diff options
Diffstat (limited to 'net')
133 files changed, 5050 insertions, 2756 deletions
diff --git a/net/8021q/vlan.c b/net/8021q/vlan.c index a1e273a..82a116b 100644 --- a/net/8021q/vlan.c +++ b/net/8021q/vlan.c @@ -290,6 +290,10 @@ static void vlan_sync_address(struct net_device *dev, if (ether_addr_equal(vlan->real_dev_addr, dev->dev_addr)) return; + /* vlan continues to inherit address of lower device */ + if (vlan_dev_inherit_address(vlandev, dev)) + goto out; + /* vlan address was different from the old address and is equal to * the new address */ if (!ether_addr_equal(vlandev->dev_addr, vlan->real_dev_addr) && @@ -302,6 +306,7 @@ static void vlan_sync_address(struct net_device *dev, !ether_addr_equal(vlandev->dev_addr, dev->dev_addr)) dev_uc_add(dev, vlandev->dev_addr); +out: ether_addr_copy(vlan->real_dev_addr, dev->dev_addr); } diff --git a/net/8021q/vlan.h b/net/8021q/vlan.h index 9d010a09..cc15579 100644 --- a/net/8021q/vlan.h +++ b/net/8021q/vlan.h @@ -109,6 +109,8 @@ int vlan_check_real_dev(struct net_device *real_dev, void vlan_setup(struct net_device *dev); int register_vlan_dev(struct net_device *dev); void unregister_vlan_dev(struct net_device *dev, struct list_head *head); +bool vlan_dev_inherit_address(struct net_device *dev, + struct net_device *real_dev); static inline u32 vlan_get_ingress_priority(struct net_device *dev, u16 vlan_tci) diff --git a/net/8021q/vlan_dev.c b/net/8021q/vlan_dev.c index e7e6257..86ae75b 100644 --- a/net/8021q/vlan_dev.c +++ b/net/8021q/vlan_dev.c @@ -245,6 +245,17 @@ void vlan_dev_get_realdev_name(const struct net_device *dev, char *result) strncpy(result, vlan_dev_priv(dev)->real_dev->name, 23); } +bool vlan_dev_inherit_address(struct net_device *dev, + struct net_device *real_dev) +{ + if (dev->addr_assign_type != NET_ADDR_STOLEN) + return false; + + ether_addr_copy(dev->dev_addr, real_dev->dev_addr); + call_netdevice_notifiers(NETDEV_CHANGEADDR, dev); + return true; +} + static int vlan_dev_open(struct net_device *dev) { struct vlan_dev_priv *vlan = vlan_dev_priv(dev); @@ -255,7 +266,8 @@ static int vlan_dev_open(struct net_device *dev) !(vlan->flags & VLAN_FLAG_LOOSE_BINDING)) return -ENETDOWN; - if (!ether_addr_equal(dev->dev_addr, real_dev->dev_addr)) { + if (!ether_addr_equal(dev->dev_addr, real_dev->dev_addr) && + !vlan_dev_inherit_address(dev, real_dev)) { err = dev_uc_add(real_dev, dev->dev_addr); if (err < 0) goto out; @@ -560,8 +572,10 @@ static int vlan_dev_init(struct net_device *dev) /* ipv6 shared card related stuff */ dev->dev_id = real_dev->dev_id; - if (is_zero_ether_addr(dev->dev_addr)) - eth_hw_addr_inherit(dev, real_dev); + if (is_zero_ether_addr(dev->dev_addr)) { + ether_addr_copy(dev->dev_addr, real_dev->dev_addr); + dev->addr_assign_type = NET_ADDR_STOLEN; + } if (is_zero_ether_addr(dev->broadcast)) memcpy(dev->broadcast, real_dev->broadcast, dev->addr_len); diff --git a/net/9p/client.c b/net/9p/client.c index ea79ee9..3fc94a4 100644 --- a/net/9p/client.c +++ b/net/9p/client.c @@ -518,10 +518,10 @@ static int p9_check_errors(struct p9_client *c, struct p9_req_t *req) if (err) goto out_err; - if (p9_is_proto_dotu(c)) + if (p9_is_proto_dotu(c) && ecode < 512) err = -ecode; - if (!err || !IS_ERR_VALUE(err)) { + if (!err) { err = p9_errstr2errno(ename, strlen(ename)); p9_debug(P9_DEBUG_9P, "<<< RERROR (%d) %s\n", @@ -605,10 +605,10 @@ static int p9_check_zc_errors(struct p9_client *c, struct p9_req_t *req, if (err) goto out_err; - if (p9_is_proto_dotu(c)) + if (p9_is_proto_dotu(c) && ecode < 512) err = -ecode; - if (!err || !IS_ERR_VALUE(err)) { + if (!err) { err = p9_errstr2errno(ename, strlen(ename)); p9_debug(P9_DEBUG_9P, "<<< RERROR (%d) %s\n", diff --git a/net/ax25/af_ax25.c b/net/ax25/af_ax25.c index fbd0acf..2fdebab 100644 --- a/net/ax25/af_ax25.c +++ b/net/ax25/af_ax25.c @@ -976,7 +976,8 @@ static int ax25_release(struct socket *sock) release_sock(sk); ax25_disconnect(ax25, 0); lock_sock(sk); - ax25_destroy_socket(ax25); + if (!sock_flag(ax25->sk, SOCK_DESTROY)) + ax25_destroy_socket(ax25); break; case AX25_STATE_3: diff --git a/net/ax25/ax25_ds_timer.c b/net/ax25/ax25_ds_timer.c index 951cd57..5237dff 100644 --- a/net/ax25/ax25_ds_timer.c +++ b/net/ax25/ax25_ds_timer.c @@ -102,6 +102,7 @@ void ax25_ds_heartbeat_expiry(ax25_cb *ax25) switch (ax25->state) { case AX25_STATE_0: + case AX25_STATE_2: /* Magic here: If we listen() and a new link dies before it is accepted() it isn't 'dead' so doesn't get removed. */ if (!sk || sock_flag(sk, SOCK_DESTROY) || @@ -111,6 +112,7 @@ void ax25_ds_heartbeat_expiry(ax25_cb *ax25) sock_hold(sk); ax25_destroy_socket(ax25); bh_unlock_sock(sk); + /* Ungrab socket and destroy it */ sock_put(sk); } else ax25_destroy_socket(ax25); @@ -213,7 +215,8 @@ void ax25_ds_t1_timeout(ax25_cb *ax25) case AX25_STATE_2: if (ax25->n2count == ax25->n2) { ax25_send_control(ax25, AX25_DISC, AX25_POLLON, AX25_COMMAND); - ax25_disconnect(ax25, ETIMEDOUT); + if (!sock_flag(ax25->sk, SOCK_DESTROY)) + ax25_disconnect(ax25, ETIMEDOUT); return; } else { ax25->n2count++; diff --git a/net/ax25/ax25_std_timer.c b/net/ax25/ax25_std_timer.c index 004467c9..2c0d6ef 100644 --- a/net/ax25/ax25_std_timer.c +++ b/net/ax25/ax25_std_timer.c @@ -38,6 +38,7 @@ void ax25_std_heartbeat_expiry(ax25_cb *ax25) switch (ax25->state) { case AX25_STATE_0: + case AX25_STATE_2: /* Magic here: If we listen() and a new link dies before it is accepted() it isn't 'dead' so doesn't get removed. */ if (!sk || sock_flag(sk, SOCK_DESTROY) || @@ -47,6 +48,7 @@ void ax25_std_heartbeat_expiry(ax25_cb *ax25) sock_hold(sk); ax25_destroy_socket(ax25); bh_unlock_sock(sk); + /* Ungrab socket and destroy it */ sock_put(sk); } else ax25_destroy_socket(ax25); @@ -144,7 +146,8 @@ void ax25_std_t1timer_expiry(ax25_cb *ax25) case AX25_STATE_2: if (ax25->n2count == ax25->n2) { ax25_send_control(ax25, AX25_DISC, AX25_POLLON, AX25_COMMAND); - ax25_disconnect(ax25, ETIMEDOUT); + if (!sock_flag(ax25->sk, SOCK_DESTROY)) + ax25_disconnect(ax25, ETIMEDOUT); return; } else { ax25->n2count++; diff --git a/net/ax25/ax25_subr.c b/net/ax25/ax25_subr.c index 3b78e84..655a7d4 100644 --- a/net/ax25/ax25_subr.c +++ b/net/ax25/ax25_subr.c @@ -264,7 +264,8 @@ void ax25_disconnect(ax25_cb *ax25, int reason) { ax25_clear_queues(ax25); - ax25_stop_heartbeat(ax25); + if (!sock_flag(ax25->sk, SOCK_DESTROY)) + ax25_stop_heartbeat(ax25); ax25_stop_t1timer(ax25); ax25_stop_t2timer(ax25); ax25_stop_t3timer(ax25); diff --git a/net/batman-adv/routing.c b/net/batman-adv/routing.c index e3857ed..6c2901a 100644 --- a/net/batman-adv/routing.c +++ b/net/batman-adv/routing.c @@ -374,6 +374,7 @@ int batadv_recv_icmp_packet(struct sk_buff *skb, if (skb_cow(skb, ETH_HLEN) < 0) goto out; + ethhdr = eth_hdr(skb); icmph = (struct batadv_icmp_header *)skb->data; icmp_packet_rr = (struct batadv_icmp_packet_rr *)icmph; if (icmp_packet_rr->rr_cur >= BATADV_RR_LEN) diff --git a/net/batman-adv/soft-interface.c b/net/batman-adv/soft-interface.c index 343d2c9..287a387 100644 --- a/net/batman-adv/soft-interface.c +++ b/net/batman-adv/soft-interface.c @@ -1033,7 +1033,9 @@ void batadv_softif_destroy_sysfs(struct net_device *soft_iface) static void batadv_softif_destroy_netlink(struct net_device *soft_iface, struct list_head *head) { + struct batadv_priv *bat_priv = netdev_priv(soft_iface); struct batadv_hard_iface *hard_iface; + struct batadv_softif_vlan *vlan; list_for_each_entry(hard_iface, &batadv_hardif_list, list) { if (hard_iface->soft_iface == soft_iface) @@ -1041,6 +1043,13 @@ static void batadv_softif_destroy_netlink(struct net_device *soft_iface, BATADV_IF_CLEANUP_KEEP); } + /* destroy the "untagged" VLAN */ + vlan = batadv_softif_vlan_get(bat_priv, BATADV_NO_FLAGS); + if (vlan) { + batadv_softif_destroy_vlan(bat_priv, vlan); + batadv_softif_vlan_put(vlan); + } + batadv_sysfs_del_meshif(soft_iface); unregister_netdevice_queue(soft_iface, head); } diff --git a/net/batman-adv/translation-table.c b/net/batman-adv/translation-table.c index feaf492b..57ec87f 100644 --- a/net/batman-adv/translation-table.c +++ b/net/batman-adv/translation-table.c @@ -650,8 +650,10 @@ bool batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, /* increase the refcounter of the related vlan */ vlan = batadv_softif_vlan_get(bat_priv, vid); - if (WARN(!vlan, "adding TT local entry %pM to non-existent VLAN %d", - addr, BATADV_PRINT_VID(vid))) { + if (!vlan) { + net_ratelimited_function(batadv_info, soft_iface, + "adding TT local entry %pM to non-existent VLAN %d\n", + addr, BATADV_PRINT_VID(vid)); kfree(tt_local); tt_local = NULL; goto out; @@ -691,7 +693,6 @@ bool batadv_tt_local_add(struct net_device *soft_iface, const u8 *addr, if (unlikely(hash_added != 0)) { /* remove the reference for the hash */ batadv_tt_local_entry_put(tt_local); - batadv_softif_vlan_put(vlan); goto out; } @@ -2269,6 +2270,29 @@ static u32 batadv_tt_local_crc(struct batadv_priv *bat_priv, return crc; } +/** + * batadv_tt_req_node_release - free tt_req node entry + * @ref: kref pointer of the tt req_node entry + */ +static void batadv_tt_req_node_release(struct kref *ref) +{ + struct batadv_tt_req_node *tt_req_node; + + tt_req_node = container_of(ref, struct batadv_tt_req_node, refcount); + + kfree(tt_req_node); +} + +/** + * batadv_tt_req_node_put - decrement the tt_req_node refcounter and + * possibly release it + * @tt_req_node: tt_req_node to be free'd + */ +static void batadv_tt_req_node_put(struct batadv_tt_req_node *tt_req_node) +{ + kref_put(&tt_req_node->refcount, batadv_tt_req_node_release); +} + static void batadv_tt_req_list_free(struct batadv_priv *bat_priv) { struct batadv_tt_req_node *node; @@ -2278,7 +2302,7 @@ static void batadv_tt_req_list_free(struct batadv_priv *bat_priv) hlist_for_each_entry_safe(node, safe, &bat_priv->tt.req_list, list) { hlist_del_init(&node->list); - kfree(node); + batadv_tt_req_node_put(node); } spin_unlock_bh(&bat_priv->tt.req_list_lock); @@ -2315,7 +2339,7 @@ static void batadv_tt_req_purge(struct batadv_priv *bat_priv) if (batadv_has_timed_out(node->issued_at, BATADV_TT_REQUEST_TIMEOUT)) { hlist_del_init(&node->list); - kfree(node); + batadv_tt_req_node_put(node); } } spin_unlock_bh(&bat_priv->tt.req_list_lock); @@ -2347,9 +2371,11 @@ batadv_tt_req_node_new(struct batadv_priv *bat_priv, if (!tt_req_node) goto unlock; + kref_init(&tt_req_node->refcount); ether_addr_copy(tt_req_node->addr, orig_node->orig); tt_req_node->issued_at = jiffies; + kref_get(&tt_req_node->refcount); hlist_add_head(&tt_req_node->list, &bat_priv->tt.req_list); unlock: spin_unlock_bh(&bat_priv->tt.req_list_lock); @@ -2613,13 +2639,19 @@ static bool batadv_send_tt_request(struct batadv_priv *bat_priv, out: if (primary_if) batadv_hardif_put(primary_if); + if (ret && tt_req_node) { spin_lock_bh(&bat_priv->tt.req_list_lock); - /* hlist_del_init() verifies tt_req_node still is in the list */ - hlist_del_init(&tt_req_node->list); + if (!hlist_unhashed(&tt_req_node->list)) { + hlist_del_init(&tt_req_node->list); + batadv_tt_req_node_put(tt_req_node); + } spin_unlock_bh(&bat_priv->tt.req_list_lock); - kfree(tt_req_node); } + + if (tt_req_node) + batadv_tt_req_node_put(tt_req_node); + kfree(tvlv_tt_data); return ret; } @@ -3055,7 +3087,7 @@ static void batadv_handle_tt_response(struct batadv_priv *bat_priv, if (!batadv_compare_eth(node->addr, resp_src)) continue; hlist_del_init(&node->list); - kfree(node); + batadv_tt_req_node_put(node); } spin_unlock_bh(&bat_priv->tt.req_list_lock); diff --git a/net/batman-adv/types.h b/net/batman-adv/types.h index 6a577f4..ba846b0 100644 --- a/net/batman-adv/types.h +++ b/net/batman-adv/types.h @@ -1137,11 +1137,13 @@ struct batadv_tt_change_node { * struct batadv_tt_req_node - data to keep track of the tt requests in flight * @addr: mac address address of the originator this request was sent to * @issued_at: timestamp used for purging stale tt requests + * @refcount: number of contexts the object is used by * @list: list node for batadv_priv_tt::req_list */ struct batadv_tt_req_node { u8 addr[ETH_ALEN]; unsigned long issued_at; + struct kref refcount; struct hlist_node list; }; diff --git a/net/bridge/br_fdb.c b/net/bridge/br_fdb.c index dcea4f4..c18080a 100644 --- a/net/bridge/br_fdb.c +++ b/net/bridge/br_fdb.c @@ -279,6 +279,8 @@ void br_fdb_change_mac_address(struct net_bridge *br, const u8 *newaddr) * change from under us. */ list_for_each_entry(v, &vg->vlan_list, vlist) { + if (!br_vlan_should_use(v)) + continue; f = __br_fdb_get(br, br->dev->dev_addr, v->vid); if (f && f->is_local && !f->dst) fdb_delete_local(br, NULL, f); diff --git a/net/bridge/br_input.c b/net/bridge/br_input.c index 1607977..43d2cd8 100644 --- a/net/bridge/br_input.c +++ b/net/bridge/br_input.c @@ -213,8 +213,7 @@ drop: } EXPORT_SYMBOL_GPL(br_handle_frame_finish); -/* note: already called with rcu_read_lock */ -static int br_handle_local_finish(struct net *net, struct sock *sk, struct sk_buff *skb) +static void __br_handle_local_finish(struct sk_buff *skb) { struct net_bridge_port *p = br_port_get_rcu(skb->dev); u16 vid = 0; @@ -222,6 +221,14 @@ static int br_handle_local_finish(struct net *net, struct sock *sk, struct sk_bu /* check if vlan is allowed, to avoid spoofing */ if (p->flags & BR_LEARNING && br_should_learn(p, skb, &vid)) br_fdb_update(p->br, p, eth_hdr(skb)->h_source, vid, false); +} + +/* note: already called with rcu_read_lock */ +static int br_handle_local_finish(struct net *net, struct sock *sk, struct sk_buff *skb) +{ + struct net_bridge_port *p = br_port_get_rcu(skb->dev); + + __br_handle_local_finish(skb); BR_INPUT_SKB_CB(skb)->brdev = p->br->dev; br_pass_frame_up(skb); @@ -274,7 +281,9 @@ rx_handler_result_t br_handle_frame(struct sk_buff **pskb) if (p->br->stp_enabled == BR_NO_STP || fwd_mask & (1u << dest[5])) goto forward; - break; + *pskb = skb; + __br_handle_local_finish(skb); + return RX_HANDLER_PASS; case 0x01: /* IEEE MAC (Pause) */ goto drop; diff --git a/net/bridge/br_multicast.c b/net/bridge/br_multicast.c index 6852f3c..4384414 100644 --- a/net/bridge/br_multicast.c +++ b/net/bridge/br_multicast.c @@ -464,8 +464,11 @@ static struct sk_buff *br_ip6_multicast_alloc_query(struct net_bridge *br, if (ipv6_dev_get_saddr(dev_net(br->dev), br->dev, &ip6h->daddr, 0, &ip6h->saddr)) { kfree_skb(skb); + br->has_ipv6_addr = 0; return NULL; } + + br->has_ipv6_addr = 1; ipv6_eth_mc_map(&ip6h->daddr, eth->h_dest); hopopt = (u8 *)(ip6h + 1); @@ -1745,6 +1748,7 @@ void br_multicast_init(struct net_bridge *br) br->ip6_other_query.delay_time = 0; br->ip6_querier.port = NULL; #endif + br->has_ipv6_addr = 1; spin_lock_init(&br->multicast_lock); setup_timer(&br->multicast_router_timer, diff --git a/net/bridge/br_netlink.c b/net/bridge/br_netlink.c index a5343c7..85e89f6 100644 --- a/net/bridge/br_netlink.c +++ b/net/bridge/br_netlink.c @@ -1273,7 +1273,7 @@ static int br_fill_linkxstats(struct sk_buff *skb, const struct net_device *dev, struct bridge_vlan_xstats vxi; struct br_vlan_stats stats; - if (vl_idx++ < *prividx) + if (++vl_idx < *prividx) continue; memset(&vxi, 0, sizeof(vxi)); vxi.vid = v->vid; diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h index c7fb5d7..52edecf 100644 --- a/net/bridge/br_private.h +++ b/net/bridge/br_private.h @@ -314,6 +314,7 @@ struct net_bridge u8 multicast_disabled:1; u8 multicast_querier:1; u8 multicast_query_use_ifaddr:1; + u8 has_ipv6_addr:1; u32 hash_elasticity; u32 hash_max; @@ -588,10 +589,22 @@ static inline bool br_multicast_is_router(struct net_bridge *br) static inline bool __br_multicast_querier_exists(struct net_bridge *br, - struct bridge_mcast_other_query *querier) + struct bridge_mcast_other_query *querier, + const bool is_ipv6) { + bool own_querier_enabled; + + if (br->multicast_querier) { + if (is_ipv6 && !br->has_ipv6_addr) + own_querier_enabled = false; + else + own_querier_enabled = true; + } else { + own_querier_enabled = false; + } + return time_is_before_jiffies(querier->delay_time) && - (br->multicast_querier || timer_pending(&querier->timer)); + (own_querier_enabled || timer_pending(&querier->timer)); } static inline bool br_multicast_querier_exists(struct net_bridge *br, @@ -599,10 +612,12 @@ static inline bool br_multicast_querier_exists(struct net_bridge *br, { switch (eth->h_proto) { case (htons(ETH_P_IP)): - return __br_multicast_querier_exists(br, &br->ip4_other_query); + return __br_multicast_querier_exists(br, + &br->ip4_other_query, false); #if IS_ENABLED(CONFIG_IPV6) case (htons(ETH_P_IPV6)): - return __br_multicast_querier_exists(br, &br->ip6_other_query); + return __br_multicast_querier_exists(br, + &br->ip6_other_query, true); #endif default: return false; diff --git a/net/ceph/ceph_common.c b/net/ceph/ceph_common.c index dcc18c6..55d2bfe 100644 --- a/net/ceph/ceph_common.c +++ b/net/ceph/ceph_common.c @@ -651,7 +651,7 @@ EXPORT_SYMBOL(ceph_destroy_client); /* * true if we have the mon map (and have thus joined the cluster) */ -static int have_mon_and_osd_map(struct ceph_client *client) +static bool have_mon_and_osd_map(struct ceph_client *client) { return client->monc.monmap && client->monc.monmap->epoch && client->osdc.osdmap && client->osdc.osdmap->epoch; diff --git a/net/ceph/ceph_strings.c b/net/ceph/ceph_strings.c index 139a9cb..3773a4f 100644 --- a/net/ceph/ceph_strings.c +++ b/net/ceph/ceph_strings.c @@ -27,6 +27,22 @@ __CEPH_FORALL_OSD_OPS(GENERATE_CASE) } } +const char *ceph_osd_watch_op_name(int o) +{ + switch (o) { + case CEPH_OSD_WATCH_OP_UNWATCH: + return "unwatch"; + case CEPH_OSD_WATCH_OP_WATCH: + return "watch"; + case CEPH_OSD_WATCH_OP_RECONNECT: + return "reconnect"; + case CEPH_OSD_WATCH_OP_PING: + return "ping"; + default: + return "???"; + } +} + const char *ceph_osd_state_name(int s) { switch (s) { diff --git a/net/ceph/debugfs.c b/net/ceph/debugfs.c index b902fbc..e77b04c 100644 --- a/net/ceph/debugfs.c +++ b/net/ceph/debugfs.c @@ -54,24 +54,25 @@ static int osdmap_show(struct seq_file *s, void *p) { int i; struct ceph_client *client = s->private; - struct ceph_osdmap *map = client->osdc.osdmap; + struct ceph_osd_client *osdc = &client->osdc; + struct ceph_osdmap *map = osdc->osdmap; struct rb_node *n; if (map == NULL) return 0; - seq_printf(s, "epoch %d\n", map->epoch); - seq_printf(s, "flags%s%s\n", - (map->flags & CEPH_OSDMAP_NEARFULL) ? " NEARFULL" : "", - (map->flags & CEPH_OSDMAP_FULL) ? " FULL" : ""); + down_read(&osdc->lock); + seq_printf(s, "epoch %d flags 0x%x\n", map->epoch, map->flags); for (n = rb_first(&map->pg_pools); n; n = rb_next(n)) { - struct ceph_pg_pool_info *pool = + struct ceph_pg_pool_info *pi = rb_entry(n, struct ceph_pg_pool_info, node); - seq_printf(s, "pool %lld pg_num %u (%d) read_tier %lld write_tier %lld\n", - pool->id, pool->pg_num, pool->pg_num_mask, - pool->read_tier, pool->write_tier); + seq_printf(s, "pool %lld '%s' type %d size %d min_size %d pg_num %u pg_num_mask %d flags 0x%llx lfor %u read_tier %lld write_tier %lld\n", + pi->id, pi->name, pi->type, pi->size, pi->min_size, + pi->pg_num, pi->pg_num_mask, pi->flags, + pi->last_force_request_resend, pi->read_tier, + pi->write_tier); } for (i = 0; i < map->max_osd; i++) { struct ceph_entity_addr *addr = &map->osd_addr[i]; @@ -103,6 +104,7 @@ static int osdmap_show(struct seq_file *s, void *p) pg->pgid.seed, pg->primary_temp.osd); } + up_read(&osdc->lock); return 0; } @@ -126,6 +128,7 @@ static int monc_show(struct seq_file *s, void *p) CEPH_SUBSCRIBE_ONETIME ? "" : "+")); seq_putc(s, '\n'); } + seq_printf(s, "fs_cluster_id %d\n", monc->fs_cluster_id); for (rp = rb_first(&monc->generic_request_tree); rp; rp = rb_next(rp)) { __u16 op; @@ -143,43 +146,113 @@ static int monc_show(struct seq_file *s, void *p) return 0; } -static int osdc_show(struct seq_file *s, void *pp) +static void dump_target(struct seq_file *s, struct ceph_osd_request_target *t) { - struct ceph_client *client = s->private; - struct ceph_osd_client *osdc = &client->osdc; - struct rb_node *p; + int i; - mutex_lock(&osdc->request_mutex); - for (p = rb_first(&osdc->requests); p; p = rb_next(p)) { - struct ceph_osd_request *req; - unsigned int i; - int opcode; + seq_printf(s, "osd%d\t%llu.%x\t[", t->osd, t->pgid.pool, t->pgid.seed); + for (i = 0; i < t->up.size; i++) + seq_printf(s, "%s%d", (!i ? "" : ","), t->up.osds[i]); + seq_printf(s, "]/%d\t[", t->up.primary); + for (i = 0; i < t->acting.size; i++) + seq_printf(s, "%s%d", (!i ? "" : ","), t->acting.osds[i]); + seq_printf(s, "]/%d\t%*pE\t0x%x", t->acting.primary, + t->target_oid.name_len, t->target_oid.name, t->flags); + if (t->paused) + seq_puts(s, "\tP"); +} - req = rb_entry(p, struct ceph_osd_request, r_node); +static void dump_request(struct seq_file *s, struct ceph_osd_request *req) +{ + int i; - seq_printf(s, "%lld\tosd%d\t%lld.%x\t", req->r_tid, - req->r_osd ? req->r_osd->o_osd : -1, - req->r_pgid.pool, req->r_pgid.seed); + seq_printf(s, "%llu\t", req->r_tid); + dump_target(s, &req->r_t); - seq_printf(s, "%.*s", req->r_base_oid.name_len, - req->r_base_oid.name); + seq_printf(s, "\t%d\t%u'%llu", req->r_attempts, + le32_to_cpu(req->r_replay_version.epoch), + le64_to_cpu(req->r_replay_version.version)); - if (req->r_reassert_version.epoch) - seq_printf(s, "\t%u'%llu", - (unsigned int)le32_to_cpu(req->r_reassert_version.epoch), - le64_to_cpu(req->r_reassert_version.version)); - else - seq_printf(s, "\t"); + for (i = 0; i < req->r_num_ops; i++) { + struct ceph_osd_req_op *op = &req->r_ops[i]; + + seq_printf(s, "%s%s", (i == 0 ? "\t" : ","), + ceph_osd_op_name(op->op)); + if (op->op == CEPH_OSD_OP_WATCH) + seq_printf(s, "-%s", + ceph_osd_watch_op_name(op->watch.op)); + } + + seq_putc(s, '\n'); +} + +static void dump_requests(struct seq_file *s, struct ceph_osd *osd) +{ + struct rb_node *n; + + mutex_lock(&osd->lock); + for (n = rb_first(&osd->o_requests); n; n = rb_next(n)) { + struct ceph_osd_request *req = + rb_entry(n, struct ceph_osd_request, r_node); + + dump_request(s, req); + } + + mutex_unlock(&osd->lock); +} - for (i = 0; i < req->r_num_ops; i++) { - opcode = req->r_ops[i].op; - seq_printf(s, "%s%s", (i == 0 ? "\t" : ","), - ceph_osd_op_name(opcode)); - } +static void dump_linger_request(struct seq_file *s, + struct ceph_osd_linger_request *lreq) +{ + seq_printf(s, "%llu\t", lreq->linger_id); + dump_target(s, &lreq->t); + + seq_printf(s, "\t%u\t%s%s/%d\n", lreq->register_gen, + lreq->is_watch ? "W" : "N", lreq->committed ? "C" : "", + lreq->last_error); +} + +static void dump_linger_requests(struct seq_file *s, struct ceph_osd *osd) +{ + struct rb_node *n; + + mutex_lock(&osd->lock); + for (n = rb_first(&osd->o_linger_requests); n; n = rb_next(n)) { + struct ceph_osd_linger_request *lreq = + rb_entry(n, struct ceph_osd_linger_request, node); + + dump_linger_request(s, lreq); + } + + mutex_unlock(&osd->lock); +} - seq_printf(s, "\n"); +static int osdc_show(struct seq_file *s, void *pp) +{ + struct ceph_client *client = s->private; + struct ceph_osd_client *osdc = &client->osdc; + struct rb_node *n; + + down_read(&osdc->lock); + seq_printf(s, "REQUESTS %d homeless %d\n", + atomic_read(&osdc->num_requests), + atomic_read(&osdc->num_homeless)); + for (n = rb_first(&osdc->osds); n; n = rb_next(n)) { + struct ceph_osd *osd = rb_entry(n, struct ceph_osd, o_node); + + dump_requests(s, osd); } - mutex_unlock(&osdc->request_mutex); + dump_requests(s, &osdc->homeless_osd); + + seq_puts(s, "LINGER REQUESTS\n"); + for (n = rb_first(&osdc->osds); n; n = rb_next(n)) { + struct ceph_osd *osd = rb_entry(n, struct ceph_osd, o_node); + + dump_linger_requests(s, osd); + } + dump_linger_requests(s, &osdc->homeless_osd); + + up_read(&osdc->lock); return 0; } diff --git a/net/ceph/mon_client.c b/net/ceph/mon_client.c index cf638c0..37c38a7 100644 --- a/net/ceph/mon_client.c +++ b/net/ceph/mon_client.c @@ -260,20 +260,26 @@ static void __send_subscribe(struct ceph_mon_client *monc) BUG_ON(num < 1); /* monmap sub is always there */ ceph_encode_32(&p, num); for (i = 0; i < ARRAY_SIZE(monc->subs); i++) { - const char *s = ceph_sub_str[i]; + char buf[32]; + int len; if (!monc->subs[i].want) continue; - dout("%s %s start %llu flags 0x%x\n", __func__, s, + len = sprintf(buf, "%s", ceph_sub_str[i]); + if (i == CEPH_SUB_MDSMAP && + monc->fs_cluster_id != CEPH_FS_CLUSTER_ID_NONE) + len += sprintf(buf + len, ".%d", monc->fs_cluster_id); + + dout("%s %s start %llu flags 0x%x\n", __func__, buf, le64_to_cpu(monc->subs[i].item.start), monc->subs[i].item.flags); - ceph_encode_string(&p, end, s, strlen(s)); + ceph_encode_string(&p, end, buf, len); memcpy(p, &monc->subs[i].item, sizeof(monc->subs[i].item)); p += sizeof(monc->subs[i].item); } - BUG_ON(p != (end - 35 - (ARRAY_SIZE(monc->subs) - num) * 19)); + BUG_ON(p > end); msg->front.iov_len = p - msg->front.iov_base; msg->hdr.front_len = cpu_to_le32(msg->front.iov_len); ceph_msg_revoke(msg); @@ -376,19 +382,13 @@ void ceph_monc_got_map(struct ceph_mon_client *monc, int sub, u32 epoch) } EXPORT_SYMBOL(ceph_monc_got_map); -/* - * Register interest in the next osdmap - */ -void ceph_monc_request_next_osdmap(struct ceph_mon_client *monc) +void ceph_monc_renew_subs(struct ceph_mon_client *monc) { - dout("%s have %u\n", __func__, monc->subs[CEPH_SUB_OSDMAP].have); mutex_lock(&monc->mutex); - if (__ceph_monc_want_map(monc, CEPH_SUB_OSDMAP, - monc->subs[CEPH_SUB_OSDMAP].have + 1, false)) - __send_subscribe(monc); + __send_subscribe(monc); mutex_unlock(&monc->mutex); } -EXPORT_SYMBOL(ceph_monc_request_next_osdmap); +EXPORT_SYMBOL(ceph_monc_renew_subs); /* * Wait for an osdmap with a given epoch. @@ -478,51 +478,17 @@ out: /* * generic requests (currently statfs, mon_get_version) */ -static struct ceph_mon_generic_request *__lookup_generic_req( - struct ceph_mon_client *monc, u64 tid) -{ - struct ceph_mon_generic_request *req; - struct rb_node *n = monc->generic_request_tree.rb_node; - - while (n) { - req = rb_entry(n, struct ceph_mon_generic_request, node); - if (tid < req->tid) - n = n->rb_left; - else if (tid > req->tid) - n = n->rb_right; - else - return req; - } - return NULL; -} - -static void __insert_generic_request(struct ceph_mon_client *monc, - struct ceph_mon_generic_request *new) -{ - struct rb_node **p = &monc->generic_request_tree.rb_node; - struct rb_node *parent = NULL; - struct ceph_mon_generic_request *req = NULL; - - while (*p) { - parent = *p; - req = rb_entry(parent, struct ceph_mon_generic_request, node); - if (new->tid < req->tid) - p = &(*p)->rb_left; - else if (new->tid > req->tid) - p = &(*p)->rb_right; - else - BUG(); - } - - rb_link_node(&new->node, parent, p); - rb_insert_color(&new->node, &monc->generic_request_tree); -} +DEFINE_RB_FUNCS(generic_request, struct ceph_mon_generic_request, tid, node) static void release_generic_request(struct kref *kref) { struct ceph_mon_generic_request *req = container_of(kref, struct ceph_mon_generic_request, kref); + dout("%s greq %p request %p reply %p\n", __func__, req, req->request, + req->reply); + WARN_ON(!RB_EMPTY_NODE(&req->node)); + if (req->reply) ceph_msg_put(req->reply); if (req->request) @@ -533,7 +499,8 @@ static void release_generic_request(struct kref *kref) static void put_generic_request(struct ceph_mon_generic_request *req) { - kref_put(&req->kref, release_generic_request); + if (req) + kref_put(&req->kref, release_generic_request); } static void get_generic_request(struct ceph_mon_generic_request *req) @@ -541,6 +508,103 @@ static void get_generic_request(struct ceph_mon_generic_request *req) kref_get(&req->kref); } +static struct ceph_mon_generic_request * +alloc_generic_request(struct ceph_mon_client *monc, gfp_t gfp) +{ + struct ceph_mon_generic_request *req; + + req = kzalloc(sizeof(*req), gfp); + if (!req) + return NULL; + + req->monc = monc; + kref_init(&req->kref); + RB_CLEAR_NODE(&req->node); + init_completion(&req->completion); + + dout("%s greq %p\n", __func__, req); + return req; +} + +static void register_generic_request(struct ceph_mon_generic_request *req) +{ + struct ceph_mon_client *monc = req->monc; + + WARN_ON(req->tid); + + get_generic_request(req); + req->tid = ++monc->last_tid; + insert_generic_request(&monc->generic_request_tree, req); +} + +static void send_generic_request(struct ceph_mon_client *monc, + struct ceph_mon_generic_request *req) +{ + WARN_ON(!req->tid); + + dout("%s greq %p tid %llu\n", __func__, req, req->tid); + req->request->hdr.tid = cpu_to_le64(req->tid); + ceph_con_send(&monc->con, ceph_msg_get(req->request)); +} + +static void __finish_generic_request(struct ceph_mon_generic_request *req) +{ + struct ceph_mon_client *monc = req->monc; + + dout("%s greq %p tid %llu\n", __func__, req, req->tid); + erase_generic_request(&monc->generic_request_tree, req); + + ceph_msg_revoke(req->request); + ceph_msg_revoke_incoming(req->reply); +} + +static void finish_generic_request(struct ceph_mon_generic_request *req) +{ + __finish_generic_request(req); + put_generic_request(req); +} + +static void complete_generic_request(struct ceph_mon_generic_request *req) +{ + if (req->complete_cb) + req->complete_cb(req); + else + complete_all(&req->completion); + put_generic_request(req); +} + +void cancel_generic_request(struct ceph_mon_generic_request *req) +{ + struct ceph_mon_client *monc = req->monc; + struct ceph_mon_generic_request *lookup_req; + + dout("%s greq %p tid %llu\n", __func__, req, req->tid); + + mutex_lock(&monc->mutex); + lookup_req = lookup_generic_request(&monc->generic_request_tree, + req->tid); + if (lookup_req) { + WARN_ON(lookup_req != req); + finish_generic_request(req); + } + + mutex_unlock(&monc->mutex); +} + +static int wait_generic_request(struct ceph_mon_generic_request *req) +{ + int ret; + + dout("%s greq %p tid %llu\n", __func__, req, req->tid); + ret = wait_for_completion_interruptible(&req->completion); + if (ret) + cancel_generic_request(req); + else + ret = req->result; /* completed */ + + return ret; +} + static struct ceph_msg *get_generic_reply(struct ceph_connection *con, struct ceph_msg_header *hdr, int *skip) @@ -551,7 +615,7 @@ static struct ceph_msg *get_generic_reply(struct ceph_connection *con, struct ceph_msg *m; mutex_lock(&monc->mutex); - req = __lookup_generic_req(monc, tid); + req = lookup_generic_request(&monc->generic_request_tree, tid); if (!req) { dout("get_generic_reply %lld dne\n", tid); *skip = 1; @@ -570,42 +634,6 @@ static struct ceph_msg *get_generic_reply(struct ceph_connection *con, return m; } -static int __do_generic_request(struct ceph_mon_client *monc, u64 tid, - struct ceph_mon_generic_request *req) -{ - int err; - - /* register request */ - req->tid = tid != 0 ? tid : ++monc->last_tid; - req->request->hdr.tid = cpu_to_le64(req->tid); - __insert_generic_request(monc, req); - monc->num_generic_requests++; - ceph_con_send(&monc->con, ceph_msg_get(req->request)); - mutex_unlock(&monc->mutex); - - err = wait_for_completion_interruptible(&req->completion); - - mutex_lock(&monc->mutex); - rb_erase(&req->node, &monc->generic_request_tree); - monc->num_generic_requests--; - - if (!err) - err = req->result; - return err; -} - -static int do_generic_request(struct ceph_mon_client *monc, - struct ceph_mon_generic_request *req) -{ - int err; - - mutex_lock(&monc->mutex); - err = __do_generic_request(monc, 0, req); - mutex_unlock(&monc->mutex); - - return err; -} - /* * statfs */ @@ -616,22 +644,24 @@ static void handle_statfs_reply(struct ceph_mon_client *monc, struct ceph_mon_statfs_reply *reply = msg->front.iov_base; u64 tid = le64_to_cpu(msg->hdr.tid); + dout("%s msg %p tid %llu\n", __func__, msg, tid); + if (msg->front.iov_len != sizeof(*reply)) goto bad; - dout("handle_statfs_reply %p tid %llu\n", msg, tid); mutex_lock(&monc->mutex); - req = __lookup_generic_req(monc, tid); - if (req) { - *(struct ceph_statfs *)req->buf = reply->st; - req->result = 0; - get_generic_request(req); + req = lookup_generic_request(&monc->generic_request_tree, tid); + if (!req) { + mutex_unlock(&monc->mutex); + return; } + + req->result = 0; + *req->u.st = reply->st; /* struct */ + __finish_generic_request(req); mutex_unlock(&monc->mutex); - if (req) { - complete_all(&req->completion); - put_generic_request(req); - } + + complete_generic_request(req); return; bad: @@ -646,38 +676,38 @@ int ceph_monc_do_statfs(struct ceph_mon_client *monc, struct ceph_statfs *buf) { struct ceph_mon_generic_request *req; struct ceph_mon_statfs *h; - int err; + int ret = -ENOMEM; - req = kzalloc(sizeof(*req), GFP_NOFS); + req = alloc_generic_request(monc, GFP_NOFS); if (!req) - return -ENOMEM; - - kref_init(&req->kref); - req->buf = buf; - init_completion(&req->completion); + goto out; - err = -ENOMEM; req->request = ceph_msg_new(CEPH_MSG_STATFS, sizeof(*h), GFP_NOFS, true); if (!req->request) goto out; - req->reply = ceph_msg_new(CEPH_MSG_STATFS_REPLY, 1024, GFP_NOFS, - true); + + req->reply = ceph_msg_new(CEPH_MSG_STATFS_REPLY, 64, GFP_NOFS, true); if (!req->reply) goto out; + req->u.st = buf; + + mutex_lock(&monc->mutex); + register_generic_request(req); /* fill out request */ h = req->request->front.iov_base; h->monhdr.have_version = 0; h->monhdr.session_mon = cpu_to_le16(-1); h->monhdr.session_mon_tid = 0; h->fsid = monc->monmap->fsid; + send_generic_request(monc, req); + mutex_unlock(&monc->mutex); - err = do_generic_request(monc, req); - + ret = wait_generic_request(req); out: put_generic_request(req); - return err; + return ret; } EXPORT_SYMBOL(ceph_monc_do_statfs); @@ -690,7 +720,7 @@ static void handle_get_version_reply(struct ceph_mon_client *monc, void *end = p + msg->front_alloc_len; u64 handle; - dout("%s %p tid %llu\n", __func__, msg, tid); + dout("%s msg %p tid %llu\n", __func__, msg, tid); ceph_decode_need(&p, end, 2*sizeof(u64), bad); handle = ceph_decode_64(&p); @@ -698,77 +728,111 @@ static void handle_get_version_reply(struct ceph_mon_client *monc, goto bad; mutex_lock(&monc->mutex); - req = __lookup_generic_req(monc, handle); - if (req) { - *(u64 *)req->buf = ceph_decode_64(&p); - req->result = 0; - get_generic_request(req); + req = lookup_generic_request(&monc->generic_request_tree, handle); + if (!req) { + mutex_unlock(&monc->mutex); + return; } + + req->result = 0; + req->u.newest = ceph_decode_64(&p); + __finish_generic_request(req); mutex_unlock(&monc->mutex); - if (req) { - complete_all(&req->completion); - put_generic_request(req); - } + complete_generic_request(req); return; + bad: pr_err("corrupt mon_get_version reply, tid %llu\n", tid); ceph_msg_dump(msg); } -/* - * Send MMonGetVersion and wait for the reply. - * - * @what: one of "mdsmap", "osdmap" or "monmap" - */ -int ceph_monc_do_get_version(struct ceph_mon_client *monc, const char *what, - u64 *newest) +static struct ceph_mon_generic_request * +__ceph_monc_get_version(struct ceph_mon_client *monc, const char *what, + ceph_monc_callback_t cb, u64 private_data) { struct ceph_mon_generic_request *req; - void *p, *end; - u64 tid; - int err; - req = kzalloc(sizeof(*req), GFP_NOFS); + req = alloc_generic_request(monc, GFP_NOIO); if (!req) - return -ENOMEM; - - kref_init(&req->kref); - req->buf = newest; - init_completion(&req->completion); + goto err_put_req; req->request = ceph_msg_new(CEPH_MSG_MON_GET_VERSION, sizeof(u64) + sizeof(u32) + strlen(what), - GFP_NOFS, true); - if (!req->request) { - err = -ENOMEM; - goto out; - } + GFP_NOIO, true); + if (!req->request) + goto err_put_req; - req->reply = ceph_msg_new(CEPH_MSG_MON_GET_VERSION_REPLY, 1024, - GFP_NOFS, true); - if (!req->reply) { - err = -ENOMEM; - goto out; - } + req->reply = ceph_msg_new(CEPH_MSG_MON_GET_VERSION_REPLY, 32, GFP_NOIO, + true); + if (!req->reply) + goto err_put_req; - p = req->request->front.iov_base; - end = p + req->request->front_alloc_len; + req->complete_cb = cb; + req->private_data = private_data; - /* fill out request */ mutex_lock(&monc->mutex); - tid = ++monc->last_tid; - ceph_encode_64(&p, tid); /* handle */ - ceph_encode_string(&p, end, what, strlen(what)); + register_generic_request(req); + { + void *p = req->request->front.iov_base; + void *const end = p + req->request->front_alloc_len; + + ceph_encode_64(&p, req->tid); /* handle */ + ceph_encode_string(&p, end, what, strlen(what)); + WARN_ON(p != end); + } + send_generic_request(monc, req); + mutex_unlock(&monc->mutex); - err = __do_generic_request(monc, tid, req); + return req; - mutex_unlock(&monc->mutex); -out: +err_put_req: put_generic_request(req); - return err; + return ERR_PTR(-ENOMEM); +} + +/* + * Send MMonGetVersion and wait for the reply. + * + * @what: one of "mdsmap", "osdmap" or "monmap" + */ +int ceph_monc_get_version(struct ceph_mon_client *monc, const char *what, + u64 *newest) +{ + struct ceph_mon_generic_request *req; + int ret; + + req = __ceph_monc_get_version(monc, what, NULL, 0); + if (IS_ERR(req)) + return PTR_ERR(req); + + ret = wait_generic_request(req); + if (!ret) + *newest = req->u.newest; + + put_generic_request(req); + return ret; } -EXPORT_SYMBOL(ceph_monc_do_get_version); +EXPORT_SYMBOL(ceph_monc_get_version); + +/* + * Send MMonGetVersion, + * + * @what: one of "mdsmap", "osdmap" or "monmap" + */ +int ceph_monc_get_version_async(struct ceph_mon_client *monc, const char *what, + ceph_monc_callback_t cb, u64 private_data) +{ + struct ceph_mon_generic_request *req; + + req = __ceph_monc_get_version(monc, what, cb, private_data); + if (IS_ERR(req)) + return PTR_ERR(req); + + put_generic_request(req); + return 0; +} +EXPORT_SYMBOL(ceph_monc_get_version_async); /* * Resend pending generic requests. @@ -890,7 +954,7 @@ int ceph_monc_init(struct ceph_mon_client *monc, struct ceph_client *cl) if (!monc->m_subscribe_ack) goto out_auth; - monc->m_subscribe = ceph_msg_new(CEPH_MSG_MON_SUBSCRIBE, 96, GFP_NOFS, + monc->m_subscribe = ceph_msg_new(CEPH_MSG_MON_SUBSCRIBE, 128, GFP_NOFS, true); if (!monc->m_subscribe) goto out_subscribe_ack; @@ -914,9 +978,10 @@ int ceph_monc_init(struct ceph_mon_client *monc, struct ceph_client *cl) INIT_DELAYED_WORK(&monc->delayed_work, delayed_work); monc->generic_request_tree = RB_ROOT; - monc->num_generic_requests = 0; monc->last_tid = 0; + monc->fs_cluster_id = CEPH_FS_CLUSTER_ID_NONE; + return 0; out_auth_reply: @@ -954,6 +1019,8 @@ void ceph_monc_stop(struct ceph_mon_client *monc) ceph_auth_destroy(monc->auth); + WARN_ON(!RB_EMPTY_ROOT(&monc->generic_request_tree)); + ceph_msg_put(monc->m_auth); ceph_msg_put(monc->m_auth_reply); ceph_msg_put(monc->m_subscribe); diff --git a/net/ceph/osd_client.c b/net/ceph/osd_client.c index 40a53a7..8946959 100644 --- a/net/ceph/osd_client.c +++ b/net/ceph/osd_client.c @@ -19,25 +19,12 @@ #include <linux/ceph/auth.h> #include <linux/ceph/pagelist.h> -#define OSD_OP_FRONT_LEN 4096 #define OSD_OPREPLY_FRONT_LEN 512 static struct kmem_cache *ceph_osd_request_cache; static const struct ceph_connection_operations osd_con_ops; -static void __send_queued(struct ceph_osd_client *osdc); -static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd); -static void __register_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req); -static void __unregister_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req); -static void __unregister_linger_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req); -static void __enqueue_request(struct ceph_osd_request *req); -static void __send_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req); - /* * Implement client access to distributed object storage cluster. * @@ -56,6 +43,52 @@ static void __send_request(struct ceph_osd_client *osdc, * channel with an OSD is reset. */ +static void link_request(struct ceph_osd *osd, struct ceph_osd_request *req); +static void unlink_request(struct ceph_osd *osd, struct ceph_osd_request *req); +static void link_linger(struct ceph_osd *osd, + struct ceph_osd_linger_request *lreq); +static void unlink_linger(struct ceph_osd *osd, + struct ceph_osd_linger_request *lreq); + +#if 1 +static inline bool rwsem_is_wrlocked(struct rw_semaphore *sem) +{ + bool wrlocked = true; + + if (unlikely(down_read_trylock(sem))) { + wrlocked = false; + up_read(sem); + } + + return wrlocked; +} +static inline void verify_osdc_locked(struct ceph_osd_client *osdc) +{ + WARN_ON(!rwsem_is_locked(&osdc->lock)); +} +static inline void verify_osdc_wrlocked(struct ceph_osd_client *osdc) +{ + WARN_ON(!rwsem_is_wrlocked(&osdc->lock)); +} +static inline void verify_osd_locked(struct ceph_osd *osd) +{ + struct ceph_osd_client *osdc = osd->o_osdc; + + WARN_ON(!