2; - spin_lock_bh(&net->ksnn_lock); - while (net->ksnn_npeers != 0) { - spin_unlock_bh(&net->ksnn_lock); - - i++; - CDEBUG(((i & (-i)) == i) ? D_WARNING : D_NET, /* power of 2? */ - "waiting for %d peers to disconnect\n", - net->ksnn_npeers); - cfs_pause(cfs_time_seconds(1)); + /* Wait for all peer_ni state to clean up */ + wait_var_event_warning(&net->ksnn_npeers, + atomic_read(&net->ksnn_npeers) == + SOCKNAL_SHUTDOWN_BIAS, + "waiting for %d peers to disconnect\n", + ksocknal_debug_peerhash(ni) + + atomic_read(&net->ksnn_npeers) - + SOCKNAL_SHUTDOWN_BIAS); - ksocknal_debug_peerhash(ni); - - spin_lock_bh(&net->ksnn_lock); + for (i = 0; i < net->ksnn_ninterfaces; i++) { + LASSERT(net->ksnn_interfaces[i].ksni_npeers == 0); + LASSERT(net->ksnn_interfaces[i].ksni_nroutes == 0); } - spin_unlock_bh(&net->ksnn_lock); - for (i = 0; i < net->ksnn_ninterfaces; i++) { - LASSERT (net->ksnn_interfaces[i].ksni_npeers == 0); - LASSERT (net->ksnn_interfaces[i].ksni_nroutes == 0); - } - - cfs_list_del(&net->ksnn_list); + list_del(&net->ksnn_list); LIBCFS_FREE(net, sizeof(*net)); - ksocknal_data.ksnd_nnets--; - if (ksocknal_data.ksnd_nnets == 0) - ksocknal_base_shutdown(); + ksocknal_data.ksnd_nnets--; + if (ksocknal_data.ksnd_nnets == 0) + ksocknal_base_shutdown(); } -int -ksocknal_enumerate_interfaces(ksock_net_t *net) +static int +ksocknal_search_new_ipif(struct ksock_net *net) { - char **names; - int i; - int j; - int rc; - int n; - - n = libcfs_ipif_enumerate(&names); - if (n <= 0) { - CERROR("Can't enumerate interfaces: %d\n", n); - return n; - } - - for (i = j = 0; i < n; i++) { - int up; - __u32 ip; - __u32 mask; - - if (!strcmp(names[i], "lo")) /* skip the loopback IF */ - continue; - - rc = libcfs_ipif_query(names[i], &up, &ip, &mask); - if (rc != 0) { - CWARN("Can't get interface %s info: %d\n", - names[i], rc); - continue; - } - - if (!up) { - CWARN("Ignoring interface %s (down)\n", - names[i]); - continue; - } - - if (j == LNET_MAX_INTERFACES) { - CWARN("Ignoring interface %s (too many interfaces)\n", - names[i]); - continue; - } - - net->ksnn_interfaces[j].ksni_ipaddr = ip; - net->ksnn_interfaces[j].ksni_netmask = mask; - strncpy(&net->ksnn_interfaces[j].ksni_name[0], - names[i], IFNAMSIZ); - j++; - } - - libcfs_ipif_free_enumeration(names, n); - - if (j == 0) - CERROR("Can't find any usable interfaces\n"); - - return j; -} - -int -ksocknal_search_new_ipif(ksock_net_t *net) -{ - int new_ipif = 0; - int i; + int new_ipif = 0; + int i; for (i = 0; i < net->ksnn_ninterfaces; i++) { - char *ifnam = &net->ksnn_interfaces[i].ksni_name[0]; - char *colon = strchr(ifnam, ':'); - int found = 0; - ksock_net_t *tmp; - int j; + char *ifnam = &net->ksnn_interfaces[i].ksni_name[0]; + char *colon = strchr(ifnam, ':'); + bool found = false; + struct ksock_net *tmp; + int j; if (colon != NULL) /* ignore alias device */ *colon = 0; - cfs_list_for_each_entry(tmp, &ksocknal_data.ksnd_nets, - ksnn_list) { + list_for_each_entry(tmp, &ksocknal_data.ksnd_nets, + ksnn_list) { for (j = 0; !found && j < tmp->ksnn_ninterfaces; j++) { char *ifnam2 = &tmp->ksnn_interfaces[j].