(mutex_is_locked(&osd->lock) && + rwsem_is_locked(&osdc->lock)) && + !rwsem_is_wrlocked(&osdc->lock)); +} +static inline void verify_lreq_locked(struct ceph_osd_linger_request *lreq) +{ + WARN_ON(!mutex_is_locked(&lreq->lock)); +} +#else +static inline void verify_osdc_locked(struct ceph_osd_client *osdc) { } +static inline void verify_osdc_wrlocked(struct ceph_osd_client *osdc) { } +static inline void verify_osd_locked(struct ceph_osd *osd) { } +static inline void verify_lreq_locked(struct ceph_osd_linger_request *lreq) { } +#endif + /* * calculate the mapping of a file extent onto an object, and fill out the * request accordingly. shorten extent as necessary if it crosses an @@ -144,14 +177,6 @@ osd_req_op_extent_osd_data(struct ceph_osd_request *osd_req, } EXPORT_SYMBOL(osd_req_op_extent_osd_data); -struct ceph_osd_data * -osd_req_op_cls_response_data(struct ceph_osd_request *osd_req, - unsigned int which) -{ - return osd_req_op_data(osd_req, which, cls, response_data); -} -EXPORT_SYMBOL(osd_req_op_cls_response_data); /* ??? */ - void osd_req_op_raw_data_in_pages(struct ceph_osd_request *osd_req, unsigned int which, struct page **pages, u64 length, u32 alignment, @@ -218,6 +243,8 @@ void osd_req_op_cls_request_data_pagelist( osd_data = osd_req_op_data(osd_req, which, cls, request_data); ceph_osd_data_pagelist_init(osd_data, pagelist); + osd_req->r_ops[which].cls.indata_len += pagelist->length; + osd_req->r_ops[which].indata_len += pagelist->length; } EXPORT_SYMBOL(osd_req_op_cls_request_data_pagelist); @@ -230,6 +257,8 @@ void osd_req_op_cls_request_data_pages(struct ceph_osd_request *osd_req, osd_data = osd_req_op_data(osd_req, which, cls, request_data); ceph_osd_data_pages_init(osd_data, pages, length, alignment, pages_from_pool, own_pages); + osd_req->r_ops[which].cls.indata_len += length; + osd_req->r_ops[which].indata_len += length; } EXPORT_SYMBOL(osd_req_op_cls_request_data_pages); @@ -302,14 +331,76 @@ static void osd_req_op_data_release(struct ceph_osd_request *osd_req, case CEPH_OSD_OP_STAT: ceph_osd_data_release(&op->raw_data_in); break; + case CEPH_OSD_OP_NOTIFY_ACK: + ceph_osd_data_release(&op->notify_ack.request_data); + break; + case CEPH_OSD_OP_NOTIFY: + ceph_osd_data_release(&op->notify.request_data); + ceph_osd_data_release(&op->notify.response_data); + break; default: break; } } /* + * Assumes @t is zero-initialized. + */ +static void target_init(struct ceph_osd_request_target *t) +{ + ceph_oid_init(&t->base_oid); + ceph_oloc_init(&t->base_oloc); + ceph_oid_init(&t->target_oid); + ceph_oloc_init(&t->target_oloc); + + ceph_osds_init(&t->acting); + ceph_osds_init(&t->up); + t->size = -1; + t->min_size = -1; + + t->osd = CEPH_HOMELESS_OSD; +} + +static void target_copy(struct ceph_osd_request_target *dest, + const struct ceph_osd_request_target *src) +{ + ceph_oid_copy(&dest->base_oid, &src->base_oid); + ceph_oloc_copy(&dest->base_oloc, &src->base_oloc); + ceph_oid_copy(&dest->target_oid, &src->target_oid); + ceph_oloc_copy(&dest->target_oloc, &src->target_oloc); + + dest->pgid = src->pgid; /* struct */ + dest->pg_num = src->pg_num; + dest->pg_num_mask = src->pg_num_mask; + ceph_osds_copy(&dest->acting, &src->acting); + ceph_osds_copy(&dest->up, &src->up); + dest->size = src->size; + dest->min_size = src->min_size; + dest->sort_bitwise = src->sort_bitwise; + + dest->flags = src->flags; + dest->paused = src->paused; + + dest->osd = src->osd; +} + +static void target_destroy(struct ceph_osd_request_target *t) +{ + ceph_oid_destroy(&t->base_oid); + ceph_oid_destroy(&t->target_oid); +} + +/* * requests */ +static void request_release_checks(struct ceph_osd_request *req) +{ + WARN_ON(!RB_EMPTY_NODE(&req->r_node)); + WARN_ON(!RB_EMPTY_NODE(&req->r_mc_node)); + WARN_ON(!list_empty(&req->r_unsafe_item)); + WARN_ON(req->r_osd); +} + static void ceph_osdc_release_request(struct kref *kref) { struct ceph_osd_request *req = container_of(kref, @@ -318,24 +409,19 @@ static void ceph_osdc_release_request(struct kref *kref) dout("%s %p (r_request %p r_reply %p)\n", __func__, req, req->r_request, req->r_reply); - WARN_ON(!RB_EMPTY_NODE(&req->r_node)); - WARN_ON(!list_empty(&req->r_req_lru_item)); - WARN_ON(!list_empty(&req->r_osd_item)); - WARN_ON(!list_empty(&req->r_linger_item)); - WARN_ON(!list_empty(&req->r_linger_osd_item)); - WARN_ON(req->r_osd); + request_release_checks(req); if (req->r_request) ceph_msg_put(req->r_request); - if (req->r_reply) { - ceph_msg_revoke_incoming(req->r_reply); + if (req->r_reply) ceph_msg_put(req->r_reply); - } for (which = 0; which < req->r_num_ops; which++) osd_req_op_data_release(req, which); + target_destroy(&req->r_t); ceph_put_snap_context(req->r_snapc); + if (req->r_mempool) mempool_free(req, req->r_osdc->req_mempool); else if (req->r_num_ops <= CEPH_OSD_SLAB_OPS) @@ -354,12 +440,66 @@ EXPORT_SYMBOL(ceph_osdc_get_request); void ceph_osdc_put_request(struct ceph_osd_request *req) { - dout("%s %p (was %d)\n", __func__, req, - atomic_read(&req->r_kref.refcount)); - kref_put(&req->r_kref, ceph_osdc_release_request); + if (req) { + dout("%s %p (was %d)\n", __func__, req, + atomic_read(&req->r_kref.refcount)); + kref_put(&req->r_kref, ceph_osdc_release_request); + } } EXPORT_SYMBOL(ceph_osdc_put_request); +static void request_init(struct ceph_osd_request *req) +{ + /* req only, each op is zeroed in _osd_req_op_init() */ + memset(req, 0, sizeof(*req)); + + kref_init(&req->r_kref); + init_completion(&req->r_completion); + init_completion(&req->r_safe_completion); + RB_CLEAR_NODE(&req->r_node); + RB_CLEAR_NODE(&req->r_mc_node); + INIT_LIST_HEAD(&req->r_unsafe_item); + + target_init(&req->r_t); +} + +/* + * This is ugly, but it allows us to reuse linger registration and ping + * requests, keeping the structure of the code around send_linger{_ping}() + * reasonable. Setting up a min_nr=2 mempool for each linger request + * and dealing with copying ops (this blasts req only, watch op remains + * intact) isn't any better. + */ +static void request_reinit(struct ceph_osd_request *req) +{ + struct ceph_osd_client *osdc = req->r_osdc; + bool mempool = req->r_mempool; + unsigned int num_ops = req->r_num_ops; + u64 snapid = req->r_snapid; + struct ceph_snap_context *snapc = req->r_snapc; + bool linger = req->r_linger; + struct ceph_msg *request_msg = req->r_request; + struct ceph_msg *reply_msg = req->r_reply; + + dout("%s req %p\n", __func__, req); + WARN_ON(atomic_read(&req->r_kref.refcount) != 1); + request_release_checks(req); + + WARN_ON(atomic_read(&request_msg->kref.refcount) != 1); + WARN_ON(atomic_read(&reply_msg->kref.refcount) != 1); + target_destroy(&req->r_t); + + request_init(req); + req->r_osdc = osdc; + req->r_mempool = mempool; + req->r_num_ops = num_ops; + req->r_snapid = snapid; + req->r_snapc = snapc; + req->r_linger = linger; + req->r_request = request_msg; + req->r_reply = reply_msg; +} + struct ceph_osd_request *ceph_osdc_alloc_request(struct ceph_osd_client *osdc, struct ceph_snap_context *snapc, unsigned int num_ops, @@ -367,8 +507,6 @@ struct ceph_osd_request *ceph_osdc_alloc_request(struct ceph_osd_client *osdc, gfp_t gfp_flags) { struct ceph_osd_request *req; - struct ceph_msg *msg; - size_t msg_size; if (use_mempool) { BUG_ON(num_ops > CEPH_OSD_SLAB_OPS); @@ -383,73 +521,65 @@ struct ceph_osd_request *ceph_osdc_alloc_request(struct ceph_osd_client *osdc, if (unlikely(!req)) return NULL; - /* req only, each op is zeroed in _osd_req_op_init() */ - memset(req, 0, sizeof(*req)); - + request_init(req); req->r_osdc = osdc; req->r_mempool = use_mempool; req->r_num_ops = num_ops; + req->r_snapid = CEPH_NOSNAP; + req->r_snapc = ceph_get_snap_context(snapc); - kref_init(&req->r_kref); - init_completion(&req->r_completion); - init_completion(&req->r_safe_completion); - RB_CLEAR_NODE(&req->r_node); - INIT_LIST_HEAD(&req->r_unsafe_item); - INIT_LIST_HEAD(&req->r_linger_item); - INIT_LIST_HEAD(&req->r_linger_osd_item); - INIT_LIST_HEAD(&req->r_req_lru_item); - INIT_LIST_HEAD(&req->r_osd_item); - - req->r_base_oloc.pool = -1; - req->r_target_oloc.pool = -1; + dout("%s req %p\n", __func__, req); + return req; +} +EXPORT_SYMBOL(ceph_osdc_alloc_request); - msg_size = OSD_OPREPLY_FRONT_LEN; - if (num_ops > CEPH_OSD_SLAB_OPS) { - /* ceph_osd_op and rval */ - msg_size += (num_ops - CEPH_OSD_SLAB_OPS) * - (sizeof(struct ceph_osd_op) + 4); - } +int ceph_osdc_alloc_messages(struct ceph_osd_request *req, gfp_t gfp) +{ + struct ceph_osd_client *osdc = req->r_osdc; + struct ceph_msg *msg; + int msg_size; - /* create reply message */ - if (use_mempool) - msg = ceph_msgpool_get(&osdc->msgpool_op_reply, 0); - else - msg = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, msg_size, - gfp_flags, true); - if (!msg) { - ceph_osdc_put_request(req); - return NULL; - } - req->r_reply = msg; + WARN_ON(ceph_oid_empty(&req->r_base_oid)); + /* create request message */ msg_size = 4 + 4 + 4; /* client_inc, osdmap_epoch, flags */ msg_size += 4 + 4 + 4 + 8; /* mtime, reassert_version */ msg_size += 2 + 4 + 8 + 4 + 4; /* oloc */ msg_size += 1 + 8 + 4 + 4; /* pgid */ - msg_size += 4 + CEPH_MAX_OID_NAME_LEN; /* oid */ - msg_size += 2 + num_ops * sizeof(struct ceph_osd_op); + msg_size += 4 + req->r_base_oid.name_len; /* oid */ + msg_size += 2 + req->r_num_ops * sizeof(struct ceph_osd_op); msg_size += 8; /* snapid */ msg_size += 8; /* snap_seq */ - msg_size += 4 + 8 * (snapc ? snapc->num_snaps : 0); /* snaps */ + msg_size += 4 + 8 * (req->r_snapc ? req->r_snapc->num_snaps : 0); msg_size += 4; /* retry_attempt */ - /* create request message; allow space for oid */ - if (use_mempool) + if (req->r_mempool) msg = ceph_msgpool_get(&osdc->msgpool_op, 0); else - msg = ceph_msg_new(CEPH_MSG_OSD_OP, msg_size, gfp_flags, true); - if (!msg) { - ceph_osdc_put_request(req); - return NULL; - } + msg = ceph_msg_new(CEPH_MSG_OSD_OP, msg_size, gfp, true); + if (!msg) + return -ENOMEM; memset(msg->front.iov_base, 0, msg->front.iov_len); - req->r_request = msg; - return req; + /* create reply message */ + msg_size = OSD_OPREPLY_FRONT_LEN; + msg_size += req->r_base_oid.name_len; + msg_size += req->r_num_ops * sizeof(struct ceph_osd_op); + + if (req->r_mempool) + msg = ceph_msgpool_get(&osdc->msgpool_op_reply, 0); + else + msg = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, msg_size, gfp, true); + if (!msg) + return -ENOMEM; + + req->r_reply = msg; + + return 0; } -EXPORT_SYMBOL(ceph_osdc_alloc_request); +EXPORT_SYMBOL(ceph_osdc_alloc_messages); static bool osd_req_opcode_valid(u16 opcode) { @@ -587,8 +717,6 @@ void osd_req_op_cls_init(struct ceph_osd_request *osd_req, unsigned int which, osd_req_op_cls_request_info_pagelist(osd_req, which, pagelist); - op->cls.argc = 0; /* currently unused */ - op->indata_len = payload_len; } EXPORT_SYMBOL(osd_req_op_cls_init); @@ -627,21 +755,19 @@ int osd_req_op_xattr_init(struct ceph_osd_request *osd_req, unsigned int which, } EXPORT_SYMBOL(osd_req_op_xattr_init); -void osd_req_op_watch_init(struct ceph_osd_request *osd_req, - unsigned int which, u16 opcode, - u64 cookie, u64 version, int flag) +/* + * @watch_opcode: CEPH_OSD_WATCH_OP_* + */ +static void osd_req_op_watch_init(struct ceph_osd_request *req, int which, + u64 cookie, u8 watch_opcode) { - struct ceph_osd_req_op *op = _osd_req_op_init(osd_req, which, - opcode, 0); - - BUG_ON(opcode != CEPH_OSD_OP_NOTIFY_ACK && opcode != CEPH_OSD_OP_WATCH); + struct ceph_osd_req_op *op; + op = _osd_req_op_init(req, which, CEPH_OSD_OP_WATCH, 0); op->watch.cookie = cookie; - op->watch.ver = version; - if (opcode == CEPH_OSD_OP_WATCH && flag) - op->watch.flag = (u8)1; + op->watch.op = watch_opcode; + op->watch.gen = 0; } -EXPORT_SYMBOL(osd_req_op_watch_init); void osd_req_op_alloc_hint_init(struct ceph_osd_request *osd_req, unsigned int which, @@ -686,16 +812,9 @@ static void ceph_osdc_msg_data_add(struct ceph_msg *msg, } } -static u64 osd_req_encode_op(struct ceph_osd_request *req, - struct ceph_osd_op *dst, unsigned int which) +static u32 osd_req_encode_op(struct ceph_osd_op *dst, + const struct ceph_osd_req_op *src) { - struct ceph_osd_req_op *src; - struct ceph_osd_data *osd_data; - u64 request_data_len = 0; - u64 data_length; - - BUG_ON(which >= req->r_num_ops); - src = &req->r_ops[which]; if (WARN_ON(!osd_req_opcode_valid(src->op))) { pr_err("unrecognized osd opcode %d\n", src->op); @@ -704,57 +823,36 @@ static u64 osd_req_encode_op(struct ceph_osd_request *req, switch (src->op) { case CEPH_OSD_OP_STAT: - osd_data = &src->raw_data_in; - ceph_osdc_msg_data_add(req->r_reply, osd_data); break; case CEPH_OSD_OP_READ: case CEPH_OSD_OP_WRITE: case CEPH_OSD_OP_WRITEFULL: case CEPH_OSD_OP_ZERO: case CEPH_OSD_OP_TRUNCATE: - if (src->op == CEPH_OSD_OP_WRITE || - src->op == CEPH_OSD_OP_WRITEFULL) - request_data_len = src->extent.length; dst->extent.offset = cpu_to_le64(src->extent.offset); dst->extent.length = cpu_to_le64(src->extent.length); dst->extent.truncate_size = cpu_to_le64(src->extent.truncate_size); dst->extent.truncate_seq = cpu_to_le32(src->extent.truncate_seq); - osd_data = &src->extent.osd_data; - if (src->op == CEPH_OSD_OP_WRITE || - src->op == CEPH_OSD_OP_WRITEFULL) - ceph_osdc_msg_data_add(req->r_request, osd_data); - else - ceph_osdc_msg_data_add(req->r_reply, osd_data); break; case CEPH_OSD_OP_CALL: dst->cls.class_len = src->cls.class_len; dst->cls.method_len = src->cls.method_len; - osd_data = &src->cls.request_info; - ceph_osdc_msg_data_add(req->r_request, osd_data); - BUG_ON(osd_data->type != CEPH_OSD_DATA_TYPE_PAGELIST); - request_data_len = osd_data->pagelist->length; - - osd_data = &src->cls.request_data; - data_length = ceph_osd_data_length(osd_data); - if (data_length) { - BUG_ON(osd_data->type == CEPH_OSD_DATA_TYPE_NONE); - dst->cls.indata_len = cpu_to_le32(data_length); - ceph_osdc_msg_data_add(req->r_request, osd_data); - src->indata_len += data_length; - request_data_len += data_length; - } - osd_data = &src->cls.response_data; - ceph_osdc_msg_data_add(req->r_reply, osd_data); + dst->cls.indata_len = cpu_to_le32(src->cls.indata_len); break; case CEPH_OSD_OP_STARTSYNC: break; - case CEPH_OSD_OP_NOTIFY_ACK: case CEPH_OSD_OP_WATCH: dst->watch.cookie = cpu_to_le64(src->watch.cookie); - dst->watch.ver = cpu_to_le64(src->watch.ver); - dst->watch.flag = src->watch.flag; + dst->watch.ver = cpu_to_le64(0); + dst->watch.op = src->watch.op; + dst->watch.gen = cpu_to_le32(src->watch.gen); + break; + case CEPH_OSD_OP_NOTIFY_ACK: + break; + case CEPH_OSD_OP_NOTIFY: + dst->notify.cookie = cpu_to_le64(src->notify.cookie); break; case CEPH_OSD_OP_SETALLOCHINT: dst->alloc_hint.expected_object_size = @@ -768,9 +866,6 @@ static u64 osd_req_encode_op(struct ceph_osd_request *req, dst->xattr.value_len = cpu_to_le32(src->xattr.value_len); dst->xattr.cmp_op = src->xattr.cmp_op; dst->xattr.cmp_mode = src->xattr.cmp_mode; - osd_data = &src->xattr.osd_data; - ceph_osdc_msg_data_add(req->r_request, osd_data); - request_data_len = osd_data->pagelist->length; break; case CEPH_OSD_OP_CREATE: case CEPH_OSD_OP_DELETE: @@ -787,7 +882,7 @@ static u64 osd_req_encode_op(struct ceph_osd_request *req, dst->flags = cpu_to_le32(src->flags); dst->payload_len = cpu_to_le32(src->indata_len); - return request_data_len; + return src->indata_len; } /* @@ -824,17 +919,15 @@ struct ceph_osd_request *ceph_osdc_new_request(struct ceph_osd_client *osdc, req = ceph_osdc_alloc_request(osdc, snapc, num_ops, use_mempool, GFP_NOFS); - if (!req) - return ERR_PTR(-ENOMEM); - - req->r_flags = flags; + if (!req) { + r = -ENOMEM; + goto fail; + } /* calculate max write size */ r = calc_layout(layout, off, plen, &objnum, &objoff, &objlen); - if (r < 0) { - ceph_osdc_put_request(req); - return ERR_PTR(r); - } + if (r) + goto fail; if (opcode == CEPH_OSD_OP_CREATE || opcode == CEPH_OSD_OP_DELETE) { osd_req_op_init(req, which, opcode, 0); @@ -854,194 +947,71 @@ struct ceph_osd_request *ceph_osdc_new_request(struct ceph_osd_client *osdc, truncate_size, truncate_seq); } + req->r_flags = flags; req->r_base_oloc.pool = ceph_file_layout_pg_pool(*layout); + ceph_oid_printf(&req->r_base_oid, "%llx.%08llx", vino.ino, objnum); + + req->r_snapid = vino.snap; + if (flags & CEPH_OSD_FLAG_WRITE) + req->r_data_offset = off; - snprintf(req->r_base_oid.name, sizeof(req->r_base_oid.name), - "%llx.%08llx", vino.ino, objnum); - req->r_base_oid.name_len = strlen(req->r_base_oid.name); + r = ceph_osdc_alloc_messages(req, GFP_NOFS); + if (r) + goto fail; return req; + +fail: + ceph_osdc_put_request(req); + return ERR_PTR(r); } EXPORT_SYMBOL(ceph_osdc_new_request); /* * We keep osd requests in an rbtree, sorted by ->r_tid. */ -static void __insert_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *new) -{ - struct rb_node **p = &osdc->requests.rb_node; - struct rb_node *parent = NULL; - struct ceph_osd_request *req = NULL; - - while (*p) { - parent = *p; - req = rb_entry(parent, struct ceph_osd_request, r_node); - if (new->r_tid < req->r_tid) - p = &(*p)->rb_left; - else if (new->r_tid > req->r_tid) - p = &(*p)->rb_right; - else - BUG(); - } - - rb_link_node(&new->r_node, parent, p); - rb_insert_color(&new->r_node, &osdc->requests); -} +DEFINE_RB_FUNCS(request, struct ceph_osd_request, r_tid, r_node) +DEFINE_RB_FUNCS(request_mc, struct ceph_osd_request, r_tid, r_mc_node) -static struct ceph_osd_request *__lookup_request(struct ceph_osd_client *osdc, - u64 tid) +static bool osd_homeless(struct ceph_osd *osd) { - struct ceph_osd_request *req; - struct rb_node *n = osdc->requests.rb_node; - - while (n) { - req = rb_entry(n, struct ceph_osd_request, r_node); - if (tid < req->r_tid) - n = n->rb_left; - else if (tid > req->r_tid) - n = n->rb_right; - else - return req; - } - return NULL; + return osd->o_osd == CEPH_HOMELESS_OSD; } -static struct ceph_osd_request * -__lookup_request_ge(struct ceph_osd_client *osdc, - u64 tid) +static bool osd_registered(struct ceph_osd *osd) { - struct ceph_osd_request *req; - struct rb_node *n = osdc->requests.rb_node; - - while (n) { - req = rb_entry(n, struct ceph_osd_request, r_node); - if (tid < req->r_tid) { - if (!n->rb_left) - return req; - n = n->rb_left; - } else if (tid > req->r_tid) { - n = n->rb_right; - } else { - return req; - } - } - return NULL; -} - -static void __kick_linger_request(struct ceph_osd_request *req) -{ - struct ceph_osd_client *osdc = req->r_osdc; - struct ceph_osd *osd = req->r_osd; + verify_osdc_locked(osd->o_osdc); - /* - * Linger requests need to be resent with a new tid to avoid - * the dup op detection logic on the OSDs. Achieve this with - * a re-register dance instead of open-coding. - */ - ceph_osdc_get_request(req); - if (!list_empty(&req->r_linger_item)) - __unregister_linger_request(osdc, req); - else - __unregister_request(osdc, req); - __register_request(osdc, req); - ceph_osdc_put_request(req); - - /* - * Unless request has been registered as both normal and - * lingering, __unregister{,_linger}_request clears r_osd. - * However, here we need to preserve r_osd to make sure we - * requeue on the same OSD. - */ - WARN_ON(req->r_osd || !osd); - req->r_osd = osd; - - dout("%s requeueing %p tid %llu\n", __func__, req, req->r_tid); - __enqueue_request(req); + return !RB_EMPTY_NODE(&osd->o_node); } /* - * Resubmit requests pending on the given osd. + * Assumes @osd is zero-initialized. */ -static void __kick_osd_requests(struct ceph_osd_client *osdc, - struct ceph_osd *osd) +static void osd_init(struct ceph_osd *osd) { - struct ceph_osd_request *req, *nreq; - LIST_HEAD(resend); - LIST_HEAD(resend_linger); - int err; - - dout("%s osd%d\n", __func__, osd->o_osd); - err = __reset_osd(osdc, osd); - if (err) - return; - - /* - * Build up a list of requests to resend by traversing the - * osd's list of requests. Requests for a given object are - * sent in tid order, and that is also the order they're - * kept on this list. Therefore all requests that are in - * flight will be found first, followed by all requests that - * have not yet been sent. And to resend requests while - * preserving this order we will want to put any sent - * requests back on the front of the osd client's unsent - * list. - * - * So we build a separate ordered list of already-sent - * requests for the affected osd and splice it onto the - * front of the osd client's unsent list. Once we've seen a - * request that has not yet been sent we're done. Those - * requests are already sitting right where they belong. - */ - list_for_each_entry(req, &osd->o_requests, r_osd_item) { - if (!req->r_sent) - break; - - if (!req->r_linger) { - dout("%s requeueing %p tid %llu\n", __func__, req, - req->r_tid); - list_move_tail(&req->r_req_lru_item, &resend); - req->r_flags |= CEPH_OSD_FLAG_RETRY; - } else { - list_move_tail(&req->r_req_lru_item, &resend_linger); - } - } - list_splice(&resend, &osdc->req_unsent); - - /* - * Both registered and not yet registered linger requests are - * enqueued with a new tid on the same OSD. We add/move them - * to req_unsent/o_requests at the end to keep things in tid - * order. - */ - list_for_each_entry_safe(req, nreq, &osd->o_linger_requests, - r_linger_osd_item) { - WARN_ON(!list_empty(&req->r_req_lru_item)); - __kick_linger_request(req); - } - - list_for_each_entry_safe(req, nreq, &resend_linger, r_req_lru_item) - __kick_linger_request(req); + atomic_set(&osd->o_ref, 1); + RB_CLEAR_NODE(&osd->o_node); + osd->o_requests = RB_ROOT; + osd->o_linger_requests = RB_ROOT; + INIT_LIST_HEAD(&osd->o_osd_lru); + INIT_LIST_HEAD(&osd->o_keepalive_item); + osd->o_incarnation = 1; + mutex_init(&osd->lock); } -/* - * If the osd connection drops, we need to resubmit all requests. - */ -static void osd_reset(struct ceph_connection *con) +static void osd_cleanup(struct ceph_osd *osd) { - struct ceph_osd *osd = con->private; - struct ceph_osd_client *osdc; - - if (!osd) - return; - dout("osd_reset osd%d\n", osd->o_osd); - osdc = osd->o_osdc; - down_read(&osdc->map_sem); - mutex_lock(&osdc->request_mutex); - __kick_osd_requests(osdc, osd); - __send_queued(osdc); - mutex_unlock(&osdc->request_mutex); - up_read(&osdc->map_sem); + WARN_ON(!RB_EMPTY_NODE(&osd->o_node)); + WARN_ON(!RB_EMPTY_ROOT(&osd->o_requests)); + WARN_ON(!RB_EMPTY_ROOT(&osd->o_linger_requests)); + WARN_ON(!list_empty(&osd->o_osd_lru)); + WARN_ON(!list_empty(&osd->o_keepalive_item)); + + if (osd->o_auth.authorizer) { + WARN_ON(osd_homeless(osd)); + ceph_auth_destroy_authorizer(osd->o_auth.authorizer); + } } /* @@ -1051,22 +1021,15 @@ static struct ceph_osd *create_osd(struct ceph_osd_client *osdc, int onum) { struct ceph_osd *osd; - osd = kzalloc(sizeof(*osd), GFP_NOFS); - if (!osd) - return NULL; + WARN_ON(onum == CEPH_HOMELESS_OSD); - atomic_set(&osd->o_ref, 1); + osd = kzalloc(sizeof(*osd), GFP_NOIO | __GFP_NOFAIL); + osd_init(osd); osd->o_osdc = osdc; osd->o_osd = onum; - RB_CLEAR_NODE(&osd->o_node); - INIT_LIST_HEAD(&osd->o_requests); - INIT_LIST_HEAD(&osd->o_linger_requests); - INIT_LIST_HEAD(&osd->o_osd_lru); - osd->o_incarnation = 1; ceph_con_init(&osd->o_con, osd, &osd_con_ops, &osdc->client->msgr); - INIT_LIST_HEAD(&osd->o_keepalive_item); return osd; } @@ -1087,114 +1050,115 @@ static void put_osd(struct ceph_osd *osd) dout("put_osd %p %d -> %d\n", osd, atomic_read(&osd->o_ref), atomic_read(&osd->o_ref) - 1); if (atomic_dec_and_test(&osd->o_ref)) { - if (osd->o_auth.authorizer) - ceph_auth_destroy_authorizer(osd->o_auth.authorizer); + osd_cleanup(osd); kfree(osd); } } -/* - * remove an osd from our map - */ -static void __remove_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd) -{ - dout("%s %p osd%d\n", __func__, osd, osd->o_osd); - WARN_ON(!list_empty(&osd->o_requests)); - WARN_ON(!list_empty(&osd->o_linger_requests)); - - list_del_init(&osd->o_osd_lru); - rb_erase(&osd->o_node, &osdc->osds); - RB_CLEAR_NODE(&osd->o_node); -} - -static void remove_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd) -{ - dout("%s %p osd%d\n", __func__, osd, osd->o_osd); - - if (!RB_EMPTY_NODE(&osd->o_node)) { - ceph_con_close(&osd->o_con); - __remove_osd(osdc, osd); - put_osd(osd); - } -} +DEFINE_RB_FUNCS(osd, struct ceph_osd, o_osd, o_node) -static void remove_all_osds(struct ceph_osd_client *osdc) +static void __move_osd_to_lru(struct ceph_osd *osd) { - dout("%s %p\n", __func__, osdc); - mutex_lock(&osdc->request_mutex); - while (!RB_EMPTY_ROOT(&osdc->osds)) { - struct ceph_osd *osd = rb_entry(rb_first(&osdc->osds), - struct ceph_osd, o_node); - remove_osd(osdc, osd); - } - mutex_unlock(&osdc->request_mutex); -} + struct ceph_osd_client *osdc = osd->o_osdc; -static void __move_osd_to_lru(struct ceph_osd_client *osdc, - struct ceph_osd *osd) -{ - dout("%s %p\n", __func__, osd); + dout("%s osd %p osd%d\n", __func__, osd, osd->o_osd); BUG_ON(!list_empty(&osd->o_osd_lru)); + spin_lock(&osdc->osd_lru_lock); list_add_tail(&osd->o_osd_lru, &osdc->osd_lru); + spin_unlock(&osdc->osd_lru_lock); + osd->lru_ttl = jiffies + osdc->client->options->osd_idle_ttl; } -static void maybe_move_osd_to_lru(struct ceph_osd_client *osdc, - struct ceph_osd *osd) +static void maybe_move_osd_to_lru(struct ceph_osd *osd) { - dout("%s %p\n", __func__, osd); - - if (list_empty(&osd->o_requests) && - list_empty(&osd->o_linger_requests)) - __move_osd_to_lru(osdc, osd); + if (RB_EMPTY_ROOT(&osd->o_requests) && + RB_EMPTY_ROOT(&osd->o_linger_requests)) + __move_osd_to_lru(osd); } static void __remove_osd_from_lru(struct ceph_osd *osd) { - dout("__remove_osd_from_lru %p\n", osd); + struct ceph_osd_client *osdc = osd->o_osdc; + + dout("%s osd %p osd%d\n", __func__, osd, osd->o_osd); + + spin_lock(&osdc->osd_lru_lock); if (!list_empty(&osd->o_osd_lru)) list_del_init(&osd->o_osd_lru); + spin_unlock(&osdc->osd_lru_lock); } -static void remove_old_osds(struct ceph_osd_client *osdc) +/* + * Close the connection and assign any leftover requests to the + * homeless session. + */ +static void close_osd(struct ceph_osd *osd) { - struct ceph_osd *osd, *nosd; + struct ceph_osd_client *osdc = osd->o_osdc; + struct rb_node *n; - dout("__remove_old_osds %p\n", osdc); - mutex_lock(&osdc->request_mutex); - list_for_each_entry_safe(osd, nosd, &osdc->osd_lru, o_osd_lru) { - if (time_before(jiffies, osd->lru_ttl)) - break; - remove_osd(osdc, osd); + verify_osdc_wrlocked(osdc); + dout("%s osd %p osd%d\n", __func__, osd, osd->o_osd); + + ceph_con_close(&osd->o_con); + + for (n = rb_first(&osd->o_requests); n; ) { + struct ceph_osd_request *req = + rb_entry(n, struct ceph_osd_request, r_node); + + n = rb_next(n); /* unlink_request() */ + + dout(" reassigning req %p tid %llu\n", req, req->r_tid); + unlink_request(osd, req); + link_request(&osdc->homeless_osd, req); + } + for (n = rb_first(&osd->o_linger_requests); n; ) { + struct ceph_osd_linger_request *lreq = + rb_entry(n, struct ceph_osd_linger_request, node); + + n = rb_next(n); /* unlink_linger() */ + + dout(" reassigning lreq %p linger_id %llu\n", lreq, + lreq->linger_id); + unlink_linger(osd, lreq); + link_linger(&osdc->homeless_osd, lreq); } - mutex_unlock(&osdc->request_mutex); + + __remove_osd_from_lru(osd); + erase_osd(&osdc->osds, osd); + put_osd(osd); } /* * reset osd connect */ -static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd) +static int reopen_osd(struct ceph_osd *osd) { struct ceph_entity_addr *peer_addr; - dout("__reset_osd %p osd%d\n", osd, osd->o_osd); - if (list_empty(&osd->o_requests) && - list_empty(&osd->o_linger_requests)) { - remove_osd(osdc, osd); + dout("%s osd %p osd%d\n", __func__, osd, osd->o_osd); + + if (RB_EMPTY_ROOT(&osd->o_requests) && + RB_EMPTY_ROOT(&osd->o_linger_requests)) { + close_osd(osd); return -ENODEV; } - peer_addr = &osdc->osdmap->osd_addr[osd->o_osd]; + peer_addr = &osd->o_osdc->osdmap->osd_addr[osd->o_osd]; if (!memcmp(peer_addr, &osd->o_con.peer_addr, sizeof (*peer_addr)) && !ceph_con_opened(&osd->o_con)) { - struct ceph_osd_request *req; + struct rb_node *n; dout("osd addr hasn't changed and connection never opened, " "letting msgr retry\n"); /* touch each r_stamp for handle_timeout()'s benfit */ - list_for_each_entry(req, &osd->o_requests, r_osd_item) + for (n = rb_first(&osd->o_requests); n; n = rb_next(n)) { + struct ceph_osd_request *req = + rb_entry(n, struct ceph_osd_request, r_node); req->r_stamp = jiffies; + } return -EAGAIN; } @@ -1206,455 +1170,1369 @@ static int __reset_osd(struct ceph_osd_client *osdc, struct ceph_osd *osd) return 0; } -static void __insert_osd(struct ceph_osd_client *osdc, struct ceph_osd *new) +static struct ceph_osd *lookup_create_osd(struct ceph_osd_client *osdc, int o, + bool wrlocked) { - struct rb_node **p = &osdc->osds.rb_node; - struct rb_node *parent = NULL; - struct ceph_osd *osd = NULL; + struct ceph_osd *osd; - dout("__insert_osd %p osd%d\n", new, new->o_osd); - while (*p) { - parent = *p; - osd = rb_entry(parent, struct ceph_osd, o_node); - if (new->o_osd < osd->o_osd) - p = &(*p)->rb_left; - else if (new->o_osd > osd->o_osd) - p = &(*p)->rb_right; - else - BUG(); + if (wrlocked) + verify_osdc_wrlocked(osdc); + else + verify_osdc_locked(osdc); + + if (o != CEPH_HOMELESS_OSD) + osd = lookup_osd(&osdc->osds, o); + else + osd = &osdc->homeless_osd; + if (!osd) { + if (!wrlocked) + return ERR_PTR(-EAGAIN); + + osd = create_osd(osdc, o); + insert_osd(&osdc->osds, osd); + ceph_con_open(&osd->o_con, CEPH_ENTITY_TYPE_OSD, osd->o_osd, + &osdc->osdmap->osd_addr[osd->o_osd]); } - rb_link_node(&new->o_node, parent, p); - rb_insert_color(&new->o_node, &osdc->osds); + dout("%s osdc %p osd%d -> osd %p\n", __func__, osdc, o, osd); + return osd; } -static struct ceph_osd *__lookup_osd(struct ceph_osd_client *osdc, int o) +/* + * Create request <-> OSD session relation. + * + * @req has to be assigned a tid, @osd may be homeless. + */ +static void link_request(struct ceph_osd *osd, struct ceph_osd_request *req) { - struct ceph_osd *osd; - struct rb_node *n = osdc->osds.rb_node; - - while (n) { - osd = rb_entry(n, struct ceph_osd, o_node); - if (o < osd->o_osd) - n = n->rb_left; - else if (o > osd->o_osd) - n = n->rb_right; - else - return osd; - } - return NULL; + verify_osd_locked(osd); + WARN_ON(!req->r_tid || req->r_osd); + dout("%s osd %p osd%d req %p tid %llu\n", __func__, osd, osd->o_osd, + req, req->r_tid); + + if (!osd_homeless(osd)) + __remove_osd_from_lru(osd); + else + atomic_inc(&osd->o_osdc->num_homeless); + + get_osd(osd); + insert_request(&osd->o_requests, req); + req->r_osd = osd; } -static void __schedule_osd_timeout(struct ceph_osd_client *osdc) +static void unlink_request(struct ceph_osd *osd, struct ceph_osd_request *req) { - schedule_delayed_work(&osdc->timeout_work, - osdc->client->options->osd_keepalive_timeout); + verify_osd_locked(osd); + WARN_ON(req->r_osd != osd); + dout("%s osd %p osd%d req %p tid %llu\n", __func__, osd, osd->o_osd, + req, req->r_tid); + + req->r_osd = NULL; + erase_request(&osd->o_requests, req); + put_osd(osd); + + if (!osd_homeless(osd)) + maybe_move_osd_to_lru(osd); + else + atomic_dec(&osd->o_osdc->num_homeless); } -static void __cancel_osd_timeout(struct ceph_osd_client *osdc) +static bool __pool_full(struct ceph_pg_pool_info *pi) { - cancel_delayed_work(&osdc->timeout_work); + return pi->flags & CEPH_POOL_FLAG_FULL; } -/* - * Register request, assign tid. If this is the first request, set up - * the timeout event. - */ -static void __register_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req) +static bool have_pool_full(struct ceph_osd_client *osdc) { - req->r_tid = ++osdc->last_tid; - req->r_request->hdr.tid = cpu_to_le64(req->r_tid); - dout("__register_request %p tid %lld\n", req, req->r_tid); - __insert_request(osdc, req); - ceph_osdc_get_request(req); - osdc->num_requests++; - if (osdc->num_requests == 1) { - dout(" first request, scheduling timeout\n"); - __schedule_osd_timeout(osdc); + struct rb_node *n; + + for (n = rb_first(&osdc->osdmap->pg_pools); n; n = rb_next(n)) { + struct ceph_pg_pool_info *pi = + rb_entry(n, struct ceph_pg_pool_info, node); + + if (__pool_full(pi)) + return true; } + + return false; +} + +static bool pool_full(struct ceph_osd_client *osdc, s64 pool_id) +{ + struct ceph_pg_pool_info *pi; + + pi = ceph_pg_pool_by_id(osdc->osdmap, pool_id); + if (!pi) + return false; + + return __pool_full(pi); } /* - * called under osdc->request_mutex + * Returns whether a request should be blocked from being sent + * based on the current osdmap and osd_client settings. */ -static void __unregister_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req) +static bool target_should_be_paused(struct ceph_osd_client *osdc, + const struct ceph_osd_request_target *t, + struct ceph_pg_pool_info *pi) { - if (RB_EMPTY_NODE(&req->r_node)) { - dout("__unregister_request %p tid %lld not registered\n", - req, req->r_tid); - return; + bool pauserd = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD); + bool pausewr = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR) || + ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) || + __pool_full(pi); + + WARN_ON(pi->id != t->base_oloc.pool); + return (t->flags & CEPH_OSD_FLAG_READ && pauserd) || + (t->flags & CEPH_OSD_FLAG_WRITE && pausewr); +} + +enum calc_target_result { + CALC_TARGET_NO_ACTION = 0, + CALC_TARGET_NEED_RESEND, + CALC_TARGET_POOL_DNE, +}; + +static enum calc_target_result calc_target(struct ceph_osd_client *osdc, + struct ceph_osd_request_target *t, + u32 *last_force_resend, + bool any_change) +{ + struct ceph_pg_pool_info *pi; + struct ceph_pg pgid, last_pgid; + struct ceph_osds up, acting; + bool force_resend = false; + bool need_check_tiering = false; + bool need_resend = false; + bool sort_bitwise = ceph_osdmap_flag(osdc, CEPH_OSDMAP_SORTBITWISE); + enum calc_target_result ct_res; + int ret; + + pi = ceph_pg_pool_by_id(osdc->osdmap, t->base_oloc.pool); + if (!pi) { + t->osd = CEPH_HOMELESS_OSD; + ct_res = CALC_TARGET_POOL_DNE; + goto out; } - dout("__unregister_request %p tid %lld\n", req, req->r_tid); - rb_erase(&req->r_node, &osdc->requests); - RB_CLEAR_NODE(&req->r_node); - osdc->num_requests--; + if (osdc->osdmap->epoch == pi->last_force_request_resend) { + if (last_force_resend && + *last_force_resend < pi->last_force_request_resend) { + *last_force_resend = pi->last_force_request_resend; + force_resend = true; + } else if (!last_force_resend) { + force_resend = true; + } + } + if (ceph_oid_empty(&t->target_oid) || force_resend) { + ceph_oid_copy(&t->target_oid, &t->base_oid); + need_check_tiering = true; + } + if (ceph_oloc_empty(&t->target_oloc) || force_resend) { + ceph_oloc_copy(&t->target_oloc, &t->base_oloc); + need_check_tiering = true; + } - if (req->r_osd) { - /* make sure the original request isn't in flight. */ - ceph_msg_revoke(req->r_request); + if (need_check_tiering && + (t->flags & CEPH_OSD_FLAG_IGNORE_OVERLAY) == 0) { + if (t->flags & CEPH_OSD_FLAG_READ && pi->read_tier >= 0) + t->target_oloc.pool = pi->read_tier; + if (t->flags & CEPH_OSD_FLAG_WRITE && pi->write_tier >= 0) + t->target_oloc.pool = pi->write_tier; + } - list_del_init(&req->r_osd_item); - maybe_move_osd_to_lru(osdc, req->r_osd); - if (list_empty(&req->r_linger_osd_item)) - req->r_osd = NULL; + ret = ceph_object_locator_to_pg(osdc->osdmap, &t->target_oid, + &t->target_oloc, &pgid); + if (ret) { + WARN_ON(ret != -ENOENT); + t->osd = CEPH_HOMELESS_OSD; + ct_res = CALC_TARGET_POOL_DNE; + goto out; + } + last_pgid.pool = pgid.pool; + last_pgid.seed = ceph_stable_mod(pgid.seed, t->pg_num, t->pg_num_mask); + + ceph_pg_to_up_acting_osds(osdc->osdmap, &pgid, &up, &acting); + if (any_change && + ceph_is_new_interval(&t->acting, + &acting, + &t->up, + &up, + t->size, + pi->size, + t->min_size, + pi->min_size, + t->pg_num, + pi->pg_num, + t->sort_bitwise, + sort_bitwise, + &last_pgid)) + force_resend = true; + + if (t->paused && !target_should_be_paused(osdc, t, pi)) { + t->paused = false; + need_resend = true; } - list_del_init(&req->r_req_lru_item); - ceph_osdc_put_request(req); + if (ceph_pg_compare(&t->pgid, &pgid) || + ceph_osds_changed(&t->acting, &acting, any_change) || + force_resend) { + t->pgid = pgid; /* struct */ + ceph_osds_copy(&t->acting, &acting); + ceph_osds_copy(&t->up, &up); + t->size = pi->size; + t->min_size = pi->min_size; + t->pg_num = pi->pg_num; + t->pg_num_mask = pi->pg_num_mask; + t->sort_bitwise = sort_bitwise; + + t->osd = acting.primary; + need_resend = true; + } + + ct_res = need_resend ? CALC_TARGET_NEED_RESEND : CALC_TARGET_NO_ACTION; +out: + dout("%s t %p -> ct_res %d osd %d\n", __func__, t, ct_res, t->osd); + return ct_res; +} + +static void setup_request_data(struct ceph_osd_request *req, + struct ceph_msg *msg) +{ + u32 data_len = 0; + int i; + + if (!list_empty(&msg->data)) + return; + + WARN_ON(msg->data_length); + for (i = 0; i < req->r_num_ops; i++) { + struct ceph_osd_req_op *op = &req->r_ops[i]; + + switch (op->op) { + /* request */ + case CEPH_OSD_OP_WRITE: + case CEPH_OSD_OP_WRITEFULL: + WARN_ON(op->indata_len != op->extent.length); + ceph_osdc_msg_data_add(msg, &op->extent.osd_data); + break; + case CEPH_OSD_OP_SETXATTR: + case CEPH_OSD_OP_CMPXATTR: + WARN_ON(op->indata_len != op->xattr.name_len + + op->xattr.value_len); + ceph_osdc_msg_data_add(msg, &op->xattr.osd_data); + break; + case CEPH_OSD_OP_NOTIFY_ACK: + ceph_osdc_msg_data_add(msg, + &op->notify_ack.request_data); + break; + + /* reply */ + case CEPH_OSD_OP_STAT: + ceph_osdc_msg_data_add(req->r_reply, + &op->raw_data_in); + break; + case CEPH_OSD_OP_READ: + ceph_osdc_msg_data_add(req->r_reply, + &op->extent.osd_data); + break; + + /* both */ + case CEPH_OSD_OP_CALL: + WARN_ON(op->indata_len != op->cls.class_len + + op->cls.method_len + + op->cls.indata_len); + ceph_osdc_msg_data_add(msg, &op->cls.request_info); + /* optional, can be NONE */ + ceph_osdc_msg_data_add(msg, &op->cls.request_data); + /* optional, can be NONE */ + ceph_osdc_msg_data_add(req->r_reply, + &op->cls.response_data); + break; + case CEPH_OSD_OP_NOTIFY: + ceph_osdc_msg_data_add(msg, + &op->notify.request_data); + ceph_osdc_msg_data_add(req->r_reply, + &op->notify.response_data); + break; + } + + data_len += op->indata_len; + } + + WARN_ON(data_len != msg->data_length); +} + +static void encode_request(struct ceph_osd_request *req, struct ceph_msg *msg) +{ + void *p = msg->front.iov_base; + void *const end = p + msg->front_alloc_len; + u32 data_len = 0; + int i; + + if (req->r_flags & CEPH_OSD_FLAG_WRITE) { + /* snapshots aren't writeable */ + WARN_ON(req->r_snapid != CEPH_NOSNAP); + } else { + WARN_ON(req->r_mtime.tv_sec || req->r_mtime.tv_nsec || + req->r_data_offset || req->r_snapc); + } + + setup_request_data(req, msg); + + ceph_encode_32(&p, 1); /* client_inc, always 1 */ + ceph_encode_32(&p, req->r_osdc->osdmap->epoch); + ceph_encode_32(&p, req->r_flags); + ceph_encode_timespec(p, &req->r_mtime); + p += sizeof(struct ceph_timespec); + /* aka reassert_version */ + memcpy(p, &req->r_replay_version, sizeof(req->r_replay_version)); + p += sizeof(req->r_replay_version); + + /* oloc */ + ceph_encode_8(&p, 4); + ceph_encode_8(&p, 4); + ceph_encode_32(&p, 8 + 4 + 4); + ceph_encode_64(&p, req->r_t.target_oloc.pool); + ceph_encode_32(&p, -1); /* preferred */ + ceph_encode_32(&p, 0); /* key len */ + + /* pgid */ + ceph_encode_8(&p, 1); + ceph_encode_64(&p, req->r_t.pgid.pool); + ceph_encode_32(&p, req->r_t.pgid.seed); + ceph_encode_32(&p, -1); /* preferred */ + + /* oid */ + ceph_encode_32(&p, req->r_t.target_oid.name_len); + memcpy(p, req->r_t.target_oid.name, req->r_t.target_oid.name_len); + p += req->r_t.target_oid.name_len; - if (osdc->num_requests == 0) { - dout(" no requests, canceling timeout\n"); - __cancel_osd_timeout(osdc); + /* ops, can imply data */ + ceph_encode_16(&p, req->r_num_ops); + for (i = 0; i < req->r_num_ops; i++) { + data_len += osd_req_encode_op(p, &req->r_ops[i]); + p += sizeof(struct ceph_osd_op); } + + ceph_encode_64(&p, req->r_snapid); /* snapid */ + if (req->r_snapc) { + ceph_encode_64(&p, req->r_snapc->seq); + ceph_encode_32(&p, req->r_snapc->num_snaps); + for (i = 0; i < req->r_snapc->num_snaps; i++) + ceph_encode_64(&p, req->r_snapc->snaps[i]); + } else { + ceph_encode_64(&p, 0); /* snap_seq */ + ceph_encode_32(&p, 0); /* snaps len */ + } + + ceph_encode_32(&p, req->r_attempts); /* retry_attempt */ + + BUG_ON(p > end); + msg->front.iov_len = p - msg->front.iov_base; + msg->hdr.version = cpu_to_le16(4); /* MOSDOp v4 */ + msg->hdr.front_len = cpu_to_le32(msg->front.iov_len); + msg->hdr.data_len = cpu_to_le32(data_len); + /* + * The header "data_off" is a hint to the receiver allowing it + * to align received data into its buffers such that there's no + * need to re-copy it before writing it to disk (direct I/O). + */ + msg->hdr.data_off = cpu_to_le16(req->r_data_offset); + + dout("%s req %p oid %s oid_len %d front %zu data %u\n", __func__, + req, req->r_t.target_oid.name, req->r_t.target_oid.name_len, + msg->front.iov_len, data_len); } /* - * Cancel a previously queued request message + * @req has to be assigned a tid and registered. */ -static void __cancel_request(struct ceph_osd_request *req) +static void send_request(struct ceph_osd_request *req) { - if (req->r_sent && req->r_osd) { + struct ceph_osd *osd = req->r_osd; + + verify_osd_locked(osd); + WARN_ON(osd->o_osd != req->r_t.osd); + + /* + * We may have a previously queued request message hanging + * around. Cancel it to avoid corrupting the msgr. + */ + if (req->r_sent) ceph_msg_revoke(req->r_request); - req->r_sent = 0; + + req->r_flags |= CEPH_OSD_FLAG_KNOWN_REDIR; + if (req->r_attempts) + req->r_flags |= CEPH_OSD_FLAG_RETRY; + else + WARN_ON(req->r_flags & CEPH_OSD_FLAG_RETRY); + + encode_request(req, req->r_request); + + dout("%s req %p tid %llu to pg %llu.%x osd%d flags 0x%x attempt %d\n", + __func__, req, req->r_tid, req->r_t.pgid.pool, req->r_t.pgid.seed, + req->r_t.osd, req->r_flags, req->r_attempts); + + req->r_t.paused = false; + req->r_stamp = jiffies; + req->r_attempts++; + + req->r_sent = osd->o_incarnation; + req->r_request->hdr.tid = cpu_to_le64(req->r_tid); + ceph_con_send(&osd->o_con, ceph_msg_get(req->r_request)); +} + +static void maybe_request_map(struct ceph_osd_client *osdc) +{ + bool continuous = false; + + verify_osdc_locked(osdc); + WARN_ON(!osdc->osdmap->epoch); + + if (ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) || + ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD) || + ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR)) { + dout("%s osdc %p continuous\n", __func__, osdc); + continuous = true; + } else { + dout("%s osdc %p onetime\n", __func__, osdc); } + + if (ceph_monc_want_map(&osdc->client->monc, CEPH_SUB_OSDMAP, + osdc->osdmap->epoch + 1, continuous)) + ceph_monc_renew_subs(&osdc->client->monc); } -static void __register_linger_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req) +static void send_map_check(struct ceph_osd_request *req); + +static void __submit_request(struct ceph_osd_request *req, bool wrlocked) { - dout("%s %p tid %llu\n", __func__, req, req->r_tid); - WARN_ON(!req->r_linger); + struct ceph_osd_client *osdc = req->r_osdc; + struct ceph_osd *osd; + enum calc_target_result ct_res; + bool need_send = false; + bool promoted = false; + + WARN_ON(req->r_tid || req->r_got_reply); + dout("%s req %p wrlocked %d\n", __func__, req, wrlocked); + +again: + ct_res = calc_target(osdc, &req->r_t, &req->r_last_force_resend, false); + if (ct_res == CALC_TARGET_POOL_DNE && !wrlocked) + goto promote; + + osd = lookup_create_osd(osdc, req->r_t.osd, wrlocked); + if (IS_ERR(osd)) { + WARN_ON(PTR_ERR(osd) != -EAGAIN || wrlocked); + goto promote; + } + + if ((req->r_flags & CEPH_OSD_FLAG_WRITE) && + ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR)) { + dout("req %p pausewr\n", req); + req->r_t.paused = true; + maybe_request_map(osdc); + } else if ((req->r_flags & CEPH_OSD_FLAG_READ) && + ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD)) { + dout("req %p pauserd\n", req); + req->r_t.paused = true; + maybe_request_map(osdc); + } else if ((req->r_flags & CEPH_OSD_FLAG_WRITE) && + !(req->r_flags & (CEPH_OSD_FLAG_FULL_TRY | + CEPH_OSD_FLAG_FULL_FORCE)) && + (ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) || + pool_full(osdc, req->r_t.base_oloc.pool))) { + dout("req %p full/pool_full\n", req); + pr_warn_ratelimited("FULL or reached pool quota\n"); + req->r_t.paused = true; + maybe_request_map(osdc); + } else if (!osd_homeless(osd)) { + need_send = true; + } else { + maybe_request_map(osdc); + } + + mutex_lock(&osd->lock); + /* + * Assign the tid atomically with send_request() to protect + * multiple writes to the same object from racing with each + * other, resulting in out of order ops on the OSDs. + */ + req->r_tid = atomic64_inc_return(&osdc->last_tid); + link_request(osd, req); + if (need_send) + send_request(req); + mutex_unlock(&osd->lock); + if (ct_res == CALC_TARGET_POOL_DNE) + send_map_check(req); + + if (promoted) + downgrade_write(&osdc->lock); + return; + +promote: + up_read(&osdc->lock); + down_write(&osdc->lock); + wrlocked = true; + promoted = true; + goto again; +} + +static void account_request(struct ceph_osd_request *req) +{ + unsigned int mask = CEPH_OSD_FLAG_ACK | CEPH_OSD_FLAG_ONDISK; + + if (req->r_flags & CEPH_OSD_FLAG_READ) { + WARN_ON(req->r_flags & mask); + req->r_flags |= CEPH_OSD_FLAG_ACK; + } else if (req->r_flags & CEPH_OSD_FLAG_WRITE) + WARN_ON(!(req->r_flags & mask)); + else + WARN_ON(1); + + WARN_ON(req->r_unsafe_callback && (req->r_flags & mask) != mask); + atomic_inc(&req->r_osdc->num_requests); +} + +static void submit_request(struct ceph_osd_request *req, bool wrlocked) +{ ceph_osdc_get_request(req); - list_add_tail(&req->r_linger_item, &osdc->req_linger); - if (req->r_osd) - list_add_tail(&req->r_linger_osd_item, - &req->r_osd->o_linger_requests); + account_request(req); + __submit_request(req, wrlocked); } -static void __unregister_linger_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req) +static void __finish_request(struct ceph_osd_request *req) { - WARN_ON(!req->r_linger); + struct ceph_osd_client *osdc = req->r_osdc; + struct ceph_osd *osd = req->r_osd; - if (list_empty(&req->r_linger_item)) { - dout("%s %p tid %llu not registered\n", __func__, req, - req->r_tid); + verify_osd_locked(osd); + dout("%s req %p tid %llu\n", __func__, req, req->r_tid); + + WARN_ON(lookup_request_mc(&osdc->map_checks, req->r_tid)); + unlink_request(osd, req); + atomic_dec(&osdc->num_requests); + + /* + * If an OSD has failed or returned and a request has been sent + * twice, it's possible to get a reply and end up here while the + * request message is queued for delivery. We will ignore the + * reply, so not a big deal, but better to try and catch it. + */ + ceph_msg_revoke(req->r_request); + ceph_msg_revoke_incoming(req->r_reply); +} + +static void finish_request(struct ceph_osd_request *req) +{ + __finish_request(req); + ceph_osdc_put_request(req); +} + +static void __complete_request(struct ceph_osd_request *req) +{ + if (req->r_callback) + req->r_callback(req); + else + complete_all(&req->r_completion); +} + +/* + * Note that this is open-coded in handle_reply(), which has to deal + * with ack vs commit, dup acks, etc. + */ +static void complete_request(struct ceph_osd_request *req, int err) +{ + dout("%s req %p tid %llu err %d\n", __func__, req, req->r_tid, err); + + req->r_result = err; + __finish_request(req); + __complete_request(req); + complete_all(&req->r_safe_completion); + ceph_osdc_put_request(req); +} + +static void cancel_map_check(struct ceph_osd_request *req) +{ + struct ceph_osd_client *osdc = req->r_osdc; + struct ceph_osd_request *lookup_req; + + verify_osdc_wrlocked(osdc); + + lookup_req = lookup_request_mc(&osdc->map_checks, req->r_tid); + if (!lookup_req) return; + + WARN_ON(lookup_req != req); + erase_request_mc(&osdc->map_checks, req); + ceph_osdc_put_request(req); +} + +static void cancel_request(struct ceph_osd_request *req) +{ + dout("%s req %p tid %llu\n", __func__, req, req->r_tid); + + cancel_map_check(req); + finish_request(req); +} + +static void check_pool_dne(struct ceph_osd_request *req) +{ + struct ceph_osd_client *osdc = req->r_osdc; + struct ceph_osdmap *map = osdc->osdmap; + + verify_osdc_wrlocked(osdc); + WARN_ON(!map->epoch); + + if (req->r_attempts) { + /* + * We sent a request earlier, which means that + * previously the pool existed, and now it does not + * (i.e., it was deleted). + */ + req->r_map_dne_bound = map->epoch; + dout("%s req %p tid %llu pool disappeared\n", __func__, req, + req->r_tid); + } else { + dout("%s req %p tid %llu map_dne_bound %u have %u\n", __func__, + req, req->r_tid, req->r_map_dne_bound, map->epoch); + } + + if (req->r_map_dne_bound) { + if (map->epoch >= req->r_map_dne_bound) { + /* we had a new enough map */ + pr_info_ratelimited("tid %llu pool does not exist\n", + req->r_tid); + complete_request(req, -ENOENT); + } + } else { + send_map_check(req); } +} + +static void map_check_cb(struct ceph_mon_generic_request *greq) +{ + struct ceph_osd_client *osdc = &greq->monc->client->osdc; + struct ceph_osd_request *req; + u64 tid = greq->private_data; - dout("%s %p tid %llu\n", __func__, req, req->r_tid); - list_del_init(&req->r_linger_item); + WARN_ON(greq->result || !greq->u.newest); - if (req->r_osd) { - list_del_init(&req->r_linger_osd_item); - maybe_move_osd_to_lru(osdc, req->r_osd); - if (list_empty(&req->r_osd_item)) - req->r_osd = NULL; + down_write(&osdc->lock); + req = lookup_request_mc(&osdc->map_checks, tid); + if (!req) { + dout("%s tid %llu dne\n", __func__, tid); + goto out_unlock; } + + dout("%s req %p tid %llu map_dne_bound %u newest %llu\n", __func__, + req, req->r_tid, req->r_map_dne_bound, greq->u.newest); + if (!req->r_map_dne_bound) + req->r_map_dne_bound = greq->u.newest; + erase_request_mc(&osdc->map_checks, req); + check_pool_dne(req); + ceph_osdc_put_request(req); +out_unlock: + up_write(&osdc->lock); } -void ceph_osdc_set_request_linger(struct ceph_osd_client *osdc, - struct ceph_osd_request *req) +static void send_map_check(struct ceph_osd_request *req) { - if (!req->r_linger) { - dout("set_request_linger %p\n", req); - req->r_linger = 1; + struct ceph_osd_client *osdc = req->r_osdc; + struct ceph_osd_request *lookup_req; + int ret; + + verify_osdc_wrlocked(osdc); + + lookup_req = lookup_request_mc(&osdc->map_checks, req->r_tid); + if (lookup_req) { + WARN_ON(lookup_req != req); + return; } + + ceph_osdc_get_request(req); + insert_request_mc(&osdc->map_checks, req); + ret = ceph_monc_get_version_async(&osdc->client->monc, "osdmap", + map_check_cb, req->r_tid); + WARN_ON(ret); } -EXPORT_SYMBOL(ceph_osdc_set_request_linger); /* - * Returns whether a request should be blocked from being sent - * based on the current osdmap and osd_client settings. - * - * Caller should hold map_sem for read. + * lingering requests, watch/notify v2 infrastructure */ -static bool __req_should_be_paused(struct ceph_osd_client *osdc, - struct ceph_osd_request *req) +static void linger_release(struct kref *kref) +{ + struct ceph_osd_linger_request *lreq = + container_of(kref, struct ceph_osd_linger_request, kref); + + dout("%s lreq %p reg_req %p ping_req %p\n", __func__, lreq, + lreq->reg_req, lreq->ping_req); + WARN_ON(!RB_EMPTY_NODE(&lreq->node)); + WARN_ON(!RB_EMPTY_NODE(&lreq->osdc_node)); + WARN_ON(!RB_EMPTY_NODE(&lreq->mc_node)); + WARN_ON(!list_empty(&lreq->scan_item)); + WARN_ON(!list_empty(&lreq->pending_lworks)); + WARN_ON(lreq->osd); + + if (lreq->reg_req) + ceph_osdc_put_request(lreq->reg_req); + if (lreq->ping_req) + ceph_osdc_put_request(lreq->ping_req); + target_destroy(&lreq->t); + kfree(lreq); +} + +static void linger_put(struct ceph_osd_linger_request *lreq) { - bool pauserd = ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSERD); - bool pausewr = ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSEWR) || - ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL); - return (req->r_flags & CEPH_OSD_FLAG_READ && pauserd) || - (req->r_flags & CEPH_OSD_FLAG_WRITE && pausewr); + if (lreq) + kref_put(&lreq->kref, linger_release); +} + +static struct ceph_osd_linger_request * +linger_get(struct ceph_osd_linger_request *lreq) +{ + kref_get(&lreq->kref); + return lreq; +} + +static struct ceph_osd_linger_request * +linger_alloc(struct ceph_osd_client *osdc) +{ + struct ceph_osd_linger_request *lreq; + + lreq = kzalloc(sizeof(*lreq), GFP_NOIO); + if (!lreq) + return NULL; + + kref_init(&lreq->kref); + mutex_init(&lreq->lock); + RB_CLEAR_NODE(&lreq->node); + RB_CLEAR_NODE(&lreq->osdc_node); + RB_CLEAR_NODE(&lreq->mc_node); + INIT_LIST_HEAD(&lreq->scan_item); + INIT_LIST_HEAD(&lreq->pending_lworks); + init_completion(&lreq->reg_commit_wait); + init_completion(&lreq->notify_finish_wait); + + lreq->osdc = osdc; + target_init(&lreq->t); + + dout("%s lreq %p\n", __func__, lreq); + return lreq; } +DEFINE_RB_INSDEL_FUNCS(linger, struct ceph_osd_linger_request, linger_id, node) +DEFINE_RB_FUNCS(linger_osdc, struct ceph_osd_linger_request, linger_id, osdc_node) +DEFINE_RB_FUNCS(linger_mc, struct ceph_osd_linger_request, linger_id, mc_node) + /* - * Calculate mapping of a request to a PG. Takes tiering into account. + * Create linger request <-> OSD session relation. + * + * @lreq has to be registered, @osd may be homeless. */ -static int __calc_request_pg(struct ceph_osdmap *osdmap, - struct ceph_osd_request *req, - struct ceph_pg *pg_out) +static void link_linger(struct ceph_osd *osd, + struct ceph_osd_linger_request *lreq) { - bool need_check_tiering; + verify_osd_locked(osd); + WARN_ON(!lreq->linger_id || lreq->osd); + dout("%s osd %p osd%d lreq %p linger_id %llu\n", __func__, osd, + osd->o_osd, lreq, lreq->linger_id); - need_check_tiering = false; - if (req->r_target_oloc.pool == -1) { - req->r_target_oloc = req->r_base_oloc; /* struct */ - need_check_tiering = true; + if (!osd_homeless(osd)) + __remove_osd_from_lru(osd); + else + atomic_inc(&osd->o_osdc->num_homeless); + + get_osd(osd); + insert_linger(&osd->o_linger_requests, lreq); + lreq->osd = osd; +} + +static void unlink_linger(struct ceph_osd *osd, + struct ceph_osd_linger_request *lreq) +{ + verify_osd_locked(osd); + WARN_ON(lreq->osd != osd); + dout("%s osd %p osd%d lreq %p linger_id %llu\n", __func__, osd, + osd->o_osd, lreq, lreq->linger_id); + + lreq->osd = NULL; + erase_linger(&osd->o_linger_requests, lreq); + put_osd(osd); + + if (!osd_homeless(osd)) + maybe_move_osd_to_lru(osd); + else + atomic_dec(&osd->o_osdc->num_homeless); +} + +static bool __linger_registered(struct ceph_osd_linger_request *lreq) +{ + verify_osdc_locked(lreq->osdc); + + return !RB_EMPTY_NODE(&lreq->osdc_node); +} + +static bool linger_registered(struct ceph_osd_linger_request *lreq) +{ + struct ceph_osd_client *osdc = lreq->osdc; + bool registered; + + down_read(&osdc->lock); + registered = __linger_registered(lreq); + up_read(&osdc->lock); + + return registered; +} + +static void linger_register(struct ceph_osd_linger_request *lreq) +{ + struct ceph_osd_client *osdc = lreq->osdc; + + verify_osdc_wrlocked(osdc); + WARN_ON(lreq->linger_id); + + linger_get(lreq); + lreq->linger_id = ++osdc->last_linger_id; + insert_linger_osdc(&osdc->linger_requests, lreq); +} + +static void linger_unregister(struct ceph_osd_linger_request *lreq) +{ + struct ceph_osd_client *osdc = lreq->osdc; + + verify_osdc_wrlocked(osdc); + + erase_linger_osdc(&osdc->linger_requests, lreq); + linger_put(lreq); +} + +static void cancel_linger_request(struct ceph_osd_request *req) +{ + struct ceph_osd_linger_request *lreq = req->r_priv; + + WARN_ON(!req->r_linger); + cancel_request(req); + linger_put(lreq); +} + +struct linger_work { + struct work_struct work; + struct ceph_osd_linger_request *lreq; + struct list_head pending_item; + unsigned long queued_stamp; + + union { + struct { + u64 notify_id; + u64 notifier_id; + void *payload; /* points into @msg front */ + size_t payload_len; + + struct ceph_msg *msg; /* for ceph_msg_put() */ + } notify; + struct { + int err; + } error; + }; +}; + +static struct linger_work *lwork_alloc(struct ceph_osd_linger_request *lreq, + work_func_t workfn) +{ + struct linger_work *lwork; + + lwork = kzalloc(sizeof(*lwork), GFP_NOIO); + if (!lwork) + return NULL; + + INIT_WORK(&lwork->work, workfn); + INIT_LIST_HEAD(&lwork->pending_item); + lwork->lreq = linger_get(lreq); + + return lwork; +} + +static void lwork_free(struct linger_work *lwork) +{ + struct ceph_osd_linger_request *lreq = lwork->lreq; + + mutex_lock(&lreq->lock); + list_del(&lwork->pending_item); + mutex_unlock(&lreq->lock); + + linger_put(lreq); + kfree(lwork); +} + +static void lwork_queue(struct linger_work *lwork) +{ + struct ceph_osd_linger_request *lreq = lwork->lreq; + struct ceph_osd_client *osdc = lreq->osdc; + + verify_lreq_locked(lreq); + WARN_ON(!list_empty(&lwork->pending_item)); + + lwork->queued_stamp = jiffies; + list_add_tail(&lwork->pending_item, &lreq->pending_lworks); + queue_work(osdc->notify_wq, &lwork->work); +} + +static void do_watch_notify(struct work_struct *w) +{ + struct linger_work *lwork = container_of(w, struct linger_work, work); + struct ceph_osd_linger_request *lreq = lwork->lreq; + + if (!linger_registered(lreq)) { + dout("%s lreq %p not registered\n", __func__, lreq); + goto out; } - if (req->r_target_oid.name_len == 0) { - ceph_oid_copy(&req->r_target_oid, &req->r_base_oid); - need_check_tiering = true; + + WARN_ON(!lreq->is_watch); + dout("%s lreq %p notify_id %llu notifier_id %llu payload_len %zu\n", + __func__, lreq, lwork->notify.notify_id, lwork->notify.notifier_id, + lwork->notify.payload_len); + lreq->wcb(lreq->data, lwork->notify.notify_id, lreq->linger_id, + lwork->notify.notifier_id, lwork->notify.payload, + lwork->notify.payload_len); + +out: + ceph_msg_put(lwork->notify.msg); + lwork_free(lwork); +} + +static void do_watch_error(struct work_struct *w) +{ + struct linger_work *lwork = container_of(w, struct linger_work, work); + struct ceph_osd_linger_request *lreq = lwork->lreq; + + if (!linger_registered(lreq)) { + dout("%s lreq %p not registered\n", __func__, lreq); + goto out; } - if (need_check_tiering && - (req->r_flags & CEPH_OSD_FLAG_IGNORE_OVERLAY) == 0) { - struct ceph_pg_pool_info *pi; - - pi = ceph_pg_pool_by_id(osdmap, req->r_target_oloc.pool); - if (pi) { - if ((req->r_flags & CEPH_OSD_FLAG_READ) && - pi->read_tier >= 0) - req->r_target_oloc.pool = pi->read_tier; - if ((req->r_flags & CEPH_OSD_FLAG_WRITE) && - pi->write_tier >= 0) - req->r_target_oloc.pool = pi->write_tier; + dout("%s lreq %p err %d\n", __func__, lreq, lwork->error.err); + lreq->errcb(lreq->data, lreq->linger_id, lwork->error.err); + +out: + lwork_free(lwork); +} + +static void queue_watch_error(struct ceph_osd_linger_request *lreq) +{ + struct linger_work *lwork; + + lwork = lwork_alloc(lreq, do_watch_error); + if (!lwork) { + pr_err("failed to allocate error-lwork\n"); + return; + } + + lwork->error.err = lreq->last_error; + lwork_queue(lwork); +} + +static void linger_reg_commit_complete(struct ceph_osd_linger_request *lreq, + int result) +{ + if (!completion_done(&lreq->reg_commit_wait)) { + lreq->reg_commit_error = (result <= 0 ? result : 0); + complete_all(&lreq->reg_commit_wait); + } +} + +static void linger_commit_cb(struct ceph_osd_request *req) +{ + struct ceph_osd_linger_request *lreq = req->r_priv; + + mutex_lock(&lreq->lock); + dout("%s lreq %p linger_id %llu result %d\n", __func__, lreq, + lreq->linger_id, req->r_result); + WARN_ON(!__linger_registered(lreq)); + linger_reg_commit_complete(lreq, req->r_result); + lreq->committed = true; + + if (!lreq->is_watch) { + struct ceph_osd_data *osd_data = + osd_req_op_data(req, 0, notify, response_data); + void *p = page_address(osd_data->pages[0]); + + WARN_ON(req->r_ops[0].op != CEPH_OSD_OP_NOTIFY || + osd_data->type != CEPH_OSD_DATA_TYPE_PAGES); + + /* make note of the notify_id */ + if (req->r_ops[0].outdata_len >= sizeof(u64)) { + lreq->notify_id = ceph_decode_64(&p); + dout("lreq %p notify_id %llu\n", lreq, + lreq->notify_id); + } else { + dout("lreq %p no notify_id\n", lreq); } - /* !pi is caught in ceph_oloc_oid_to_pg() */ } - return ceph_oloc_oid_to_pg(osdmap, &req->r_target_oloc, - &req->r_target_oid, pg_out); + mutex_unlock(&lreq->lock); + linger_put(lreq); } -static void __enqueue_request(struct ceph_osd_request *req) +static int normalize_watch_error(int err) { - struct ceph_osd_client *osdc = req->r_osdc; + /* + * Translate ENOENT -> ENOTCONN so that a delete->disconnection + * notification and a failure to reconnect because we raced with + * the delete appear the same to the user. + */ + if (err == -ENOENT) + err = -ENOTCONN; + + return err; +} + +static void linger_reconnect_cb(struct ceph_osd_request *req) +{ + struct ceph_osd_linger_request *lreq = req->r_priv; + + mutex_lock(&lreq->lock); + dout("%s lreq %p linger_id %llu result %d last_error %d\n", __func__, + lreq, lreq->linger_id, req->r_result, lreq->last_error); + if (req->r_result < 0) { + if (!lreq->last_error) { + lreq->last_error = normalize_watch_error(req->r_result); + queue_watch_error(lreq); + } + } + + mutex_unlock(&lreq->lock); + linger_put(lreq); +} + +static void send_linger(struct ceph_osd_linger_request *lreq) +{ + struct ceph_osd_request *req = lreq->reg_req; + struct ceph_osd_req_op *op = &req->r_ops[0]; - dout("%s %p tid %llu to osd%d\n", __func__, req, req->r_tid, - req->r_osd ? req->r_osd->o_osd : -1); + verify_osdc_wrlocked(req->r_osdc); + dout("%s lreq %p linger_id %llu\n", __func__, lreq, lreq->linger_id); - if (req->r_osd) { - __remove_osd_from_lru(req->r_osd); - list_add_tail(&req->r_osd_item, &req->r_osd->o_requests); - list_move_tail(&req->r_req_lru_item, &osdc->req_unsent); + if (req->r_osd) + cancel_linger_request(req); + + request_reinit(req); + ceph_oid_copy(&req->r_base_oid, &lreq->t.base_oid); + ceph_oloc_copy(&req->r_base_oloc, &lreq->t.base_oloc); + req->r_flags = lreq->t.flags; + req->r_mtime = lreq->mtime; + + mutex_lock(&lreq->lock); + if (lreq->is_watch && lreq->committed) { + WARN_ON(op->op != CEPH_OSD_OP_WATCH || + op->watch.cookie != lreq->linger_id); + op->watch.op = CEPH_OSD_WATCH_OP_RECONNECT; + op->watch.gen = ++lreq->register_gen; + dout("lreq %p reconnect register_gen %u\n", lreq, + op->watch.gen); + req->r_callback = linger_reconnect_cb; } else { - list_move_tail(&req->r_req_lru_item, &osdc->req_notarget); + if (!lreq->is_watch) + lreq->notify_id = 0; + else + WARN_ON(op->watch.op != CEPH_OSD_WATCH_OP_WATCH); + dout("lreq %p register\n", lreq); + req->r_callback = linger_commit_cb; } + mutex_unlock(&lreq->lock); + + req->r_priv = linger_get(lreq); + req->r_linger = true; + + submit_request(req, true); } -/* - * Pick an osd (the first 'up' osd in the pg), allocate the osd struct - * (as needed), and set the request r_osd appropriately. If there is - * no up osd, set r_osd to NULL. Move the request to the appropriate list - * (unsent, homeless) or leave on in-flight lru. - * - * Return 0 if unchanged, 1 if changed, or negative on error. - * - * Caller should hold map_sem for read and request_mutex. - */ -static int __map_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req, int force_resend) +static void linger_ping_cb(struct ceph_osd_request *req) { - struct ceph_pg pgid; - int acting[CEPH_PG_MAX_SIZE]; - int num, o; - int err; - bool was_paused; - - dout("map_request %p tid %lld\n", req, req->r_tid); - - err = __calc_request_pg(osdc->osdmap, req, &pgid); - if (err) { - list_move(&req->r_req_lru_item, &osdc->req_notarget); - return err; - } - req->r_pgid = pgid; - - num = ceph_calc_pg_acting(osdc->osdmap, pgid, acting, &o); - if (num < 0) - num = 0; - - was_paused = req->r_paused; - req->r_paused = __req_should_be_paused(osdc, req); - if (was_paused && !req->r_paused) - force_resend = 1; - - if ((!force_resend && - req->r_osd && req->r_osd->o_osd == o && - req->r_sent >= req->r_osd->o_incarnation && - req->r_num_pg_osds == num && - memcmp(req->r_pg_osds, acting, sizeof(acting[0])*num) == 0) || - (req->r_osd == NULL && o == -1) || - req->r_paused) - return 0; /* no change */ - - dout("map_request tid %llu pgid %lld.%x osd%d (was osd%d)\n", - req->r_tid, pgid.pool, pgid.seed, o, - req->r_osd ? req->r_osd->o_osd : -1); - - /* record full pg acting set */ - memcpy(req->r_pg_osds, acting, sizeof(acting[0]) * num); - req->r_num_pg_osds = num; - - if (req->r_osd) { - __cancel_request(req); - list_del_init(&req->r_osd_item); - list_del_init(&req->r_linger_osd_item); - req->r_osd = NULL; - } - - req->r_osd = __lookup_osd(osdc, o); - if (!req->r_osd && o >= 0) { - err = -ENOMEM; - req->r_osd = create_osd(osdc, o); - if (!req->r_osd) { - list_move(&req->r_req_lru_item, &osdc->req_notarget); - goto out; + struct ceph_osd_linger_request *lreq = req->r_priv; + + mutex_lock(&lreq->lock); + dout("%s lreq %p linger_id %llu result %d ping_sent %lu last_error %d\n", + __func__, lreq, lreq->linger_id, req->r_result, lreq->ping_sent, + lreq->last_error); + if (lreq->register_gen == req->r_ops[0].watch.gen) { + if (!req->r_result) { + lreq->watch_valid_thru = lreq->ping_sent; + } else if (!lreq->last_error) { + lreq->last_error = normalize_watch_error(req->r_result); + queue_watch_error(lreq); } + } else { + dout("lreq %p register_gen %u ignoring old pong %u\n", lreq, + lreq->register_gen, req->r_ops[0].watch.gen); + } + + mutex_unlock(&lreq->lock); + linger_put(lreq); +} - dout("map_request osd %p is osd%d\n", req->r_osd, o); - __insert_osd(osdc, req->r_osd); +static void send_linger_ping(struct ceph_osd_linger_request *lreq) +{ + struct ceph_osd_client *osdc = lreq->osdc; + struct ceph_osd_request *req = lreq->ping_req; + struct ceph_osd_req_op *op = &req->r_ops[0]; - ceph_con_open(&req->r_osd->o_con, - CEPH_ENTITY_TYPE_OSD, o, - &osdc->osdmap->osd_addr[o]); + if (ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD)) { + dout("%s PAUSERD\n", __func__); + return; } - __enqueue_request(req); - err = 1; /* osd or pg changed */ + lreq->ping_sent = jiffies; + dout("%s lreq %p linger_id %llu ping_sent %lu register_gen %u\n", + __func__, lreq, lreq->linger_id, lreq->ping_sent, + lreq->register_gen); -out: - return err; + if (req->r_osd) + cancel_linger_request(req); + + request_reinit(req); + target_copy(&req->r_t, &lreq->t); + + WARN_ON(op->op != CEPH_OSD_OP_WATCH || + op->watch.cookie != lreq->linger_id || + op->watch.op != CEPH_OSD_WATCH_OP_PING); + op->watch.gen = lreq->register_gen; + req->r_callback = linger_ping_cb; + req->r_priv = linger_get(lreq); + req->r_linger = true; + + ceph_osdc_get_request(req); + account_request(req); + req->r_tid = atomic64_inc_return(&osdc->last_tid); + link_request(lreq->osd, req); + send_request(req); } -/* - * caller should hold map_sem (for read) and request_mutex - */ -static void __send_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req) +static void linger_submit(struct ceph_osd_linger_request *lreq) { - void *p; + struct ceph_osd_client *osdc = lreq->osdc; + struct ceph_osd *osd; - dout("send_request %p tid %llu to osd%d flags %d pg %lld.%x\n", - req, req->r_tid, req->r_osd->o_osd, req->r_flags, - (unsigned long long)req->r_pgid.pool, req->r_pgid.seed); + calc_target(osdc, &lreq->t, &lreq->last_force_resend, false); + osd = lookup_create_osd(osdc, lreq->t.osd, true); + link_linger(osd, lreq); - /* fill in message content that changes each time we send it */ - put_unaligned_le32(osdc->osdmap->epoch, req->r_request_osdmap_epoch); - put_unaligned_le32(req->r_flags, req->r_request_flags); - put_unaligned_le64(req->r_target_oloc.pool, req->r_request_pool); - p = req->r_request_pgid; - ceph_encode_64(&p, req->r_pgid.pool); - ceph_encode_32(&p, req->r_pgid.seed); - put_unaligned_le64(1, req->r_request_attempts); /* FIXME */ - memcpy(req->r_request_reassert_version, &req->r_reassert_version, - sizeof(req->r_reassert_version)); + send_linger(lreq); +} - req->r_stamp = jiffies; - list_move_tail(&req->r_req_lru_item, &osdc->req_lru); +static void cancel_linger_map_check(struct ceph_osd_linger_request *lreq) +{ + struct ceph_osd_client *osdc = lreq->osdc; + struct ceph_osd_linger_request *lookup_lreq; - ceph_msg_get(req->r_request); /* send consumes a ref */ + verify_osdc_wrlocked(osdc); - req->r_sent = req->r_osd->o_incarnation; + lookup_lreq = lookup_linger_mc(&osdc->linger_map_checks, + lreq->linger_id); + if (!lookup_lreq) + return; - ceph_con_send(&req->r_osd->o_con, req->r_request); + WARN_ON(lookup_lreq != lreq); + erase_linger_mc(&osdc->linger_map_checks, lreq); + linger_put(lreq); } /* - * Send any requests in the queue (req_unsent). + * @lreq has to be both registered and linked. */ -static void __send_queued(struct ceph_osd_client *osdc) +static void __linger_cancel(struct ceph_osd_linger_request *lreq) { - struct ceph_osd_request *req, *tmp; + if (lreq->is_watch && lreq->ping_req->r_osd) + cancel_linger_request(lreq->ping_req); + if (lreq->reg_req->r_osd) + cancel_linger_request(lreq->reg_req); + cancel_linger_map_check(lreq); + unlink_linger(lreq->osd, lreq); + linger_unregister(lreq); +} + +static void linger_cancel(struct ceph_osd_linger_request *lreq) +{ + struct ceph_osd_client *osdc = lreq->osdc; - dout("__send_queued\n"); - list_for_each_entry_safe(req, tmp, &osdc->req_unsent, r_req_lru_item) - __send_request(osdc, req); + down_write(&osdc->lock); + if (__linger_registered(lreq)) + __linger_cancel(lreq); + up_write(&osdc->lock); } -/* - * Caller should hold map_sem for read and request_mutex. - */ -static int __ceph_osdc_start_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req, - bool nofail) -{ - int rc; - - __register_request(osdc, req); - req->r_sent = 0; - req->r_got_reply = 0; - rc = __map_request(osdc, req, 0); - if (rc < 0) { - if (nofail) { - dout("osdc_start_request failed map, " - " will retry %lld\n", req->r_tid); - rc = 0; - } else { - __unregister_request(osdc, req); - } - return rc; +static void send_linger_map_check(struct ceph_osd_linger_request *lreq); + +static void check_linger_pool_dne(struct ceph_osd_linger_request *lreq) +{ + struct ceph_osd_client *osdc = lreq->osdc; + struct ceph_osdmap *map = osdc->osdmap; + + verify_osdc_wrlocked(osdc); + WARN_ON(!map->epoch); + + if (lreq->register_gen) { + lreq->map_dne_bound = map->epoch; + dout("%s lreq %p linger_id %llu pool disappeared\n", __func__, + lreq, lreq->linger_id); + } else { + dout("%s lreq %p linger_id %llu map_dne_bound %u have %u\n", + __func__, lreq, lreq->linger_id, lreq->map_dne_bound, + map->epoch); } - if (req->r_osd == NULL) { - dout("send_request %p no up osds in pg\n", req); - ceph_monc_request_next_osdmap(&osdc->client->monc); + if (lreq->map_dne_bound) { + if (map->epoch >= lreq->map_dne_bound) { + /* we had a new enough map */ + pr_info("linger_id %llu pool does not exist\n", + lreq->linger_id); + linger_reg_commit_complete(lreq, -ENOENT); + __linger_cancel(lreq); + } } else { - __send_queued(osdc); + send_linger_map_check(lreq); } +} - return 0; +static void linger_map_check_cb(struct ceph_mon_generic_request *greq) +{ + struct ceph_osd_client *osdc = &greq->monc->client->osdc; + struct ceph_osd_linger_request *lreq; + u64 linger_id = greq->private_data; + + WARN_ON(greq->result || !greq->u.newest); + + down_write(&osdc->lock); + lreq = lookup_linger_mc(&osdc->linger_map_checks, linger_id); + if (!lreq) { + dout("%s linger_id %llu dne\n", __func__, linger_id); + goto out_unlock; + } + + dout("%s lreq %p linger_id %llu map_dne_bound %u newest %llu\n", + __func__, lreq, lreq->linger_id, lreq->map_dne_bound, + greq->u.newest); + if (!lreq->map_dne_bound) + lreq->map_dne_bound = greq->u.newest; + erase_linger_mc(&osdc->linger_map_checks, lreq); + check_linger_pool_dne(lreq); + + linger_put(lreq); +out_unlock: + up_write(&osdc->lock); +} + +static void send_linger_map_check(struct ceph_osd_linger_request *lreq) +{ + struct ceph_osd_client *osdc = lreq->osdc; + struct ceph_osd_linger_request *lookup_lreq; + int ret; + + verify_osdc_wrlocked(osdc); + + lookup_lreq = lookup_linger_mc(&osdc->linger_map_checks, + lreq->linger_id); + if (lookup_lreq) { + WARN_ON(lookup_lreq != lreq); + return; + } + + linger_get(lreq); + insert_linger_mc(&osdc->linger_map_checks, lreq); + ret = ceph_monc_get_version_async(&osdc->client->monc, "osdmap", + linger_map_check_cb, lreq->linger_id); + WARN_ON(ret); +} + +static int linger_reg_commit_wait(struct ceph_osd_linger_request *lreq) +{ + int ret; + + dout("%s lreq %p linger_id %llu\n", __func__, lreq, lreq->linger_id); + ret = wait_for_completion_interruptible(&lreq->reg_commit_wait); + return ret ?: lreq->reg_commit_error; +} + +static int linger_notify_finish_wait(struct ceph_osd_linger_request *lreq) +{ + int ret; + + dout("%s lreq %p linger_id %llu\n", __func__, lreq, lreq->linger_id); + ret = wait_for_completion_interruptible(&lreq->notify_finish_wait); + return ret ?: lreq->notify_finish_error; } /* - * Timeout callback, called every N seconds when 1 or more osd - * requests has been active for more than N seconds. When this - * happens, we ping all OSDs with requests who have timed out to - * ensure any communications channel reset is detected. Reset the - * request timeouts another N seconds in the future as we go. - * Reschedule the timeout event another N seconds in future (unless - * there are no open requests). + * Timeout callback, called every N seconds. When 1 or more OSD + * requests has been active for more than N seconds, we send a keepalive + * (tag + timestamp) to its OSD to ensure any communications channel + * reset is detected. */ static void handle_timeout(struct work_struct *work) { struct ceph_osd_client *osdc = container_of(work, struct ceph_osd_client, timeout_work.work); struct ceph_options *opts = osdc->client->options; - struct ceph_osd_request *req; - struct ceph_osd *osd; - struct list_head slow_osds; - dout("timeout\n"); - down_read(&osdc->map_sem); - - ceph_monc_request_next_osdmap(&osdc->client->monc); + unsigned long cutoff = jiffies - opts->osd_keepalive_timeout; + LIST_HEAD(slow_osds); + struct rb_node *n, *p; - mutex_lock(&osdc->request_mutex); + dout("%s osdc %p\n", __func__, osdc); + down_write(&osdc->lock); /* * ping osds that are a bit slow. this ensures that if there * is a break in the TCP connection we will notice, and reopen * a connection with that osd (from the fault callback). */ - INIT_LIST_HEAD(&slow_osds); - list_for_each_entry(req, &osdc->req_lru, r_req_lru_item) { - if (time_before(jiffies, - req->r_stamp + opts->osd_keepalive_timeout)) - break; + for (n = rb_first(&osdc->osds); n; n = rb_next(n)) { + struct ceph_osd *osd = rb_entry(n, struct ceph_osd, o_node); + bool found = false; + + for (p = rb_first(&osd->o_requests); p; p = rb_next(p)) { + struct ceph_osd_request *req = + rb_entry(p, struct ceph_osd_request, r_node); + + if (time_before(req->r_stamp, cutoff)) { + dout(" req %p tid %llu on osd%d is laggy\n", + req, req->r_tid, osd->o_osd); + found = true; + } + } + for (p = rb_first(&osd->o_linger_requests); p; p = rb_next(p)) { + struct ceph_osd_linger_request *lreq = + rb_entry(p, struct ceph_osd_linger_request, node); + + dout(" lreq %p linger_id %llu is served by osd%d\n", + lreq, lreq->linger_id, osd->o_osd); + found = true; + + mutex_lock(&lreq->lock); + if (lreq->is_watch && lreq->committed && !lreq->last_error) + send_linger_ping(lreq); + mutex_unlock(&lreq->lock); + } - osd = req->r_osd; - BUG_ON(!osd); - dout(" tid %llu is slow, will send keepalive on osd%d\n", - req->r_tid, osd->o_osd); - list_move_tail(&osd->o_keepalive_item, &slow_osds); + if (found) + list_move_tail(&osd->o_keepalive_item, &slow_osds); } + + if (atomic_read(&osdc->num_homeless) || !list_empty(&slow_osds)) + maybe_request_map(osdc); + while (!list_empty(&slow_osds)) { - osd = list_entry(slow_osds.next, struct ceph_osd, - o_keepalive_item); + struct ceph_osd *osd = list_first_entry(&slow_osds, + struct ceph_osd, + o_keepalive_item); list_del_init(&osd->o_keepalive_item); ceph_con_keepalive(&osd->o_con); } - __schedule_osd_timeout(osdc); - __send_queued(osdc); - mutex_unlock(&osdc->request_mutex); - up_read(&osdc->map_sem); + up_write(&osdc->lock); + schedule_delayed_work(&osdc->timeout_work, + osdc->client->options->osd_keepalive_timeout); } static void handle_osds_timeout(struct work_struct *work) @@ -1663,12 +2541,20 @@ static void handle_osds_timeout(struct work_struct *work) container_of(work, struct ceph_osd_client, osds_timeout_work.work); unsigned long delay = osdc->client->options->osd_idle_ttl / 4; + struct ceph_osd *osd, *nosd; - dout("osds timeout\n"); - down_read(&osdc->map_sem); - remove_old_osds(osdc); - up_read(&osdc->map_sem); + dout("%s osdc %p\n", __func__, osdc); + down_write(&osdc->lock); + list_for_each_entry_safe(osd, nosd, &osdc->osd_lru, o_osd_lru) { + if (time_before(jiffies, osd->lru_ttl)) + break; + WARN_ON(!RB_EMPTY_ROOT(&osd->o_requests)); + WARN_ON(!RB_EMPTY_ROOT(&osd->o_linger_requests)); + close_osd(osd); + } + + up_write(&osdc->lock); schedule_delayed_work(&osdc->osds_timeout_work, round_jiffies_relative(delay)); } @@ -1776,107 +2662,76 @@ e_inval: goto out; } -static void complete_request(struct ceph_osd_request *req) -{ - complete_all(&req->r_safe_completion); /* fsync waiter */ -} +struct MOSDOpReply { + struct ceph_pg pgid; + u64 flags; + int result; + u32 epoch; + int num_ops; + u32 outdata_len[CEPH_OSD_MAX_OPS]; + s32 rval[CEPH_OSD_MAX_OPS]; + int retry_attempt; + struct ceph_eversion replay_version; + u64 user_version; + struct ceph_request_redirect redirect; +}; -/* - * handle osd op reply. either call the callback if it is specified, - * or do the completion to wake up the waiting thread. - */ -static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg) +static int decode_MOSDOpReply(const struct ceph_msg *msg, struct MOSDOpReply *m) { - void *p, *end; - struct ceph_osd_request *req; - struct ceph_request_redirect redir; - u64 tid; - int object_len; - unsigned int numops; - int payload_len, flags; - s32 result; - s32 retry_attempt; - struct ceph_pg pg; - int err; - u32 reassert_epoch; - u64 reassert_version; - u32 osdmap_epoch; - int already_completed; - u32 bytes; + void *p = msg->front.iov_base; + void *const end = p + msg->front.iov_len; + u16 version = le16_to_cpu(msg->hdr.version); + struct ceph_eversion bad_replay_version; u8 decode_redir; - unsigned int i; - - tid = le64_to_cpu(msg->hdr.tid); - dout("handle_reply %p tid %llu\n", msg, tid); + u32 len; + int ret; + int i; - p = msg->front.iov_base; - end = p + msg->front.iov_len; + ceph_decode_32_safe(&p, end, len, e_inval); + ceph_decode_need(&p, end, len, e_inval); + p += len; /* skip oid */ - ceph_decode_need(&p, end, 4, bad); - object_len = ceph_decode_32(&p); - ceph_decode_need(&p, end, object_len, bad); - p += object_len; + ret = ceph_decode_pgid(&p, end, &m->pgid); + if (ret) + return ret; - err = ceph_decode_pgid(&p, end, &pg); - if (err) - goto bad; + ceph_decode_64_safe(&p, end, m->flags, e_inval); + ceph_decode_32_safe(&p, end, m->result, e_inval); + ceph_decode_need(&p, end, sizeof(bad_replay_version), e_inval); + memcpy(&bad_replay_version, p, sizeof(bad_replay_version)); + p += sizeof(bad_replay_version); + ceph_decode_32_safe(&p, end, m->epoch, e_inval); - ceph_decode_need(&p, end, 8 + 4 + 4 + 8 + 4, bad); - flags = ceph_decode_64(&p); - result = ceph_decode_32(&p); - reassert_epoch = ceph_decode_32(&p); - reassert_version = ceph_decode_64(&p); - osdmap_epoch = ceph_decode_32(&p); - - /* lookup */ - down_read(&osdc->map_sem); - mutex_lock(&osdc->request_mutex); - req = __lookup_request(osdc, tid); - if (req == NULL) { - dout("handle_reply tid %llu dne\n", tid); - goto bad_mutex; - } - ceph_osdc_get_request(req); + ceph_decode_32_safe(&p, end, m->num_ops, e_inval); + if (m->num_ops > ARRAY_SIZE(m->outdata_len)) + goto e_inval; - dout("handle_reply %p tid %llu req %p result %d\n", msg, tid, - req, result); - - ceph_decode_need(&p, end, 4, bad_put); - numops = ceph_decode_32(&p); - if (numops > CEPH_OSD_MAX_OPS) - goto bad_put; - if (numops != req->r_num_ops) - goto bad_put; - payload_len = 0; - ceph_decode_need(&p, end, numops * sizeof(struct ceph_osd_op), bad_put); - for (i = 0; i < numops; i++) { + ceph_decode_need(&p, end, m->num_ops * sizeof(struct ceph_osd_op), + e_inval); + for (i = 0; i < m->num_ops; i++) { struct ceph_osd_op *op = p; - int len; - len = le32_to_cpu(op->payload_len); - req->r_ops[i].outdata_len = len; - dout(" op %d has %d bytes\n", i, len); - payload_len += len; + m->outdata_len[i] = le32_to_cpu(op->payload_len); p += sizeof(*op); } - bytes = le32_to_cpu(msg->hdr.data_len); - if (payload_len != bytes) { - pr_warn("sum of op payload lens %d != data_len %d\n", - payload_len, bytes); - goto bad_put; - } - ceph_decode_need(&p, end, 4 + numops * 4, bad_put); - retry_attempt = ceph_decode_32(&p); - for (i = 0; i < numops; i++) - req->r_ops[i].rval = ceph_decode_32(&p); + ceph_decode_32_safe(&p, end, m->retry_attempt, e_inval); + for (i = 0; i < m->num_ops; i++) + ceph_decode_32_safe(&p, end, m->rval[i], e_inval); - if (le16_to_cpu(msg->hdr.version) >= 6) { - p += 8 + 4; /* skip replay_version */ - p += 8; /* skip user_version */ + if (version >= 5) { + ceph_decode_need(&p, end, sizeof(m->replay_version), e_inval); + memcpy(&m->replay_version, p, sizeof(m->replay_version)); + p += sizeof(m->replay_version); + ceph_decode_64_safe(&p, end, m->user_version, e_inval); + } else { + m->replay_version = bad_replay_version; /* struct */ + m->user_version = le64_to_cpu(m->replay_version.version); + } - if (le16_to_cpu(msg->hdr.version) >= 7) - ceph_decode_8_safe(&p, end, decode_redir, bad_put); + if (version >= 6) { + if (version >= 7) + ceph_decode_8_safe(&p, end, decode_redir, e_inval); else decode_redir = 1; } else { @@ -1884,228 +2739,410 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg) } if (decode_redir) { - err = ceph_redirect_decode(&p, end, &redir); - if (err) - goto bad_put; + ret = ceph_redirect_decode(&p, end, &m->redirect); + if (ret) + return ret; } else { - redir.oloc.pool = -1; + ceph_oloc_init(&m->redirect.oloc); } - if (redir.oloc.pool != -1) { - dout("redirect pool %lld\n", redir.oloc.pool); + return 0; - __unregister_request(osdc, req); +e_inval: + return -EINVAL; +} - req->r_target_oloc = redir.oloc; /* struct */ +/* + * We are done with @req if + * - @m is a safe reply, or + * - @m is an unsafe reply and we didn't want a safe one + */ +static bool done_request(const struct ceph_osd_request *req, + const struct MOSDOpReply *m) +{ + return (m->result < 0 || + (m->flags & CEPH_OSD_FLAG_ONDISK) || + !