\ - ksni_name[0]; + ksni_name[0]; char *colon2 = strchr(ifnam2, ':'); if (colon2 != NULL) @@ -2702,37 +2636,36 @@ ksocknal_search_new_ipif(ksock_net_t *net) return new_ipif; } -int -ksocknal_start_schedulers(struct ksock_sched_info *info) +static int +ksocknal_start_schedulers(struct ksock_sched *sched) { int nthrs; int rc = 0; int i; - if (info->ksi_nthreads == 0) { + if (sched->kss_nthreads == 0) { if (*ksocknal_tunables.ksnd_nscheds > 0) { - nthrs = info->ksi_nthreads_max; + nthrs = sched->kss_nthreads_max; } else { nthrs = cfs_cpt_weight(lnet_cpt_table(), - info->ksi_cpt); + sched->kss_cpt); nthrs = min(max(SOCKNAL_NSCHEDS, nthrs >> 1), nthrs); nthrs = min(SOCKNAL_NSCHEDS_HIGH, nthrs); } - nthrs = min(nthrs, info->ksi_nthreads_max); + nthrs = min(nthrs, sched->kss_nthreads_max); } else { - LASSERT(info->ksi_nthreads <= info->ksi_nthreads_max); + LASSERT(sched->kss_nthreads <= sched->kss_nthreads_max); /* increase two threads if there is new interface */ - nthrs = min(2, info->ksi_nthreads_max - info->ksi_nthreads); + nthrs = min(2, sched->kss_nthreads_max - sched->kss_nthreads); } for (i = 0; i < nthrs; i++) { - long id; - char name[20]; - ksock_sched_t *sched; - id = KSOCK_THREAD_ID(info->ksi_cpt, info->ksi_nthreads + i); - sched = &info->ksi_scheds[KSOCK_THREAD_SID(id)]; + long id; + char name[20]; + + id = KSOCK_THREAD_ID(sched->kss_cpt, sched->kss_nthreads + i); snprintf(name, sizeof(name), "socknal_sd%02d_%02d", - info->ksi_cpt, (int)(sched - &info->ksi_scheds[0])); + sched->kss_cpt, (int)KSOCK_THREAD_SID(id)); rc = ksocknal_thread_start(ksocknal_scheduler, (void *)id, name); @@ -2740,34 +2673,35 @@ ksocknal_start_schedulers(struct ksock_sched_info *info) continue; CERROR("Can't spawn thread %d for scheduler[%d]: %d\n", - info->ksi_cpt, info->ksi_nthreads + i, rc); + sched->kss_cpt, (int) KSOCK_THREAD_SID(id), rc); break; } - info->ksi_nthreads += i; + sched->kss_nthreads += i; return rc; } -int -ksocknal_net_start_threads(ksock_net_t *net, __u32 *cpts, int ncpts) +static int +ksocknal_net_start_threads(struct ksock_net *net, __u32 *cpts, int ncpts) { - int newif = ksocknal_search_new_ipif(net); - int rc; - int i; + int newif = ksocknal_search_new_ipif(net); + int rc; + int i; - LASSERT(ncpts > 0 && ncpts <= cfs_cpt_number(lnet_cpt_table())); + if (ncpts > 0 && ncpts > cfs_cpt_number(lnet_cpt_table())) + return -EINVAL; for (i = 0; i < ncpts; i++) { - struct ksock_sched_info *info; + struct ksock_sched *sched; int cpt = (cpts == NULL) ? i : cpts[i]; LASSERT(cpt < cfs_cpt_number(lnet_cpt_table())); - info = ksocknal_data.ksnd_sched_info[cpt]; + sched = ksocknal_data.ksnd_schedulers[cpt]; - if (!newif && info->ksi_nthreads > 0) + if (!newif && sched->kss_nthreads > 0) continue; - rc = ksocknal_start_schedulers(info); + rc = ksocknal_start_schedulers(sched); if (rc != 0) return rc; } @@ -2775,13 +2709,16 @@ ksocknal_net_start_threads(ksock_net_t *net, __u32 *cpts, int ncpts) } int -ksocknal_startup (lnet_ni_t *ni) +ksocknal_startup(struct lnet_ni *ni) { - ksock_net_t *net; - int rc; - int i; + struct ksock_net *net; + struct lnet_ioctl_config_lnd_cmn_tunables *net_tunables; + struct ksock_interface *ksi = NULL; + struct lnet_inetdev *ifaces = NULL; + int i = 0; + int rc; - LASSERT (ni->ni_lnd == &the_ksocklnd); + LASSERT (ni->ni_net->net_lnd == &the_ksocklnd); if (ksocknal_data.ksnd_init == SOCKNAL_INIT_NOTHING) { rc = ksocknal_base_startup(); @@ -2789,52 +2726,113 @@ ksocknal_startup (lnet_ni_t *ni) return rc; } - LIBCFS_ALLOC(net, sizeof(*net)); - if (net == NULL) - goto fail_0; + LIBCFS_ALLOC(net, sizeof(*net)); + if (net == NULL) + goto fail_0; - spin_lock_init(&net->ksnn_lock); - net->ksnn_incarnation = ksocknal_new_incarnation(); - ni->ni_data = net; - ni->ni_peertimeout = *ksocknal_tunables.ksnd_peertimeout; - ni->ni_maxtxcredits = *ksocknal_tunables.ksnd_credits; - ni->ni_peertxcredits = *ksocknal_tunables.ksnd_peertxcredits; - ni->ni_peerrtrcredits = *ksocknal_tunables.ksnd_peerrtrcredits; + net->ksnn_incarnation = ktime_get_real_ns(); + ni->ni_data = net; + net_tunables = &ni->ni_net->net_tunables; - if (ni->ni_interfaces[0] == NULL) { - rc = ksocknal_enumerate_interfaces(net); - if (rc <= 0) - goto fail_1; + if (net_tunables->lct_peer_timeout == -1) + net_tunables->lct_peer_timeout = + *ksocknal_tunables.ksnd_peertimeout; - net->ksnn_ninterfaces = 1; - } else { - for (i = 0; i < LNET_MAX_INTERFACES; i++) { - int up; + if (net_tunables->lct_max_tx_credits == -1) + net_tunables->lct_max_tx_credits = + *ksocknal_tunables.ksnd_credits; - if (ni->ni_interfaces[i] == NULL) - break; + if (net_tunables->lct_peer_tx_credits == -1) + net_tunables->lct_peer_tx_credits = + *ksocknal_tunables.ksnd_peertxcredits; - rc = libcfs_ipif_query( - ni->ni_interfaces[i], &up, - &net->ksnn_interfaces[i].ksni_ipaddr, - &net->ksnn_interfaces[i].ksni_netmask); + if (net_tunables->lct_peer_tx_credits > + net_tunables->lct_max_tx_credits) + net_tunables->lct_peer_tx_credits = + net_tunables->lct_max_tx_credits; - if (rc != 0) { - CERROR("Can't get interface %s info: %d\n", - ni->ni_interfaces[i], rc); - goto fail_1; - } + if (net_tunables->lct_peer_rtr_credits == -1) + net_tunables->lct_peer_rtr_credits = + *ksocknal_tunables.ksnd_peerrtrcredits; - if (!up) { - CERROR("Interface %s is down\n", - ni->ni_interfaces[i]); - goto fail_1; - } + rc = lnet_inet_enumerate(&ifaces, ni->ni_net_ns); + if (rc < 0) + goto fail_1; - strncpy(&net->ksnn_interfaces[i].ksni_name[0], - ni->ni_interfaces[i], IFNAMSIZ); + if (!ni->ni_interfaces[0]) { + struct sockaddr_in *sa; + + ksi = &net->ksnn_interfaces[0]; + sa = (void *)&ksi->ksni_addr; + + /* Use the first discovered interface */ + net->ksnn_ninterfaces = 1; + ni->ni_dev_cpt = ifaces[0].li_cpt; + memset(sa, 0, sizeof(*sa)); + sa->sin_family = AF_INET; + sa->sin_addr.s_addr = htonl(ifaces[0].li_ipaddr); + ksi->ksni_index = ksocknal_ip2index((struct sockaddr *)sa, ni); + ksi->ksni_netmask = ifaces[0].li_netmask; + strlcpy(ksi->ksni_name, ifaces[0].li_name, + sizeof(ksi->ksni_name)); + } else { + /* Before Multi-Rail ksocklnd would manage + * multiple interfaces with its own tcp bonding. + * If we encounter an old configuration using + * this tcp bonding approach then we need to + * handle more than one ni_interfaces. + * + * In Multi-Rail configuration only ONE ni_interface + * should exist. Each IP alias should be mapped to + * each 'struct net_ni'. + */ + for (i = 0; i < LNET_INTERFACES_NUM; i++) { + int j; + + if (!ni->ni_interfaces[i]) + break; + + for (j = 0; j < LNET_INTERFACES_NUM; j++) { + if (i != j && ni->ni_interfaces[j] && + strcmp(ni->ni_interfaces[i], + ni->ni_interfaces[j]) == 0) { + rc = -EEXIST; + CERROR("ksocklnd: found duplicate %s at %d and %d, rc = %d\n", + ni->ni_interfaces[i], i, j, rc); + goto fail_1; + } + } + + for (j = 0; j < rc; j++) { + struct sockaddr_in *sa; + + if (strcmp(ifaces[j].li_name, + ni->ni_interfaces[i]) != 0) + continue; + + ksi = + &net->ksnn_interfaces[net->ksnn_ninterfaces]; + sa = (void *)&ksi->ksni_addr; + ni->ni_dev_cpt = ifaces[j].li_cpt; + memset(sa, 