(req->r_flags & CEPH_OSD_FLAG_ONDISK)); +} - /* - * Start redirect requests with nofail=true. If - * mapping fails, request will end up on the notarget - * list, waiting for the new osdmap (which can take - * a while), even though the original request mapped - * successfully. In the future we might want to follow - * original request's nofail setting here. - */ - err = __ceph_osdc_start_request(osdc, req, true); - BUG_ON(err); +/* + * handle osd op reply. either call the callback if it is specified, + * or do the completion to wake up the waiting thread. + * + * ->r_unsafe_callback is set? yes no + * + * first reply is OK (needed r_cb/r_completion, r_cb/r_completion, + * any or needed/got safe) r_safe_completion r_safe_completion + * + * first reply is unsafe r_unsafe_cb(true) (nothing) + * + * when we get the safe reply r_unsafe_cb(false), r_cb/r_completion, + * r_safe_completion r_safe_completion + */ +static void handle_reply(struct ceph_osd *osd, struct ceph_msg *msg) +{ + struct ceph_osd_client *osdc = osd->o_osdc; + struct ceph_osd_request *req; + struct MOSDOpReply m; + u64 tid = le64_to_cpu(msg->hdr.tid); + u32 data_len = 0; + bool already_acked; + int ret; + int i; - goto out_unlock; - } + dout("%s msg %p tid %llu\n", __func__, msg, tid); - already_completed = req->r_got_reply; - if (!req->r_got_reply) { - req->r_result = result; - dout("handle_reply result %d bytes %d\n", req->r_result, - bytes); - if (req->r_result == 0) - req->r_result = bytes; + down_read(&osdc->lock); + if (!osd_registered(osd)) { + dout("%s osd%d unknown\n", __func__, osd->o_osd); + goto out_unlock_osdc; + } + WARN_ON(osd->o_osd != le64_to_cpu(msg->hdr.src.num)); - /* in case this is a write and we need to replay, */ - req->r_reassert_version.epoch = cpu_to_le32(reassert_epoch); - req->r_reassert_version.version = cpu_to_le64(reassert_version); + mutex_lock(&osd->lock); + req = lookup_request(&osd->o_requests, tid); + if (!req) { + dout("%s osd%d tid %llu unknown\n", __func__, osd->o_osd, tid); + goto out_unlock_session; + } - req->r_got_reply = 1; - } else if ((flags & CEPH_OSD_FLAG_ONDISK) == 0) { - dout("handle_reply tid %llu dup ack\n", tid); - goto out_unlock; + ret = decode_MOSDOpReply(msg, &m); + if (ret) { + pr_err("failed to decode MOSDOpReply for tid %llu: %d\n", + req->r_tid, ret); + ceph_msg_dump(msg); + goto fail_request; + } + dout("%s req %p tid %llu flags 0x%llx pgid %llu.%x epoch %u attempt %d v %u'%llu uv %llu\n", + __func__, req, req->r_tid, m.flags, m.pgid.pool, m.pgid.seed, + m.epoch, m.retry_attempt, le32_to_cpu(m.replay_version.epoch), + le64_to_cpu(m.replay_version.version), m.user_version); + + if (m.retry_attempt >= 0) { + if (m.retry_attempt != req->r_attempts - 1) { + dout("req %p tid %llu retry_attempt %d != %d, ignoring\n", + req, req->r_tid, m.retry_attempt, + req->r_attempts - 1); + goto out_unlock_session; + } + } else { + WARN_ON(1); /* MOSDOpReply v4 is assumed */ } - dout("handle_reply tid %llu flags %d\n", tid, flags); + if (!ceph_oloc_empty(&m.redirect.oloc)) { + dout("req %p tid %llu redirect pool %lld\n", req, req->r_tid, + m.redirect.oloc.pool); + unlink_request(osd, req); + mutex_unlock(&osd->lock); + + ceph_oloc_copy(&req->r_t.target_oloc, &m.redirect.oloc); + req->r_flags |= CEPH_OSD_FLAG_REDIRECTED; + req->r_tid = 0; + __submit_request(req, false); + goto out_unlock_osdc; + } - if (req->r_linger && (flags & CEPH_OSD_FLAG_ONDISK)) - __register_linger_request(osdc, req); + if (m.num_ops != req->r_num_ops) { + pr_err("num_ops %d != %d for tid %llu\n", m.num_ops, + req->r_num_ops, req->r_tid); + goto fail_request; + } + for (i = 0; i < req->r_num_ops; i++) { + dout(" req %p tid %llu op %d rval %d len %u\n", req, + req->r_tid, i, m.rval[i], m.outdata_len[i]); + req->r_ops[i].rval = m.rval[i]; + req->r_ops[i].outdata_len = m.outdata_len[i]; + data_len += m.outdata_len[i]; + } + if (data_len != le32_to_cpu(msg->hdr.data_len)) { + pr_err("sum of lens %u != %u for tid %llu\n", data_len, + le32_to_cpu(msg->hdr.data_len), req->r_tid); + goto fail_request; + } + dout("%s req %p tid %llu acked %d result %d data_len %u\n", __func__, + req, req->r_tid, req->r_got_reply, m.result, data_len); + + already_acked = req->r_got_reply; + if (!already_acked) { + req->r_result = m.result ?: data_len; + req->r_replay_version = m.replay_version; /* struct */ + req->r_got_reply = true; + } else if (!(m.flags & CEPH_OSD_FLAG_ONDISK)) { + dout("req %p tid %llu dup ack\n", req, req->r_tid); + goto out_unlock_session; + } - /* either this is a read, or we got the safe response */ - if (result < 0 || - (flags & CEPH_OSD_FLAG_ONDISK) || - ((flags & CEPH_OSD_FLAG_WRITE) == 0)) - __unregister_request(osdc, req); + if (done_request(req, &m)) { + __finish_request(req); + if (req->r_linger) { + WARN_ON(req->r_unsafe_callback); + dout("req %p tid %llu cb (locked)\n", req, req->r_tid); + __complete_request(req); + } + } - mutex_unlock(&osdc->request_mutex); - up_read(&osdc->map_sem); + mutex_unlock(&osd->lock); + up_read(&osdc->lock); - if (!already_completed) { - if (req->r_unsafe_callback && - result >= 0 && !(flags & CEPH_OSD_FLAG_ONDISK)) + if (done_request(req, &m)) { + if (already_acked && req->r_unsafe_callback) { + dout("req %p tid %llu safe-cb\n", req, req->r_tid); + req->r_unsafe_callback(req, false); + } else if (!req->r_linger) { + dout("req %p tid %llu cb\n", req, req->r_tid); + __complete_request(req); + } + if (m.flags & CEPH_OSD_FLAG_ONDISK) + complete_all(&req->r_safe_completion); + ceph_osdc_put_request(req); + } else { + if (req->r_unsafe_callback) { + dout("req %p tid %llu unsafe-cb\n", req, req->r_tid); req->r_unsafe_callback(req, true); - if (req->r_callback) - req->r_callback(req, msg); - else - complete_all(&req->r_completion); + } else { + WARN_ON(1); + } } - if (flags & CEPH_OSD_FLAG_ONDISK) { - if (req->r_unsafe_callback && already_completed) - req->r_unsafe_callback(req, false); - complete_request(req); + return; + +fail_request: + complete_request(req, -EIO); +out_unlock_session: + mutex_unlock(&osd->lock); +out_unlock_osdc: + up_read(&osdc->lock); +} + +static void set_pool_was_full(struct ceph_osd_client *osdc) +{ + struct rb_node *n; + + for (n = rb_first(&osdc->osdmap->pg_pools); n; n = rb_next(n)) { + struct ceph_pg_pool_info *pi = + rb_entry(n, struct ceph_pg_pool_info, node); + + pi->was_full = __pool_full(pi); } +} -out: - dout("req=%p req->r_linger=%d\n", req, req->r_linger); - ceph_osdc_put_request(req); - return; -out_unlock: - mutex_unlock(&osdc->request_mutex); - up_read(&osdc->map_sem); - goto out; +static bool pool_cleared_full(struct ceph_osd_client *osdc, s64 pool_id) +{ + struct ceph_pg_pool_info *pi; -bad_put: - req->r_result = -EIO; - __unregister_request(osdc, req); - if (req->r_callback) - req->r_callback(req, msg); - else - complete_all(&req->r_completion); - complete_request(req); - ceph_osdc_put_request(req); -bad_mutex: - mutex_unlock(&osdc->request_mutex); - up_read(&osdc->map_sem); -bad: - pr_err("corrupt osd_op_reply got %d %d\n", - (int)msg->front.iov_len, le32_to_cpu(msg->hdr.front_len)); - ceph_msg_dump(msg); + pi = ceph_pg_pool_by_id(osdc->osdmap, pool_id); + if (!pi) + return false; + + return pi->was_full && !__pool_full(pi); } -static void reset_changed_osds(struct ceph_osd_client *osdc) +static enum calc_target_result +recalc_linger_target(struct ceph_osd_linger_request *lreq) { - struct rb_node *p, *n; + struct ceph_osd_client *osdc = lreq->osdc; + enum calc_target_result ct_res; - dout("%s %p\n", __func__, osdc); - for (p = rb_first(&osdc->osds); p; p = n) { - struct ceph_osd *osd = rb_entry(p, struct ceph_osd, o_node); + ct_res = calc_target(osdc, &lreq->t, &lreq->last_force_resend, true); + if (ct_res == CALC_TARGET_NEED_RESEND) { + struct ceph_osd *osd; - n = rb_next(p); - if (!ceph_osd_is_up(osdc->osdmap, osd->o_osd) || - memcmp(&osd->o_con.peer_addr, - ceph_osd_addr(osdc->osdmap, - osd->o_osd), - sizeof(struct ceph_entity_addr)) != 0) - __reset_osd(osdc, osd); + osd = lookup_create_osd(osdc, lreq->t.osd, true); + if (osd != lreq->osd) { + unlink_linger(lreq->osd, lreq); + link_linger(osd, lreq); + } } + + return ct_res; } /* - * Requeue requests whose mapping to an OSD has changed. If requests map to - * no osd, request a new map. - * - * Caller should hold map_sem for read. + * Requeue requests whose mapping to an OSD has changed. */ -static void kick_requests(struct ceph_osd_client *osdc, bool force_resend, - bool force_resend_writes) +static void scan_requests(struct ceph_osd *osd, + bool force_resend, + bool cleared_full, + bool check_pool_cleared_full, + struct rb_root *need_resend, + struct list_head *need_resend_linger) { - struct ceph_osd_request *req, *nreq; - struct rb_node *p; - int needmap = 0; - int err; - bool force_resend_req; + struct ceph_osd_client *osdc = osd->o_osdc; + struct rb_node *n; + bool force_resend_writes; + + for (n = rb_first(&osd->o_linger_requests); n; ) { + struct ceph_osd_linger_request *lreq = + rb_entry(n, struct ceph_osd_linger_request, node); + enum calc_target_result ct_res; + + n = rb_next(n); /* recalc_linger_target() */ + + dout("%s lreq %p linger_id %llu\n", __func__, lreq, + lreq->linger_id); + ct_res = recalc_linger_target(lreq); + switch (ct_res) { + case CALC_TARGET_NO_ACTION: + force_resend_writes = cleared_full || + (check_pool_cleared_full && + pool_cleared_full(osdc, lreq->t.base_oloc.pool)); + if (!force_resend && !force_resend_writes) + break; + + /* fall through */ + case CALC_TARGET_NEED_RESEND: + cancel_linger_map_check(lreq); + /* + * scan_requests() for the previous epoch(s) + * may have already added it to the list, since + * it's not unlinked here. + */ + if (list_empty(&lreq->scan_item)) + list_add_tail(&lreq->scan_item, need_resend_linger); + break; + case CALC_TARGET_POOL_DNE: + check_linger_pool_dne(lreq); + break; + } + } - dout("kick_requests %s %s\n", force_resend ? " (force resend)" : "", - force_resend_writes ? " (force resend writes)" : ""); - mutex_lock(&osdc->request_mutex); - for (p = rb_first(&osdc->requests); p; ) { - req = rb_entry(p, struct ceph_osd_request, r_node); - p = rb_next(p); + for (n = rb_first(&osd->o_requests); n; ) { + struct ceph_osd_request *req = + rb_entry(n, struct ceph_osd_request, r_node); + enum calc_target_result ct_res; + + n = rb_next(n); /* unlink_request(), check_pool_dne() */ + + dout("%s req %p tid %llu\n", __func__, req, req->r_tid); + ct_res = calc_target(osdc, &req->r_t, + &req->r_last_force_resend, false); + switch (ct_res) { + case CALC_TARGET_NO_ACTION: + force_resend_writes = cleared_full || + (check_pool_cleared_full && + pool_cleared_full(osdc, req->r_t.base_oloc.pool)); + if (!force_resend && + (!(req->r_flags & CEPH_OSD_FLAG_WRITE) || + !force_resend_writes)) + break; + + /* fall through */ + case CALC_TARGET_NEED_RESEND: + cancel_map_check(req); + unlink_request(osd, req); + insert_request(need_resend, req); + break; + case CALC_TARGET_POOL_DNE: + check_pool_dne(req); + break; + } + } +} +static int handle_one_map(struct ceph_osd_client *osdc, + void *p, void *end, bool incremental, + struct rb_root *need_resend, + struct list_head *need_resend_linger) +{ + struct ceph_osdmap *newmap; + struct rb_node *n; + bool skipped_map = false; + bool was_full; + + was_full = ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL); + set_pool_was_full(osdc); + + if (incremental) + newmap = osdmap_apply_incremental(&p, end, osdc->osdmap); + else + newmap = ceph_osdmap_decode(&p, end); + if (IS_ERR(newmap)) + return PTR_ERR(newmap); + + if (newmap != osdc->osdmap) { /* - * For linger requests that have not yet been - * registered, move them to the linger list; they'll - * be sent to the osd in the loop below. Unregister - * the request before re-registering it as a linger - * request to ensure the __map_request() below - * will decide it needs to be sent. + * Preserve ->was_full before destroying the old map. + * For pools that weren't in the old map, ->was_full + * should be false. */ - if (req->r_linger && list_empty(&req->r_linger_item)) { - dout("%p tid %llu restart on osd%d\n", - req, req->r_tid, - req->r_osd ? req->r_osd->o_osd : -1); - ceph_osdc_get_request(req); - __unregister_request(osdc, req); - __register_linger_request(osdc, req); - ceph_osdc_put_request(req); - continue; + for (n = rb_first(&newmap->pg_pools); n; n = rb_next(n)) { + struct ceph_pg_pool_info *pi = + rb_entry(n, struct ceph_pg_pool_info, node); + struct ceph_pg_pool_info *old_pi; + + old_pi = ceph_pg_pool_by_id(osdc->osdmap, pi->id); + if (old_pi) + pi->was_full = old_pi->was_full; + else + WARN_ON(pi->was_full); } - force_resend_req = force_resend || - (force_resend_writes && - req->r_flags & CEPH_OSD_FLAG_WRITE); - err = __map_request(osdc, req, force_resend_req); - if (err < 0) - continue; /* error */ - if (req->r_osd == NULL) { - dout("%p tid %llu maps to no osd\n", req, req->r_tid); - needmap++; /* request a newer map */ - } else if (err > 0) { - if (!req->r_linger) { - dout("%p tid %llu requeued on osd%d\n", req, - req->r_tid, - req->r_osd ? req->r_osd->o_osd : -1); - req->r_flags |= CEPH_OSD_FLAG_RETRY; - } + if (osdc->osdmap->epoch && + osdc->osdmap->epoch + 1 < newmap->epoch) { + WARN_ON(incremental); + skipped_map = true; } + + ceph_osdmap_destroy(osdc->osdmap); + osdc->osdmap = newmap; } - list_for_each_entry_safe(req, nreq, &osdc->req_linger, - r_linger_item) { - dout("linger req=%p req->r_osd=%p\n", req, req->r_osd); - - err = __map_request(osdc, req, - force_resend || force_resend_writes); - dout("__map_request returned %d\n", err); - if (err < 0) - continue; /* hrm! */ - if (req->r_osd == NULL || err > 0) { - if (req->r_osd == NULL) { - dout("lingering %p tid %llu maps to no osd\n", - req, req->r_tid); - /* - * A homeless lingering request makes - * no sense, as it's job is to keep - * a particular OSD connection open. - * Request a newer map and kick the - * request, knowing that it won't be - * resent until we actually get a map - * that can tell us where to send it. - */ - needmap++; - } + was_full &= !ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL); + scan_requests(&osdc->homeless_osd, skipped_map, was_full, true, + need_resend, need_resend_linger); + + for (n = rb_first(&osdc->osds); n; ) { + struct ceph_osd *osd = rb_entry(n, struct ceph_osd, o_node); + + n = rb_next(n); /* close_osd() */ + + scan_requests(osd, skipped_map, was_full, true, need_resend, + need_resend_linger); + if (!ceph_osd_is_up(osdc->osdmap, osd->o_osd) || + memcmp(&osd->o_con.peer_addr, + ceph_osd_addr(osdc->osdmap, osd->o_osd), + sizeof(struct ceph_entity_addr))) + close_osd(osd); + } - dout("kicking lingering %p tid %llu osd%d\n", req, - req->r_tid, req->r_osd ? req->r_osd->o_osd : -1); - __register_request(osdc, req); - __unregister_linger_request(osdc, req); + return 0; +} + +static void kick_requests(struct ceph_osd_client *osdc, + struct rb_root *need_resend, + struct list_head *need_resend_linger) +{ + struct ceph_osd_linger_request *lreq, *nlreq; + struct rb_node *n; + + for (n = rb_first(need_resend); n; ) { + struct ceph_osd_request *req = + rb_entry(n, struct ceph_osd_request, r_node); + struct ceph_osd *osd; + + n = rb_next(n); + erase_request(need_resend, req); /* before link_request() */ + + WARN_ON(req->r_osd); + calc_target(osdc, &req->r_t, NULL, false); + osd = lookup_create_osd(osdc, req->r_t.osd, true); + link_request(osd, req); + if (!req->r_linger) { + if (!osd_homeless(osd) && !req->r_t.paused) + send_request(req); + } else { + cancel_linger_request(req); } } - reset_changed_osds(osdc); - mutex_unlock(&osdc->request_mutex); - if (needmap) { - dout("%d requests for down osds, need new map\n", needmap); - ceph_monc_request_next_osdmap(&osdc->client->monc); + list_for_each_entry_safe(lreq, nlreq, need_resend_linger, scan_item) { + if (!osd_homeless(lreq->osd)) + send_linger(lreq); + + list_del_init(&lreq->scan_item); } } - /* * Process updated osd map. * @@ -2115,27 +3152,31 @@ static void kick_requests(struct ceph_osd_client *osdc, bool force_resend, */ void ceph_osdc_handle_map(struct ceph_osd_client *osdc, struct ceph_msg *msg) { - void *p, *end, *next; + void *p = msg->front.iov_base; + void *const end = p + msg->front.iov_len; u32 nr_maps, maplen; u32 epoch; - struct ceph_osdmap *newmap = NULL, *oldmap; - int err; struct ceph_fsid fsid; - bool was_full; + struct rb_root need_resend = RB_ROOT; + LIST_HEAD(need_resend_linger); + bool handled_incremental = false; + bool was_pauserd, was_pausewr; + bool pauserd, pausewr; + int err; - dout("handle_map have %u\n", osdc->osdmap ? osdc->osdmap->epoch : 0); - p = msg->front.iov_base; - end = p + msg->front.iov_len; + dout("%s have %u\n", __func__, osdc->osdmap->epoch); + down_write(&osdc->lock); /* verify fsid */ ceph_decode_need(&p, end, sizeof(fsid), bad); ceph_decode_copy(&p, &fsid, sizeof(fsid)); if (ceph_check_fsid(osdc->client, &fsid) < 0) - return; - - down_write(&osdc->map_sem); + goto bad; - was_full = ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL); + was_pauserd = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD); + was_pausewr = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR) || + ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) || + have_pool_full(osdc); /* incremental maps */ ceph_decode_32_safe(&p, end, nr_maps, bad); @@ -2145,34 +3186,23 @@ void ceph_osdc_handle_map(struct ceph_osd_client *osdc, struct ceph_msg *msg) epoch = ceph_decode_32(&p); maplen = ceph_decode_32(&p); ceph_decode_need(&p, end, maplen, bad); - next = p + maplen; - if (osdc->osdmap && osdc->osdmap->epoch+1 == epoch) { + if (osdc->osdmap->epoch && + osdc->osdmap->epoch + 1 == epoch) { dout("applying incremental map %u len %d\n", epoch, maplen); - newmap = osdmap_apply_incremental(&p, next, - osdc->osdmap, - &osdc->client->msgr); - if (IS_ERR(newmap)) { - err = PTR_ERR(newmap); + err = handle_one_map(osdc, p, p + maplen, true, + &need_resend, &need_resend_linger); + if (err) goto bad; - } - BUG_ON(!newmap); - if (newmap != osdc->osdmap) { - ceph_osdmap_destroy(osdc->osdmap); - osdc->osdmap = newmap; - } - was_full = was_full || - ceph_osdmap_flag(osdc->osdmap, - CEPH_OSDMAP_FULL); - kick_requests(osdc, 0, was_full); + handled_incremental = true; } else { dout("ignoring incremental map %u len %d\n", epoch, maplen); } - p = next; + p += maplen; nr_maps--; } - if (newmap) + if (handled_incremental) goto done; /* full maps */ @@ -2186,455 +3216,647 @@ void ceph_osdc_handle_map(struct ceph_osd_client *osdc, struct ceph_msg *msg) if (nr_maps > 1) { dout("skipping non-latest full map %u len %d\n", epoch, maplen); - } else if (osdc->osdmap && osdc->osdmap->epoch >= epoch) { + } else if (osdc->osdmap->epoch >= epoch) { dout("skipping full map %u len %d, " "older than our %u\n", epoch, maplen, osdc->osdmap->epoch); } else { - int skipped_map = 0; - dout("taking full map %u len %d\n", epoch, maplen); - newmap = ceph_osdmap_decode(&p, p+maplen); - if (IS_ERR(newmap)) { - err = PTR_ERR(newmap); + err = handle_one_map(osdc, p, p + maplen, false, + &need_resend, &need_resend_linger); + if (err) goto bad; - } - BUG_ON(!newmap); - oldmap = osdc->osdmap; - osdc->osdmap = newmap; - if (oldmap) { - if (oldmap->epoch + 1 < newmap->epoch) - skipped_map = 1; - ceph_osdmap_destroy(oldmap); - } - was_full = was_full || - ceph_osdmap_flag(osdc->osdmap, - CEPH_OSDMAP_FULL); - kick_requests(osdc, skipped_map, was_full); } p += maplen; nr_maps--; } - if (!osdc->osdmap) - goto bad; done: - downgrade_write(&osdc->map_sem); - ceph_monc_got_map(&osdc->client->monc, CEPH_SUB_OSDMAP, - osdc->osdmap->epoch); - /* * subscribe to subsequent osdmap updates if full to ensure * we find out when we are no longer full and stop returning * ENOSPC. */ - if (ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL) || - ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSERD) || - ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_PAUSEWR)) - ceph_monc_request_next_osdmap(&osdc->client->monc); - - mutex_lock(&osdc->request_mutex); - __send_queued(osdc); - mutex_unlock(&osdc->request_mutex); - up_read(&osdc->map_sem); + pauserd = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSERD); + pausewr = ceph_osdmap_flag(osdc, CEPH_OSDMAP_PAUSEWR) || + ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) || + have_pool_full(osdc); + if (was_pauserd || was_pausewr || pauserd || pausewr) + maybe_request_map(osdc); + + kick_requests(osdc, &need_resend, &need_resend_linger); + + ceph_monc_got_map(&osdc->client->monc, CEPH_SUB_OSDMAP, + osdc->osdmap->epoch); + up_write(&osdc->lock); wake_up_all(&osdc->client->auth_wq); return; bad: pr_err("osdc handle_map corrupt msg\n"); ceph_msg_dump(msg); - up_write(&osdc->map_sem); + up_write(&osdc->lock); } /* - * watch/notify callback event infrastructure - * - * These callbacks are used both for watch and notify operations. + * Resubmit requests pending on the given osd. */ -static void __release_event(struct kref *kref) +static void kick_osd_requests(struct ceph_osd *osd) { - struct ceph_osd_event *event = - container_of(kref, struct ceph_osd_event, kref); + struct rb_node *n; - dout("__release_event %p\n", event); - kfree(event); -} + for (n = rb_first(&osd->o_requests); n; ) { + struct ceph_osd_request *req = + rb_entry(n, struct ceph_osd_request, r_node); -static void get_event(struct ceph_osd_event *event) -{ - kref_get(&event->kref); -} + n = rb_next(n); /* cancel_linger_request() */ -void ceph_osdc_put_event(struct ceph_osd_event *event) -{ - kref_put(&event->kref, __release_event); + if (!req->r_linger) { + if (!req->r_t.paused) + send_request(req); + } else { + cancel_linger_request(req); + } + } + for (n = rb_first(&osd->o_linger_requests); n; n = rb_next(n)) { + struct ceph_osd_linger_request *lreq = + rb_entry(n, struct ceph_osd_linger_request, node); + + send_linger(lreq); + } } -EXPORT_SYMBOL(ceph_osdc_put_event); -static void __insert_event(struct ceph_osd_client *osdc, - struct ceph_osd_event *new) +/* + * If the osd connection drops, we need to resubmit all requests. + */ +static void osd_fault(struct ceph_connection *con) { - struct rb_node **p = &osdc->event_tree.rb_node; - struct rb_node *parent = NULL; - struct ceph_osd_event *event = NULL; + struct ceph_osd *osd = con->private; + struct ceph_osd_client *osdc = osd->o_osdc; - while (*p) { - parent = *p; - event = rb_entry(parent, struct ceph_osd_event, node); - if (new->cookie < event->cookie) - p = &(*p)->rb_left; - else if (new->cookie > event->cookie) - p = &(*p)->rb_right; - else - BUG(); + dout("%s osd %p osd%d\n", __func__, osd, osd->o_osd); + + down_write(&osdc->lock); + if (!osd_registered(osd)) { + dout("%s osd%d unknown\n", __func__, osd->o_osd); + goto out_unlock; } - rb_link_node(&new->node, parent, p); - rb_insert_color(&new->node, &osdc->event_tree); + if (!reopen_osd(osd)) + kick_osd_requests(osd); + maybe_request_map(osdc); + +out_unlock: + up_write(&osdc->lock); } -static struct ceph_osd_event *__find_event(struct ceph_osd_client *osdc, - u64 cookie) +/* + * Process osd watch notifications + */ +static void handle_watch_notify(struct ceph_osd_client *osdc, + struct ceph_msg *msg) { - struct rb_node **p = &osdc->event_tree.rb_node; - struct rb_node *parent = NULL; - struct ceph_osd_event *event = NULL; + void *p = msg->front.iov_base; + void *const end = p + msg->front.iov_len; + struct ceph_osd_linger_request *lreq; + struct linger_work *lwork; + u8 proto_ver, opcode; + u64 cookie, notify_id; + u64 notifier_id = 0; + s32 return_code = 0; + void *payload = NULL; + u32 payload_len = 0; - while (*p) { - parent = *p; - event = rb_entry(parent, struct ceph_osd_event, node); - if (cookie < event->cookie) - p = &(*p)->rb_left; - else if (cookie > event->cookie) - p = &(*p)->rb_right; - else - return event; + ceph_decode_8_safe(&p, end, proto_ver, bad); + ceph_decode_8_safe(&p, end, opcode, bad); + ceph_decode_64_safe(&p, end, cookie, bad); + p += 8; /* skip ver */ + ceph_decode_64_safe(&p, end, notify_id, bad); + + if (proto_ver >= 1) { + ceph_decode_32_safe(&p, end, payload_len, bad); + ceph_decode_need(&p, end, payload_len, bad); + payload = p; + p += payload_len; } - return NULL; -} -static void __remove_event(struct ceph_osd_event *event) -{ - struct ceph_osd_client *osdc = event->osdc; + if (le16_to_cpu(msg->hdr.version) >= 2) + ceph_decode_32_safe(&p, end, return_code, bad); + + if (le16_to_cpu(msg->hdr.version) >= 3) + ceph_decode_64_safe(&p, end, notifier_id, bad); - if (!RB_EMPTY_NODE(&event->node)) { - dout("__remove_event removed %p\n", event); - rb_erase(&event->node, &osdc->event_tree); - ceph_osdc_put_event(event); + down_read(&osdc->lock); + lreq = lookup_linger_osdc(&osdc->linger_requests, cookie); + if (!lreq) { + dout("%s opcode %d cookie %llu dne\n", __func__, opcode, + cookie); + goto out_unlock_osdc; + } + + mutex_lock(&lreq->lock); + dout("%s opcode %d cookie %llu lreq %p is_watch %d\n", __func__, + opcode, cookie, lreq, lreq->is_watch); + if (opcode == CEPH_WATCH_EVENT_DISCONNECT) { + if (!lreq->last_error) { + lreq->last_error = -ENOTCONN; + queue_watch_error(lreq); + } + } else if (!lreq->is_watch) { + /* CEPH_WATCH_EVENT_NOTIFY_COMPLETE */ + if (lreq->notify_id && lreq->notify_id != notify_id) { + dout("lreq %p notify_id %llu != %llu, ignoring\n", lreq, + lreq->notify_id, notify_id); + } else if (!completion_done(&lreq->notify_finish_wait)) { + struct ceph_msg_data *data = + list_first_entry_or_null(&msg->data, + struct ceph_msg_data, + links); + + if (data) { + if (lreq->preply_pages) { + WARN_ON(data->type != + CEPH_MSG_DATA_PAGES); + *lreq->preply_pages = data->pages; + *lreq->preply_len = data->length; + } else { + ceph_release_page_vector(data->pages, + calc_pages_for(0, data->length)); + } + } + lreq->notify_finish_error = return_code; + complete_all(&lreq->notify_finish_wait); + } } else { - dout("__remove_event didn't remove %p\n", event); + /* CEPH_WATCH_EVENT_NOTIFY */ + lwork = lwork_alloc(lreq, do_watch_notify); + if (!lwork) { + pr_err("failed to allocate notify-lwork\n"); + goto out_unlock_lreq; + } + + lwork->notify.notify_id = notify_id; + lwork->notify.notifier_id = notifier_id; + lwork->notify.payload = payload; + lwork->notify.payload_len = payload_len; + lwork->notify.msg = ceph_msg_get(msg); + lwork_queue(lwork); } + +out_unlock_lreq: + mutex_unlock(&lreq->lock); +out_unlock_osdc: + up_read(&osdc->lock); + return; + +bad: + pr_err("osdc handle_watch_notify corrupt msg\n"); } -int ceph_osdc_create_event(struct ceph_osd_client *osdc, - void (*event_cb)(u64, u64, u8, void *), - void *data, struct ceph_osd_event **pevent) +/* + * Register request, send initial attempt. + */ +int ceph_osdc_start_request(struct ceph_osd_client *osdc, + struct ceph_osd_request *req, + bool nofail) { - struct ceph_osd_event *event; + down_read(&osdc->lock); + submit_request(req, false); + up_read(&osdc->lock); - event = kmalloc(sizeof(*event), GFP_NOIO); - if (!event) - return -ENOMEM; - - dout("create_event %p\n", event); - event->cb = event_cb; - event->one_shot = 0; - event->data = data; - event->osdc = osdc; - INIT_LIST_HEAD(&event->osd_node); - RB_CLEAR_NODE(&event->node); - kref_init(&event->kref); /* one ref for us */ - kref_get(&event->kref); /* one ref for the caller */ - - spin_lock(&osdc->event_lock); - event->cookie = ++osdc->event_count; - __insert_event(osdc, event); - spin_unlock(&osdc->event_lock); - - *pevent = event; return 0; } -EXPORT_SYMBOL(ceph_osdc_create_event); +EXPORT_SYMBOL(ceph_osdc_start_request); -void ceph_osdc_cancel_event(struct ceph_osd_event *event) +/* + * Unregister a registered request. The request is not completed (i.e. + * no callbacks or wakeups) - higher layers are supposed to know what + * they are canceling. + */ +void ceph_osdc_cancel_request(struct ceph_osd_request *req) { - struct ceph_osd_client *osdc = event->osdc; + struct ceph_osd_client *osdc = req->r_osdc; - dout("cancel_event %p\n", event); - spin_lock(&osdc->event_lock); - __remove_event(event); - spin_unlock(&osdc->event_lock); - ceph_osdc_put_event(event); /* caller's */ + down_write(&osdc->lock); + if (req->r_osd) + cancel_request(req); + up_write(&osdc->lock); } -EXPORT_SYMBOL(ceph_osdc_cancel_event); - +EXPORT_SYMBOL(ceph_osdc_cancel_request); -static void do_event_work(struct work_struct *work) +/* + * @timeout: in jiffies, 0 means "wait forever" + */ +static int wait_request_timeout(struct ceph_osd_request *req, + unsigned long timeout) { - struct ceph_osd_event_work *event_work = - container_of(work, struct ceph_osd_event_work, work); - struct ceph_osd_event *event = event_work->event; - u64 ver = event_work->ver; - u64 notify_id = event_work->notify_id; - u8 opcode = event_work->opcode; + long left; + + dout("%s req %p tid %llu\n", __func__, req, req->r_tid); + left = wait_for_completion_killable_timeout(&req->r_completion, + ceph_timeout_jiffies(timeout)); + if (left <= 0) { + left = left ?: -ETIMEDOUT; + ceph_osdc_cancel_request(req); - dout("do_event_work completing %p\n", event); - event->cb(ver, notify_id, opcode, event->data); - dout("do_event_work completed %p\n", event); - ceph_osdc_put_event(event); - kfree(event_work); + /* kludge - need to to wake ceph_osdc_sync() */ + complete_all(&req->r_safe_completion); + } else { + left = req->r_result; /* completed */ + } + + return left; } +/* + * wait for a request to complete + */ +int ceph_osdc_wait_request(struct ceph_osd_client *osdc, + struct ceph_osd_request *req) +{ + return wait_request_timeout(req, 0); +} +EXPORT_SYMBOL(ceph_osdc_wait_request); /* - * Process osd watch notifications + * sync - wait for all in-flight requests to flush. avoid starvation. */ -static void handle_watch_notify(struct ceph_osd_client *osdc, - struct ceph_msg *msg) +void ceph_osdc_sync(struct ceph_osd_client *osdc) { - void *p, *end; - u8 proto_ver; - u64 cookie, ver, notify_id; - u8 opcode; - struct ceph_osd_event *event; - struct ceph_osd_event_work *event_work; + struct rb_node *n, *p; + u64 last_tid = atomic64_read(&osdc->last_tid); - p = msg->front.iov_base; - end = p + msg->front.iov_len; +again: + down_read(&osdc->lock); + for (n = rb_first(&osdc->osds); n; n = rb_next(n)) { + struct ceph_osd *osd = rb_entry(n, struct ceph_osd, o_node); - ceph_decode_8_safe(&p, end, proto_ver, bad); - ceph_decode_8_safe(&p, end, opcode, bad); - ceph_decode_64_safe(&p, end, cookie, bad); - ceph_decode_64_safe(&p, end, ver, bad); - ceph_decode_64_safe(&p, end, notify_id, bad); + mutex_lock(&osd->lock); + for (p = rb_first(&osd->o_requests); p; p = rb_next(p)) { + struct ceph_osd_request *req = + rb_entry(p, struct ceph_osd_request, r_node); + + if (req->r_tid > last_tid) + break; - spin_lock(&osdc->event_lock); - event = __find_event(osdc, cookie); - if (event) { - BUG_ON(event->one_shot); - get_event(event); - } - spin_unlock(&osdc->event_lock); - dout("handle_watch_notify cookie %lld ver %lld event %p\n", - cookie, ver, event); - if (event) { - event_work = kmalloc(sizeof(*event_work), GFP_NOIO); - if (!event_work) { - pr_err("couldn't allocate event_work\n"); - ceph_osdc_put_event(event); - return; + if (!(req->r_flags & CEPH_OSD_FLAG_WRITE)) + continue; + + ceph_osdc_get_request(req); + mutex_unlock(&osd->lock); + up_read(&osdc->lock); + dout("%s waiting on req %p tid %llu last_tid %llu\n", + __func__, req, req->r_tid, last_tid); + wait_for_completion(&req->r_safe_completion); + ceph_osdc_put_request(req); + goto again; } - INIT_WORK(&event_work->work, do_event_work); - event_work->event = event; - event_work->ver = ver; - event_work->notify_id = notify_id; - event_work->opcode = opcode; - queue_work(osdc->notify_wq, &event_work->work); + mutex_unlock(&osd->lock); } - return; + up_read(&osdc->lock); + dout("%s done last_tid %llu\n", __func__, last_tid); +} +EXPORT_SYMBOL(ceph_osdc_sync); -bad: - pr_err("osdc handle_watch_notify corrupt msg\n"); +static struct ceph_osd_request * +alloc_linger_request(struct ceph_osd_linger_request *lreq) +{ + struct ceph_osd_request *req; + + req = ceph_osdc_alloc_request(lreq->osdc, NULL, 1, false, GFP_NOIO); + if (!req) + return NULL; + + ceph_oid_copy(&req->r_base_oid, &lreq->t.base_oid); + ceph_oloc_copy(&req->r_base_oloc, &lreq->t.base_oloc); + + if (ceph_osdc_alloc_messages(req, GFP_NOIO)) { + ceph_osdc_put_request(req); + return NULL; + } + + return req; } /* - * build new request AND message - * + * Returns a handle, caller owns a ref. */ -void ceph_osdc_build_request(struct ceph_osd_request *req, u64 off, - struct ceph_snap_context *snapc, u64 snap_id, - struct timespec *mtime) -{ - struct ceph_msg *msg = req->r_request; - void *p; - size_t msg_size; - int flags = req->r_flags; - u64 data_len; - unsigned int i; - - req->r_snapid = snap_id; - req->r_snapc = ceph_get_snap_context(snapc); - - /* encode request */ - msg->hdr.version = cpu_to_le16(4); - - p = msg->front.iov_base; - ceph_encode_32(&p, 1); /* client_inc is always 1 */ - req->r_request_osdmap_epoch = p; - p += 4; - req->r_request_flags = p; - p += 4; - if (req->r_flags & CEPH_OSD_FLAG_WRITE) - ceph_encode_timespec(p, mtime); - p += sizeof(struct ceph_timespec); - req->r_request_reassert_version = p; - p += sizeof(struct ceph_eversion); /* will get filled in */ - - /* oloc */ - ceph_encode_8(&p, 4); - ceph_encode_8(&p, 4); - ceph_encode_32(&p, 8 + 4 + 4); - req->r_request_pool = p; - p += 8; - ceph_encode_32(&p, -1); /* preferred */ - ceph_encode_32(&p, 0); /* key len */ +struct ceph_osd_linger_request * +ceph_osdc_watch(struct ceph_osd_client *osdc, + struct ceph_object_id *oid, + struct ceph_object_locator *oloc, + rados_watchcb2_t wcb, + rados_watcherrcb_t errcb, + void *data) +{ + struct ceph_osd_linger_request *lreq; + int ret; - ceph_encode_8(&p, 1); - req->r_request_pgid = p; - p += 8 + 4; - ceph_encode_32(&p, -1); /* preferred */ + lreq = linger_alloc(osdc); + if (!lreq) + return ERR_PTR(-ENOMEM); - /* oid */ - ceph_encode_32(&p, req->r_base_oid.name_len); - memcpy(p, req->r_base_oid.name, req->r_base_oid.name_len); - dout("oid '%.*s' len %d\n", req->r_base_oid.name_len, - req->r_base_oid.name, req->r_base_oid.name_len); - p += req->r_base_oid.name_len; - - /* ops--can imply data */ - ceph_encode_16(&p, (u16)req->r_num_ops); - data_len = 0; - for (i = 0; i < req->r_num_ops; i++) { - data_len += osd_req_encode_op(req, p, i); - p += sizeof(struct ceph_osd_op); + lreq->is_watch = true; + lreq->wcb = wcb; + lreq->errcb = errcb; + lreq->data = data; + lreq->watch_valid_thru = jiffies; + + ceph_oid_copy(&lreq->t.base_oid, oid); + ceph_oloc_copy(&lreq->t.base_oloc, oloc); + lreq->t.flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK; + lreq->mtime = CURRENT_TIME; + + lreq->reg_req = alloc_linger_request(lreq); + if (!lreq->reg_req) { + ret = -ENOMEM; + goto err_put_lreq; } - /* snaps */ - ceph_encode_64(&p, req->r_snapid); - ceph_encode_64(&p, req->r_snapc ? req->r_snapc->seq : 0); - ceph_encode_32(&p, req->r_snapc ? req->r_snapc->num_snaps : 0); - if (req->r_snapc) { - for (i = 0; i < snapc->num_snaps; i++) { - ceph_encode_64(&p, req->r_snapc->snaps[i]); - } + lreq->ping_req = alloc_linger_request(lreq); + if (!lreq->ping_req) { + ret = -ENOMEM; + goto err_put_lreq; } - req->r_request_attempts = p; - p += 4; - - /* data */ - if (flags & CEPH_OSD_FLAG_WRITE) { - u16 data_off; - - /* - * The header "data_off" is a hint to the receiver - * allowing it to align received data into its - * buffers such that there's no need to re-copy - * it before writing it to disk (direct I/O). - */ - data_off = (u16) (off & 0xffff); - req->r_request->hdr.data_off = cpu_to_le16(data_off); + down_write(&osdc->lock); + linger_register(lreq); /* before osd_req_op_* */ + osd_req_op_watch_init(lreq->reg_req, 0, lreq->linger_id, + CEPH_OSD_WATCH_OP_WATCH); + osd_req_op_watch_init(lreq->ping_req, 0, lreq->linger_id, + CEPH_OSD_WATCH_OP_PING); + linger_submit(lreq); + up_write(&osdc->lock); + + ret = linger_reg_commit_wait(lreq); + if (ret) { + linger_cancel(lreq); + goto err_put_lreq; } - req->r_request->hdr.data_len = cpu_to_le32(data_len); - BUG_ON(p > msg->front.iov_base + msg->front.iov_len); - msg_size = p - msg->front.iov_base; - msg->front.iov_len = msg_size; - msg->hdr.front_len = cpu_to_le32(msg_size); + return lreq; - dout("build_request msg_size was %d\n", (int)msg_size); +err_put_lreq: + linger_put(lreq); + return ERR_PTR(ret); } -EXPORT_SYMBOL(ceph_osdc_build_request); +EXPORT_SYMBOL(ceph_osdc_watch); /* - * Register request, send initial attempt. + * Releases a ref. + * + * Times out after mount_timeout to preserve rbd unmap behaviour + * introduced in 2894e1d76974 ("rbd: timeout watch teardown on unmap + * with mount_timeout"). */ -int ceph_osdc_start_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req, - bool nofail) +int ceph_osdc_unwatch(struct ceph_osd_client *osdc, + struct ceph_osd_linger_request *lreq) { - int rc; + struct ceph_options *opts = osdc->client->options; + struct ceph_osd_request *req; + int ret; - down_read(&osdc->map_sem); - mutex_lock(&osdc->request_mutex); + req = ceph_osdc_alloc_request(osdc, NULL, 1, false, GFP_NOIO); + if (!req) + return -ENOMEM; - rc = __ceph_osdc_start_request(osdc, req, nofail); + ceph_oid_copy(&req->r_base_oid, &lreq->t.base_oid); + ceph_oloc_copy(&req->r_base_oloc, &lreq->t.base_oloc); + req->r_flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK; + req->r_mtime = CURRENT_TIME; + osd_req_op_watch_init(req, 0, lreq->linger_id, + CEPH_OSD_WATCH_OP_UNWATCH); - mutex_unlock(&osdc->request_mutex); - up_read(&osdc->map_sem); + ret = ceph_osdc_alloc_messages(req, GFP_NOIO); + if (ret) + goto out_put_req; - return rc; + ceph_osdc_start_request(osdc, req, false); + linger_cancel(lreq); + linger_put(lreq); + ret = wait_request_timeout(req, opts->mount_timeout); + +out_put_req: + ceph_osdc_put_request(req); + return ret; } -EXPORT_SYMBOL(ceph_osdc_start_request); +EXPORT_SYMBOL(ceph_osdc_unwatch); -/* - * Unregister a registered request. The request is not completed (i.e. - * no callbacks or wakeups) - higher layers are supposed to know what - * they are canceling. - */ -void ceph_osdc_cancel_request(struct ceph_osd_request *req) +static int osd_req_op_notify_ack_init(struct ceph_osd_request *req, int which, + u64 notify_id, u64 cookie, void *payload, + size_t payload_len) { - struct ceph_osd_client *osdc = req->r_osdc; + struct ceph_osd_req_op *op; + struct ceph_pagelist *pl; + int ret; + + op = _osd_req_op_init(req, which, CEPH_OSD_OP_NOTIFY_ACK, 0); + + pl = kmalloc(sizeof(*pl), GFP_NOIO); + if (!pl) + return -ENOMEM; - mutex_lock(&osdc->request_mutex); - if (req->r_linger) - __unregister_linger_request(osdc, req); - __unregister_request(osdc, req); - mutex_unlock(&osdc->request_mutex); + ceph_pagelist_init(pl); + ret = ceph_pagelist_encode_64(pl, notify_id); + ret |= ceph_pagelist_encode_64(pl, cookie); + if (payload) { + ret |= ceph_pagelist_encode_32(pl, payload_len); + ret |= ceph_pagelist_append(pl, payload, payload_len); + } else { + ret |= ceph_pagelist_encode_32(pl, 0); + } + if (ret) { + ceph_pagelist_release(pl); + return -ENOMEM; + } - dout("%s %p tid %llu canceled\n", __func__, req, req->r_tid); + ceph_osd_data_pagelist_init(&op->notify_ack.request_data, pl); + op->indata_len = pl->length; + return 0; } -EXPORT_SYMBOL(ceph_osdc_cancel_request); -/* - * wait for a request to complete - */ -int ceph_osdc_wait_request(struct ceph_osd_client *osdc, - struct ceph_osd_request *req) +int ceph_osdc_notify_ack(struct ceph_osd_client *osdc, + struct ceph_object_id *oid, + struct ceph_object_locator *oloc, + u64 notify_id, + u64 cookie, + void *payload, + size_t payload_len) +{ + struct ceph_osd_request *req; + int ret; + + req = ceph_osdc_alloc_request(osdc, NULL, 1, false, GFP_NOIO); + if (!req) + return -ENOMEM; + + ceph_oid_copy(&req->r_base_oid, oid); + ceph_oloc_copy(&req->r_base_oloc, oloc); + req->r_flags = CEPH_OSD_FLAG_READ; + + ret = ceph_osdc_alloc_messages(req, GFP_NOIO); + if (ret) + goto out_put_req; + + ret = osd_req_op_notify_ack_init(req, 0, notify_id, cookie, payload, + payload_len); + if (ret) + goto out_put_req; + + ceph_osdc_start_request(osdc, req, false); + ret = ceph_osdc_wait_request(osdc, req); + +out_put_req: + ceph_osdc_put_request(req); + return ret; +} +EXPORT_SYMBOL(ceph_osdc_notify_ack); + +static int osd_req_op_notify_init(struct ceph_osd_request *req, int which, + u64 cookie, u32 prot_ver, u32 timeout, + void *payload, size_t payload_len) { - int rc; + struct ceph_osd_req_op *op; + struct ceph_pagelist *pl; + int ret; - dout("%s %p tid %llu\n", __func__, req, req->r_tid); + op = _osd_req_op_init(req, which, CEPH_OSD_OP_NOTIFY, 0); + op->notify.cookie = cookie; - rc = wait_for_completion_interruptible(&req->r_completion); - if (rc < 0) { - dout("%s %p tid %llu interrupted\n", __func__, req, req->r_tid); - ceph_osdc_cancel_request(req); - complete_request(req); - return rc; + pl = kmalloc(sizeof(*pl), GFP_NOIO); + if (!pl) + return -ENOMEM; + + ceph_pagelist_init(pl); + ret = ceph_pagelist_encode_32(pl, 1); /* prot_ver */ + ret |= ceph_pagelist_encode_32(pl, timeout); + ret |= ceph_pagelist_encode_32(pl, payload_len); + ret |= ceph_pagelist_append(pl, payload, payload_len); + if (ret) { + ceph_pagelist_release(pl); + return -ENOMEM; } - dout("%s %p tid %llu result %d\n", __func__, req, req->r_tid, - req->r_result); - return req->r_result; + ceph_osd_data_pagelist_init(&op->notify.request_data, pl); + op->indata_len = pl->length; + return 0; } -EXPORT_SYMBOL(ceph_osdc_wait_request); /* - * sync - wait for all in-flight requests to flush. avoid starvation. + * @timeout: in seconds + * + * @preply_{pages,len} are initialized both on success and error. + * The caller is responsible for: + * + * ceph_release_page_vector(reply_pages, calc_pages_for(0, reply_len)) */ -void ceph_osdc_sync(struct ceph_osd_client *osdc) +int ceph_osdc_notify(struct ceph_osd_client *osdc, + struct ceph_object_id *oid, + struct ceph_object_locator *oloc, + void *payload, + size_t payload_len, + u32 timeout, + struct page ***preply_pages, + size_t *preply_len) { - struct ceph_osd_request *req; - u64 last_tid, next_tid = 0; + struct ceph_osd_linger_request *lreq; + struct page **pages; + int ret; - mutex_lock(&osdc->request_mutex); - last_tid = osdc->last_tid; - while (1) { - req = __lookup_request_ge(osdc, next_tid); - if (!req) - break; - if (req->r_tid > last_tid) - break; + WARN_ON(!timeout); + if (preply_pages) { + *preply_pages = NULL; + *preply_len = 0; + } - next_tid = req->r_tid + 1; - if ((req->r_flags & CEPH_OSD_FLAG_WRITE) == 0) - continue; + lreq = linger_alloc(osdc); + if (!lreq) + return -ENOMEM; - ceph_osdc_get_request(req); - mutex_unlock(&osdc->request_mutex); - dout("sync waiting on tid %llu (last is %llu)\n", - req->r_tid, last_tid); - wait_for_completion(&req->r_safe_completion); - mutex_lock(&osdc->request_mutex); - ceph_osdc_put_request(req); + lreq->preply_pages = preply_pages; + lreq->preply_len = preply_len; + + ceph_oid_copy(&lreq->t.base_oid, oid); + ceph_oloc_copy(&lreq->t.base_oloc, oloc); + lreq->t.flags = CEPH_OSD_FLAG_READ; + + lreq->reg_req = alloc_linger_request(lreq); + if (!lreq->reg_req) { + ret = -ENOMEM; + goto out_put_lreq; } - mutex_unlock(&osdc->request_mutex); - dout("sync done (thru tid %llu)\n", last_tid); + + /* for notify_id */ + pages = ceph_alloc_page_vector(1, GFP_NOIO); + if (IS_ERR(pages)) { + ret = PTR_ERR(pages); + goto out_put_lreq; + } + + down_write(&osdc->lock); + linger_register(lreq); /* before osd_req_op_* */ + ret = osd_req_op_notify_init(lreq->reg_req, 0, lreq->linger_id, 1, + timeout, payload, payload_len); + if (ret) { + linger_unregister(lreq); + up_write(&osdc->lock); + ceph_release_page_vector(pages, 1); + goto out_put_lreq; + } + ceph_osd_data_pages_init(osd_req_op_data(lreq->reg_req, 0, notify, + response_data), + pages, PAGE_SIZE, 0, false, true); + linger_submit(lreq); + up_write(&osdc->lock); + + ret = linger_reg_commit_wait(lreq); + if (!ret) + ret = linger_notify_finish_wait(lreq); + else + dout("lreq %p failed to initiate notify %d\n", lreq, ret); + + linger_cancel(lreq); +out_put_lreq: + linger_put(lreq); + return ret; +} +EXPORT_SYMBOL(ceph_osdc_notify); + +/* + * Return the number of milliseconds since the watch was last + * confirmed, or an error. If there is an error, the watch is no + * longer valid, and should be destroyed with ceph_osdc_unwatch(). + */ +int ceph_osdc_watch_check(struct ceph_osd_client *osdc, + struct ceph_osd_linger_request *lreq) +{ + unsigned long stamp, age; + int ret; + + down_read(&osdc->lock); + mutex_lock(&lreq->lock); + stamp = lreq->watch_valid_thru; + if (!list_empty(&lreq->pending_lworks)) { + struct linger_work *lwork = + list_first_entry(&lreq->pending_lworks, + struct linger_work, + pending_item); + + if (time_before(lwork->queued_stamp, stamp)) + stamp = lwork->queued_stamp; + } + age = jiffies - stamp; + dout("%s lreq %p linger_id %llu age %lu last_error %d\n", __func__, + lreq, lreq->linger_id, age, lreq->last_error); + /* we are truncating to msecs, so return a safe upper bound */ + ret = lreq->last_error ?: 1 + jiffies_to_msecs(age); + + mutex_unlock(&lreq->lock); + up_read(&osdc->lock); + return ret; } -EXPORT_SYMBOL(ceph_osdc_sync); /* * Call all pending notify callbacks - for use after a watch is @@ -2646,6 +3868,13 @@ void ceph_osdc_flush_notifies(struct ceph_osd_client *osdc) } EXPORT_SYMBOL(ceph_osdc_flush_notifies); +void ceph_osdc_maybe_request_map(struct ceph_osd_client *osdc) +{ + down_read(&osdc->lock); + maybe_request_map(osdc); + up_read(&osdc->lock); +} +EXPORT_SYMBOL(ceph_osdc_maybe_request_map); /* * init, shutdown @@ -2656,43 +3885,35 @@ int ceph_osdc_init(struct ceph_osd_client *osdc, struct ceph_client *client) dout("init\n"); osdc->client = client; - osdc->osdmap = NULL; - init_rwsem(&osdc->map_sem); - init_completion(&osdc->map_waiters); - osdc->last_requested_map = 0; - mutex_init(&osdc->request_mutex); - osdc->last_tid = 0; + init_rwsem(&osdc->lock); osdc->osds = RB_ROOT; INIT_LIST_HEAD(&osdc->osd_lru); - osdc->requests = RB_ROOT; - INIT_LIST_HEAD(&osdc->req_lru); - INIT_LIST_HEAD(&osdc->req_unsent); - INIT_LIST_HEAD(&osdc->req_notarget); - INIT_LIST_HEAD(&osdc->req_linger); - osdc->num_requests = 0; + spin_lock_init(&osdc->osd_lru_lock); + osd_init(&osdc->homeless_osd); + osdc->homeless_osd.o_osdc = osdc; + osdc->homeless_osd.o_osd = CEPH_HOMELESS_OSD; + osdc->linger_requests = RB_ROOT; + osdc->map_checks = RB_ROOT; + osdc->linger_map_checks = RB_ROOT; INIT_DELAYED_WORK(&osdc->timeout_work, handle_timeout); INIT_DELAYED_WORK(&osdc->osds_timeout_work, handle_osds_timeout); - spin_lock_init(&osdc->event_lock); - osdc->event_tree = RB_ROOT; - osdc->event_count = 0; - - schedule_delayed_work(&osdc->osds_timeout_work, - round_jiffies_relative(osdc->client->options->osd_idle_ttl)); err = -ENOMEM; + osdc->osdmap = ceph_osdmap_alloc(); + if (!osdc->osdmap) + goto out; + osdc->req_mempool = mempool_create_slab_pool(10, ceph_osd_request_cache); if (!osdc->req_mempool) - goto out; + goto out_map; err = ceph_msgpool_init(&osdc->msgpool_op, CEPH_MSG_OSD_OP, - OSD_OP_FRONT_LEN, 10, true, - "osd_op"); + PAGE_SIZE, 10, true, "osd_op"); if (err < 0) goto out_mempool; err = ceph_msgpool_init(&osdc->msgpool_op_reply, CEPH_MSG_OSD_OPREPLY, - OSD_OPREPLY_FRONT_LEN, 10, true, - "osd_op_reply"); + PAGE_SIZE, 10, true, "osd_op_reply"); if (err < 0) goto out_msgpool; @@ -2701,6 +3922,11 @@ int ceph_osdc_init(struct ceph_osd_client *osdc, struct ceph_client *client) if (!osdc->notify_wq) goto out_msgpool_reply; + schedule_delayed_work(&osdc->timeout_work, + osdc->client->options->osd_keepalive_timeout); + schedule_delayed_work(&osdc->osds_timeout_work, + round_jiffies_relative(osdc->client->options->osd_idle_ttl)); + return 0; out_msgpool_reply: @@ -2709,6 +3935,8 @@ out_msgpool: ceph_msgpool_destroy(&osdc->msgpool_op); out_mempool: mempool_destroy(osdc->req_mempool); +out_map: + ceph_osdmap_destroy(osdc->osdmap); out: return err; } @@ -2719,11 +3947,25 @@ void ceph_osdc_stop(struct ceph_osd_client *osdc) destroy_workqueue(osdc->notify_wq); cancel_delayed_work_sync(&osdc->timeout_work); cancel_delayed_work_sync(&osdc->osds_timeout_work); - if (osdc->osdmap) { - ceph_osdmap_destroy(osdc->osdmap); - osdc->osdmap = NULL; + + down_write(&osdc->lock); + while (!RB_EMPTY_ROOT(&osdc->osds)) { + struct ceph_osd *osd = rb_entry(rb_first(&osdc->osds), + struct ceph_osd, o_node); + close_osd(osd); } - remove_all_osds(osdc); + up_write(&osdc->lock); + WARN_ON(atomic_read(&osdc->homeless_osd.o_ref) != 1); + osd_cleanup(&osdc->homeless_osd); + + WARN_ON(!list_empty(&osdc->osd_lru)); + WARN_ON(!RB_EMPTY_ROOT(&osdc->linger_requests)); + WARN_ON(!RB_EMPTY_ROOT(&osdc->map_checks)); + WARN_ON(!RB_EMPTY_ROOT(&osdc->linger_map_checks)); + WARN_ON(atomic_read(&osdc->num_requests)); + WARN_ON(atomic_read(&osdc->num_homeless)); + + ceph_osdmap_destroy(osdc->osdmap); mempool_destroy(osdc->req_mempool); ceph_msgpool_destroy(&osdc->msgpool_op); ceph_msgpool_destroy(&osdc->msgpool_op_reply); @@ -2752,15 +3994,12 @@ int ceph_osdc_readpages(struct ceph_osd_client *osdc, return PTR_ERR(req); /* it may be a short read due to an object boundary */ - osd_req_op_extent_osd_data_pages(req, 0, pages, *plen, page_align, false, false); dout("readpages final extent is %llu~%llu (%llu bytes align %d)\n", off, *plen, *plen, page_align); - ceph_osdc_build_request(req, off, NULL, vino.snap, NULL); - rc = ceph_osdc_start_request(osdc, req, false); if (!rc) rc = ceph_osdc_wait_request(osdc, req); @@ -2786,7 +4025,6 @@ int ceph_osdc_writepages(struct ceph_osd_client *osdc, struct ceph_vino vino, int rc = 0; int page_align = off & ~PAGE_MASK; - BUG_ON(vino.snap != CEPH_NOSNAP); /* snapshots aren't writeable */ req = ceph_osdc_new_request(osdc, layout, vino, off, &len, 0, 1, CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE, @@ -2800,8 +4038,7 @@ int ceph_osdc_writepages(struct ceph_osd_client *osdc, struct ceph_vino vino, false, false); dout("writepages %llu~%llu (%llu bytes)\n", off, len, len); - ceph_osdc_build_request(req, off, snapc, CEPH_NOSNAP, mtime); - + req->r_mtime = *mtime; rc = ceph_osdc_start_request(osdc, req, true); if (!rc) rc = ceph_osdc_wait_request(osdc, req); @@ -2841,19 +4078,15 @@ EXPORT_SYMBOL(ceph_osdc_cleanup); static void dispatch(struct ceph_connection *con, struct ceph_msg *msg) { struct ceph_osd *osd = con->private; - struct ceph_osd_client *osdc; + struct ceph_osd_client *osdc = osd->o_osdc; int type = le16_to_cpu(msg->hdr.type); - if (!osd) - goto out; - osdc = osd->o_osdc; - switch (type) { case CEPH_MSG_OSD_MAP: ceph_osdc_handle_map(osdc, msg); break; case CEPH_MSG_OSD_OPREPLY: - handle_reply(osdc, msg); + handle_reply(osd, msg); break; case CEPH_MSG_WATCH_NOTIFY: handle_watch_notify(osdc, msg); @@ -2863,7 +4096,7 @@ static void dispatch(struct ceph_connection *con, struct ceph_msg *msg) pr_err("received unknown message type %d %s\n", type, ceph_msg_type_name(type)); } -out: + ceph_msg_put(msg); } @@ -2878,21 +4111,27 @@ static struct ceph_msg *get_reply(struct ceph_connection *con, { struct ceph_osd *osd = con->private; struct ceph_osd_client *osdc = osd->o_osdc; - struct ceph_msg *m; + struct ceph_msg *m = NULL; struct ceph_osd_request *req; int front_len = le32_to_cpu(hdr->front_len); int data_len = le32_to_cpu(hdr->data_len); - u64 tid; + u64 tid = le64_to_cpu(hdr->tid); - tid = le64_to_cpu(hdr->tid); - mutex_lock(&osdc->request_mutex); - req = __lookup_request(osdc, tid); + down_read(&osdc->lock); + if (!osd_registered(osd)) { + dout("%s osd%d unknown, skipping\n", __func__, osd->o_osd); + *skip = 1; + goto out_unlock_osdc; + } + WARN_ON(osd->o_osd != le64_to_cpu(hdr->src.num)); + + mutex_lock(&osd->lock); + req = lookup_request(&osd->o_requests, tid); if (!req) { dout("%s osd%d tid %llu unknown, skipping\n", __func__, osd->o_osd, tid); - m = NULL; *skip = 1; - goto out; + goto out_unlock_session; } ceph_msg_revoke_incoming(req->r_reply); @@ -2904,7 +4143,7 @@ static struct ceph_msg *get_reply(struct ceph_connection *con, m = ceph_msg_new(CEPH_MSG_OSD_OPREPLY, front_len, GFP_NOFS, false); if (!m) - goto out; + goto out_unlock_session; ceph_msg_put(req->r_reply); req->r_reply = m; } @@ -2915,14 +4154,49 @@ static struct ceph_msg *get_reply(struct ceph_connection *con, req->r_reply->data_length); m = NULL; *skip = 1; - goto out; + goto out_unlock_session; } m = ceph_msg_get(req->r_reply); dout("get_reply tid %lld %p\n", tid, m); -out: - mutex_unlock(&osdc->request_mutex); +out_unlock_session: + mutex_unlock(&osd->lock); +out_unlock_osdc: + up_read(&osdc->lock); + return m; +} + +/* + * TODO: switch to a msg-owned pagelist + */ +static struct ceph_msg *alloc_msg_with_page_vector(struct ceph_msg_header *hdr) +{ + struct ceph_msg *m; + int type = le16_to_cpu(hdr->type); + u32 front_len = le32_to_cpu(hdr->front_len); + u32 data_len = le32_to_cpu(hdr->data_len); + + m = ceph_msg_new(type, front_len, GFP_NOIO, false); + if (!m) + return NULL; + + if (data_len) { + struct page **pages; + struct ceph_osd_data osd_data; + + pages = ceph_alloc_page_vector(calc_pages_for(0, data_len), + GFP_NOIO); + if (!pages) { + ceph_msg_put(m); + return NULL; + } + + ceph_osd_data_pages_init(&osd_data, pages, data_len, 0, false, + false); + ceph_osdc_msg_data_add(m, &osd_data); + } + return m; } @@ -2932,18 +4206,17 @@ static struct ceph_msg *alloc_msg(struct ceph_connection *con, { struct ceph_osd *osd = con->private; int type = le16_to_cpu(hdr->type); - int front = le32_to_cpu(hdr->front_len); *skip = 0; switch (type) { case CEPH_MSG_OSD_MAP: case CEPH_MSG_WATCH_NOTIFY: - return ceph_msg_new(type, front, GFP_NOFS, false); + return alloc_msg_with_page_vector(hdr); case CEPH_MSG_OSD_OPREPLY: return get_reply(con, hdr, skip); default: - pr_info("alloc_msg unexpected msg type %d from osd%d\n", type, - osd->o_osd); + pr_warn("%s osd%d unknown msg type %d, skipping\n", __func__, + osd->o_osd, type); *skip = 1; return NULL; } @@ -3047,5 +4320,5 @@ static const struct ceph_connection_operations osd_con_ops = { .alloc_msg = alloc_msg, .sign_message = osd_sign_message, .check_message_signature = osd_check_message_signature, - .fault = osd_reset, + .fault = osd_fault, }; diff --git a/net/ceph/osdmap.c b/net/ceph/osdmap.c index 243574c..03062bb 100644 --- a/net/ceph/osdmap.c +++ b/net/ceph/osdmap.c @@ -380,23 +380,24 @@ bad: return ERR_PTR(err); } -/* - * rbtree of pg_mapping for handling pg_temp (explicit mapping of pgid - * to a set of osds) and primary_temp (explicit primary setting) - */ -static int pgid_cmp(struct ceph_pg l, struct ceph_pg r) +int ceph_pg_compare(const struct ceph_pg *lhs, const struct ceph_pg *rhs) { - if (l.pool < r.pool) + if (lhs->pool < rhs->pool) return -1; - if (l.pool > r.pool) + if (lhs->pool > rhs->pool) return 1; - if (l.seed < r.seed) + if (lhs->seed < rhs->seed) return -1; - if (l.seed > r.seed) + if (lhs->seed > rhs->seed) return 1; + return 0; } +/* + * rbtree of pg_mapping for handling pg_temp (explicit mapping of pgid + * to a set of osds) and primary_temp (explicit primary setting) + */ static int __insert_pg_mapping(struct ceph_pg_mapping *new, struct rb_root *root) { @@ -409,7 +410,7 @@ static int __insert_pg_mapping(struct ceph_pg_mapping *new, while (*p) { parent = *p; pg = rb_entry(parent, struct ceph_pg_mapping, node); - c = pgid_cmp(new->pgid, pg->pgid); + c = ceph_pg_compare(&new->pgid, &pg->pgid); if (c < 0) p = &(*p)->rb_left; else if (c > 0) @@ -432,7 +433,7 @@ static struct ceph_pg_mapping *__lookup_pg_mapping(struct rb_root *root, while (n) { pg = rb_entry(n, struct ceph_pg_mapping, node); - c = pgid_cmp(pgid, pg->pgid); + c = ceph_pg_compare(&pgid, &pg->pgid); if (c < 0) { n = n->rb_left; } else if (c > 0) { @@ -596,7 +597,9 @@ static int decode_pool(void **p, void *end, struct ceph_pg_pool_info *pi) *p += 4; /* skip crash_replay_interval */ if (ev >= 7) - *p += 1; /* skip min_size */ + pi->min_size = ceph_decode_8(p); + else + pi->min_size = pi->size - pi->size / 2; if (ev >= 8) *p += 8 + 8; /* skip quota_max_* */ @@ -616,6 +619,50 @@ static int decode_pool(void **p, void *end, struct ceph_pg_pool_info *pi) pi->write_tier = -1; } + if (ev >= 10) { + /* skip properties */ + num = ceph_decode_32(p); + while (num--) { + len = ceph_decode_32(p); + *p += len; /* key */ + len = ceph_decode_32(p); + *p += len; /* val */ + } + } + + if (ev >= 11) { + /* skip hit_set_params */ + *p += 1 + 1; /* versions */ + len = ceph_decode_32(p); + *p += len; + + *p += 4; /* skip hit_set_period */ + *p += 4; /* skip hit_set_count */ + } + + if (ev >= 12) + *p += 4; /* skip stripe_width */ + + if (ev >= 13) { + *p += 8; /* skip target_max_bytes */ + *p += 8; /* skip target_max_objects */ + *p += 4; /* skip cache_target_dirty_ratio_micro */ + *p += 4; /* skip cache_target_full_ratio_micro */ + *p += 4; /* skip cache_min_flush_age */ + *p += 4; /* skip cache_min_evict_age */ + } + + if (ev >= 14) { + /* skip erasure_code_profile */ + len = ceph_decode_32(p); + *p += len; + } + + if (ev >= 15) + pi->last_force_request_resend = ceph_decode_32(p); + else + pi->last_force_request_resend = 0; + /* ignore the rest */ *p = pool_end; @@ -660,6 +707,23 @@ bad: /* * osd map */ +struct ceph_osdmap *ceph_osdmap_alloc(void) +{ + struct ceph_osdmap *map; + + map = kzalloc(sizeof(*map), GFP_NOIO); + if (!map) + return NULL; + + map->pg_pools = RB_ROOT; + map->pool_max = -1; + map->pg_temp = RB_ROOT; + map->primary_temp = RB_ROOT; + mutex_init(&map->crush_scratch_mutex); + + return map; +} + void ceph_osdmap_destroy(struct ceph_osdmap *map) { dout("osdmap_destroy %p\n", map); @@ -1183,14 +1247,10 @@ struct ceph_osdmap *ceph_osdmap_decode(void **p, void *end) struct ceph_osdmap *map; int ret; - map = kzalloc(sizeof(*map), GFP_NOFS); + map = ceph_osdmap_alloc(); if (!map) return ERR_PTR(-ENOMEM); - map->pg_temp = RB_ROOT; - map->primary_temp = RB_ROOT; - mutex_init(&map->crush_scratch_mutex); - ret = osdmap_decode(p, end, map); if (ret) { ceph_osdmap_destroy(map); @@ -1204,8 +1264,7 @@ struct ceph_osdmap *ceph_osdmap_decode(void **p, void *end) * decode and apply an incremental map update. */ struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end, - struct ceph_osdmap *map, - struct ceph_messenger *msgr) + struct ceph_osdmap *map) { struct crush_map *newcrush = NULL; struct ceph_fsid fsid; @@ -1381,8 +1440,252 @@ bad: return ERR_PTR(err); } +void ceph_oid_copy(struct ceph_object_id *dest, + const struct ceph_object_id *src) +{ + WARN_ON(!ceph_oid_empty(dest)); + + if (src->name != src->inline_name) { + /* very rare, see ceph_object_id definition */ + dest->name = kmalloc(src->name_len + 1, + GFP_NOIO | __GFP_NOFAIL); + } + + memcpy(dest->name, src->name, src->name_len + 1); + dest->name_len = src->name_len; +} +EXPORT_SYMBOL(ceph_oid_copy); + +static __printf(2, 0) +int oid_printf_vargs(struct ceph_object_id *oid, const char *fmt, va_list ap) +{ + int len; + + WARN_ON(!ceph_oid_empty(oid)); + + len = vsnprintf(oid->inline_name, sizeof(oid->inline_name), fmt, ap); + if (len >= sizeof(oid->inline_name)) + return len; + + oid->name_len = len; + return 0; +} + +/* + * If oid doesn't fit into inline buffer, BUG. + */ +void ceph_oid_printf(struct ceph_object_id *oid, const char *fmt, ...) +{ + va_list ap; + + va_start(ap, fmt); + BUG_ON(oid_printf_vargs(oid, fmt, ap)); + va_end(ap); +} +EXPORT_SYMBOL(ceph_oid_printf); + +static __printf(3, 0) +int oid_aprintf_vargs(struct ceph_object_id *oid, gfp_t gfp, + const char *fmt, va_list ap) +{ + va_list aq; + int len; + + va_copy(aq, ap); + len = oid_printf_vargs(oid, fmt, aq); + va_end(aq); + + if (len) { + char *external_name; + + external_name = kmalloc(len + 1, gfp); + if (!external_name) + return -ENOMEM; + + oid->name = external_name; + WARN_ON(vsnprintf(oid->name, len + 1, fmt, ap) != len); + oid->name_len = len; + } + + return 0; +} + +/* + * If oid doesn't fit into inline buffer, allocate. + */ +int ceph_oid_aprintf(struct ceph_object_id *oid, gfp_t gfp, + const char *fmt, ...) +{ + va_list ap; + int ret; + + va_start(ap, fmt); + ret = oid_aprintf_vargs(oid, gfp, fmt, ap); + va_end(ap); + + return ret; +} +EXPORT_SYMBOL(ceph_oid_aprintf); + +void ceph_oid_destroy(struct ceph_object_id *oid) +{ + if (oid->name != oid->inline_name) + kfree(oid->name); +} +EXPORT_SYMBOL(ceph_oid_destroy); + +/* + * osds only + */ +static bool __osds_equal(const struct ceph_osds *lhs, + const struct ceph_osds *rhs) +{ + if (lhs->size == rhs->size && + !memcmp(lhs->osds, rhs->osds, rhs->size * sizeof(rhs->osds[0]))) + return true; + + return false; +} + +/* + * osds + primary + */ +static bool osds_equal(const struct ceph_osds *lhs, + const struct ceph_osds *rhs) +{ + if (__osds_equal(lhs, rhs) && + lhs->primary == rhs->primary) + return true; + + return false; +} + +static bool osds_valid(const struct ceph_osds *set) +{ + /* non-empty set */ + if (set->size > 0 && set->primary >= 0) + return true; + + /* empty can_shift_osds set */ + if (!set->size && set->primary == -1) + return true; + + /* empty !can_shift_osds set - all NONE */ + if (set->size > 0 && set->primary == -1) { + int i; + + for (i = 0; i < set->size; i++) { + if (set->osds[i] != CRUSH_ITEM_NONE) + break; + } + if (i == set->size) + return true; + } + + return false; +} + +void ceph_osds_copy(struct ceph_osds *dest, const struct ceph_osds *src) +{ + memcpy(dest->osds, src->osds, src->size * sizeof(src->osds[0])); + dest->size = src->size; + dest->primary = src->primary; +} + +static bool is_split(const struct ceph_pg *pgid, + u32 old_pg_num, + u32 new_pg_num) +{ + int old_bits = calc_bits_of(old_pg_num); + int old_mask = (1 << old_bits) - 1; + int n; + + WARN_ON(pgid->seed >= old_pg_num); + if (new_pg_num <= old_pg_num) + return false; + + for (n = 1; ; n++) { + int next_bit = n << (old_bits - 1); + u32 s = next_bit | pgid->seed; + + if (s < old_pg_num || s == pgid->seed) + continue; + if (s >= new_pg_num) + break; + + s = ceph_stable_mod(s, old_pg_num, old_mask); + if (s == pgid->seed) + return true; + } + + return false; +} + +bool ceph_is_new_interval(const struct ceph_osds *old_acting, + const struct ceph_osds *new_acting, + const struct ceph_osds *old_up, + const struct ceph_osds *new_up, + int old_size, + int new_size, + int old_min_size, + int new_min_size, + u32 old_pg_num, + u32 new_pg_num, + bool old_sort_bitwise, + bool new_sort_bitwise, + const struct ceph_pg *pgid) +{ + return !osds_equal(old_acting, new_acting) || + !osds_equal(old_up, new_up) || + old_size != new_size || + old_min_size != new_min_size || + is_split(pgid, old_pg_num, new_pg_num) || + old_sort_bitwise != new_sort_bitwise; +} + +static int calc_pg_rank(int osd, const struct ceph_osds *acting) +{ + int i; + + for (i = 0; i < acting->size; i++) { + if (acting->osds[i] == osd) + return i; + } + + return -1; +} + +static bool primary_changed(const struct ceph_osds *old_acting, + const struct ceph_osds *new_acting) +{ + if (!old_acting->size && !new_acting->size) + return false; /* both still empty */ + if (!old_acting->size ^ !new_acting->size) + return true; /* was empty, now not, or vice versa */ + if (old_acting->primary != new_acting->primary) + return true; /* primary changed */ + + if (calc_pg_rank(old_acting->primary, old_acting) != + calc_pg_rank(new_acting->primary, new_acting)) + return true; + + return false; /* same primary (tho replicas may have changed) */ +} + +bool ceph_osds_changed(const struct ceph_osds *old_acting, + const struct ceph_osds *new_acting, + bool any_change) +{ + if (primary_changed(old_acting, new_acting)) + return true; + + if (any_change && !__osds_equal(old_acting, new_acting)) + return true; + + return false; +} /* * calculate file layout from given offset, length. @@ -1455,30 +1758,71 @@ invalid: EXPORT_SYMBOL(ceph_calc_file_object_mapping); /* - * Calculate mapping of a (oloc, oid) pair to a PG. Should only be - * called with target's (oloc, oid), since tiering isn't taken into - * account. + * Map an object into a PG. + * + * Should only be called with target_oid and target_oloc (as opposed to + * base_oid and base_oloc), since tiering isn't taken into account. */ -int ceph_oloc_oid_to_pg(struct ceph_osdmap *osdmap, - struct ceph_object_locator *oloc, - struct ceph_object_id *oid, - struct ceph_pg *pg_out) +int ceph_object_locator_to_pg(struct ceph_osdmap *osdmap, + struct ceph_object_id *oid, + struct ceph_object_locator *oloc, + struct ceph_pg *raw_pgid) { struct ceph_pg_pool_info *pi; - pi = __lookup_pg_pool(&osdmap->pg_pools, oloc->pool); + pi = ceph_pg_pool_by_id(osdmap, oloc->pool); if (!pi) - return -EIO; + return -ENOENT; - pg_out->pool = oloc->pool; - pg_out->seed = ceph_str_hash(pi->object_hash, oid->name, - oid->name_len); + raw_pgid->pool = oloc->pool; + raw_pgid->seed = ceph_str_hash(pi->object_hash, oid->name, + oid->name_len); - dout("%s '%.*s' pgid %llu.%x\n", __func__, oid->name_len, oid->name, - pg_out->pool, pg_out->seed); + dout("%s %s -> raw_pgid %llu.%x\n", __func__, oid->name, + raw_pgid->pool, raw_pgid->seed); return 0; } -EXPORT_SYMBOL(ceph_oloc_oid_to_pg); +EXPORT_SYMBOL(ceph_object_locator_to_pg); + +/* + * Map a raw PG (full precision ps) into an actual PG. + */ +static void raw_pg_to_pg(struct ceph_pg_pool_info *pi, + const struct ceph_pg *raw_pgid, + struct ceph_pg *pgid) +{ + pgid->pool = raw_pgid->pool; + pgid->seed = ceph_stable_mod(raw_pgid->seed, pi->pg_num, + pi->pg_num_mask); +} + +/* + * Map a raw PG (full precision ps) into a placement ps (placement + * seed). Include pool id in that value so that different pools don't + * use the same seeds. + */ +static u32 raw_pg_to_pps(struct ceph_pg_pool_info *pi, + const struct ceph_pg *raw_pgid) +{ + if (pi->flags & CEPH_POOL_FLAG_HASHPSPOOL) { + /* hash pool id and seed so that pool PGs do not overlap */ + return crush_hash32_2(CRUSH_HASH_RJENKINS1, + ceph_stable_mod(raw_pgid->seed, + pi->pgp_num, + pi->pgp_num_mask), + raw_pgid->pool); + } else { + /* + * legacy behavior: add ps and pool together. this is + * not a great approach because the PGs from each pool + * will overlap on top of each other: 0.5 == 1.4 == + * 2.3 == ... + */ + return ceph_stable_mod(raw_pgid->seed, pi->pgp_num, + pi->pgp_num_mask) + + (unsigned)raw_pgid->pool; + } +} static int do_crush(struct ceph_osdmap *map, int ruleno, int x, int *result, int result_max, @@ -1497,84 +1841,92 @@ static int do_crush(struct ceph_osdmap *map, int ruleno, int x, } /* - * Calculate raw (crush) set for given pgid. + * Calculate raw set (CRUSH output) for given PG. The result may + * contain nonexistent OSDs. ->primary is undefined for a raw set. * - * Return raw set length, or error. + * Placement seed (CRUSH input) is returned through @ppps. */ -static int pg_to_raw_osds(struct ceph_osdmap *osdmap, - struct ceph_pg_pool_info *pool, - struct ceph_pg pgid, u32 pps, int *osds) +static void pg_to_raw_osds(struct ceph_osdmap *osdmap, + struct ceph_pg_pool_info *pi, + const struct ceph_pg *raw_pgid, + struct ceph_osds *raw, + u32 *ppps) { + u32 pps = raw_pg_to_pps(pi, raw_pgid); int ruleno; int len; - /* crush */ - ruleno = crush_find_rule(osdmap->crush, pool->crush_ruleset, - pool->type, pool->size); + ceph_osds_init(raw); + if (ppps) + *ppps = pps; + + ruleno = crush_find_rule(osdmap->crush, pi->crush_ruleset, pi->type, + pi->size); if (ruleno < 0) { pr_err("no crush rule: pool %lld ruleset %d type %d size %d\n", - pgid.pool, pool->crush_ruleset, pool->type, - pool->size); - return -ENOENT; + pi->id, pi->crush_ruleset, pi->type, pi->size); + return; } - len = do_crush(osdmap, ruleno, pps, osds, - min_t(int, pool->size, CEPH_PG_MAX_SIZE), + len = do_crush(osdmap, ruleno, pps, raw->osds, + min_t(int, pi->size, ARRAY_SIZE(raw->osds)), osdmap->osd_weight, osdmap->max_osd); if (len < 0) { pr_err("error %d from crush rule %d: pool %lld ruleset %d type %d size %d\n", - len, ruleno, pgid.pool, pool->crush_ruleset, - pool->type, pool->size); - return len; + len, ruleno, pi->id, pi->crush_ruleset, pi->type, + pi->size); + return; } - return len; + raw->size = len; } /* - * Given raw set, calculate up set and up primary. + * Given raw set, calculate up set and up primary. By definition of an + * up set, the result won't contain nonexistent or down OSDs. * - * Return up set length. *primary is set to up primary osd id, or -1 - * if up set is empty. + * This is done in-place - on return @set is the up set. If it's + * empty, ->primary will remain undefined. */ -static int raw_to_up_osds(struct ceph_osdmap *osdmap, - struct ceph_pg_pool_info *pool, - int *osds, int len, int *primary) +static void raw_to_up_osds(struct ceph_osdmap *osdmap, + struct ceph_pg_pool_info *pi, + struct ceph_osds *set) { - int up_primary = -1; int i; - if (ceph_can_shift_osds(pool)) { + /* ->primary is undefined for a raw set */ + BUG_ON(set->primary != -1); + + if (ceph_can_shift_osds(pi)) { int removed = 0; - for (i = 0; i < len; i++) { - if (ceph_osd_is_down(osdmap, osds[i])) { + /* shift left */ + for (i = 0; i < set->size; i++) { + if (ceph_osd_is_down(osdmap, set->osds[i])) { removed++; continue; } if (removed) - osds[i - removed] = osds[i]; + set->osds[i - removed] = set->osds[i]; } - - len -= removed; - if (len > 0) - up_primary = osds[0]; + set->size -= removed; + if (set->size > 0) + set->primary = set->osds[0]; } else { - for (i = len - 1; i >= 0; i--) { - if (ceph_osd_is_down(osdmap, osds[i])) - osds[i] = CRUSH_ITEM_NONE; + /* set down/dne devices to NONE */ + for (i = set->size - 1; i >= 0; i--) { + if (ceph_osd_is_down(osdmap, set->osds[i])) + set->osds[i] = CRUSH_ITEM_NONE; else - up_primary = osds[i]; + set->primary = set->osds[i]; } } - - *primary = up_primary; - return len; } -static void apply_primary_affinity(struct ceph_osdmap *osdmap, u32 pps, - struct ceph_pg_pool_info *pool, - int *osds, int len, int *primary) +static void apply_primary_affinity(struct ceph_osdmap *osdmap, + struct ceph_pg_pool_info *pi, + u32 pps, + struct ceph_osds *up) { int i; int pos = -1; @@ -1586,8 +1938,8 @@ static void apply_primary_affinity(struct ceph_osdmap *osdmap, u32 pps, if (!osdmap->osd_primary_affinity) return; - for (i = 0; i < len; i++) { - int osd = osds[i]; + for (i = 0; i < up->size; i++) { + int osd = up->osds[i]; if (osd != CRUSH_ITEM_NONE && osdmap->osd_primary_affinity[osd] != @@ -1595,7 +1947,7 @@ static void apply_primary_affinity(struct ceph_osdmap *osdmap, u32 pps, break; } } - if (i == len) + if (i == up->size) return; /* @@ -1603,8 +1955,8 @@ static void apply_primary_affinity(struct ceph_osdmap *osdmap, u32 pps, * osd into the hash/rng so that a proportional fraction of an * osd's pgs get rejected as primary. */ - for (i = 0; i < len; i++) { - int osd = osds[i]; + for (i = 0; i < up->size; i++) { + int osd = up->osds[i]; u32 aff; if (osd == CRUSH_ITEM_NONE) @@ -1629,135 +1981,110 @@ static void apply_primary_affinity(struct ceph_osdmap *osdmap, u32 pps, if (pos < 0) return; - *primary = osds[pos]; + up->primary = up->osds[pos]; - if (ceph_can_shift_osds(pool) && pos > 0) { + if (ceph_can_shift_osds(pi) && pos > 0) { /* move the new primary to the front */ for (i = pos; i > 0; i--) - osds[i] = osds[i - 1]; - osds[0] = *primary; + up->osds[i] = up->osds[i - 1]; + up->osds[0] = up->primary; } } /* - * Given up set, apply pg_temp and primary_temp mappings. + * Get pg_temp and primary_temp mappings for given PG. * - * Return acting set length. *primary is set to acting primary osd id, - * or -1 if acting set is empty. + * Note that a PG may have none, only pg_temp, only primary_temp or + * both pg_temp and primary_temp mappings. This means @temp isn't + * always a valid OSD set on return: in the "only primary_temp" case, + * @temp will have its ->primary >= 0 but ->size == 0. */ -static int apply_temps(struct ceph_osdmap *osdmap, - struct ceph_pg_pool_info *pool, struct ceph_pg pgid, - int *osds, int len, int *primary) +static void get_temp_osds(struct ceph_osdmap *osdmap, + struct ceph_pg_pool_info *pi, + const struct ceph_pg *raw_pgid, + struct ceph_osds *temp) { + struct ceph_pg pgid; struct ceph_pg_mapping *pg; - int temp_len; - int temp_primary; int i; - /* raw_pg -> pg */ - pgid.seed = ceph_stable_mod(pgid.seed, pool->pg_num, - pool->pg_num_mask); + raw_pg_to_pg(pi, raw_pgid, &pgid); + ceph_osds_init(temp); /* pg_temp? */ pg = __lookup_pg_mapping(&osdmap->pg_temp, pgid); if (pg) { - temp_len = 0; - temp_primary = -1; - for (i = 0; i < pg->pg_temp.len; i++) { if (ceph_osd_is_down(osdmap, pg->pg_temp.osds[i])) { - if (ceph_can_shift_osds(pool)) + if (ceph_can_shift_osds(pi)) continue; - else - osds[temp_len++] = CRUSH_ITEM_NONE; + + temp->osds[temp->size++] = CRUSH_ITEM_NONE; } else { - osds[temp_len++] = pg->pg_temp.osds[i]; + temp->osds[temp->size++] = pg->pg_temp.osds[i]; } } /* apply pg_temp's primary */ - for (i = 0; i < temp_len; i++) { - if (osds[i] != CRUSH_ITEM_NONE) { - temp_primary = osds[i]; + for (i = 0; i < temp->size; i++) { + if (temp->osds[i] != CRUSH_ITEM_NONE) { + temp->primary = temp->osds[i]; break; } } - } else { - temp_len = len; - temp_primary = *primary; } /* primary_temp? */ pg = __lookup_pg_mapping(&osdmap->primary_temp, pgid); if (pg) - temp_primary = pg->primary_temp.osd; - - *primary = temp_primary; - return temp_len; + temp->primary = pg->primary_temp.osd; } /* - * Calculate acting set for given pgid. + * Map a PG to its acting set as well as its up set. * - * Return acting set length, or error. *primary is set to acting - * primary osd id, or -1 if acting set is empty or on error. + * Acting set is used for data mapping purposes, while up set can be + * recorded for detecting interval changes and deciding whether to + * resend a request. */ -int ceph_calc_pg_acting(struct ceph_osdmap *osdmap, struct ceph_pg pgid, - int *osds, int *primary) +void ceph_pg_to_up_acting_osds(struct ceph_osdmap *osdmap, + const struct ceph_pg *raw_pgid, + struct ceph_osds *up, + struct ceph_osds *acting) { - struct ceph_pg_pool_info *pool; + struct ceph_pg_pool_info *pi; u32 pps; - int len; - pool = __lookup_pg_pool(&osdmap->pg_pools, pgid.pool); - if (!pool) { - *primary = -1; - return -ENOENT; + pi = ceph_pg_pool_by_id(osdmap, raw_pgid->pool); + if (!pi) { + ceph_osds_init(up); + ceph_osds_init(acting); + goto out; } - if (pool->flags & CEPH_POOL_FLAG_HASHPSPOOL) { - /* hash pool id and seed so that pool PGs do not overlap */ - pps = crush_hash32_2(CRUSH_HASH_RJENKINS1, - ceph_stable_mod(pgid.seed, pool->pgp_num, - pool->pgp_num_mask), - pgid.pool); - } else { - /* - * legacy behavior: add ps and pool together. this is - * not a great approach because the PGs from each pool - * will overlap on top of each other: 0.5 == 1.4 == - * 2.3 == ... - */ - pps = ceph_stable_mod(pgid.seed, pool->pgp_num, - pool->pgp_num_mask) + - (unsigned)pgid.pool; - } - - len = pg_to_raw_osds(osdmap, pool, pgid, pps, osds); - if (len < 0) { - *primary = -1; - return len; + pg_to_raw_osds(osdmap, pi, raw_pgid, up, &pps); + raw_to_up_osds(osdmap, pi, up); + apply_primary_affinity(osdmap, pi, pps, up); + get_temp_osds(osdmap, pi, raw_pgid, acting); + if (!acting->size) { + memcpy(acting->osds, up->osds, up->size * sizeof(up->osds[0])); + acting->size = up->size; + if (acting->primary == -1) + acting->primary = up->primary; } - - len = raw_to_up_osds(osdmap, pool, osds, len, primary); - - apply_primary_affinity(osdmap, pps, pool, osds, len, primary); - - len = apply_temps(osdmap, pool, pgid, osds, len, primary); - - return len; +out: + WARN_ON(!osds_valid(up) || !osds_valid(acting)); } /* - * Return primary osd for given pgid, or -1 if none. + * Return acting primary for given PG, or -1 if none. */ -int ceph_calc_pg_primary(struct ceph_osdmap *osdmap, struct ceph_pg pgid) +int ceph_pg_to_acting_primary(struct ceph_osdmap *osdmap, + const struct ceph_pg *raw_pgid) { - int osds[CEPH_PG_MAX_SIZE]; - int primary; - - ceph_calc_pg_acting(osdmap, pgid, osds, &primary); + struct ceph_osds up, acting; - return primary; + ceph_pg_to_up_acting_osds(osdmap, raw_pgid, &up, &acting); + return acting.primary; } -EXPORT_SYMBOL(ceph_calc_pg_primary); +EXPORT_SYMBOL(ceph_pg_to_acting_primary); diff --git a/net/compat.c b/net/compat.c index 5cfd26a..1cd2ec0 100644 --- a/net/compat.c +++ b/net/compat.c @@ -309,8 +309,8 @@ void scm_detach_fds_compat(struct msghdr *kmsg, struct scm_cookie *scm) __scm_destroy(scm); } -static int do_set_attach_filter(struct socket *sock, int level, int optname, - char __user *optval, unsigned int optlen) +/* allocate a 64-bit sock_fprog on the user stack for duration of syscall. */ +struct sock_fprog __user *get_compat_bpf_fprog(char __user *optval) { struct compat_sock_fprog __user *fprog32 = (struct compat_sock_fprog __user *)optval; struct sock_fprog __user *kfprog = compat_alloc_user_space(sizeof(struct sock_fprog)); @@ -323,6 +323,19 @@ static int do_set_attach_filter(struct socket *sock, int level, int optname, __get_user(ptr, &fprog32->filter) || __put_user(len, &kfprog->len) || __put_user(compat_ptr(ptr), &kfprog->filter)) + return NULL; + + return kfprog; +} +EXPORT_SYMBOL_GPL(get_compat_bpf_fprog); + +static int do_set_attach_filter(struct socket *sock, int level, int optname, + char __user *optval, unsigned int optlen) +{ + struct sock_fprog __user *kfprog; + + kfprog = get_compat_bpf_fprog(optval); + if (!kfprog) return -EFAULT; return sock_setsockopt(sock, level, optname, (char __user *)kfprog, @@ -354,7 +367,8 @@ static int do_set_sock_timeout(struct socket *sock, int level, static int