0, sizeof(*sa)); + sa->sin_family = AF_INET; + sa->sin_addr.s_addr = + htonl(ifaces[j].li_ipaddr); + ksi->ksni_index = ksocknal_ip2index( + (struct sockaddr *)sa, ni); + ksi->ksni_netmask = ifaces[j].li_netmask; + strlcpy(ksi->ksni_name, ifaces[j].li_name, + sizeof(ksi->ksni_name)); + net->ksnn_ninterfaces++; + break; + } + } + /* ni_interfaces don't map to all network interfaces */ + if (!ksi || net->ksnn_ninterfaces != i) { + CERROR("ksocklnd: requested %d but only %d interfaces found\n", + i, net->ksnn_ninterfaces); + goto fail_1; } - net->ksnn_ninterfaces = i; } /* call it before add it to ksocknal_data.ksnd_nets */ @@ -2842,62 +2840,65 @@ ksocknal_startup (lnet_ni_t *ni) if (rc != 0) goto fail_1; - ni->ni_nid = LNET_MKNID(LNET_NIDNET(ni->ni_nid), - net->ksnn_interfaces[0].ksni_ipaddr); - cfs_list_add(&net->ksnn_list, &ksocknal_data.ksnd_nets); + LASSERT(ksi); + LASSERT(ksi->ksni_addr.ss_family == AF_INET); + ni->ni_nid = LNET_MKNID( + LNET_NIDNET(ni->ni_nid), + ntohl(((struct sockaddr_in *) + &ksi->ksni_addr)->sin_addr.s_addr)); + list_add(&net->ksnn_list, &ksocknal_data.ksnd_nets); - ksocknal_data.ksnd_nnets++; + ksocknal_data.ksnd_nnets++; - return 0; + return 0; - fail_1: - LIBCFS_FREE(net, sizeof(*net)); - fail_0: - if (ksocknal_data.ksnd_nnets == 0) - ksocknal_base_shutdown(); +fail_1: + LIBCFS_FREE(net, sizeof(*net)); +fail_0: + if (ksocknal_data.ksnd_nnets == 0) + ksocknal_base_shutdown(); - return -ENETDOWN; + return -ENETDOWN; } -void __exit -ksocknal_module_fini (void) +static void __exit ksocklnd_exit(void) { - lnet_unregister_lnd(&the_ksocklnd); - ksocknal_tunables_fini(); + lnet_unregister_lnd(&the_ksocklnd); } -int __init -ksocknal_module_init (void) +static const struct lnet_lnd the_ksocklnd = { + .lnd_type = SOCKLND, + .lnd_startup = ksocknal_startup, + .lnd_shutdown = ksocknal_shutdown, + .lnd_ctl = ksocknal_ctl, + .lnd_send = ksocknal_send, + .lnd_recv = ksocknal_recv, + .lnd_notify_peer_down = ksocknal_notify_gw_down, + .lnd_accept = ksocknal_accept, +}; + +static int __init ksocklnd_init(void) { - int rc; - - /* check ksnr_connected/connecting field large enough */ - CLASSERT (SOCKLND_CONN_NTYPES <= 4); - CLASSERT (SOCKLND_CONN_ACK == SOCKLND_CONN_BULK_IN); - - /* initialize the_ksocklnd */ - the_ksocklnd.lnd_type = SOCKLND; - the_ksocklnd.lnd_startup = ksocknal_startup; - the_ksocklnd.lnd_shutdown = ksocknal_shutdown; - the_ksocklnd.lnd_ctl = ksocknal_ctl; - the_ksocklnd.lnd_send = ksocknal_send; - the_ksocklnd.lnd_recv = ksocknal_recv; - the_ksocklnd.lnd_notify = ksocknal_notify; - the_ksocklnd.lnd_query = ksocknal_query; - the_ksocklnd.lnd_accept = ksocknal_accept; - - rc = ksocknal_tunables_init(); - if (rc != 0) - return rc; + int rc; - lnet_register_lnd(&the_ksocklnd); + /* check ksnr_connected/connecting field large enough */ + BUILD_BUG_ON(SOCKLND_CONN_NTYPES > 4); + BUILD_BUG_ON(SOCKLND_CONN_ACK != SOCKLND_CONN_BULK_IN); - return 0; + rc = ksocknal_tunables_init(); + if (rc != 0) + return rc; + + lnet_register_lnd(&the_ksocklnd); + + return 0; } -MODULE_AUTHOR("Sun Microsystems, Inc. "); -MODULE_DESCRIPTION("Kernel TCP Socket LND v3.0.0"); +MODULE_AUTHOR("OpenSFS, Inc. "); +MODULE_DESCRIPTION("TCP Socket LNet Network Driver"); +MODULE_VERSION("2.8.0"); MODULE_LICENSE("GPL"); -cfs_module(ksocknal, "3.0.0", ksocknal_module_init, ksocknal_module_fini); +module_init(ksocklnd_init); +module_exit(ksocklnd_exit);