compat_sock_setsockopt(struct socket *sock, int level, int optname, char __user *optval, unsigned int optlen) { - if (optname == SO_ATTACH_FILTER) + if (optname == SO_ATTACH_FILTER || + optname == SO_ATTACH_REUSEPORT_CBPF) return do_set_attach_filter(sock, level, optname, optval, optlen); if (optname == SO_RCVTIMEO || optname == SO_SNDTIMEO) diff --git a/net/core/filter.c b/net/core/filter.c index 68adb5f..c4b330c 100644 --- a/net/core/filter.c +++ b/net/core/filter.c @@ -2085,7 +2085,8 @@ static bool __is_valid_access(int off, int size, enum bpf_access_type type) } static bool sk_filter_is_valid_access(int off, int size, - enum bpf_access_type type) + enum bpf_access_type type, + enum bpf_reg_type *reg_type) { switch (off) { case offsetof(struct __sk_buff, tc_classid): @@ -2108,7 +2109,8 @@ static bool sk_filter_is_valid_access(int off, int size, } static bool tc_cls_act_is_valid_access(int off, int size, - enum bpf_access_type type) + enum bpf_access_type type, + enum bpf_reg_type *reg_type) { if (type == BPF_WRITE) { switch (off) { @@ -2123,6 +2125,16 @@ static bool tc_cls_act_is_valid_access(int off, int size, return false; } } + + switch (off) { + case offsetof(struct __sk_buff, data): + *reg_type = PTR_TO_PACKET; + break; + case offsetof(struct __sk_buff, data_end): + *reg_type = PTR_TO_PACKET_END; + break; + } + return __is_valid_access(off, size, type); } diff --git a/net/core/gen_stats.c b/net/core/gen_stats.c index f96ee8b..be873e4 100644 --- a/net/core/gen_stats.c +++ b/net/core/gen_stats.c @@ -47,6 +47,7 @@ nla_put_failure: * @xstats_type: TLV type for backward compatibility xstats TLV * @lock: statistics lock * @d: dumping handle + * @padattr: padding attribute * * Initializes the dumping handle, grabs the statistic lock and appends * an empty TLV header to the socket buffer for use a container for all @@ -87,6 +88,7 @@ EXPORT_SYMBOL(gnet_stats_start_copy_compat); * @type: TLV type for top level statistic TLV * @lock: statistics lock * @d: dumping handle + * @padattr: padding attribute * * Initializes the dumping handle, grabs the statistic lock and appends * an empty TLV header to the socket buffer for use a container for all diff --git a/net/core/neighbour.c b/net/core/neighbour.c index 29dd8cc..510cd62 100644 --- a/net/core/neighbour.c +++ b/net/core/neighbour.c @@ -2469,13 +2469,17 @@ int neigh_xmit(int index, struct net_device *dev, tbl = neigh_tables[index]; if (!tbl) goto out; + rcu_read_lock_bh(); neigh = __neigh_lookup_noref(tbl, addr, dev); if (!neigh) neigh = __neigh_create(tbl, addr, dev, false); err = PTR_ERR(neigh); - if (IS_ERR(neigh)) + if (IS_ERR(neigh)) { + rcu_read_unlock_bh(); goto out_kfree_skb; + } err = neigh->output(neigh, skb); + rcu_read_unlock_bh(); } else if (index == NEIGH_LINK_TABLE) { err = dev_hard_header(skb, dev, ntohs(skb->protocol), diff --git a/net/core/net-sysfs.c b/net/core/net-sysfs.c index 2b3f76f..7a0b616 100644 --- a/net/core/net-sysfs.c +++ b/net/core/net-sysfs.c @@ -24,6 +24,7 @@ #include <linux/jiffies.h> #include <linux/pm_runtime.h> #include <linux/of.h> +#include <linux/of_net.h> #include "net-sysfs.h" diff --git a/net/core/pktgen.c b/net/core/pktgen.c index 8604ae2..8b02df0 100644 --- a/net/core/pktgen.c +++ b/net/core/pktgen.c @@ -2245,10 +2245,8 @@ static void spin(struct pktgen_dev *pkt_dev, ktime_t spin_until) hrtimer_set_expires(&t.timer, spin_until); remaining = ktime_to_ns(hrtimer_expires_remaining(&t.timer)); - if (remaining <= 0) { - pkt_dev->next_tx = ktime_add_ns(spin_until, pkt_dev->delay); - return; - } + if (remaining <= 0) + goto out; start_time = ktime_get(); if (remaining < 100000) { @@ -2273,7 +2271,9 @@ static void spin(struct pktgen_dev *pkt_dev, ktime_t spin_until) } pkt_dev->idle_acc += ktime_to_ns(ktime_sub(end_time, start_time)); +out: pkt_dev->next_tx = ktime_add_ns(spin_until, pkt_dev->delay); + destroy_hrtimer_on_stack(&t.timer); } static inline void set_pkt_overhead(struct pktgen_dev *pkt_dev) diff --git a/net/ipv4/esp4.c b/net/ipv4/esp4.c index 4779374..d95631d 100644 --- a/net/ipv4/esp4.c +++ b/net/ipv4/esp4.c @@ -23,6 +23,11 @@ struct esp_skb_cb { void *tmp; }; +struct esp_output_extra { + __be32 seqhi; + u32 esphoff; +}; + #define ESP_SKB_CB(__skb) ((struct esp_skb_cb *)&((__skb)->cb[0])) static u32 esp4_get_mtu(struct xfrm_state *x, int mtu); @@ -35,11 +40,11 @@ static u32 esp4_get_mtu(struct xfrm_state *x, int mtu); * * TODO: Use spare space in skb for this where possible. */ -static void *esp_alloc_tmp(struct crypto_aead *aead, int nfrags, int seqhilen) +static void *esp_alloc_tmp(struct crypto_aead *aead, int nfrags, int extralen) { unsigned int len; - len = seqhilen; + len = extralen; len += crypto_aead_ivsize(aead); @@ -57,15 +62,16 @@ static void *esp_alloc_tmp(struct crypto_aead *aead, int nfrags, int seqhilen) return kmalloc(len, GFP_ATOMIC); } -static inline __be32 *esp_tmp_seqhi(void *tmp) +static inline void *esp_tmp_extra(void *tmp) { - return PTR_ALIGN((__be32 *)tmp, __alignof__(__be32)); + return PTR_ALIGN(tmp, __alignof__(struct esp_output_extra)); } -static inline u8 *esp_tmp_iv(struct crypto_aead *aead, void *tmp, int seqhilen) + +static inline u8 *esp_tmp_iv(struct crypto_aead *aead, void *tmp, int extralen) { return crypto_aead_ivsize(aead) ? - PTR_ALIGN((u8 *)tmp + seqhilen, - crypto_aead_alignmask(aead) + 1) : tmp + seqhilen; + PTR_ALIGN((u8 *)tmp + extralen, + crypto_aead_alignmask(aead) + 1) : tmp + extralen; } static inline struct aead_request *esp_tmp_req(struct crypto_aead *aead, u8 *iv) @@ -99,7 +105,7 @@ static void esp_restore_header(struct sk_buff *skb, unsigned int offset) { struct ip_esp_hdr *esph = (void *)(skb->data + offset); void *tmp = ESP_SKB_CB(skb)->tmp; - __be32 *seqhi = esp_tmp_seqhi(tmp); + __be32 *seqhi = esp_tmp_extra(tmp); esph->seq_no = esph->spi; esph->spi = *seqhi; @@ -107,7 +113,11 @@ static void esp_restore_header(struct sk_buff *skb, unsigned int offset) static void esp_output_restore_header(struct sk_buff *skb) { - esp_restore_header(skb, skb_transport_offset(skb) - sizeof(__be32)); + void *tmp = ESP_SKB_CB(skb)->tmp; + struct esp_output_extra *extra = esp_tmp_extra(tmp); + + esp_restore_header(skb, skb_transport_offset(skb) + extra->esphoff - + sizeof(__be32)); } static void esp_output_done_esn(struct crypto_async_request *base, int err) @@ -121,6 +131,7 @@ static void esp_output_done_esn(struct crypto_async_request *base, int err) static int esp_output(struct xfrm_state *x, struct sk_buff *skb) { int err; + struct esp_output_extra *extra; struct ip_esp_hdr *esph; struct crypto_aead *aead; struct aead_request *req; @@ -137,8 +148,7 @@ static int esp_output(struct xfrm_state *x, struct sk_buff *skb) int tfclen; int nfrags; int assoclen; - int seqhilen; - __be32 *seqhi; + int extralen; __be64 seqno; /* skb is pure payload to encrypt */ @@ -166,21 +176,21 @@ static int esp_output(struct xfrm_state *x, struct sk_buff *skb) nfrags = err; assoclen = sizeof(*esph); - seqhilen = 0; + extralen = 0; if (x->props.flags & XFRM_STATE_ESN) { - seqhilen += sizeof(__be32); - assoclen += seqhilen; + extralen += sizeof(*extra); + assoclen += sizeof(__be32); } - tmp = esp_alloc_tmp(aead, nfrags, seqhilen); + tmp = esp_alloc_tmp(aead, nfrags, extralen); if (!tmp) { err = -ENOMEM; goto error; } - seqhi = esp_tmp_seqhi(tmp); - iv = esp_tmp_iv(aead, tmp, seqhilen); + extra = esp_tmp_extra(tmp); + iv = esp_tmp_iv(aead, tmp, extralen); req = esp_tmp_req(aead, iv); sg = esp_req_sg(aead, req); @@ -247,8 +257,10 @@ static int esp_output(struct xfrm_state *x, struct sk_buff *skb) * encryption. */ if ((x->props.flags & XFRM_STATE_ESN)) { - esph = (void *)(skb_transport_header(skb) - sizeof(__be32)); - *seqhi = esph->spi; + extra->esphoff = (unsigned char *)esph - + skb_transport_header(skb); + esph = (struct ip_esp_hdr *)((unsigned char *)esph - 4); + extra->seqhi = esph->spi; esph->seq_no = htonl(XFRM_SKB_CB(skb)->seq.output.hi); aead_request_set_callback(req, 0, esp_output_done_esn, skb); } @@ -445,7 +457,7 @@ static int esp_input(struct xfrm_state *x, struct sk_buff *skb) goto out; ESP_SKB_CB(skb)->tmp = tmp; - seqhi = esp_tmp_seqhi(tmp); + seqhi = esp_tmp_extra(tmp); iv = esp_tmp_iv(aead, tmp, seqhilen); req = esp_tmp_req(aead, iv); sg = esp_req_sg(aead, req); diff --git a/net/ipv4/gre_demux.c b/net/ipv4/gre_demux.c index 4c39f4f..de1d119 100644 --- a/net/ipv4/gre_demux.c +++ b/net/ipv4/gre_demux.c @@ -62,26 +62,26 @@ EXPORT_SYMBOL_GPL(gre_del_protocol); /* Fills in tpi and returns header length to be pulled. */ int gre_parse_header(struct sk_buff *skb, struct tnl_ptk_info *tpi, - bool *csum_err, __be16 proto) + bool *csum_err, __be16 proto, int nhs) { const struct gre_base_hdr *greh; __be32 *options; int hdr_len; - if (unlikely(!pskb_may_pull(skb, sizeof(struct gre_base_hdr)))) + if (unlikely(!pskb_may_pull(skb, nhs + sizeof(struct gre_base_hdr)))) return -EINVAL; - greh = (struct gre_base_hdr *)skb_transport_header(skb); + greh = (struct gre_base_hdr *)(skb->data + nhs); if (unlikely(greh->flags & (GRE_VERSION | GRE_ROUTING))) return -EINVAL; tpi->flags = gre_flags_to_tnl_flags(greh->flags); hdr_len = gre_calc_hlen(tpi->flags); - if (!pskb_may_pull(skb, hdr_len)) + if (!pskb_may_pull(skb, nhs + hdr_len)) return -EINVAL; - greh = (struct gre_base_hdr *)skb_transport_header(skb); + greh = (struct gre_base_hdr *)(skb->data + nhs); tpi->proto = greh->protocol; options = (__be32 *)(greh + 1); diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c index 4d2025f..1d000af 100644 --- a/net/ipv4/ip_gre.c +++ b/net/ipv4/ip_gre.c @@ -49,12 +49,6 @@ #include <net/gre.h> #include <net/dst_metadata.h> -#if IS_ENABLED(CONFIG_IPV6) -#include <net/ipv6.h> -#include <net/ip6_fib.h> -#include <net/ip6_route.h> -#endif - /* Problems & solutions -------------------- @@ -217,12 +211,14 @@ static void gre_err(struct sk_buff *skb, u32 info) * by themselves??? */ + const struct iphdr *iph = (struct iphdr *)skb->data; const int type = icmp_hdr(skb)->type; const int code = icmp_hdr(skb)->code; struct tnl_ptk_info tpi; bool csum_err = false; - if (gre_parse_header(skb, &tpi, &csum_err, htons(ETH_P_IP)) < 0) { + if (gre_parse_header(skb, &tpi, &csum_err, htons(ETH_P_IP), + iph->ihl * 4) < 0) { if (!csum_err) /* ignore csum errors. */ return; } @@ -338,7 +334,7 @@ static int gre_rcv(struct sk_buff *skb) } #endif - hdr_len = gre_parse_header(skb, &tpi, &csum_err, htons(ETH_P_IP)); + hdr_len = gre_parse_header(skb, &tpi, &csum_err, htons(ETH_P_IP), 0); if (hdr_len < 0) goto drop; @@ -1121,6 +1117,7 @@ struct net_device *gretap_fb_dev_create(struct net *net, const char *name, { struct nlattr *tb[IFLA_MAX + 1]; struct net_device *dev; + LIST_HEAD(list_kill); struct ip_tunnel *t; int err; @@ -1136,8 +1133,10 @@ struct net_device *gretap_fb_dev_create(struct net *net, const char *name, t->collect_md = true; err = ipgre_newlink(net, dev, tb, NULL); - if (err < 0) - goto out; + if (err < 0) { + free_netdev(dev); + return ERR_PTR(err); + } /* openvswitch users expect packet sizes to be unrestricted, * so set the largest MTU we can. @@ -1146,9 +1145,14 @@ struct net_device *gretap_fb_dev_create(struct net *net, const char *name, if (err) goto out; + err = rtnl_configure_link(dev, NULL); + if (err < 0) + goto out; + return dev; out: - free_netdev(dev); + ip_tunnel_dellink(dev, &list_kill); + unregister_netdevice_many(&list_kill); return ERR_PTR(err); } EXPORT_SYMBOL_GPL(gretap_fb_dev_create); diff --git a/net/ipv4/ipconfig.c b/net/ipv4/ipconfig.c index 2ed9dd2..1d71c40 100644 --- a/net/ipv4/ipconfig.c +++ b/net/ipv4/ipconfig.c @@ -127,7 +127,9 @@ __be32 ic_myaddr = NONE; /* My IP address */ static __be32 ic_netmask = NONE; /* Netmask for local subnet */ __be32 ic_gateway = NONE; /* Gateway IP address */ -__be32 ic_addrservaddr = NONE; /* IP Address of the IP addresses'server */ +#ifdef IPCONFIG_DYNAMIC +static __be32 ic_addrservaddr = NONE; /* IP Address of the IP addresses'server */ +#endif __be32 ic_servaddr = NONE; /* Boot server IP address */ diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c index 21a38e2..5ad48ec 100644 --- a/net/ipv4/ipmr.c +++ b/net/ipv4/ipmr.c @@ -891,8 +891,10 @@ static struct mfc_cache *ipmr_cache_alloc(void) { struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL); - if (c) + if (c) { + c->mfc_un.res.last_assert = jiffies - MFC_ASSERT_THRESH - 1; c->mfc_un.res.minvif = MAXVIFS; + } return c; } diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index 8bd9911..e00e972 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -2751,7 +2751,7 @@ void tcp_xmit_retransmit_queue(struct sock *sk) struct tcp_sock *tp = tcp_sk(sk); struct sk_buff *skb; struct sk_buff *hole = NULL; - u32 last_lost; + u32 max_segs, last_lost; int mib_idx; int fwd_rexmitting = 0; @@ -2771,6 +2771,7 @@ void tcp_xmit_retransmit_queue(struct sock *sk) last_lost = tp->snd_una; } + max_segs = tcp_tso_autosize(sk, tcp_current_mss(sk)); tcp_for_write_queue_from(skb, sk) { __u8 sacked = TCP_SKB_CB(skb)->sacked; int segs; @@ -2784,6 +2785,10 @@ void tcp_xmit_retransmit_queue(struct sock *sk) segs = tp->snd_cwnd - tcp_packets_in_flight(tp); if (segs <= 0) return; + /* In case tcp_shift_skb_data() have aggregated large skbs, + * we need to make sure not sending too bigs TSO packets + */ + segs = min_t(int, segs, max_segs); if (fwd_rexmitting) { begin_fwd: diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index d56c055..ca5e8ea 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -391,9 +391,9 @@ int udp_v4_get_port(struct sock *sk, unsigned short snum) return udp_lib_get_port(sk, snum, ipv4_rcv_saddr_equal, hash2_nulladdr); } -static inline int compute_score(struct sock *sk, struct net *net, - __be32 saddr, unsigned short hnum, __be16 sport, - __be32 daddr, __be16 dport, int dif) +static int compute_score(struct sock *sk, struct net *net, + __be32 saddr, __be16 sport, + __be32 daddr, unsigned short hnum, int dif) { int score; struct inet_sock *inet; @@ -434,52 +434,6 @@ static inline int compute_score(struct sock *sk, struct net *net, return score; } -/* - * In this second variant, we check (daddr, dport) matches (inet_rcv_sadd, inet_num) - */ -static inline int compute_score2(struct sock *sk, struct net *net, - __be32 saddr, __be16 sport, - __be32 daddr, unsigned int hnum, int dif) -{ - int score; - struct inet_sock *inet; - - if (!net_eq(sock_net(sk), net) || - ipv6_only_sock(sk)) - return -1; - - inet = inet_sk(sk); - - if (inet->inet_rcv_saddr != daddr || - inet->inet_num != hnum) - return -1; - - score = (sk->sk_family == PF_INET) ? 2 : 1; - - if (inet->inet_daddr) { - if (inet->inet_daddr != saddr) - return -1; - score += 4; - } - - if (inet->inet_dport) { - if (inet->inet_dport != sport) - return -1; - score += 4; - } - - if (sk->sk_bound_dev_if) { - if (sk->sk_bound_dev_if != dif) - return -1; - score += 4; - } - - if (sk->sk_incoming_cpu == raw_smp_processor_id()) - score++; - - return score; -} - static u32 udp_ehashfn(const struct net *net, const __be32 laddr, const __u16 lport, const __be32 faddr, const __be16 fport) @@ -492,11 +446,11 @@ static u32 udp_ehashfn(const struct net *net, const __be32 laddr, udp_ehash_secret + net_hash_mix(net)); } -/* called with read_rcu_lock() */ +/* called with rcu_read_lock() */ static struct sock *udp4_lib_lookup2(struct net *net, __be32 saddr, __be16 sport, __be32 daddr, unsigned int hnum, int dif, - struct udp_hslot *hslot2, unsigned int slot2, + struct udp_hslot *hslot2, struct sk_buff *skb) { struct sock *sk, *result; @@ -506,7 +460,7 @@ static struct sock *udp4_lib_lookup2(struct net *net, result = NULL; badness = 0; udp_portaddr_for_each_entry_rcu(sk, &hslot2->head) { - score = compute_score2(sk, net, saddr, sport, + score = compute_score(sk, net, saddr, sport, daddr, hnum, dif); if (score > badness) { reuseport = sk->sk_reuseport; @@ -554,17 +508,22 @@ struct sock *__udp4_lib_lookup(struct net *net, __be32 saddr, result = udp4_lib_lookup2(net, saddr, sport, daddr, hnum, dif, - hslot2, slot2, skb); + hslot2, skb); if (!result) { + unsigned int old_slot2 = slot2; hash2 = udp4_portaddr_hash(net, htonl(INADDR_ANY), hnum); slot2 = hash2 & udptable->mask; + /* avoid searching the same slot again. */ + if (unlikely(slot2 == old_slot2)) + return result; + hslot2 = &udptable->hash2[slot2]; if (hslot->count < hslot2->count) goto begin; result = udp4_lib_lookup2(net, saddr, sport, - htonl(INADDR_ANY), hnum, dif, - hslot2, slot2, skb); + daddr, hnum, dif, + hslot2, skb); } return result; } @@ -572,8 +531,8 @@ begin: result = NULL; badness = 0; sk_for_each_rcu(sk, &hslot->head) { - score = compute_score(sk, net, saddr, hnum, sport, - daddr, dport, dif); + score = compute_score(sk, net, saddr, sport, + daddr, hnum, dif); if (score > badness) { reuseport = sk->sk_reuseport; if (reuseport) { @@ -1618,12 +1577,12 @@ int udp_queue_rcv_skb(struct sock *sk, struct sk_buff *skb) } } - if (rcu_access_pointer(sk->sk_filter)) { - if (udp_lib_checksum_complete(skb)) + if (rcu_access_pointer(sk->sk_filter) && + udp_lib_checksum_complete(skb)) goto csum_error; - if (sk_filter(sk, skb)) - goto drop; - } + + if (sk_filter(sk, skb)) + goto drop; udp_csum_pull_header(skb); if (sk_rcvqueues_full(sk, sk->sk_rcvbuf)) { @@ -1755,8 +1714,11 @@ static inline int udp4_csum_init(struct sk_buff *skb, struct udphdr *uh, return err; } - return skb_checksum_init_zero_check(skb, proto, uh->check, - inet_compute_pseudo); + /* Note, we are only interested in != 0 or == 0, thus the + * force to int. + */ + return (__force int)skb_checksum_init_zero_check(skb, proto, uh->check, + inet_compute_pseudo); } /* diff --git a/net/ipv6/Kconfig b/net/ipv6/Kconfig index 9946082..2343e4f 100644 --- a/net/ipv6/Kconfig +++ b/net/ipv6/Kconfig @@ -239,7 +239,7 @@ config IPV6_FOU config IPV6_FOU_TUNNEL tristate default NET_FOU_IP_TUNNELS && IPV6_FOU - select INET6_TUNNEL + select IPV6_TUNNEL config IPV6_MULTIPLE_TABLES bool "IPv6: Multiple Routing Tables" diff --git a/net/ipv6/Makefile b/net/ipv6/Makefile index d42ca3d..6d8ea09 100644 --- a/net/ipv6/Makefile +++ b/net/ipv6/Makefile @@ -42,7 +42,7 @@ obj-$(CONFIG_IPV6_VTI) += ip6_vti.o obj-$(CONFIG_IPV6_SIT) += sit.o obj-$(CONFIG_IPV6_TUNNEL) += ip6_tunnel.o obj-$(CONFIG_IPV6_GRE) += ip6_gre.o -obj-$(CONFIG_NET_FOU_IPV6_TUNNELS) += fou6.o +obj-$(CONFIG_IPV6_FOU) += fou6.o obj-y += addrconf_core.o exthdrs_core.o ip6_checksum.o ip6_icmp.o obj-$(CONFIG_INET) += output_core.o protocol.o $(ipv6-offload) diff --git a/net/ipv6/icmp.c b/net/ipv6/icmp.c index 4527285..a4fa840 100644 --- a/net/ipv6/icmp.c +++ b/net/ipv6/icmp.c @@ -98,7 +98,7 @@ static void icmpv6_err(struct sk_buff *skb, struct inet6_skb_parm *opt, if (!(type & ICMPV6_INFOMSG_MASK)) if (icmp6->icmp6_type == ICMPV6_ECHO_REQUEST) - ping_err(skb, offset, info); + ping_err(skb, offset, ntohl(info)); } static int icmpv6_rcv(struct sk_buff *skb); diff --git a/net/ipv6/ip6_checksum.c b/net/ipv6/ip6_checksum.c index b2025bf..c0cbcb2 100644 --- a/net/ipv6/ip6_checksum.c +++ b/net/ipv6/ip6_checksum.c @@ -78,9 +78,12 @@ int udp6_csum_init(struct sk_buff *skb, struct udphdr *uh, int proto) * we accept a checksum of zero here. When we find the socket * for the UDP packet we'll check if that socket allows zero checksum * for IPv6 (set by socket option). + * + * Note, we are only interested in != 0 or == 0, thus the + * force to int. */ - return skb_checksum_init_zero_check(skb, proto, uh->check, - ip6_compute_pseudo); + return (__force int)skb_checksum_init_zero_check(skb, proto, uh->check, + ip6_compute_pseudo); } EXPORT_SYMBOL(udp6_csum_init); diff --git a/net/ipv6/ip6_gre.c b/net/ipv6/ip6_gre.c index f4ac284..776d145 100644 --- a/net/ipv6/ip6_gre.c +++ b/net/ipv6/ip6_gre.c @@ -468,7 +468,7 @@ static int gre_rcv(struct sk_buff *skb) bool csum_err = false; int hdr_len; - hdr_len = gre_parse_header(skb, &tpi, &csum_err, htons(ETH_P_IPV6)); + hdr_len = gre_parse_header(skb, &tpi, &csum_err, htons(ETH_P_IPV6), 0); if (hdr_len < 0) goto drop; @@ -1256,6 +1256,8 @@ static int ip6gre_tap_init(struct net_device *dev) if (ret) return ret; + dev->priv_flags |= IFF_LIVE_ADDR_CHANGE; + tunnel = netdev_priv(dev); ip6gre_tnl_link_config(tunnel, 1); @@ -1289,6 +1291,7 @@ static void ip6gre_tap_setup(struct net_device *dev) dev->features |= NETIF_F_NETNS_LOCAL; dev->priv_flags &= ~IFF_TX_SKB_SHARING; + dev->priv_flags |= IFF_LIVE_ADDR_CHANGE; } static bool ip6gre_netlink_encap_parms(struct nlattr *data[], diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c index cbf127a..635b8d3 100644 --- a/net/ipv6/ip6_output.c +++ b/net/ipv6/ip6_output.c @@ -1071,17 +1071,12 @@ struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6, const struct in6_addr *final_dst) { struct dst_entry *dst = sk_dst_check(sk, inet6_sk(sk)->dst_cookie); - int err; dst = ip6_sk_dst_check(sk, dst, fl6); + if (!dst) + dst = ip6_dst_lookup_flow(sk, fl6, final_dst); - err = ip6_dst_lookup_tail(sock_net(sk), sk, &dst, fl6); - if (err) - return ERR_PTR(err); - if (final_dst) - fl6->daddr = *final_dst; - - return xfrm_lookup_route(sock_net(sk), dst, flowi6_to_flowi(fl6), sk, 0); + return dst; } EXPORT_SYMBOL_GPL(ip6_sk_dst_lookup_flow); diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c index f2e2013f8..487ef3b 100644 --- a/net/ipv6/ip6mr.c +++ b/net/ipv6/ip6mr.c @@ -1074,6 +1074,7 @@ static struct mfc6_cache *ip6mr_cache_alloc(void) struct mfc6_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL); if (!c) return NULL; + c->mfc_un.res.last_assert = jiffies - MFC_ASSERT_THRESH - 1; c->mfc_un.res.minvif = MAXMIFS; return c; } diff --git a/net/ipv6/netfilter/nf_dup_ipv6.c b/net/ipv6/netfilter/nf_dup_ipv6.c index 6989c70..4a84b5a 100644 --- a/net/ipv6/netfilter/nf_dup_ipv6.c +++ b/net/ipv6/netfilter/nf_dup_ipv6.c @@ -33,6 +33,7 @@ static bool nf_dup_ipv6_route(struct net *net, struct sk_buff *skb, fl6.daddr = *gw; fl6.flowlabel = (__force __be32)(((iph->flow_lbl[0] & 0xF) << 16) | (iph->flow_lbl[1] << 8) | iph->flow_lbl[2]); + fl6.flowi6_flags = FLOWI_FLAG_KNOWN_NH; dst = ip6_route_output(net, NULL, &fl6); if (dst->error) { dst_release(dst); diff --git a/net/ipv6/route.c b/net/ipv6/route.c index 969913d..520b788 100644 --- a/net/ipv6/route.c +++ b/net/ipv6/route.c @@ -1782,7 +1782,7 @@ static struct rt6_info *ip6_nh_lookup_table(struct net *net, }; struct fib6_table *table; struct rt6_info *rt; - int flags = 0; + int flags = RT6_LOOKUP_F_IFACE; table = fib6_get_table(net, cfg->fc_table); if (!table) diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c index 0a5a255..0619ac7 100644 --- a/net/ipv6/sit.c +++ b/net/ipv6/sit.c @@ -560,13 +560,13 @@ static int ipip6_err(struct sk_buff *skb, u32 info) if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED) { ipv4_update_pmtu(skb, dev_net(skb->dev), info, - t->parms.link, 0, IPPROTO_IPV6, 0); + t->parms.link, 0, iph->protocol, 0); err = 0; goto out; } if (type == ICMP_REDIRECT) { ipv4_redirect(skb, dev_net(skb->dev), t->parms.link, 0, - IPPROTO_IPV6, 0); + iph->protocol, 0); err = 0; goto out; } diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c index 79e33e0..2255d2b 100644 --- a/net/ipv6/tcp_ipv6.c +++ b/net/ipv6/tcp_ipv6.c @@ -738,7 +738,7 @@ static const struct tcp_request_sock_ops tcp_request_sock_ipv6_ops = { static void tcp_v6_send_response(const struct sock *sk, struct sk_buff *skb, u32 seq, u32 ack, u32 win, u32 tsval, u32 tsecr, int oif, struct tcp_md5sig_key *key, int rst, - u8 tclass, u32 label) + u8 tclass, __be32 label) { const struct tcphdr *th = tcp_hdr(skb); struct tcphdr *t1; @@ -911,7 +911,7 @@ out: static void tcp_v6_send_ack(const struct sock *sk, struct sk_buff *skb, u32 seq, u32 ack, u32 win, u32 tsval, u32 tsecr, int oif, struct tcp_md5sig_key *key, u8 tclass, - u32 label) + __be32 label) { tcp_v6_send_response(sk, skb, seq, ack, win, tsval, tsecr, oif, key, 0, tclass, label); @@ -1721,7 +1721,9 @@ static void get_tcp6_sock(struct seq_file *seq, struct sock *sp, int i) destp = ntohs(inet->inet_dport); srcp = ntohs(inet->inet_sport); - if (icsk->icsk_pending == ICSK_TIME_RETRANS) { + if (icsk->icsk_pending == ICSK_TIME_RETRANS || + icsk->icsk_pending == ICSK_TIME_EARLY_RETRANS || + icsk->icsk_pending == ICSK_TIME_LOSS_PROBE) { timer_active = 1; timer_expires = icsk->icsk_timeout; } else if (icsk->icsk_pending == ICSK_TIME_PROBE0) { diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c index 2da1896..005dc82 100644 --- a/net/ipv6/udp.c +++ b/net/ipv6/udp.c @@ -115,11 +115,10 @@ static void udp_v6_rehash(struct sock *sk) udp_lib_rehash(sk, new_hash); } -static inline int compute_score(struct sock *sk, struct net *net, - unsigned short hnum, - const struct in6_addr *saddr, __be16 sport, - const struct in6_addr *daddr, __be16 dport, - int dif) +static int compute_score(struct sock *sk, struct net *net, + const struct in6_addr *saddr, __be16 sport, + const struct in6_addr *daddr, unsigned short hnum, + int dif) { int score; struct inet_sock *inet; @@ -162,54 +161,11 @@ static inline int compute_score(struct sock *sk, struct net *net, return score; } -static inline int compute_score2(struct sock *sk, struct net *net, - const struct in6_addr *saddr, __be16 sport, - const struct in6_addr *daddr, - unsigned short hnum, int dif) -{ - int score; - struct inet_sock *inet; - - if (!net_eq(sock_net(sk), net) || - udp_sk(sk)->udp_port_hash != hnum || - sk->sk_family != PF_INET6) - return -1; - - if (!ipv6_addr_equal(&sk->sk_v6_rcv_saddr, daddr)) - return -1; - - score = 0; - inet = inet_sk(sk); - - if (inet->inet_dport) { - if (inet->inet_dport != sport) - return -1; - score++; - } - - if (!ipv6_addr_any(&sk->sk_v6_daddr)) { - if (!ipv6_addr_equal(&sk->sk_v6_daddr, saddr)) - return -1; - score++; - } - - if (sk->sk_bound_dev_if) { - if (sk->sk_bound_dev_if != dif) - return -1; - score++; - } - - if (sk->sk_incoming_cpu == raw_smp_processor_id()) - score++; - - return score; -} - -/* called with read_rcu_lock() */ +/* called with rcu_read_lock() */ static struct sock *udp6_lib_lookup2(struct net *net, const struct in6_addr *saddr, __be16 sport, const struct in6_addr *daddr, unsigned int hnum, int dif, - struct udp_hslot *hslot2, unsigned int slot2, + struct udp_hslot *hslot2, struct sk_buff *skb) { struct sock *sk, *result; @@ -219,7 +175,7 @@ static struct sock *udp6_lib_lookup2(struct net *net, result = NULL; badness = -1; udp_portaddr_for_each_entry_rcu(sk, &hslot2->head) { - score = compute_score2(sk, net, saddr, sport, + score = compute_score(sk, net, saddr, sport, daddr, hnum, dif); if (score > badness) { reuseport = sk->sk_reuseport; @@ -268,17 +224,22 @@ struct sock *__udp6_lib_lookup(struct net *net, result = udp6_lib_lookup2(net, saddr, sport, daddr, hnum, dif, - hslot2, slot2, skb); + hslot2, skb); if (!result) { + unsigned int old_slot2 = slot2; hash2 = udp6_portaddr_hash(net, &in6addr_any, hnum); slot2 = hash2 & udptable->mask; + /* avoid searching the same slot again. */ + if (unlikely(slot2 == old_slot2)) + return result; + hslot2 = &udptable->hash2[slot2]; if (hslot->count < hslot2->count) goto begin; result = udp6_lib_lookup2(net, saddr, sport, - &in6addr_any, hnum, dif, - hslot2, slot2, skb); + daddr, hnum, dif, + hslot2, skb); } return result; } @@ -286,7 +247,7 @@ begin: result = NULL; badness = -1; sk_for_each_rcu(sk, &hslot->head) { - score = compute_score(sk, net, hnum, saddr, sport, daddr, dport, dif); + score = compute_score(sk, net, saddr, sport, daddr, hnum, dif); if (score > badness) { reuseport = sk->sk_reuseport; if (reuseport) { @@ -653,12 +614,12 @@ int udpv6_queue_rcv_skb(struct sock *sk, struct sk_buff *skb) } } - if (rcu_access_pointer(sk->sk_filter)) { - if (udp_lib_checksum_complete(skb)) - goto csum_error; - if (sk_filter(sk, skb)) - goto drop; - } + if (rcu_access_pointer(sk->sk_filter) && + udp_lib_checksum_complete(skb)) + goto csum_error; + + if (sk_filter(sk, skb)) + goto drop; udp_csum_pull_header(skb); if (sk_rcvqueues_full(sk, sk->sk_rcvbuf)) { diff --git a/net/kcm/kcmproc.c b/net/kcm/kcmproc.c index 7380087..fda7f47 100644 --- a/net/kcm/kcmproc.c +++ b/net/kcm/kcmproc.c @@ -241,6 +241,7 @@ static const struct file_operations kcm_seq_fops = { .open = kcm_seq_open, .read = seq_read, .llseek = seq_lseek, + .release = seq_release_net, }; static struct kcm_seq_muxinfo kcm_seq_muxinfo = { diff --git a/net/l2tp/l2tp_core.c b/net/l2tp/l2tp_core.c index 6edfa99..1e40dac 100644 --- a/net/l2tp/l2tp_core.c +++ b/net/l2tp/l2tp_core.c @@ -1581,7 +1581,7 @@ int l2tp_tunnel_create(struct net *net, int fd, int version, u32 tunnel_id, u32 /* Mark socket as an encapsulation socket. See net/ipv4/udp.c */ tunnel->encap = encap; if (encap == L2TP_ENCAPTYPE_UDP) { - struct udp_tunnel_sock_cfg udp_cfg; + struct udp_tunnel_sock_cfg udp_cfg = { }; udp_cfg.sk_user_data = tunnel; udp_cfg.encap_type = UDP_ENCAP_L2TPINUDP; diff --git a/net/netfilter/ipvs/ip_vs_conn.c b/net/netfilter/ipvs/ip_vs_conn.c index 2cb3c62..096a451 100644 --- a/net/netfilter/ipvs/ip_vs_conn.c +++ b/net/netfilter/ipvs/ip_vs_conn.c @@ -762,7 +762,7 @@ static int expire_quiescent_template(struct netns_ipvs *ipvs, * If available, return 1, otherwise invalidate this connection * template and return 0. */ -int ip_vs_check_template(struct ip_vs_conn *ct) +int ip_vs_check_template(struct ip_vs_conn *ct, struct ip_vs_dest *cdest) { struct ip_vs_dest *dest = ct->dest; struct netns_ipvs *ipvs = ct->ipvs; @@ -772,7 +772,8 @@ int ip_vs_check_template(struct ip_vs_conn *ct) */ if ((dest == NULL) || !(dest->flags & IP_VS_DEST_F_AVAILABLE) || - expire_quiescent_template(ipvs, dest)) { + expire_quiescent_template(ipvs, dest) || + (cdest && (dest != cdest))) { IP_VS_DBG_BUF(9, "check_template: dest not available for " "protocol %s s:%s:%d v:%s:%d " "-> d:%s:%d\n", diff --git a/net/netfilter/ipvs/ip_vs_core.c b/net/netfilter/ipvs/ip_vs_core.c index 1207f20..2c1b498 100644 --- a/net/netfilter/ipvs/ip_vs_core.c +++ b/net/netfilter/ipvs/ip_vs_core.c @@ -321,7 +321,7 @@ ip_vs_sched_persist(struct ip_vs_service *svc, /* Check if a template already exists */ ct = ip_vs_ct_in_get(¶m); - if (!ct || !ip_vs_check_template(ct)) { + if (!ct || !ip_vs_check_template(ct, NULL)) { struct ip_vs_scheduler *sched; /* @@ -1154,7 +1154,8 @@ struct ip_vs_conn *ip_vs_new_conn_out(struct ip_vs_service *svc, vport, ¶m) < 0) return NULL; ct = ip_vs_ct_in_get(¶m); - if (!ct) { + /* check if template exists and points to the same dest */ + if (!ct || !ip_vs_check_template(ct, dest)) { ct = ip_vs_conn_new(¶m, dest->af, daddr, dport, IP_VS_CONN_F_TEMPLATE, dest, 0); if (!ct) { diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c index db2312e..f204274 100644 --- a/net/netfilter/nf_conntrack_core.c +++ b/net/netfilter/nf_conntrack_core.c @@ -1544,6 +1544,8 @@ void nf_conntrack_cleanup_end(void) nf_conntrack_tstamp_fini(); nf_conntrack_acct_fini(); nf_conntrack_expect_fini(); + + kmem_cache_destroy(nf_conntrack_cachep); } /* diff --git a/net/netfilter/nf_conntrack_ftp.c b/net/netfilter/nf_conntrack_ftp.c index 883c691..19efeba 100644 --- a/net/netfilter/nf_conntrack_ftp.c +++ b/net/netfilter/nf_conntrack_ftp.c @@ -632,6 +632,7 @@ static int __init nf_conntrack_ftp_init(void) if (ret) { pr_err("failed to register helper for pf: %d port: %d\n", ftp[i][j].tuple.src.l3num, ports[i]); + ports_c = i; nf_conntrack_ftp_fini(); return ret; } diff --git a/net/netfilter/nf_conntrack_helper.c b/net/netfilter/nf_conntrack_helper.c index f703adb..196cb39 100644 --- a/net/netfilter/nf_conntrack_helper.c +++ b/net/netfilter/nf_conntrack_helper.c @@ -361,9 +361,10 @@ EXPORT_SYMBOL_GPL(nf_ct_helper_log); int nf_conntrack_helper_register(struct nf_conntrack_helper *me) { - int ret = 0; - struct nf_conntrack_helper *cur; + struct nf_conntrack_tuple_mask mask = { .src.u.all = htons(0xFFFF) }; unsigned int h = helper_hash(&me->tuple); + struct nf_conntrack_helper *cur; + int ret = 0; BUG_ON(me->expect_policy == NULL); BUG_ON(me->expect_class_max >= NF_CT_MAX_EXPECT_CLASSES); @@ -371,9 +372,7 @@ int nf_conntrack_helper_register(struct nf_conntrack_helper *me) mutex_lock(&nf_ct_helper_mutex); hlist_for_each_entry(cur, &nf_ct_helper_hash[h], hnode) { - if (strncmp(cur->name, me->name, NF_CT_HELPER_NAME_LEN) == 0 && - cur->tuple.src.l3num == me->tuple.src.l3num && - cur->tuple.dst.protonum == me->tuple.dst.protonum) { + if (nf_ct_tuple_src_mask_cmp(&cur->tuple, &me->tuple, &mask)) { ret = -EEXIST; goto out; } diff --git a/net/netfilter/nf_conntrack_irc.c b/net/netfilter/nf_conntrack_irc.c index 8b6da27..f97ac61 100644 --- a/net/netfilter/nf_conntrack_irc.c +++ b/net/netfilter/nf_conntrack_irc.c @@ -271,6 +271,7 @@ static int __init nf_conntrack_irc_init(void) if (ret) { pr_err("failed to register helper for pf: %u port: %u\n", irc[i].tuple.src.l3num, ports[i]); + ports_c = i; nf_conntrack_irc_fini(); return ret; } diff --git a/net/netfilter/nf_conntrack_sane.c b/net/netfilter/nf_conntrack_sane.c index 7523a57..3fcbaab 100644 --- a/net/netfilter/nf_conntrack_sane.c +++ b/net/netfilter/nf_conntrack_sane.c @@ -223,6 +223,7 @@ static int __init nf_conntrack_sane_init(void) if (ret) { pr_err("failed to register helper for pf: %d port: %d\n", sane[i][j].tuple.src.l3num, ports[i]); + ports_c = i; nf_conntrack_sane_fini(); return ret; } diff --git a/net/netfilter/nf_conntrack_sip.c b/net/netfilter/nf_conntrack_sip.c index 3e06402..f72ba55 100644 --- a/net/netfilter/nf_conntrack_sip.c +++ b/net/netfilter/nf_conntrack_sip.c @@ -1669,6 +1669,7 @@ static int __init nf_conntrack_sip_init(void) if (ret) { pr_err("failed to register helper for pf: %u port: %u\n", sip[i][j].tuple.src.l3num, ports[i]); + ports_c = i; nf_conntrack_sip_fini(); return ret; } diff --git a/net/netfilter/nf_conntrack_standalone.c b/net/netfilter/nf_conntrack_standalone.c index f87e84e..c026c47 100644 --- a/net/netfilter/nf_conntrack_standalone.c +++ b/net/netfilter/nf_conntrack_standalone.c @@ -487,8 +487,6 @@ static struct ctl_table nf_ct_sysctl_table[] = { { } }; -#define NET_NF_CONNTRACK_MAX 2089 - static struct ctl_table nf_ct_netfilter_table[] = { { .procname = "nf_conntrack_max", diff --git a/net/netfilter/nf_conntrack_tftp.c b/net/netfilter/nf_conntrack_tftp.c index 36f9640..2e65b543 100644 --- a/net/netfilter/nf_conntrack_tftp.c +++ b/net/netfilter/nf_conntrack_tftp.c @@ -142,6 +142,7 @@ static int __init nf_conntrack_tftp_init(void) if (ret) { pr_err("failed to register helper for pf: %u port: %u\n", tftp[i][j].tuple.src.l3num, ports[i]); + ports_c = i; nf_conntrack_tftp_fini(); return ret; } diff --git a/net/netfilter/nf_queue.c b/net/netfilter/nf_queue.c index 5baa8e2..b19ad20 100644 --- a/net/netfilter/nf_queue.c +++ b/net/netfilter/nf_queue.c @@ -26,23 +26,21 @@ * Once the queue is registered it must reinject all packets it * receives, no matter what. */ -static const struct nf_queue_handler __rcu *queue_handler __read_mostly; /* return EBUSY when somebody else is registered, return EEXIST if the * same handler is registered, return 0 in case of success. */ -void nf_register_queue_handler(const struct nf_queue_handler *qh) +void nf_register_queue_handler(struct net *net, const struct nf_queue_handler *qh) { /* should never happen, we only have one queueing backend in kernel */ - WARN_ON(rcu_access_pointer(queue_handler)); - rcu_assign_pointer(queue_handler, qh); + WARN_ON(rcu_access_pointer(net->nf.queue_handler)); + rcu_assign_pointer(net->nf.queue_handler, qh); } EXPORT_SYMBOL(nf_register_queue_handler); /* The caller must flush their queue before this */ -void nf_unregister_queue_handler(void) +void nf_unregister_queue_handler(struct net *net) { - RCU_INIT_POINTER(queue_handler, NULL); - synchronize_rcu(); + RCU_INIT_POINTER(net->nf.queue_handler, NULL); } EXPORT_SYMBOL(nf_unregister_queue_handler); @@ -103,7 +101,7 @@ void nf_queue_nf_hook_drop(struct net *net, struct nf_hook_ops *ops) const struct nf_queue_handler *qh; rcu_read_lock(); - qh = rcu_dereference(queue_handler); + qh = rcu_dereference(net->nf.queue_handler); if (qh) qh->nf_hook_drop(net, ops); rcu_read_unlock(); @@ -122,9 +120,10 @@ int nf_queue(struct sk_buff *skb, struct nf_queue_entry *entry = NULL; const struct nf_afinfo *afinfo; const struct nf_queue_handler *qh; + struct net *net = state->net; /* QUEUE == DROP if no one is waiting, to be safe. */ - qh = rcu_dereference(queue_handler); + qh = rcu_dereference(net->nf.queue_handler); if (!qh) { status = -ESRCH; goto err; diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 4d292b9..2c88187 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -2647,6 +2647,8 @@ static int nf_tables_getset(struct net *net, struct sock *nlsk, /* Only accept unspec with dump */ if (nfmsg->nfgen_family == NFPROTO_UNSPEC) return -EAFNOSUPPORT; + if (!nla[NFTA_SET_TABLE]) + return -EINVAL; set = nf_tables_set_lookup(ctx.table, nla[NFTA_SET_NAME]); if (IS_ERR(set)) @@ -2944,24 +2946,20 @@ int nf_tables_bind_set(const struct nft_ctx *ctx, struct nft_set *set, * jumps are already validated for that chain. */ list_for_each_entry(i, &set->bindings, list) { - if (binding->flags & NFT_SET_MAP && + if (i->flags & NFT_SET_MAP && i->chain == binding->chain) goto bind; } + iter.genmask = nft_genmask_next(ctx->net); iter.skip = 0; iter.count = 0; iter.err = 0; iter.fn = nf_tables_bind_check_setelem; set->ops->walk(ctx, set, &iter); - if (iter.err < 0) { - /* Destroy anonymous sets if binding fails */ - if (set->flags & NFT_SET_ANONYMOUS) - nf_tables_set_destroy(ctx, set); - + if (iter.err < 0) return iter.err; - } } bind: binding->chain = ctx->chain; @@ -3190,12 +3188,13 @@ static int nf_tables_dump_set(struct sk_buff *skb, struct netlink_callback *cb) if (nest == NULL) goto nla_put_failure; - args.cb = cb; - args.skb = skb; - args.iter.skip = cb->args[0]; - args.iter.count = 0; - args.iter.err = 0; - args.iter.fn = nf_tables_dump_setelem; + args.cb = cb; + args.skb = skb; + args.iter.genmask = nft_genmask_cur(ctx.net); + args.iter.skip = cb->args[0]; + args.iter.count = 0; + args.iter.err = 0; + args.iter.fn = nf_tables_dump_setelem; set->ops->walk(&ctx, set, &args.iter); nla_nest_end(skb, nest); @@ -4282,6 +4281,7 @@ static int nf_tables_check_loops(const struct nft_ctx *ctx, binding->chain != chain) continue; + iter.genmask = nft_genmask_next(ctx->net); iter.skip = 0; iter.count = 0; iter.err = 0; diff --git a/net/netfilter/nf_tables_core.c b/net/netfilter/nf_tables_core.c index e9f8dff..fb8b589 100644 --- a/net/netfilter/nf_tables_core.c +++ b/net/netfilter/nf_tables_core.c @@ -143,7 +143,7 @@ next_rule: list_for_each_entry_continue_rcu(rule, &chain->rules, list) { /* This rule is not active, skip. */ - if (unlikely(rule->genmask & (1 << gencursor))) + if (unlikely(rule->genmask & gencursor)) continue; rulenum++; diff --git a/net/netfilter/nfnetlink_queue.c b/net/netfilter/nfnetlink_queue.c index aa93877..5d36a09 100644 --- a/net/netfilter/nfnetlink_queue.c +++ b/net/netfilter/nfnetlink_queue.c @@ -557,7 +557,7 @@ nfqnl_build_packet_message(struct net *net, struct nfqnl_instance *queue, if (entskb->tstamp.tv64) { struct nfqnl_msg_packet_timestamp ts; - struct timespec64 kts = ktime_to_timespec64(skb->tstamp); + struct timespec64 kts = ktime_to_timespec64(entskb->tstamp); ts.sec = cpu_to_be64(kts.tv_sec); ts.usec = cpu_to_be64(kts.tv_nsec / NSEC_PER_USEC); @@ -1482,21 +1482,29 @@ static int __net_init nfnl_queue_net_init(struct net *net) net->nf.proc_netfilter, &nfqnl_file_ops)) return -ENOMEM; #endif + nf_register_queue_handler(net, &nfqh); return 0; } static void __net_exit nfnl_queue_net_exit(struct net *net) { + nf_unregister_queue_handler(net); #ifdef CONFIG_PROC_FS remove_proc_entry("nfnetlink_queue", net->nf.proc_netfilter); #endif } +static void nfnl_queue_net_exit_batch(struct list_head *net_exit_list) +{ + synchronize_rcu(); +} + static struct pernet_operations nfnl_queue_net_ops = { - .init = nfnl_queue_net_init, - .exit = nfnl_queue_net_exit, - .id = &nfnl_queue_net_id, - .size = sizeof(struct nfnl_queue_net), + .init = nfnl_queue_net_init, + .exit = nfnl_queue_net_exit, + .exit_batch = nfnl_queue_net_exit_batch, + .id = &nfnl_queue_net_id, + .size = sizeof(struct nfnl_queue_net), }; static int __init nfnetlink_queue_init(void) @@ -1517,7 +1525,6 @@ static int __init nfnetlink_queue_init(void) } register_netdevice_notifier(&nfqnl_dev_notifier); - nf_register_queue_handler(&nfqh); return status; cleanup_netlink_notifier: @@ -1529,7 +1536,6 @@ out: static void __exit nfnetlink_queue_fini(void) { - nf_unregister_queue_handler(); unregister_netdevice_notifier(&nfqnl_dev_notifier); nfnetlink_subsys_unregister(&nfqnl_subsys); netlink_unregister_notifier(&nfqnl_rtnl_notifier); diff --git a/net/netfilter/nft_hash.c b/net/netfilter/nft_hash.c index 6fa0165..f39c53a 100644 --- a/net/netfilter/nft_hash.c +++ b/net/netfilter/nft_hash.c @@ -189,7 +189,6 @@ static void nft_hash_walk(const struct nft_ctx *ctx, const struct nft_set *set, struct nft_hash_elem *he; struct rhashtable_iter hti; struct nft_set_elem elem; - u8 genmask = nft_genmask_cur(read_pnet(&set->pnet)); int err; err = rhashtable_walk_init(&priv->ht, &hti, GFP_KERNEL); @@ -218,7 +217,7 @@ static void nft_hash_walk(const struct nft_ctx *ctx, const struct nft_set *set, goto cont; if (nft_set_elem_expired(&he->ext)) goto cont; - if (!nft_set_elem_active(&he->ext, genmask)) + if (!nft_set_elem_active(&he->ext, iter->genmask)) goto cont; elem.priv = he; diff --git a/net/netfilter/nft_rbtree.c b/net/netfilter/nft_rbtree.c index f762094..7201d57 100644 --- a/net/netfilter/nft_rbtree.c +++ b/net/netfilter/nft_rbtree.c @@ -211,7 +211,6 @@ static void nft_rbtree_walk(const struct nft_ctx *ctx, struct nft_rbtree_elem *rbe; struct nft_set_elem elem; struct rb_node *node; - u8 genmask = nft_genmask_cur(read_pnet(&set->pnet)); spin_lock_bh(&nft_rbtree_lock); for (node = rb_first(&priv->root); node != NULL; node = rb_next(node)) { @@ -219,7 +218,7 @@ static void nft_rbtree_walk(const struct nft_ctx *ctx, if (iter->count < iter->skip) goto cont; - if (!nft_set_elem_active(&rbe->ext, genmask)) + if (!nft_set_elem_active(&rbe->ext, iter->genmask)) goto cont; elem.priv = rbe; diff --git a/net/netfilter/x_tables.c b/net/netfilter/x_tables.c index c69c892..2675d58 100644 --- a/net/netfilter/x_tables.c +++ b/net/netfilter/x_tables.c @@ -612,7 +612,7 @@ int xt_compat_check_entry_offsets(const void *base, const char *elems, return -EINVAL; if (strcmp(t->u.user.name, XT_STANDARD_TARGET) == 0 && - target_offset + sizeof(struct compat_xt_standard_target) != next_offset) + COMPAT_XT_ALIGN(target_offset + sizeof(struct compat_xt_standard_target)) != next_offset) return -EINVAL; /* compat_xt_entry match has less strict aligment requirements, @@ -694,7 +694,7 @@ int xt_check_entry_offsets(const void *base, return -EINVAL; if (strcmp(t->u.user.name, XT_STANDARD_TARGET) == 0 && - target_offset + sizeof(struct xt_standard_target) != next_offset) + XT_ALIGN(target_offset + sizeof(struct xt_standard_target)) != next_offset) return -EINVAL; return xt_check_entry_match(elems, base + target_offset, diff --git a/net/openvswitch/actions.c b/net/openvswitch/actions.c index 879185f..9a3eb7a 100644 --- a/net/openvswitch/actions.c +++ b/net/openvswitch/actions.c @@ -137,11 +137,23 @@ static bool is_flow_key_valid(const struct sw_flow_key *key) return !!key->eth.type; } +static void update_ethertype(struct sk_buff *skb, struct ethhdr *hdr, + __be16 ethertype) +{ + if (skb->ip_summed == CHECKSUM_COMPLETE) { + __be16 diff[] = { ~(hdr->h_proto), ethertype }; + + skb->csum = ~csum_partial((char *)diff, sizeof(diff), + ~skb->csum); + } + + hdr->h_proto = ethertype; +} + static int push_mpls(struct sk_buff *skb, struct sw_flow_key *key, const struct ovs_action_push_mpls *mpls) { __be32 *new_mpls_lse; - struct ethhdr *hdr; /* Networking stack do not allow simultaneous Tunnel and MPLS GSO. */ if (skb->encapsulation) @@ -160,9 +172,7 @@ static int push_mpls(struct sk_buff *skb, struct sw_flow_key *key, skb_postpush_rcsum(skb, new_mpls_lse, MPLS_HLEN); - hdr = eth_hdr(skb); - hdr->h_proto = mpls->mpls_ethertype; - + update_ethertype(skb, eth_hdr(skb), mpls->mpls_ethertype); if (!skb->inner_protocol) skb_set_inner_protocol(skb, skb->protocol); skb->protocol = mpls->mpls_ethertype; @@ -193,7 +203,7 @@ static int pop_mpls(struct sk_buff *skb, struct sw_flow_key *key, * field correctly in the presence of VLAN tags. */ hdr = (struct ethhdr *)(skb_mpls_header(skb) - ETH_HLEN); - hdr->h_proto = ethertype; + update_ethertype(skb, hdr, ethertype); if (eth_p_mpls(skb->protocol)) skb->protocol = ethertype; diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c index 3d5feed..d843125 100644 --- a/net/openvswitch/conntrack.c +++ b/net/openvswitch/conntrack.c @@ -818,8 +818,18 @@ static int ovs_ct_lookup(struct net *net, struct sw_flow_key *key, */ state = OVS_CS_F_TRACKED | OVS_CS_F_NEW | OVS_CS_F_RELATED; __ovs_ct_update_key(key, state, &info->zone, exp->master); - } else - return __ovs_ct_lookup(net, key, info, skb); + } else { + struct nf_conn *ct; + int err; + + err = __ovs_ct_lookup(net, key, info, skb); + if (err) + return err; + + ct = (struct nf_conn *)skb->nfct; + if (ct) + nf_ct_deliver_cached_events(ct); + } return 0; } diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c index 4040eb9..9bff6ef 100644 --- a/net/packet/af_packet.c +++ b/net/packet/af_packet.c @@ -93,6 +93,7 @@ #include <net/inet_common.h> #endif #include <linux/bpf.h> +#include <net/compat.h> #include "internal.h" @@ -3940,6 +3941,27 @@ static int packet_getsockopt(struct socket *sock, int level, int optname, } +#ifdef CONFIG_COMPAT +static int compat_packet_setsockopt(struct socket *sock, int level, int optname, + char __user *optval, unsigned int optlen) +{ + struct packet_sock *po = pkt_sk(sock->sk); + + if (level != SOL_PACKET) + return -ENOPROTOOPT; + + if (optname == PACKET_FANOUT_DATA && + po->fanout && po->fanout->type == PACKET_FANOUT_CBPF) { + optval = (char __user *)get_compat_bpf_fprog(optval); + if (!optval) + return -EFAULT; + optlen = sizeof(struct sock_fprog); + } + + return packet_setsockopt(sock, level, optname, optval, optlen); +} +#endif + static int packet_notifier(struct notifier_block *this, unsigned long msg, void *ptr) { @@ -4416,6 +4438,9 @@ static const struct proto_ops packet_ops = { .shutdown = sock_no_shutdown, .setsockopt = packet_setsockopt, .getsockopt = packet_getsockopt, +#ifdef CONFIG_COMPAT + .compat_setsockopt = compat_packet_setsockopt, +#endif .sendmsg = packet_sendmsg, .recvmsg = packet_recvmsg, .mmap = packet_mmap, diff --git a/net/rds/ib_cm.c b/net/rds/ib_cm.c index 310cabc..7c2a65a 100644 --- a/net/rds/ib_cm.c +++ b/net/rds/ib_cm.c @@ -111,7 +111,7 @@ void rds_ib_cm_connect_complete(struct rds_connection *conn, struct rdma_cm_even } } - if (conn->c_version < RDS_PROTOCOL(3,1)) { + if (conn->c_version < RDS_PROTOCOL(3, 1)) { printk(KERN_NOTICE "RDS/IB: Connection to %pI4 version %u.%u failed," " no longer supported\n", &conn->c_faddr, diff --git a/net/rds/loop.c b/net/rds/loop.c index 6b12b68..814173b 100644 --- a/net/rds/loop.c +++ b/net/rds/loop.c @@ -95,8 +95,9 @@ out: */ static void rds_loop_inc_free(struct rds_incoming *inc) { - struct rds_message *rm = container_of(inc, struct rds_message, m_inc); - rds_message_put(rm); + struct rds_message *rm = container_of(inc, struct rds_message, m_inc); + + rds_message_put(rm); } /* we need to at least give the thread something to succeed */ diff --git a/net/rds/rds.h b/net/rds/rds.h index 80256b0..387df5f 100644 --- a/net/rds/rds.h +++ b/net/rds/rds.h @@ -74,6 +74,7 @@ enum { RDS_CONN_CONNECTING, RDS_CONN_DISCONNECTING, RDS_CONN_UP, + RDS_CONN_RESETTING, RDS_CONN_ERROR, }; @@ -813,6 +814,7 @@ void rds_connect_worker(struct work_struct *); void rds_shutdown_worker(struct work_struct *); void rds_send_worker(struct work_struct *); void rds_recv_worker(struct work_struct *); +void rds_connect_path_complete(struct rds_connection *conn, int curr); void rds_connect_complete(struct rds_connection *conn); /* transport.c */ diff --git a/net/rds/recv.c b/net/rds/recv.c index c0be1ec..8413f6c 100644 --- a/net/rds/recv.c +++ b/net/rds/recv.c @@ -561,5 +561,7 @@ void rds_inc_info_copy(struct rds_incoming *inc, minfo.fport = inc->i_hdr.h_dport; } + minfo.flags = 0; + rds_info_copy(iter, &minfo, sizeof(minfo)); } diff --git a/net/rds/send.c b/net/rds/send.c index c9cdb35..b1962f8 100644 --- a/net/rds/send.c +++ b/net/rds/send.c @@ -99,6 +99,7 @@ void rds_send_reset(struct rds_connection *conn) list_splice_init(&conn->c_retrans, &conn->c_send_queue); spin_unlock_irqrestore(&conn->c_lock, flags); } +EXPORT_SYMBOL_GPL(rds_send_reset); static int acquire_in_xmit(struct rds_connection *conn) { diff --git a/net/rds/sysctl.c b/net/rds/sysctl.c index c173f69..e381bbc 100644 --- a/net/rds/sysctl.c +++ b/net/rds/sysctl.c @@ -102,7 +102,8 @@ int rds_sysctl_init(void) rds_sysctl_reconnect_min = msecs_to_jiffies(1); rds_sysctl_reconnect_min_jiffies = rds_sysctl_reconnect_min; - rds_sysctl_reg_table = register_net_sysctl(&init_net,"net/rds", rds_sysctl_rds_table); + rds_sysctl_reg_table = + register_net_sysctl(&init_net, "net/rds", rds_sysctl_rds_table); if (!rds_sysctl_reg_table) return -ENOMEM; return 0; diff --git a/net/rds/tcp.c b/net/rds/tcp.c index 86187da..74ee126 100644 --- a/net/rds/tcp.c +++ b/net/rds/tcp.c @@ -126,9 +126,81 @@ void rds_tcp_restore_callbacks(struct socket *sock, } /* - * This is the only path that sets tc->t_sock. Send and receive trust that - * it is set. The RDS_CONN_UP bit protects those paths from being - * called while it isn't set. + * rds_tcp_reset_callbacks() switches the to the new sock and + * returns the existing tc->t_sock. + * + * The only functions that set tc->t_sock are rds_tcp_set_callbacks + * and rds_tcp_reset_callbacks. Send and receive trust that + * it is set. The absence of RDS_CONN_UP bit protects those paths + * from being called while it isn't set. + */ +void rds_tcp_reset_callbacks(struct socket *sock, + struct rds_connection *conn) +{ + struct rds_tcp_connection *tc = conn->c_transport_data; + struct socket *osock = tc->t_sock; + + if (!osock) + goto newsock; + + /* Need to resolve a duelling SYN between peers. + * We have an outstanding SYN to this peer, which may + * potentially have transitioned to the RDS_CONN_UP state, + * so we must quiesce any send threads before resetting + * c_transport_data. We quiesce these threads by setting + * c_state to something other than RDS_CONN_UP, and then + * waiting for any existing threads in rds_send_xmit to + * complete release_in_xmit(). (Subsequent threads entering + * rds_send_xmit() will bail on !rds_conn_up(). + * + * However an incoming syn-ack at this point would end up + * marking the conn as RDS_CONN_UP, and would again permit + * rds_send_xmi() threads through, so ideally we would + * synchronize on RDS_CONN_UP after lock_sock(), but cannot + * do that: waiting on !RDS_IN_XMIT after lock_sock() may + * end up deadlocking with tcp_sendmsg(), and the RDS_IN_XMIT + * would not get set. As a result, we set c_state to + * RDS_CONN_RESETTTING, to ensure that rds_tcp_state_change + * cannot mark rds_conn_path_up() in the window before lock_sock() + */ + atomic_set(&conn->c_state, RDS_CONN_RESETTING); + wait_event(conn->c_waitq, !test_bit(RDS_IN_XMIT, &conn->c_flags)); + lock_sock(osock->sk); + /* reset receive side state for rds_tcp_data_recv() for osock */ + if (tc->t_tinc) { + rds_inc_put(&tc->t_tinc->ti_inc); + tc->t_tinc = NULL; + } + tc->t_tinc_hdr_rem = sizeof(struct rds_header); + tc->t_tinc_data_rem = 0; + tc->t_sock = NULL; + + write_lock_bh(&osock->sk->sk_callback_lock); + + osock->sk->sk_user_data = NULL; + osock->sk->sk_data_ready = tc->t_orig_data_ready; + osock->sk->sk_write_space = tc->t_orig_write_space; + osock->sk->sk_state_change = tc->t_orig_state_change; + write_unlock_bh(&osock->sk->sk_callback_lock); + release_sock(osock->sk); + sock_release(osock); +newsock: + rds_send_reset(conn); + lock_sock(sock->sk); + write_lock_bh(&sock->sk->sk_callback_lock); + tc->t_sock = sock; + sock->sk->sk_user_data = conn; + sock->sk->sk_data_ready = rds_tcp_data_ready; + sock->sk->sk_write_space = rds_tcp_write_space; + sock->sk->sk_state_change = rds_tcp_state_change; + + write_unlock_bh(&sock->sk->sk_callback_lock); + release_sock(sock->sk); +} + +/* Add tc to rds_tcp_tc_list and set tc->t_sock. See comments + * above rds_tcp_reset_callbacks for notes about synchronization + * with data path */ void rds_tcp_set_callbacks(struct socket *sock, struct rds_connection *conn) { diff --git a/net/rds/tcp.h b/net/rds/tcp.h index 41c2283..7940bab 100644 --- a/net/rds/tcp.h +++ b/net/rds/tcp.h @@ -50,6 +50,7 @@ struct rds_tcp_statistics { void rds_tcp_tune(struct socket *sock); void rds_tcp_nonagle(struct socket *sock); void rds_tcp_set_callbacks(struct socket *sock, struct rds_connection *conn); +void rds_tcp_reset_callbacks(struct socket *sock, struct rds_connection *conn); void rds_tcp_restore_callbacks(struct socket *sock, struct rds_tcp_connection *tc); u32 rds_tcp_snd_nxt(struct rds_tcp_connection *tc); @@ -82,7 +83,7 @@ int rds_tcp_inc_copy_to_user(struct rds_incoming *inc, struct iov_iter *to); void rds_tcp_xmit_prepare(struct rds_connection *conn); void rds_tcp_xmit_complete(struct rds_connection *conn); int rds_tcp_xmit(struct rds_connection *conn, struct rds_message *rm, - unsigned int hdr_off, unsigned int sg, unsigned int off); + unsigned int hdr_off, unsigned int sg, unsigned int off); void rds_tcp_write_space(struct sock *sk); /* tcp_stats.c */ diff --git a/net/rds/tcp_connect.c b/net/rds/tcp_connect.c index fb82e0a..f6e95d6 100644 --- a/net/rds/tcp_connect.c +++ b/net/rds/tcp_connect.c @@ -54,19 +54,19 @@ void rds_tcp_state_change(struct sock *sk) rdsdebug("sock %p state_change to %d\n", tc->t_sock, sk->sk_state); - switch(sk->sk_state) { - /* ignore connecting sockets as they make progress */ - case TCP_SYN_SENT: - case TCP_SYN_RECV: - break; - case TCP_ESTABLISHED: - rds_connect_complete(conn); - break; - case TCP_CLOSE_WAIT: - case TCP_CLOSE: - rds_conn_drop(conn); - default: - break; + switch (sk->sk_state) { + /* ignore connecting sockets as they make progress */ + case TCP_SYN_SENT: + case TCP_SYN_RECV: + break; + case TCP_ESTABLISHED: + rds_connect_path_complete(conn, RDS_CONN_CONNECTING); + break; + case TCP_CLOSE_WAIT: + case TCP_CLOSE: + rds_conn_drop(conn); + default: + break; } out: read_unlock_bh(&sk->sk_callback_lock); diff --git a/net/rds/tcp_listen.c b/net/rds/tcp_listen.c index 4bf4bef..245542c 100644 --- a/net/rds/tcp_listen.c +++ b/net/rds/tcp_listen.c @@ -78,7 +78,6 @@ int rds_tcp_accept_one(struct socket *sock) struct inet_sock *inet; struct rds_tcp_connection *rs_tcp = NULL; int conn_state; - struct sock *nsk; if (!sock) /* module unload or netns delete in progress */ return -ENETUNREACH; @@ -136,26 +135,21 @@ int rds_tcp_accept_one(struct socket *sock) !conn->c_outgoing) { goto rst_nsk; } else { - atomic_set(&conn->c_state, RDS_CONN_CONNECTING); - wait_event(conn->c_waitq, - !test_bit(RDS_IN_XMIT, &conn->c_flags)); - rds_tcp_restore_callbacks(rs_tcp->t_sock, rs_tcp); + rds_tcp_reset_callbacks(new_sock, conn); conn->c_outgoing = 0; + /* rds_connect_path_complete() marks RDS_CONN_UP */ + rds_connect_path_complete(conn, RDS_CONN_RESETTING); } + } else { + rds_tcp_set_callbacks(new_sock, conn); + rds_connect_path_complete(conn, RDS_CONN_CONNECTING); } - rds_tcp_set_callbacks(new_sock, conn); - rds_connect_complete(conn); /* marks RDS_CONN_UP */ new_sock = NULL; ret = 0; goto out; rst_nsk: /* reset the newly returned accept sock and bail */ - nsk = new_sock->sk; - rds_tcp_stats_inc(s_tcp_listen_closed_stale); - nsk->sk_user_data = NULL; - nsk->sk_prot->disconnect(nsk, 0); - tcp_done(nsk); - new_sock = NULL; + kernel_sock_shutdown(new_sock, SHUT_RDWR); ret = 0; out: if (rs_tcp) diff --git a/net/rds/tcp_recv.c b/net/rds/tcp_recv.c index c3196f9..6e6a711 100644 --- a/net/rds/tcp_recv.c +++ b/net/rds/tcp_recv.c @@ -171,7 +171,7 @@ static int rds_tcp_data_recv(read_descriptor_t *desc, struct sk_buff *skb, while (left) { if (!tinc) { tinc = kmem_cache_alloc(rds_tcp_incoming_slab, - arg->gfp); + arg->gfp); if (!tinc) { desc->error = -ENOMEM; goto out; diff --git a/net/rds/tcp_send.c b/net/rds/tcp_send.c index 22d0f20..618be69 100644 --- a/net/rds/tcp_send.c +++ b/net/rds/tcp_send.c @@ -66,19 +66,19 @@ void rds_tcp_xmit_complete(struct rds_connection *conn) static int rds_tcp_sendmsg(struct socket *sock, void *data, unsigned int len) { struct kvec vec = { - .iov_base = data, - .iov_len = len, + .iov_base = data, + .iov_len = len, + }; + struct msghdr msg = { + .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL, }; - struct msghdr msg = { - .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL, - }; return kernel_sendmsg(sock, &msg, &vec, 1, vec.iov_len); } /* the core send_sem serializes this with other xmit and shutdown */ int rds_tcp_xmit(struct rds_connection *conn, struct rds_message *rm, - unsigned int hdr_off, unsigned int sg, unsigned int off) + unsigned int hdr_off, unsigned int sg, unsigned int off) { struct rds_tcp_connection *tc = conn->c_transport_data; int done = 0; @@ -196,7 +196,7 @@ void rds_tcp_write_space(struct sock *sk) tc->t_last_seen_una = rds_tcp_snd_una(tc); rds_send_drop_acked(conn, rds_tcp_snd_una(tc), rds_tcp_is_acked); - if ((atomic_read(&sk->sk_wmem_alloc) << 1) <= sk->sk_sndbuf) + if ((atomic_read(&sk->sk_wmem_alloc) << 1) <= sk->sk_sndbuf) queue_delayed_work(rds_wq, &conn->c_send_w, 0); out: diff --git a/net/rds/threads.c b/net/rds/threads.c index 454aa6d..4a32304 100644 --- a/net/rds/threads.c +++ b/net/rds/threads.c @@ -71,9 +71,9 @@ struct workqueue_struct *rds_wq; EXPORT_SYMBOL_GPL(rds_wq); -void rds_connect_complete(struct rds_connection *conn) +void rds_connect_path_complete(struct rds_connection *conn, int curr) { - if (!rds_conn_transition(conn, RDS_CONN_CONNECTING, RDS_CONN_UP)) { + if (!rds_conn_transition(conn, curr, RDS_CONN_UP)) { printk(KERN_WARNING "%s: Cannot transition to state UP, " "current state is %d\n", __func__, @@ -90,6 +90,12 @@ void rds_connect_complete(struct rds_connection *conn) queue_delayed_work(rds_wq, &conn->c_send_w, 0); queue_delayed_work(rds_wq, &conn->c_recv_w, 0); } +EXPORT_SYMBOL_GPL(rds_connect_path_complete); + +void rds_connect_complete(struct rds_connection *conn) +{ + rds_connect_path_complete(conn, RDS_CONN_CONNECTING); +} EXPORT_SYMBOL_GPL(rds_connect_complete); /* diff --git a/net/rds/transport.c b/net/rds/transport.c index f3afd1d..2ffd3e30 100644 --- a/net/rds/transport.c +++ b/net/rds/transport.c @@ -140,8 +140,7 @@ unsigned int rds_trans_stats_info_copy(struct rds_info_iterator *iter, rds_info_iter_unmap(iter); down_read(&rds_trans_sem); - for (i = 0; i < RDS_TRANS_COUNT; i++) - { + for (i = 0; i < RDS_TRANS_COUNT; i++) { trans = transports[i]; if (!trans || !trans->stats_info_copy) continue; diff --git a/net/rxrpc/rxkad.c b/net/rxrpc/rxkad.c index 6b726a0..bab56ed 100644 --- a/net/rxrpc/rxkad.c +++ b/net/rxrpc/rxkad.c @@ -1162,9 +1162,7 @@ static int rxkad_init(void) /* pin the cipher we need so that the crypto layer doesn't invoke * keventd to go get it */ rxkad_ci = crypto_alloc_skcipher("pcbc(fcrypt)", 0, CRYPTO_ALG_ASYNC); - if (IS_ERR(rxkad_ci)) - return PTR_ERR(rxkad_ci); - return 0; + return PTR_ERR_OR_ZERO(rxkad_ci); } /* diff --git a/net/sched/act_api.c b/net/sched/act_api.c index 336774a..c7a0b0d 100644 --- a/net/sched/act_api.c +++ b/net/sched/act_api.c @@ -1118,7 +1118,7 @@ tc_dump_action(struct sk_buff *skb, struct netlink_callback *cb) nla_nest_end(skb, nest); ret = skb->len; } else - nla_nest_cancel(skb, nest); + nlmsg_trim(skb, b); nlh->nlmsg_len = skb_tail_pointer(skb) - b; if (NETLINK_CB(cb->skb).portid && ret) diff --git a/net/sched/act_ife.c b/net/sched/act_ife.c index 658046d..ea4a2fe 100644 --- a/net/sched/act_ife.c +++ b/net/sched/act_ife.c @@ -106,9 +106,9 @@ int ife_get_meta_u16(struct sk_buff *skb, struct tcf_meta_info *mi) } EXPORT_SYMBOL_GPL(ife_get_meta_u16); -int ife_alloc_meta_u32(struct tcf_meta_info *mi, void *metaval) +int ife_alloc_meta_u32(struct tcf_meta_info *mi, void *metaval, gfp_t gfp) { - mi->metaval = kmemdup(metaval, sizeof(u32), GFP_KERNEL); + mi->metaval = kmemdup(metaval, sizeof(u32), gfp); if (!mi->metaval) return -ENOMEM; @@ -116,9 +116,9 @@ int ife_alloc_meta_u32(struct tcf_meta_info *mi, void *metaval) } EXPORT_SYMBOL_GPL(ife_alloc_meta_u32); -int ife_alloc_meta_u16(struct tcf_meta_info *mi, void *metaval) +int ife_alloc_meta_u16(struct tcf_meta_info *mi, void *metaval, gfp_t gfp) { - mi->metaval = kmemdup(metaval, sizeof(u16), GFP_KERNEL); + mi->metaval = kmemdup(metaval, sizeof(u16), gfp); if (!mi->metaval) return -ENOMEM; @@ -240,10 +240,10 @@ static int ife_validate_metatype(struct tcf_meta_ops *ops, void *val, int len) } /* called when adding new meta information - * under ife->tcf_lock + * under ife->tcf_lock for existing action */ static int load_metaops_and_vet(struct tcf_ife_info *ife, u32 metaid, - void *val, int len) + void *val, int len, bool exists) { struct tcf_meta_ops *ops = find_ife_oplist(metaid); int ret = 0; @@ -251,11 +251,13 @@ static int load_metaops_and_vet(struct tcf_ife_info *ife, u32 metaid, if (!ops) { ret = -ENOENT; #ifdef CONFIG_MODULES - spin_unlock_bh(&ife->tcf_lock); + if (exists) + spin_unlock_bh(&ife->tcf_lock); rtnl_unlock(); request_module("ifemeta%u", metaid); rtnl_lock(); - spin_lock_bh(&ife->tcf_lock); + if (exists) + spin_lock_bh(&ife->tcf_lock); ops = find_ife_oplist(metaid); #endif } @@ -272,10 +274,10 @@ static int load_metaops_and_vet(struct tcf_ife_info *ife, u32 metaid, } /* called when adding new meta information - * under ife->tcf_lock + * under ife->tcf_lock for existing action */ static int add_metainfo(struct tcf_ife_info *ife, u32 metaid, void *metaval, - int len) + int len, bool atomic) { struct tcf_meta_info *mi = NULL; struct tcf_meta_ops *ops = find_ife_oplist(metaid); @@ -284,7 +286,7 @@ static int add_metainfo(struct tcf_ife_info *ife, u32 metaid, void *metaval, if (!ops) return -ENOENT; - mi = kzalloc(sizeof(*mi), GFP_KERNEL); + mi = kzalloc(sizeof(*mi), atomic ? GFP_ATOMIC : GFP_KERNEL); if (!mi) { /*put back what find_ife_oplist took */ module_put(ops->owner); @@ -294,7 +296,7 @@ static int add_metainfo(struct tcf_ife_info *ife, u32 metaid, void *metaval, mi->metaid = metaid; mi->ops = ops; if (len > 0) { - ret = ops->alloc(mi, metaval); + ret = ops->alloc(mi, metaval, atomic ? GFP_ATOMIC : GFP_KERNEL); if (ret != 0) { kfree(mi); module_put(ops->owner); @@ -313,11 +315,13 @@ static int use_all_metadata(struct tcf_ife_info *ife) int rc = 0; int installed = 0; + read_lock(&ife_mod_lock); list_for_each_entry(o, &ifeoplist, list) { - rc = add_metainfo(ife, o->metaid, NULL, 0); + rc = add_metainfo(ife, o->metaid, NULL, 0, true); if (rc == 0) installed += 1; } + read_unlock(&ife_mod_lock); if (installed) return 0; @@ -385,8 +389,9 @@ static void tcf_ife_cleanup(struct tc_action *a, int bind) spin_unlock_bh(&ife->tcf_lock); } -/* under ife->tcf_lock */ -static int populate_metalist(struct tcf_ife_info *ife, struct nlattr **tb) +/* under ife->tcf_lock for existing action */ +static int populate_metalist(struct tcf_ife_info *ife, struct nlattr **tb, + bool exists) { int len = 0; int rc = 0; @@ -398,11 +403,11 @@ static int populate_metalist(struct tcf_ife_info *ife, struct nlattr **tb) val = nla_data(tb[i]); len = nla_len(tb[i]); - rc = load_metaops_and_vet(ife, i, val, len); + rc = load_metaops_and_vet(ife, i, val, len, exists); if (rc != 0) return rc; - rc = add_metainfo(ife, i, val, len); + rc = add_metainfo(ife, i, val, len, exists); if (rc) return rc; } @@ -474,7 +479,8 @@ static int tcf_ife_init(struct net *net, struct nlattr *nla, saddr = nla_data(tb[TCA_IFE_SMAC]); } - spin_lock_bh(&ife->tcf_lock); + if (exists) + spin_lock_bh(&ife->tcf_lock); ife->tcf_action = parm->action; if (parm->flags & IFE_ENCODE) { @@ -504,11 +510,12 @@ metadata_parse_err: if (ret == ACT_P_CREATED) _tcf_ife_cleanup(a, bind); - spin_unlock_bh(&ife->tcf_lock); + if (exists) + spin_unlock_bh(&ife->tcf_lock); return err; } - err = populate_metalist(ife, tb2); + err = populate_metalist(ife, tb2, exists); if (err) goto metadata_parse_err; @@ -523,12 +530,14 @@ metadata_parse_err: if (ret == ACT_P_CREATED) _tcf_ife_cleanup(a, bind); - spin_unlock_bh(&ife->tcf_lock); + if (exists) + spin_unlock_bh(&ife->tcf_lock); return err; } } - spin_unlock_bh(&ife->tcf_lock); + if (exists) + spin_unlock_bh(&ife->tcf_lock); if (ret == ACT_P_CREATED) tcf_hash_insert(tn, a); diff --git a/net/sched/act_ipt.c b/net/sched/act_ipt.c index 9f002ad..d4bd19e 100644 --- a/net/sched/act_ipt.c +++ b/net/sched/act_ipt.c @@ -121,10 +121,13 @@ static int __tcf_ipt_init(struct tc_action_net *tn, struct nlattr *nla, } td = (struct xt_entry_target *)nla_data(tb[TCA_IPT_TARG]); - if (nla_len(tb[TCA_IPT_TARG]) < td->u.target_size) + if (nla_len(tb[TCA_IPT_TARG]) < td->u.target_size) { + if (exists) + tcf_hash_release(a, bind); return -EINVAL; + } - if (!tcf_hash_check(tn, index, a, bind)) { + if (!exists) { ret = tcf_hash_create(tn, index, est, a, sizeof(*ipt), bind, false); if (ret) diff --git a/net/sched/act_police.c b/net/sched/act_police.c index b884dae..c557789 100644 --- a/net/sched/act_police.c +++ b/net/sched/act_police.c @@ -38,7 +38,7 @@ struct tcf_police { bool peak_present; }; #define to_police(pc) \ - container_of(pc, struct tcf_police, common) + container_of(pc->priv, struct tcf_police, common) #define POL_TAB_MASK 15 @@ -119,14 +119,12 @@ static int tcf_act_police_locate(struct net *net, struct nlattr *nla, struct nlattr *est, struct tc_action *a, int ovr, int bind) { - unsigned int h; int ret = 0, err; struct nlattr *tb[TCA_POLICE_MAX + 1]; struct tc_police *parm; struct tcf_police *police; struct qdisc_rate_table *R_tab = NULL, *P_tab = NULL; struct tc_action_net *tn = net_generic(net, police_net_id); - struct tcf_hashinfo *hinfo = tn->hinfo; int size; if (nla == NULL) @@ -145,7 +143,7 @@ static int tcf_act_police_locate(struct net *net, struct nlattr *nla, if (parm->index) { if (tcf_hash_search(tn, a, parm->index)) { - police = to_police(a->priv); + police = to_police(a); if (bind) { police->tcf_bindcnt += 1; police->tcf_refcnt += 1; @@ -156,16 +154,15 @@ static int tcf_act_police_locate(struct net *net, struct nlattr *nla, /* not replacing */ return -EEXIST; } + } else { + ret = tcf_hash_create(tn, parm->index, NULL, a, + sizeof(*police), bind, false); + if (ret) + return ret; + ret = ACT_P_CREATED; } - police = kzalloc(sizeof(*police), GFP_KERNEL); - if (police == NULL) - return -ENOMEM; - ret = ACT_P_CREATED; - police->tcf_refcnt = 1; - spin_lock_init(&police->tcf_lock); - if (bind) - police->tcf_bindcnt = 1; + police = to_police(a); override: if (parm->rate.rate) { err = -ENOMEM; @@ -237,16 +234,8 @@ override: return ret; police->tcfp_t_c = ktime_get_ns(); - police->tcf_index = parm->index ? parm->index : - tcf_hash_new_index(tn); - police->tcf_tm.install = jiffies; - police->tcf_tm.lastuse = jiffies; - h = tcf_hash(police->tcf_index, POL_TAB_MASK); - spin_lock_bh(&hinfo->lock); - hlist_add_head(&police->tcf_head, &hinfo->htab[h]); - spin_unlock_bh(&hinfo->lock); + tcf_hash_insert(tn, a); - a->priv = police; return ret; failure_unlock: @@ -255,7 +244,7 @@ failure: qdisc_put_rtab(P_tab); qdisc_put_rtab(R_tab); if (ret == ACT_P_CREATED) - kfree(police); + tcf_hash_cleanup(a, est); return err; } diff --git a/net/sched/cls_flower.c b/net/sched/cls_flower.c index 730aaca..b3b7978 100644 --- a/net/sched/cls_flower.c +++ b/net/sched/cls_flower.c @@ -171,7 +171,7 @@ static void fl_hw_destroy_filter(struct tcf_proto *tp, unsigned long cookie) struct tc_cls_flower_offload offload = {0}; struct tc_to_netdev tc; - if (!tc_should_offload(dev, 0)) + if (!tc_should_offload(dev, tp, 0)) return; offload.command = TC_CLSFLOWER_DESTROY; @@ -194,7 +194,7 @@ static void fl_hw_replace_filter(struct tcf_proto *tp, struct tc_cls_flower_offload offload = {0}; struct tc_to_netdev tc; - if (!tc_should_offload(dev, flags)) + if (!tc_should_offload(dev, tp, flags)) return; offload.command = TC_CLSFLOWER_REPLACE; @@ -216,7 +216,7 @@ static void fl_hw_update_stats(struct tcf_proto *tp, struct cls_fl_filter *f) struct tc_cls_flower_offload offload = {0}; struct tc_to_netdev tc; - if (!tc_should_offload(dev, 0)) + if (!tc_should_offload(dev, tp, 0)) return; offload.command = TC_CLSFLOWER_STATS; diff --git a/net/sched/cls_u32.c b/net/sched/cls_u32.c index 079b43b..ffe593e 100644 --- a/net/sched/cls_u32.c +++ b/net/sched/cls_u32.c @@ -440,7 +440,7 @@ static void u32_remove_hw_knode(struct tcf_proto *tp, u32 handle) offload.type = TC_SETUP_CLSU32; offload.cls_u32 = &u32_offload; - if (tc_should_offload(dev, 0)) { + if (tc_should_offload(dev, tp, 0)) { offload.cls_u32->command = TC_CLSU32_DELETE_KNODE; offload.cls_u32->knode.handle = handle; dev->netdev_ops->ndo_setup_tc(dev, tp->q->handle, @@ -457,20 +457,21 @@ static int u32_replace_hw_hnode(struct tcf_proto *tp, struct tc_to_netdev offload; int err; + if (!tc_should_offload(dev, tp, flags)) + return tc_skip_sw(flags) ? -EINVAL : 0; + offload.type = TC_SETUP_CLSU32; offload.cls_u32 = &u32_offload; - if (tc_should_offload(dev, flags)) { - offload.cls_u32->command = TC_CLSU32_NEW_HNODE; - offload.cls_u32->hnode.divisor = h->divisor; - offload.cls_u32->hnode.handle = h->handle; - offload.cls_u32->hnode.prio = h->prio; + offload.cls_u32->command = TC_CLSU32_NEW_HNODE; + offload.cls_u32->hnode.divisor = h->divisor; + offload.cls_u32->hnode.handle = h->handle; + offload.cls_u32->hnode.prio = h->prio; - err = dev->netdev_ops->ndo_setup_tc(dev, tp->q->handle, - tp->protocol, &offload); - if (tc_skip_sw(flags)) - return err; - } + err = dev->netdev_ops->ndo_setup_tc(dev, tp->q->handle, + tp->protocol, &offload); + if (tc_skip_sw(flags)) + return err; return 0; } @@ -484,7 +485,7 @@ static void u32_clear_hw_hnode(struct tcf_proto *tp, struct tc_u_hnode *h) offload.type = TC_SETUP_CLSU32; offload.cls_u32 = &u32_offload; - if (tc_should_offload(dev, 0)) { + if (tc_should_offload(dev, tp, 0)) { offload.cls_u32->command = TC_CLSU32_DELETE_HNODE; offload.cls_u32->hnode.divisor = h->divisor; offload.cls_u32->hnode.handle = h->handle; @@ -507,27 +508,28 @@ static int u32_replace_hw_knode(struct tcf_proto *tp, offload.type = TC_SETUP_CLSU32; offload.cls_u32 = &u32_offload; - if (tc_should_offload(dev, flags)) { - offload.cls_u32->command = TC_CLSU32_REPLACE_KNODE; - offload.cls_u32->knode.handle = n->handle; - offload.cls_u32->knode.fshift = n->fshift; + if (!tc_should_offload(dev, tp, flags)) + return tc_skip_sw(flags) ? -EINVAL : 0; + + offload.cls_u32->command = TC_CLSU32_REPLACE_KNODE; + offload.cls_u32->knode.handle = n->handle; + offload.cls_u32->knode.fshift = n->fshift; #ifdef CONFIG_CLS_U32_MARK - offload.cls_u32->knode.val = n->val; - offload.cls_u32->knode.mask = n->mask; + offload.cls_u32->knode.val = n->val; + offload.cls_u32->knode.mask = n->mask; #else - offload.cls_u32->knode.val = 0; - offload.cls_u32->knode.mask = 0; + offload.cls_u32->knode.val = 0; + offload.cls_u32->knode.mask = 0; #endif - offload.cls_u32->knode.sel = &n->sel; - offload.cls_u32->knode.exts = &n->exts; - if (n->ht_down) - offload.cls_u32->knode.link_handle = n->ht_down->handle; - - err = dev->netdev_ops->ndo_setup_tc(dev, tp->q->handle, - tp->protocol, &offload); - if (tc_skip_sw(flags)) - return err; - } + offload.cls_u32->knode.sel = &n->sel; + offload.cls_u32->knode.exts = &n->exts; + if (n->ht_down) + offload.cls_u32->knode.link_handle = n->ht_down->handle; + + err = dev->netdev_ops->ndo_setup_tc(dev, tp->q->handle, + tp->protocol, &offload); + if (tc_skip_sw(flags)) + return err; return 0; } @@ -863,7 +865,7 @@ static int u32_change(struct net *net, struct sk_buff *in_skb, if (tb[TCA_U32_FLAGS]) { flags = nla_get_u32(tb[TCA_U32_FLAGS]); if (!tc_flags_valid(flags)) - return err; + return -EINVAL; } n = (struct tc_u_knode *)*arg; @@ -921,11 +923,17 @@ static int u32_change(struct net *net, struct sk_buff *in_skb, ht->divisor = divisor; ht->handle = handle; ht->prio = tp->prio; + + err = u32_replace_hw_hnode(tp, ht, flags); + if (err) { + kfree(ht); + return err; + } + RCU_INIT_POINTER(ht->next, tp_c->hlist); rcu_assign_pointer(tp_c->hlist, ht); *arg = (unsigned long)ht; - u32_replace_hw_hnode(tp, ht, flags); return 0; } diff --git a/net/sched/sch_drr.c b/net/sched/sch_drr.c index a63e879..bf8af2c 100644 --- a/net/sched/sch_drr.c +++ b/net/sched/sch_drr.c @@ -375,6 +375,7 @@ static int drr_enqueue(struct sk_buff *skb, struct Qdisc *sch) cl->deficit = cl->quantum; } + qdisc_qstats_backlog_inc(sch, skb); sch->q.qlen++; return err; } @@ -407,6 +408,7 @@ static struct sk_buff *drr_dequeue(struct Qdisc *sch) bstats_update(&cl->bstats, skb); qdisc_bstats_update(sch, skb); + qdisc_qstats_backlog_dec(sch, skb); sch->q.qlen--; return skb; } @@ -428,6 +430,7 @@ static unsigned int drr_drop(struct Qdisc *sch) if (cl->qdisc->ops->drop) { len = cl->qdisc->ops->drop(cl->qdisc); if (len > 0) { + sch->qstats.backlog -= len; sch->q.qlen--; if (cl->qdisc->q.qlen == 0) list_del(&cl->alist); @@ -463,6 +466,7 @@ static void drr_reset_qdisc(struct Qdisc *sch) qdisc_reset(cl->qdisc); } } + sch->qstats.backlog = 0; sch->q.qlen = 0; } diff --git a/net/sched/sch_fifo.c b/net/sched/sch_fifo.c index 2177eac..2e4bd2c 100644 --- a/net/sched/sch_fifo.c +++ b/net/sched/sch_fifo.c @@ -37,14 +37,18 @@ static int pfifo_enqueue(struct sk_buff *skb, struct Qdisc *sch) static int pfifo_tail_enqueue(struct sk_buff *skb, struct Qdisc *sch) { + unsigned int prev_backlog; + if (likely(skb_queue_len(&sch->q) < sch->limit)) return qdisc_enqueue_tail(skb, sch); + prev_backlog = sch->qstats.backlog; /* queue full, remove one skb to fulfill the limit */ __qdisc_queue_drop_head(sch, &sch->q); qdisc_qstats_drop(sch); qdisc_enqueue_tail(skb, sch); + qdisc_tree_reduce_backlog(sch, 0, prev_backlog - sch->qstats.backlog); return NET_XMIT_CN; } diff --git a/net/sched/sch_fq_codel.c b/net/sched/sch_fq_codel.c index 6883a89..da250b2 100644 --- a/net/sched/sch_fq_codel.c +++ b/net/sched/sch_fq_codel.c @@ -199,6 +199,7 @@ static int fq_codel_enqueue(struct sk_buff *skb, struct Qdisc *sch) unsigned int idx, prev_backlog, prev_qlen; struct fq_codel_flow *flow; int uninitialized_var(ret); + unsigned int pkt_len; bool memory_limited; idx = fq_codel_classify(skb, sch, &ret); @@ -230,6 +231,8 @@ static int fq_codel_enqueue(struct sk_buff *skb, struct Qdisc *sch) prev_backlog = sch->qstats.backlog; prev_qlen = sch->q.qlen; + /* save this packet length as it might be dropped by fq_codel_drop() */ + pkt_len = qdisc_pkt_len(skb); /* fq_codel_drop() is quite expensive, as it performs a linear search * in q->backlogs[] to find a fat flow. * So instead of dropping a single packet, drop half of its backlog @@ -237,14 +240,23 @@ static int fq_codel_enqueue(struct sk_buff *skb, struct Qdisc *sch) */ ret = fq_codel_drop(sch, q->drop_batch_size); - q->drop_overlimit += prev_qlen - sch->q.qlen; + prev_qlen -= sch->q.qlen; + prev_backlog -= sch->qstats.backlog; + q->drop_overlimit += prev_qlen; if (memory_limited) - q->drop_overmemory += prev_qlen - sch->q.qlen; - /* As we dropped packet(s), better let upper stack know this */ - qdisc_tree_reduce_backlog(sch, prev_qlen - sch->q.qlen, - prev_backlog - sch->qstats.backlog); + q->drop_overmemory += prev_qlen; - return ret == idx ? NET_XMIT_CN : NET_XMIT_SUCCESS; + /* As we dropped packet(s), better let upper stack know this. + * If we dropped a packet for this flow, return NET_XMIT_CN, + * but in this case, our parents wont increase their backlogs. + */ + if (ret == idx) { + qdisc_tree_reduce_backlog(sch, prev_qlen - 1, + prev_backlog - pkt_len); + return NET_XMIT_CN; + } + qdisc_tree_reduce_backlog(sch, prev_qlen, prev_backlog); + return NET_XMIT_SUCCESS; } /* This is the specific function called from codel_dequeue() @@ -649,7 +661,7 @@ static int fq_codel_dump_class_stats(struct Qdisc *sch, unsigned long cl, qs.backlog = q->backlogs[idx]; qs.drops = flow->dropped; } - if (gnet_stats_copy_queue(d, NULL, &qs, 0) < 0) + if (gnet_stats_copy_queue(d, NULL, &qs, qs.qlen) < 0) return -1; if (idx < q->flows_cnt) return gnet_stats_copy_app(d, &xstats, sizeof(xstats)); diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c index 269dd71..f9e0e9c 100644 --- a/net/sched/sch_generic.c +++ b/net/sched/sch_generic.c @@ -49,6 +49,7 @@ static inline int dev_requeue_skb(struct sk_buff *skb, struct Qdisc *q) { q->gso_skb = skb; q->qstats.requeues++; + qdisc_qstats_backlog_inc(q, skb); q->q.qlen++; /* it's still part of the queue */ __netif_schedule(q); @@ -92,6 +93,7 @@ static struct sk_buff *dequeue_skb(struct Qdisc *q, bool *validate, txq = skb_get_tx_queue(txq->dev, skb); if (!netif_xmit_frozen_or_stopped(txq)) { q->gso_skb = NULL; + qdisc_qstats_backlog_dec(q, skb); q->q.qlen--; } else skb = NULL; diff --git a/net/sched/sch_hfsc.c b/net/sched/sch_hfsc.c index d783d7c..1ac9f9f 100644 --- a/net/sched/sch_hfsc.c +++ b/net/sched/sch_hfsc.c @@ -1529,6 +1529,7 @@ hfsc_reset_qdisc(struct Qdisc *sch) q->eligible = RB_ROOT; INIT_LIST_HEAD(&q->droplist); qdisc_watchdog_cancel(&q->watchdog); + sch->qstats.backlog = 0; sch->q.qlen = 0; } @@ -1559,14 +1560,6 @@ hfsc_dump_qdisc(struct Qdisc *sch, struct sk_buff *skb) struct hfsc_sched *q = qdisc_priv(sch); unsigned char *b = skb_tail_pointer(skb); struct tc_hfsc_qopt qopt; - struct hfsc_class *cl; - unsigned int i; - - sch->qstats.backlog = 0; - for (i = 0; i < q->clhash.hashsize; i++) { - hlist_for_each_entry(cl, &q->clhash.hash[i], cl_common.hnode) - sch->qstats.backlog += cl->qdisc->qstats.backlog; - } qopt.defcls = q->defcls; if (nla_put(skb, TCA_OPTIONS, sizeof(qopt), &qopt)) @@ -1604,6 +1597,7 @@ hfsc_enqueue(struct sk_buff *skb, struct Qdisc *sch) if (cl->qdisc->q.qlen == 1) set_active(cl, qdisc_pkt_len(skb)); + qdisc_qstats_backlog_inc(sch, skb); sch->q.qlen++; return NET_XMIT_SUCCESS; @@ -1672,6 +1666,7 @@ hfsc_dequeue(struct Qdisc *sch) qdisc_unthrottled(sch); qdisc_bstats_update(sch, skb); + qdisc_qstats_backlog_dec(sch, skb); sch->q.qlen--; return skb; @@ -1695,6 +1690,7 @@ hfsc_drop(struct Qdisc *sch) } cl->qstats.drops++; qdisc_qstats_drop(sch); + sch->qstats.backlog -= len; sch->q.qlen--; return len; } diff --git a/net/sched/sch_htb.c b/net/sched/sch_htb.c index d4b4218..62f9d81 100644 --- a/net/sched/sch_htb.c +++ b/net/sched/sch_htb.c @@ -1007,7 +1007,9 @@ static void htb_work_func(struct work_struct *work) struct htb_sched *q = container_of(work, struct htb_sched, work); struct Qdisc *sch = q->watchdog.qdisc; + rcu_read_lock(); __netif_schedule(qdisc_root(sch)); + rcu_read_unlock(); } static int htb_init(struct Qdisc *sch, struct nlattr *opt) diff --git a/net/sched/sch_ingress.c b/net/sched/sch_ingress.c index 10adbc6..8fe6999 100644 --- a/net/sched/sch_ingress.c +++ b/net/sched/sch_ingress.c @@ -27,6 +27,11 @@ static unsigned long ingress_get(struct Qdisc *sch, u32 classid) return TC_H_MIN(classid) + 1; } +static bool ingress_cl_offload(u32 classid) +{ + return true; +} + static unsigned long ingress_bind_filter(struct Qdisc *sch, unsigned long parent, u32 classid) { @@ -86,6 +91,7 @@ static const struct Qdisc_class_ops ingress_class_ops = { .put = ingress_put, .walk = ingress_walk, .tcf_chain = ingress_find_tcf, + .tcf_cl_offload = ingress_cl_offload, .bind_tcf = ingress_bind_filter, .unbind_tcf = ingress_put, }; @@ -110,6 +116,11 @@ static unsigned long clsact_get(struct Qdisc *sch, u32 classid) } } +static bool clsact_cl_offload(u32 classid) +{ + return TC_H_MIN(classid) == TC_H_MIN(TC_H_MIN_INGRESS); +} + static unsigned long clsact_bind_filter(struct Qdisc *sch, unsigned long parent, u32 classid) { @@ -158,6 +169,7 @@ static const struct Qdisc_class_ops clsact_class_ops = { .put = ingress_put, .walk = ingress_walk, .tcf_chain = clsact_find_tcf, + .tcf_cl_offload = clsact_cl_offload, .bind_tcf = clsact_bind_filter, .unbind_tcf = ingress_put, }; diff --git a/net/sched/sch_netem.c b/net/sched/sch_netem.c index 205bed0..178f163 100644 --- a/net/sched/sch_netem.c +++ b/net/sched/sch_netem.c @@ -650,14 +650,14 @@ deliver: #endif if (q->qdisc) { + unsigned int pkt_len = qdisc_pkt_len(skb); int err = qdisc_enqueue(skb, q->qdisc); - if (unlikely(err != NET_XMIT_SUCCESS)) { - if (net_xmit_drop_count(err)) { - qdisc_qstats_drop(sch); - qdisc_tree_reduce_backlog(sch, 1, - qdisc_pkt_len(skb)); - } + if (err != NET_XMIT_SUCCESS && + net_xmit_drop_count(err)) { + qdisc_qstats_drop(sch); + qdisc_tree_reduce_backlog(sch, 1, + pkt_len); } goto tfifo_dequeue; } diff --git a/net/sched/sch_prio.c b/net/sched/sch_prio.c index fee1b15..a356450 100644 --- a/net/sched/sch_prio.c +++ b/net/sched/sch_prio.c @@ -85,6 +85,7 @@ prio_enqueue(struct sk_buff *skb, struct Qdisc *sch) ret = qdisc_enqueue(skb, qdisc); if (ret == NET_XMIT_SUCCESS) { + qdisc_qstats_backlog_inc(sch, skb); sch->q.qlen++; return NET_XMIT_SUCCESS; } @@ -117,6 +118,7 @@ static struct sk_buff *prio_dequeue(struct Qdisc *sch) struct sk_buff *skb = qdisc_dequeue_peeked(qdisc); if (skb) { qdisc_bstats_update(sch, skb); + qdisc_qstats_backlog_dec(sch, skb); sch->q.qlen--; return skb; } @@ -135,6 +137,7 @@ static unsigned int prio_drop(struct Qdisc *sch) for (prio = q->bands-1; prio >= 0; prio--) { qdisc = q->queues[prio]; if (qdisc->ops->drop && (len = qdisc->ops->drop(qdisc)) != 0) { + sch->qstats.backlog -= len; sch->q.qlen--; return len; } @@ -151,6 +154,7 @@ prio_reset(struct Qdisc *sch) for (prio = 0; prio < q->bands; prio++) qdisc_reset(q->queues[prio]); + sch->qstats.backlog = 0; sch->q.qlen = 0; } @@ -168,8 +172,9 @@ prio_destroy(struct Qdisc *sch) static int prio_tune(struct Qdisc *sch, struct nlattr *opt) { struct prio_sched_data *q = qdisc_priv(sch); + struct Qdisc *queues[TCQ_PRIO_BANDS]; + int oldbands = q->bands, i; struct tc_prio_qopt *qopt; - int i; if (nla_len(opt) < sizeof(*qopt)) return -EINVAL; @@ -183,62 +188,42 @@ static int prio_tune(struct Qdisc *sch, struct nlattr *opt) return -EINVAL; } + /* Before commit, make sure we can allocate all new qdiscs */ + for (i = oldbands; i < qopt->bands; i++) { + queues[i] = qdisc_create_dflt(sch->dev_queue, &pfifo_qdisc_ops, + TC_H_MAKE(sch->handle, i + 1)); + if (!queues[i]) { + while (i > oldbands) + qdisc_destroy(queues[--i]); + return -ENOMEM; + } + } + sch_tree_lock(sch); q->bands = qopt->bands; memcpy(q->prio2band, qopt->priomap, TC_PRIO_MAX+1); - for (i = q->bands; i < TCQ_PRIO_BANDS; i++) { + for (i = q->bands; i < oldbands; i++) { struct Qdisc *child = q->queues[i]; - q->queues[i] = &noop_qdisc; - if (child != &noop_qdisc) { - qdisc_tree_reduce_backlog(child, child->q.qlen, child->qstats.backlog); - qdisc_destroy(child); - } - } - sch_tree_unlock(sch); - for (i = 0; i < q->bands; i++) { - if (q->queues[i] == &noop_qdisc) { - struct Qdisc *child, *old; - - child = qdisc_create_dflt(sch->dev_queue, - &pfifo_qdisc_ops, - TC_H_MAKE(sch->handle, i + 1)); - if (child) { - sch_tree_lock(sch); - old = q->queues[i]; - q->queues[i] = child; - - if (old != &noop_qdisc) { - qdisc_tree_reduce_backlog(old, - old->q.qlen, - old->qstats.backlog); - qdisc_destroy(old); - } - sch_tree_unlock(sch); - } - } + qdisc_tree_reduce_backlog(child, child->q.qlen, + child->qstats.backlog); + qdisc_destroy(child); } + + for (i = oldbands; i < q->bands; i++) + q->queues[i] = queues[i]; + + sch_tree_unlock(sch); return 0; } static int prio_init(struct Qdisc *sch, struct nlattr *opt) { - struct prio_sched_data *q = qdisc_priv(sch); - int i; - - for (i = 0; i < TCQ_PRIO_BANDS; i++) - q->queues[i] = &noop_qdisc; - - if (opt == NULL) { + if (!opt) return -EINVAL; - } else { - int err; - if ((err = prio_tune(sch, opt)) != 0) - return err; - } - return 0; + return prio_tune(sch, opt); } static int prio_dump(struct Qdisc *sch, struct sk_buff *skb) diff --git a/net/sched/sch_qfq.c b/net/sched/sch_qfq.c index 8d2d8d9..f18857f 100644 --- a/net/sched/sch_qfq.c +++ b/net/sched/sch_qfq.c @@ -1235,8 +1235,10 @@ static int qfq_enqueue(struct sk_buff *skb, struct Qdisc *sch) cl->agg->lmax, qdisc_pkt_len(skb), cl->common.classid); err = qfq_change_agg(sch, cl, cl->agg->class_weight, qdisc_pkt_len(skb)); - if (err) - return err; + if (err) { + cl->qstats.drops++; + return qdisc_drop(skb, sch); + } } err = qdisc_enqueue(skb, cl->qdisc); diff --git a/net/sched/sch_red.c b/net/sched/sch_red.c index 8c0508c..91578bd 100644 --- a/net/sched/sch_red.c +++ b/net/sched/sch_red.c @@ -97,6 +97,7 @@ static int red_enqueue(struct sk_buff *skb, struct Qdisc *sch) ret = qdisc_enqueue(skb, child); if (likely(ret == NET_XMIT_SUCCESS)) { + qdisc_qstats_backlog_inc(sch, skb); sch->q.qlen++; } else if (net_xmit_drop_count(ret)) { q->stats.pdrop++; @@ -118,6 +119,7 @@ static struct sk_buff *red_dequeue(struct Qdisc *sch) skb = child->dequeue(child); if (skb) { qdisc_bstats_update(sch, skb); + qdisc_qstats_backlog_dec(sch, skb); sch->q.qlen--; } else { if (!red_is_idling(&q->vars)) @@ -143,6 +145,7 @@ static unsigned int red_drop(struct Qdisc *sch) if (child->ops->drop && (len = child->ops->drop(child)) > 0) { q->stats.other++; qdisc_qstats_drop(sch); + sch->qstats.backlog -= len; sch->q.qlen--; return len; } @@ -158,6 +161,7 @@ static void red_reset(struct Qdisc *sch) struct red_sched_data *q = qdisc_priv(sch); qdisc_reset(q->qdisc); + sch->qstats.backlog = 0; sch->q.qlen = 0; red_restart(&q->vars); } diff --git a/net/sched/sch_tbf.c b/net/sched/sch_tbf.c index 83b90b5..3161e49 100644 --- a/net/sched/sch_tbf.c +++ b/net/sched/sch_tbf.c @@ -207,6 +207,7 @@ static int tbf_enqueue(struct sk_buff *skb, struct Qdisc *sch) return ret; } + qdisc_qstats_backlog_inc(sch, skb); sch->q.qlen++; return NET_XMIT_SUCCESS; } @@ -217,6 +218,7 @@ static unsigned int tbf_drop(struct Qdisc *sch) unsigned int len = 0; if (q->qdisc->ops->drop && (len = q->qdisc->ops->drop(q->qdisc)) != 0) { + sch->qstats.backlog -= len; sch->q.qlen--; qdisc_qstats_drop(sch); } @@ -263,6 +265,7 @@ static struct sk_buff *tbf_dequeue(struct Qdisc *sch) q->t_c = now; q->tokens = toks; q->ptokens = ptoks; + qdisc_qstats_backlog_dec(sch, skb); sch->q.qlen--; qdisc_unthrottled(sch); qdisc_bstats_update(sch, skb); @@ -294,6 +297,7 @@ static void tbf_reset(struct Qdisc *sch) struct tbf_sched_data *q = qdisc_priv(sch); qdisc_reset(q->qdisc); + sch->qstats.backlog = 0; sch->q.qlen = 0; q->t_c = ktime_get_ns(); q->tokens = q->buffer; diff --git a/net/sctp/sctp_diag.c b/net/sctp/sctp_diag.c index 1ce724b..f69edcf 100644 --- a/net/sctp/sctp_diag.c +++ b/net/sctp/sctp_diag.c @@ -3,12 +3,6 @@ #include <linux/sock_diag.h> #include <net/sctp/sctp.h> -extern void inet_diag_msg_common_fill(struct inet_diag_msg *r, - struct sock *sk); -extern int inet_diag_msg_attrs_fill(struct sock *sk, struct sk_buff *skb, - struct inet_diag_msg *r, int ext, - struct user_namespace *user_ns); - static void sctp_diag_get_info(struct sock *sk, struct inet_diag_msg *r, void *info); diff --git a/net/sctp/socket.c b/net/sctp/socket.c index 777d032..67154b8 100644 --- a/net/sctp/socket.c +++ b/net/sctp/socket.c @@ -4220,6 +4220,7 @@ int sctp_get_sctp_info(struct sock *sk, struct sctp_association *asoc, info->sctpi_s_disable_fragments = sp->disable_fragments; info->sctpi_s_v4mapped = sp->v4mapped; info->sctpi_s_frag_interleave = sp->frag_interleave; + info->sctpi_s_type = sp->type; return 0; } diff --git a/net/sunrpc/auth.c b/net/sunrpc/auth.c index 02f5367..040ff62 100644 --- a/net/sunrpc/auth.c +++ b/net/sunrpc/auth.c @@ -543,7 +543,7 @@ rpcauth_cache_enforce_limit(void) */ struct rpc_cred * rpcauth_lookup_credcache(struct rpc_auth *auth, struct auth_cred * acred, - int flags) + int flags, gfp_t gfp) { LIST_HEAD(free); struct rpc_cred_cache *cache = auth->au_credcache; @@ -580,7 +580,7 @@ rpcauth_lookup_credcache(struct rpc_auth *auth, struct auth_cred * acred, if (flags & RPCAUTH_LOOKUP_RCU) return ERR_PTR(-ECHILD); - new = auth->au_ops->crcreate(auth, acred, flags); + new = auth->au_ops->crcreate(auth, acred, flags, gfp); if (IS_ERR(new)) { cred = new; goto out; @@ -703,8 +703,7 @@ rpcauth_bindcred(struct rpc_task *task, struct rpc_cred *cred, int flags) new = rpcauth_bind_new_cred(task, lookupflags); if (IS_ERR(new)) return PTR_ERR(new); - if (req->rq_cred != NULL) - put_rpccred(req->rq_cred); + put_rpccred(req->rq_cred); req->rq_cred = new; return 0; } @@ -712,6 +711,8 @@ rpcauth_bindcred(struct rpc_task *task, struct rpc_cred *cred, int flags) void put_rpccred(struct rpc_cred *cred) { + if (cred == NULL) + return; /* Fast path for unhashed credentials */ if (test_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags) == 0) { if (atomic_dec_and_test(&cred->cr_count)) diff --git a/net/sunrpc/auth_generic.c b/net/sunrpc/auth_generic.c index 41248b1..54dd3fd 100644 --- a/net/sunrpc/auth_generic.c +++ b/net/sunrpc/auth_generic.c @@ -38,6 +38,13 @@ struct rpc_cred *rpc_lookup_cred(void) } EXPORT_SYMBOL_GPL(rpc_lookup_cred); +struct rpc_cred * +rpc_lookup_generic_cred(struct auth_cred *acred, int flags, gfp_t gfp) +{ + return rpcauth_lookup_credcache(&generic_auth, acred, flags, gfp); +} +EXPORT_SYMBOL_GPL(rpc_lookup_generic_cred); + struct rpc_cred *rpc_lookup_cred_nonblock(void) { return rpcauth_lookupcred(&generic_auth, RPCAUTH_LOOKUP_RCU); @@ -77,15 +84,15 @@ static struct rpc_cred *generic_bind_cred(struct rpc_task *task, static struct rpc_cred * generic_lookup_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) { - return rpcauth_lookup_credcache(&generic_auth, acred, flags); + return rpcauth_lookup_credcache(&generic_auth, acred, flags, GFP_KERNEL); } static struct rpc_cred * -generic_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) +generic_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags, gfp_t gfp) { struct generic_cred *gcred; - gcred = kmalloc(sizeof(*gcred), GFP_KERNEL); + gcred = kmalloc(sizeof(*gcred), gfp); if (gcred == NULL) return ERR_PTR(-ENOMEM); diff --git a/net/sunrpc/auth_gss/auth_gss.c b/net/sunrpc/auth_gss/auth_gss.c index 15612ff..e64ae93 100644 --- a/net/sunrpc/auth_gss/auth_gss.c +++ b/net/sunrpc/auth_gss/auth_gss.c @@ -1299,11 +1299,11 @@ gss_destroy_cred(struct rpc_cred *cred) static struct rpc_cred * gss_lookup_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) { - return rpcauth_lookup_credcache(auth, acred, flags); + return rpcauth_lookup_credcache(auth, acred, flags, GFP_NOFS); } static struct rpc_cred * -gss_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) +gss_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags, gfp_t gfp) { struct gss_auth *gss_auth = container_of(auth, struct gss_auth, rpc_auth); struct gss_cred *cred = NULL; @@ -1313,7 +1313,7 @@ gss_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) __func__, from_kuid(&init_user_ns, acred->uid), auth->au_flavor); - if (!(cred = kzalloc(sizeof(*cred), GFP_NOFS))) + if (!(cred = kzalloc(sizeof(*cred), gfp))) goto out_err; rpcauth_init_cred(&cred->gc_base, acred, auth, &gss_credops); diff --git a/net/sunrpc/auth_gss/svcauth_gss.c b/net/sunrpc/auth_gss/svcauth_gss.c index 1095be9..e085f5a 100644 --- a/net/sunrpc/auth_gss/svcauth_gss.c +++ b/net/sunrpc/auth_gss/svcauth_gss.c @@ -569,10 +569,9 @@ gss_svc_searchbyctx(struct cache_detail *cd, struct xdr_netobj *handle) struct rsc *found; memset(&rsci, 0, sizeof(rsci)); - if (dup_to_netobj(&rsci.handle, handle->data, handle->len)) - return NULL; + rsci.handle.data = handle->data; + rsci.handle.len = handle->len; found = rsc_lookup(cd, &rsci); - rsc_free(&rsci); if (!found) return NULL; if (cache_check(cd, &found->h, NULL)) @@ -857,8 +856,8 @@ unwrap_integ_data(struct svc_rqst *rqstp, struct xdr_buf *buf, u32 seq, struct g goto out; if (svc_getnl(&buf->head[0]) != seq) goto out; - /* trim off the mic at the end before returning */ - xdr_buf_trim(buf, mic.len + 4); + /* trim off the mic and padding at the end before returning */ + xdr_buf_trim(buf, round_up_to_quad(mic.len) + 4); stat = 0; out: kfree(mic.data); diff --git a/net/sunrpc/auth_unix.c b/net/sunrpc/auth_unix.c index 0d3dd36..9f65452 100644 --- a/net/sunrpc/auth_unix.c +++ b/net/sunrpc/auth_unix.c @@ -52,11 +52,11 @@ unx_destroy(struct rpc_auth *auth) static struct rpc_cred * unx_lookup_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) { - return rpcauth_lookup_credcache(auth, acred, flags); + return rpcauth_lookup_credcache(auth, acred, flags, GFP_NOFS); } static struct rpc_cred * -unx_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) +unx_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags, gfp_t gfp) { struct unx_cred *cred; unsigned int groups = 0; @@ -66,7 +66,7 @@ unx_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags) from_kuid(&init_user_ns, acred->uid), from_kgid(&init_user_ns, acred->gid)); - if (!(cred = kmalloc(sizeof(*cred), GFP_NOFS))) + if (!(cred = kmalloc(sizeof(*cred), gfp))) return ERR_PTR(-ENOMEM); rpcauth_init_cred(&cred->uc_base, acred, auth, &unix_credops); diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c index 7e0c9bf..06b4df9 100644 --- a/net/sunrpc/clnt.c +++ b/net/sunrpc/clnt.c @@ -1414,6 +1414,23 @@ size_t rpc_max_payload(struct rpc_clnt *clnt) EXPORT_SYMBOL_GPL(rpc_max_payload); /** + * rpc_max_bc_payload - Get maximum backchannel payload size, in bytes + * @clnt: RPC client to query + */ +size_t rpc_max_bc_payload(struct rpc_clnt *clnt) +{ + struct rpc_xprt *xprt; + size_t ret; + + rcu_read_lock(); + xprt = rcu_dereference(clnt->cl_xprt); + ret = xprt->ops->bc_maxpayload(xprt); + rcu_read_unlock(); + return ret; +} +EXPORT_SYMBOL_GPL(rpc_max_bc_payload); + +/** * rpc_get_timeout - Get timeout for transport in units of HZ * @clnt: RPC client to query */ diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c index 7422f28..f5572e3 100644 --- a/net/sunrpc/svc_xprt.c +++ b/net/sunrpc/svc_xprt.c @@ -244,13 +244,12 @@ void svc_add_new_perm_xprt(struct svc_serv *serv, struct svc_xprt *new) svc_xprt_received(new); } -int svc_create_xprt(struct svc_serv *serv, const char *xprt_name, +int _svc_create_xprt(struct svc_serv *serv, const char *xprt_name, struct net *net, const int family, const unsigned short port, int flags) { struct svc_xprt_class *xcl; - dprintk("svc: creating transport %s[%d]\n", xprt_name, port); spin_lock(&svc_xprt_class_lock); list_for_each_entry(xcl, &svc_xprt_class_list, xcl_list) { struct svc_xprt *newxprt; @@ -274,12 +273,28 @@ int svc_create_xprt(struct svc_serv *serv, const char *xprt_name, } err: spin_unlock(&svc_xprt_class_lock); - dprintk("svc: transport %s not found\n", xprt_name); - /* This errno is exposed to user space. Provide a reasonable * perror msg for a bad transport. */ return -EPROTONOSUPPORT; } + +int svc_create_xprt(struct svc_serv *serv, const char *xprt_name, + struct net *net, const int family, + const unsigned short port, int flags) +{ + int err; + + dprintk("svc: creating transport %s[%d]\n", xprt_name, port); + err = _svc_create_xprt(serv, xprt_name, net, family, port, flags); + if (err == -EPROTONOSUPPORT) { + request_module("svc%s", xprt_name); + err = _svc_create_xprt(serv, xprt_name, net, family, port, flags); + } + if (err) + dprintk("svc: transport %s not found, err %d\n", + xprt_name, err); + return err; +} EXPORT_SYMBOL_GPL(svc_create_xprt); /* diff --git a/net/sunrpc/xdr.c b/net/sunrpc/xdr.c index 6bdb386..c4f3cc0 100644 --- a/net/sunrpc/xdr.c +++ b/net/sunrpc/xdr.c @@ -797,6 +797,8 @@ void xdr_init_decode(struct xdr_stream *xdr, struct xdr_buf *buf, __be32 *p) xdr_set_iov(xdr, buf->head, buf->len); else if (buf->page_len != 0) xdr_set_page_base(xdr, 0, buf->len); + else + xdr_set_iov(xdr, buf->head, buf->len); if (p != NULL && p > xdr->p && xdr->end >= p) { xdr->nwords -= p - xdr->p; xdr->p = p; diff --git a/net/sunrpc/xprtrdma/backchannel.c b/net/sunrpc/xprtrdma/backchannel.c index 2dcd764..87762d9 100644 --- a/net/sunrpc/xprtrdma/backchannel.c +++ b/net/sunrpc/xprtrdma/backchannel.c @@ -192,6 +192,22 @@ int xprt_rdma_bc_up(struct svc_serv *serv, struct net *net) } /** + * xprt_rdma_bc_maxpayload - Return maximum backchannel message size + * @xprt: transport + * + * Returns maximum size, in bytes, of a backchannel message + */ +size_t xprt_rdma_bc_maxpayload(struct rpc_xprt *xprt) +{ + struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); + struct rpcrdma_create_data_internal *cdata = &r_xprt->rx_data; + size_t maxmsg; + + maxmsg = min_t(unsigned int, cdata->inline_rsize, cdata->inline_wsize); + return maxmsg - RPCRDMA_HDRLEN_MIN; +} + +/** * rpcrdma_bc_marshal_reply - Send backwards direction reply * @rqst: buffer containing RPC reply data * diff --git a/net/sunrpc/xprtrdma/fmr_ops.c b/net/sunrpc/xprtrdma/fmr_ops.c index b289e10..6326ebe 100644 --- a/net/sunrpc/xprtrdma/fmr_ops.c +++ b/net/sunrpc/xprtrdma/fmr_ops.c @@ -35,10 +35,71 @@ /* Maximum scatter/gather per FMR */ #define RPCRDMA_MAX_FMR_SGES (64) +static struct workqueue_struct *fmr_recovery_wq; + +#define FMR_RECOVERY_WQ_FLAGS (WQ_UNBOUND) + +int +fmr_alloc_recovery_wq(void) +{ + fmr_recovery_wq = alloc_workqueue("fmr_recovery", WQ_UNBOUND, 0); + return !fmr_recovery_wq ? -ENOMEM : 0; +} + +void +fmr_destroy_recovery_wq(void) +{ + struct workqueue_struct *wq; + + if (!fmr_recovery_wq) + return; + + wq = fmr_recovery_wq; + fmr_recovery_wq = NULL; + destroy_workqueue(wq); +} + +static int +__fmr_unmap(struct rpcrdma_mw *mw) +{ + LIST_HEAD(l); + + list_add(&mw->fmr.fmr->list, &l); + return ib_unmap_fmr(&l); +} + +/* Deferred reset of a single FMR. Generate a fresh rkey by + * replacing the MR. There's no recovery if this fails. + */ +static void +__fmr_recovery_worker(struct work_struct *work) +{ + struct rpcrdma_mw *mw = container_of(work, struct rpcrdma_mw, + mw_work); + struct rpcrdma_xprt *r_xprt = mw->mw_xprt; + + __fmr_unmap(mw); + rpcrdma_put_mw(r_xprt, mw); + return; +} + +/* A broken MR was discovered in a context that can't sleep. + * Defer recovery to the recovery worker. + */ +static void +__fmr_queue_recovery(struct rpcrdma_mw *mw) +{ + INIT_WORK(&mw->mw_work, __fmr_recovery_worker); + queue_work(fmr_recovery_wq, &mw->mw_work); +} + static int fmr_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, struct rpcrdma_create_data_internal *cdata) { + rpcrdma_set_max_header_sizes(ia, cdata, max_t(unsigned int, 1, + RPCRDMA_MAX_DATA_SEGS / + RPCRDMA_MAX_FMR_SGES)); return 0; } @@ -48,7 +109,7 @@ static size_t fmr_op_maxpages(struct rpcrdma_xprt *r_xprt) { return min_t(unsigned int, RPCRDMA_MAX_DATA_SEGS, - rpcrdma_max_segments(r_xprt) * RPCRDMA_MAX_FMR_SGES); + RPCRDMA_MAX_HDR_SEGS * RPCRDMA_MAX_FMR_SGES); } static int @@ -89,6 +150,7 @@ fmr_op_init(struct rpcrdma_xprt *r_xprt) if (IS_ERR(r->fmr.fmr)) goto out_fmr_err; + r->mw_xprt = r_xprt; list_add(&r->mw_list, &buf->rb_mws); list_add(&r->mw_all, &buf->rb_all); } @@ -104,15 +166,6 @@ out: return rc; } -static int -__fmr_unmap(struct rpcrdma_mw *r) -{ - LIST_HEAD(l); - - list_add(&r->fmr.fmr->list, &l); - return ib_unmap_fmr(&l); -} - /* Use the ib_map_phys_fmr() verb to register a memory region * for remote access via RDMA READ or RDMA WRITE. */ @@ -183,15 +236,10 @@ static void __fmr_dma_unmap(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg) { struct ib_device *device = r_xprt->rx_ia.ri_device; - struct rpcrdma_mw *mw = seg->rl_mw; int nsegs = seg->mr_nsegs; - seg->rl_mw = NULL; - while (nsegs--) rpcrdma_unmap_one(device, seg++); - - rpcrdma_put_mw(r_xprt, mw); } /* Invalidate all memory regions that were registered for "req". @@ -234,42 +282,50 @@ fmr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) seg = &req->rl_segments[i]; __fmr_dma_unmap(r_xprt, seg); + rpcrdma_put_mw(r_xprt, seg->rl_mw); i += seg->mr_nsegs; seg->mr_nsegs = 0; + seg->rl_mw = NULL; } req->rl_nchunks = 0; } -/* Use the ib_unmap_fmr() verb to prevent further remote - * access via RDMA READ or RDMA WRITE. +/* Use a slow, safe mechanism to invalidate all memory regions + * that were registered for "req". + * + * In the asynchronous case, DMA unmapping occurs first here + * because the rpcrdma_mr_seg is released immediately after this + * call. It's contents won't be available in __fmr_dma_unmap later. + * FIXME. */ -static int -fmr_op_unmap(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg) +static void +fmr_op_unmap_safe(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req, + bool sync) { - struct rpcrdma_ia *ia = &r_xprt->rx_ia; - struct rpcrdma_mr_seg *seg1 = seg; - struct rpcrdma_mw *mw = seg1->rl_mw; - int rc, nsegs = seg->mr_nsegs; + struct rpcrdma_mr_seg *seg; + struct rpcrdma_mw *mw; + unsigned int i; - dprintk("RPC: %s: FMR %p\n", __func__, mw); + for (i = 0; req->rl_nchunks; req->rl_nchunks--) { + seg = &req->rl_segments[i]; + mw = seg->rl_mw; - seg1->rl_mw = NULL; - while (seg1->mr_nsegs--) - rpcrdma_unmap_one(ia->ri_device, seg++); - rc = __fmr_unmap(mw); - if (rc) - goto out_err; - rpcrdma_put_mw(r_xprt, mw); - return nsegs; + if (sync) { + /* ORDER */ + __fmr_unmap(mw); + __fmr_dma_unmap(r_xprt, seg); + rpcrdma_put_mw(r_xprt, mw); + } else { + __fmr_dma_unmap(r_xprt, seg); + __fmr_queue_recovery(mw); + } -out_err: - /* The FMR is abandoned, but remains in rb_all. fmr_op_destroy - * will attempt to release it when the transport is destroyed. - */ - dprintk("RPC: %s: ib_unmap_fmr status %i\n", __func__, rc); - return nsegs; + i += seg->mr_nsegs; + seg->mr_nsegs = 0; + seg->rl_mw = NULL; + } } static void @@ -295,7 +351,7 @@ fmr_op_destroy(struct rpcrdma_buffer *buf) const struct rpcrdma_memreg_ops rpcrdma_fmr_memreg_ops = { .ro_map = fmr_op_map, .ro_unmap_sync = fmr_op_unmap_sync, - .ro_unmap = fmr_op_unmap, + .ro_unmap_safe = fmr_op_unmap_safe, .ro_open = fmr_op_open, .ro_maxpages = fmr_op_maxpages, .ro_init = fmr_op_init, diff --git a/net/sunrpc/xprtrdma/frwr_ops.c b/net/sunrpc/xprtrdma/frwr_ops.c index 94c3fa9..c094754 100644 --- a/net/sunrpc/xprtrdma/frwr_ops.c +++ b/net/sunrpc/xprtrdma/frwr_ops.c @@ -98,6 +98,47 @@ frwr_destroy_recovery_wq(void) destroy_workqueue(wq); } +static int +__frwr_reset_mr(struct rpcrdma_ia *ia, struct rpcrdma_mw *r) +{ + struct rpcrdma_frmr *f = &r->frmr; + int rc; + + rc = ib_dereg_mr(f->fr_mr); + if (rc) { + pr_warn("rpcrdma: ib_dereg_mr status %d, frwr %p orphaned\n", + rc, r); + return rc; + } + + f->fr_mr = ib_alloc_mr(ia->ri_pd, IB_MR_TYPE_MEM_REG, + ia->ri_max_frmr_depth); + if (IS_ERR(f->fr_mr)) { + pr_warn("rpcrdma: ib_alloc_mr status %ld, frwr %p orphaned\n", + PTR_ERR(f->fr_mr), r); + return PTR_ERR(f->fr_mr); + } + + dprintk("RPC: %s: recovered FRMR %p\n", __func__, r); + f->fr_state = FRMR_IS_INVALID; + return 0; +} + +static void +__frwr_reset_and_unmap(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mw *mw) +{ + struct rpcrdma_ia *ia = &r_xprt->rx_ia; + struct rpcrdma_frmr *f = &mw->frmr; + int rc; + + rc = __frwr_reset_mr(ia, mw); + ib_dma_unmap_sg(ia->ri_device, f->fr_sg, f->fr_nents, f->fr_dir); + if (rc) + return; + + rpcrdma_put_mw(r_xprt, mw); +} + /* Deferred reset of a single FRMR. Generate a fresh rkey by * replacing the MR. * @@ -109,26 +150,10 @@ static void __frwr_recovery_worker(struct work_struct *work) { struct rpcrdma_mw *r = container_of(work, struct rpcrdma_mw, - frmr.fr_work); - struct rpcrdma_xprt *r_xprt = r->frmr.fr_xprt; - unsigned int depth = r_xprt->rx_ia.ri_max_frmr_depth; - struct ib_pd *pd = r_xprt->rx_ia.ri_pd; - - if (ib_dereg_mr(r->frmr.fr_mr)) - goto out_fail; + mw_work); - r->frmr.fr_mr = ib_alloc_mr(pd, IB_MR_TYPE_MEM_REG, depth); - if (IS_ERR(r->frmr.fr_mr)) - goto out_fail; - - dprintk("RPC: %s: recovered FRMR %p\n", __func__, r); - r->frmr.fr_state = FRMR_IS_INVALID; - rpcrdma_put_mw(r_xprt, r); + __frwr_reset_and_unmap(r->mw_xprt, r); return; - -out_fail: - pr_warn("RPC: %s: FRMR %p unrecovered\n", - __func__, r); } /* A broken MR was discovered in a context that can't sleep. @@ -137,8 +162,8 @@ out_fail: static void __frwr_queue_recovery(struct rpcrdma_mw *r) { - INIT_WORK(&r->frmr.fr_work, __frwr_recovery_worker); - queue_work(frwr_recovery_wq, &r->frmr.fr_work); + INIT_WORK(&r->mw_work, __frwr_recovery_worker); + queue_work(frwr_recovery_wq, &r->mw_work); } static int @@ -152,11 +177,11 @@ __frwr_init(struct rpcrdma_mw *r, struct ib_pd *pd, struct ib_device *device, if (IS_ERR(f->fr_mr)) goto out_mr_err; - f->sg = kcalloc(depth, sizeof(*f->sg), GFP_KERNEL); - if (!f->sg) + f->fr_sg = kcalloc(depth, sizeof(*f->fr_sg), GFP_KERNEL); + if (!f->fr_sg) goto out_list_err; - sg_init_table(f->sg, depth); + sg_init_table(f->fr_sg, depth); init_completion(&f->fr_linv_done); @@ -185,7 +210,7 @@ __frwr_release(struct rpcrdma_mw *r) if (rc) dprintk("RPC: %s: ib_dereg_mr status %i\n", __func__, rc); - kfree(r->frmr.sg); + kfree(r->frmr.fr_sg); } static int @@ -231,6 +256,9 @@ frwr_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, depth; } + rpcrdma_set_max_header_sizes(ia, cdata, max_t(unsigned int, 1, + RPCRDMA_MAX_DATA_SEGS / + ia->ri_max_frmr_depth)); return 0; } @@ -243,7 +271,7 @@ frwr_op_maxpages(struct rpcrdma_xprt *r_xprt) struct rpcrdma_ia *ia = &r_xprt->rx_ia; return min_t(unsigned int, RPCRDMA_MAX_DATA_SEGS, - rpcrdma_max_segments(r_xprt) * ia->ri_max_frmr_depth); + RPCRDMA_MAX_HDR_SEGS * ia->ri_max_frmr_depth); } static void @@ -350,9 +378,9 @@ frwr_op_init(struct rpcrdma_xprt *r_xprt) return rc; } + r->mw_xprt = r_xprt; list_add(&r->mw_list, &buf->rb_mws); list_add(&r->mw_all, &buf->rb_all); - r->frmr.fr_xprt = r_xprt; } return 0; @@ -396,12 +424,12 @@ frwr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg, for (i = 0; i < nsegs;) { if (seg->mr_page) - sg_set_page(&frmr->sg[i], + sg_set_page(&frmr->fr_sg[i], seg->mr_page, seg->mr_len, offset_in_page(seg->mr_offset)); else - sg_set_buf(&frmr->sg[i], seg->mr_offset, + sg_set_buf(&frmr->fr_sg[i], seg->mr_offset, seg->mr_len); ++seg; @@ -412,25 +440,26 @@ frwr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg, offset_in_page((seg-1)->mr_offset + (seg-1)->mr_len)) break; } - frmr->sg_nents = i; + frmr->fr_nents = i; + frmr->fr_dir = direction; - dma_nents = ib_dma_map_sg(device, frmr->sg, frmr->sg_nents, direction); + dma_nents = ib_dma_map_sg(device, frmr->fr_sg, frmr->fr_nents, direction); if (!dma_nents) { pr_err("RPC: %s: failed to dma map sg %p sg_nents %u\n", - __func__, frmr->sg, frmr->sg_nents); + __func__, frmr->fr_sg, frmr->fr_nents); return -ENOMEM; } - n = ib_map_mr_sg(mr, frmr->sg, frmr->sg_nents, NULL, PAGE_SIZE); - if (unlikely(n != frmr->sg_nents)) { + n = ib_map_mr_sg(mr, frmr->fr_sg, frmr->fr_nents, NULL, PAGE_SIZE); + if (unlikely(n != frmr->fr_nents)) { pr_err("RPC: %s: failed to map mr %p (%u/%u)\n", - __func__, frmr->fr_mr, n, frmr->sg_nents); + __func__, frmr->fr_mr, n, frmr->fr_nents); rc = n < 0 ? n : -EINVAL; goto out_senderr; } dprintk("RPC: %s: Using frmr %p to map %u segments (%u bytes)\n", - __func__, mw, frmr->sg_nents, mr->length); + __func__, mw, frmr->fr_nents, mr->length); key = (u8)(mr->rkey & 0x000000FF); ib_update_fast_reg_key(mr, ++key); @@ -452,18 +481,16 @@ frwr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg, if (rc) goto out_senderr; - seg1->mr_dir = direction; seg1->rl_mw = mw; seg1->mr_rkey = mr->rkey; seg1->mr_base = mr->iova; - seg1->mr_nsegs = frmr->sg_nents; + seg1->mr_nsegs = frmr->fr_nents; seg1->mr_len = mr->length; - return frmr->sg_nents; + return frmr->fr_nents; out_senderr: dprintk("RPC: %s: ib_post_send status %i\n", __func__, rc); - ib_dma_unmap_sg(device, frmr->sg, dma_nents, direction); __frwr_queue_recovery(mw); return rc; } @@ -487,24 +514,6 @@ __frwr_prepare_linv_wr(struct rpcrdma_mr_seg *seg) return invalidate_wr; } -static void -__frwr_dma_unmap(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg, - int rc) -{ - struct ib_device *device = r_xprt->rx_ia.ri_device; - struct rpcrdma_mw *mw = seg->rl_mw; - struct rpcrdma_frmr *f = &mw->frmr; - - seg->rl_mw = NULL; - - ib_dma_unmap_sg(device, f->sg, f->sg_nents, seg->mr_dir); - - if (!rc) - rpcrdma_put_mw(r_xprt, mw); - else - __frwr_queue_recovery(mw); -} - /* Invalidate all memory regions that were registered for "req". * * Sleeps until it is safe for the host CPU to access the @@ -518,6 +527,7 @@ frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) struct rpcrdma_mr_seg *seg; unsigned int i, nchunks; struct rpcrdma_frmr *f; + struct rpcrdma_mw *mw; int rc; dprintk("RPC: %s: req %p\n", __func__, req); @@ -558,11 +568,8 @@ frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) * unless ri_id->qp is a valid pointer. */ rc = ib_post_send(ia->ri_id->qp, invalidate_wrs, &bad_wr); - if (rc) { - pr_warn("%s: ib_post_send failed %i\n", __func__, rc); - rdma_disconnect(ia->ri_id); - goto unmap; - } + if (rc) + goto reset_mrs; wait_for_completion(&f->fr_linv_done); @@ -572,56 +579,65 @@ frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) unmap: for (i = 0, nchunks = req->rl_nchunks; nchunks; nchunks--) { seg = &req->rl_segments[i]; + mw = seg->rl_mw; + seg->rl_mw = NULL; - __frwr_dma_unmap(r_xprt, seg, rc); + ib_dma_unmap_sg(ia->ri_device, f->fr_sg, f->fr_nents, + f->fr_dir); + rpcrdma_put_mw(r_xprt, mw); i += seg->mr_nsegs; seg->mr_nsegs = 0; } req->rl_nchunks = 0; -} + return; -/* Post a LOCAL_INV Work Request to prevent further remote access - * via RDMA READ or RDMA WRITE. - */ -static int -frwr_op_unmap(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg) -{ - struct rpcrdma_mr_seg *seg1 = seg; - struct rpcrdma_ia *ia = &r_xprt->rx_ia; - struct rpcrdma_mw *mw = seg1->rl_mw; - struct rpcrdma_frmr *frmr = &mw->frmr; - struct ib_send_wr *invalidate_wr, *bad_wr; - int rc, nsegs = seg->mr_nsegs; +reset_mrs: + pr_warn("%s: ib_post_send failed %i\n", __func__, rc); - dprintk("RPC: %s: FRMR %p\n", __func__, mw); + /* Find and reset the MRs in the LOCAL_INV WRs that did not + * get posted. This is synchronous, and slow. + */ + for (i = 0, nchunks = req->rl_nchunks; nchunks; nchunks--) { + seg = &req->rl_segments[i]; + mw = seg->rl_mw; + f = &mw->frmr; - seg1->rl_mw = NULL; - frmr->fr_state = FRMR_IS_INVALID; - invalidate_wr = &mw->frmr.fr_invwr; + if (mw->frmr.fr_mr->rkey == bad_wr->ex.invalidate_rkey) { + __frwr_reset_mr(ia, mw); + bad_wr = bad_wr->next; + } - memset(invalidate_wr, 0, sizeof(*invalidate_wr)); - frmr->fr_cqe.done = frwr_wc_localinv; - invalidate_wr->wr_cqe = &frmr->fr_cqe; - invalidate_wr->opcode = IB_WR_LOCAL_INV; - invalidate_wr->ex.invalidate_rkey = frmr->fr_mr->rkey; - DECR_CQCOUNT(&r_xprt->rx_ep); + i += seg->mr_nsegs; + } + goto unmap; +} - ib_dma_unmap_sg(ia->ri_device, frmr->sg, frmr->sg_nents, seg1->mr_dir); - read_lock(&ia->ri_qplock); - rc = ib_post_send(ia->ri_id->qp, invalidate_wr, &bad_wr); - read_unlock(&ia->ri_qplock); - if (rc) - goto out_err; +/* Use a slow, safe mechanism to invalidate all memory regions + * that were registered for "req". + */ +static void +frwr_op_unmap_safe(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req, + bool sync) +{ + struct rpcrdma_mr_seg *seg; + struct rpcrdma_mw *mw; + unsigned int i; - rpcrdma_put_mw(r_xprt, mw); - return nsegs; + for (i = 0; req->rl_nchunks; req->rl_nchunks--) { + seg = &req->rl_segments[i]; + mw = seg->rl_mw; -out_err: - dprintk("RPC: %s: ib_post_send status %i\n", __func__, rc); - __frwr_queue_recovery(mw); - return nsegs; + if (sync) + __frwr_reset_and_unmap(r_xprt, mw); + else + __frwr_queue_recovery(mw); + + i += seg->mr_nsegs; + seg->mr_nsegs = 0; + seg->rl_mw = NULL; + } } static void @@ -643,7 +659,7 @@ frwr_op_destroy(struct rpcrdma_buffer *buf) const struct rpcrdma_memreg_ops rpcrdma_frwr_memreg_ops = { .ro_map = frwr_op_map, .ro_unmap_sync = frwr_op_unmap_sync, - .ro_unmap = frwr_op_unmap, + .ro_unmap_safe = frwr_op_unmap_safe, .ro_open = frwr_op_open, .ro_maxpages = frwr_op_maxpages, .ro_init = frwr_op_init, diff --git a/net/sunrpc/xprtrdma/physical_ops.c b/net/sunrpc/xprtrdma/physical_ops.c index 481b9b6..3750596 100644 --- a/net/sunrpc/xprtrdma/physical_ops.c +++ b/net/sunrpc/xprtrdma/physical_ops.c @@ -36,8 +36,11 @@ physical_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep, __func__, PTR_ERR(mr)); return -ENOMEM; } - ia->ri_dma_mr = mr; + + rpcrdma_set_max_header_sizes(ia, cdata, min_t(unsigned int, + RPCRDMA_MAX_DATA_SEGS, + RPCRDMA_MAX_HDR_SEGS)); return 0; } @@ -47,7 +50,7 @@ static size_t physical_op_maxpages(struct rpcrdma_xprt *r_xprt) { return min_t(unsigned int, RPCRDMA_MAX_DATA_SEGS, - rpcrdma_max_segments(r_xprt)); + RPCRDMA_MAX_HDR_SEGS); } static int @@ -71,17 +74,6 @@ physical_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg, return 1; } -/* Unmap a memory region, but leave it registered. - */ -static int -physical_op_unmap(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg) -{ - struct rpcrdma_ia *ia = &r_xprt->rx_ia; - - rpcrdma_unmap_one(ia->ri_device, seg); - return 1; -} - /* DMA unmap all memory regions that were mapped for "req". */ static void @@ -94,6 +86,25 @@ physical_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) rpcrdma_unmap_one(device, &req->rl_segments[i++]); } +/* Use a slow, safe mechanism to invalidate all memory regions + * that were registered for "req". + * + * For physical memory registration, there is no good way to + * fence a single MR that has been advertised to the server. The + * client has already handed the server an R_key that cannot be + * invalidated and is shared by all MRs on this connection. + * Tearing down the PD might be the only safe choice, but it's + * not clear that a freshly acquired DMA R_key would be different + * than the one used by the PD that was just destroyed. + * FIXME. + */ +static void +physical_op_unmap_safe(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req, + bool sync) +{ + physical_op_unmap_sync(r_xprt, req); +} + static void physical_op_destroy(struct rpcrdma_buffer *buf) { @@ -102,7 +113,7 @@ physical_op_destroy(struct rpcrdma_buffer *buf) const struct rpcrdma_memreg_ops rpcrdma_physical_memreg_ops = { .ro_map = physical_op_map, .ro_unmap_sync = physical_op_unmap_sync, - .ro_unmap = physical_op_unmap, + .ro_unmap_safe = physical_op_unmap_safe, .ro_open = physical_op_open, .ro_maxpages = physical_op_maxpages, .ro_init = physical_op_init, diff --git a/net/sunrpc/xprtrdma/rpc_rdma.c b/net/sunrpc/xprtrdma/rpc_rdma.c index 888823b..35a8109 100644 --- a/net/sunrpc/xprtrdma/rpc_rdma.c +++ b/net/sunrpc/xprtrdma/rpc_rdma.c @@ -61,26 +61,84 @@ enum rpcrdma_chunktype { rpcrdma_replych }; -#if IS_ENABLED(CONFIG_SUNRPC_DEBUG) static const char transfertypes[][12] = { - "pure inline", /* no chunks */ - " read chunk", /* some argument via rdma read */ - "*read chunk", /* entire request via rdma read */ - "write chunk", /* some result via rdma write */ + "inline", /* no chunks */ + "read list", /* some argument via rdma read */ + "*read list", /* entire request via rdma read */ + "write list", /* some result via rdma write */ "reply chunk" /* entire reply via rdma write */ }; -#endif + +/* Returns size of largest RPC-over-RDMA header in a Call message + * + * The largest Call header contains a full-size Read list and a + * minimal Reply chunk. + */ +static unsigned int rpcrdma_max_call_header_size(unsigned int maxsegs) +{ + unsigned int size; + + /* Fixed header fields and list discriminators */ + size = RPCRDMA_HDRLEN_MIN; + + /* Maximum Read list size */ + maxsegs += 2; /* segment for head and tail buffers */ + size = maxsegs * sizeof(struct rpcrdma_read_chunk); + + /* Minimal Read chunk size */ + size += sizeof(__be32); /* segment count */ + size += sizeof(struct rpcrdma_segment); + size += sizeof(__be32); /* list discriminator */ + + dprintk("RPC: %s: max call header size = %u\n", + __func__, size); + return size; +} + +/* Returns size of largest RPC-over-RDMA header in a Reply message + * + * There is only one Write list or one Reply chunk per Reply + * message. The larger list is the Write list. + */ +static unsigned int rpcrdma_max_reply_header_size(unsigned int maxsegs) +{ + unsigned int size; + + /* Fixed header fields and list discriminators */ + size = RPCRDMA_HDRLEN_MIN; + + /* Maximum Write list size */ + maxsegs += 2; /* segment for head and tail buffers */ + size = sizeof(__be32); /* segment count */ + size += maxsegs * sizeof(struct rpcrdma_segment); + size += sizeof(__be32); /* list discriminator */ + + dprintk("RPC: %s: max reply header size = %u\n", + __func__, size); + return size; +} + +void rpcrdma_set_max_header_sizes(struct rpcrdma_ia *ia, + struct rpcrdma_create_data_internal *cdata, + unsigned int maxsegs) +{ + ia->ri_max_inline_write = cdata->inline_wsize - + rpcrdma_max_call_header_size(maxsegs); + ia->ri_max_inline_read = cdata->inline_rsize - + rpcrdma_max_reply_header_size(maxsegs); +} /* The client can send a request inline as long as the RPCRDMA header * plus the RPC call fit under the transport's inline limit. If the * combined call message size exceeds that limit, the client must use * the read chunk list for this operation. */ -static bool rpcrdma_args_inline(struct rpc_rqst *rqst) +static bool rpcrdma_args_inline(struct rpcrdma_xprt *r_xprt, + struct rpc_rqst *rqst) { - unsigned int callsize = RPCRDMA_HDRLEN_MIN + rqst->rq_snd_buf.len; + struct rpcrdma_ia *ia = &r_xprt->rx_ia; - return callsize <= RPCRDMA_INLINE_WRITE_THRESHOLD(rqst); + return rqst->rq_snd_buf.len <= ia->ri_max_inline_write; } /* The client can't know how large the actual reply will be. Thus it @@ -89,11 +147,12 @@ static bool rpcrdma_args_inline(struct rpc_rqst *rqst) * limit, the client must provide a write list or a reply chunk for * this request. */ -static bool rpcrdma_results_inline(struct rpc_rqst *rqst) +static bool rpcrdma_results_inline(struct rpcrdma_xprt *r_xprt, + struct rpc_rqst *rqst) { - unsigned int repsize = RPCRDMA_HDRLEN_MIN + rqst->rq_rcv_buf.buflen; + struct rpcrdma_ia *ia = &r_xprt->rx_ia; - return repsize <= RPCRDMA_INLINE_READ_THRESHOLD(rqst); + return rqst->rq_rcv_buf.buflen <= ia->ri_max_inline_read; } static int @@ -226,23 +285,16 @@ rpcrdma_convert_iovs(struct xdr_buf *xdrbuf, unsigned int pos, return n; } -/* - * Create read/write chunk lists, and reply chunks, for RDMA - * - * Assume check against THRESHOLD has been done, and chunks are required. - * Assume only encoding one list entry for read|write chunks. The NFSv3 - * protocol is simple enough to allow this as it only has a single "bulk - * result" in each procedure - complicated NFSv4 COMPOUNDs are not. (The - * RDMA/Sessions NFSv4 proposal addresses this for future v4 revs.) - * - * When used for a single reply chunk (which is a special write - * chunk used for the entire reply, rather than just the data), it - * is used primarily for READDIR and READLINK which would otherwise - * be severely size-limited by a small rdma inline read max. The server - * response will come back as an RDMA Write, followed by a message - * of type RDMA_NOMSG carrying the xid and length. As a result, reply - * chunks do not provide data alignment, however they do not require - * "fixup" (moving the response to the upper layer buffer) either. +static inline __be32 * +xdr_encode_rdma_segment(__be32 *iptr, struct rpcrdma_mr_seg *seg) +{ + *iptr++ = cpu_to_be32(seg->mr_rkey); + *iptr++ = cpu_to_be32(seg->mr_len); + return xdr_encode_hyper(iptr, seg->mr_base); +} + +/* XDR-encode the Read list. Supports encoding a list of read + * segments that belong to a single read chunk. * * Encoding key for single-list chunks (HLOO = Handle32 Length32 Offset64): * @@ -250,131 +302,190 @@ rpcrdma_convert_iovs(struct xdr_buf *xdrbuf, unsigned int pos, * N elements, position P (same P for all chunks of same arg!): * 1 - PHLOO - 1 - PHLOO - ... - 1 - PHLOO - 0 * + * Returns a pointer to the XDR word in the RDMA header following + * the end of the Read list, or an error pointer. + */ +static __be32 * +rpcrdma_encode_read_list(struct rpcrdma_xprt *r_xprt, + struct rpcrdma_req *req, struct rpc_rqst *rqst, + __be32 *iptr, enum rpcrdma_chunktype rtype) +{ + struct rpcrdma_mr_seg *seg = req->rl_nextseg; + unsigned int pos; + int n, nsegs; + + if (rtype == rpcrdma_noch) { + *iptr++ = xdr_zero; /* item not present */ + return iptr; + } + + pos = rqst->rq_snd_buf.head[0].iov_len; + if (rtype == rpcrdma_areadch) + pos = 0; + nsegs = rpcrdma_convert_iovs(&rqst->rq_snd_buf, pos, rtype, seg, + RPCRDMA_MAX_SEGS - req->rl_nchunks); + if (nsegs < 0) + return ERR_PTR(nsegs); + + do { + n = r_xprt->rx_ia.ri_ops->ro_map(r_xprt, seg, nsegs, false); + if (n <= 0) + return ERR_PTR(n); + + *iptr++ = xdr_one; /* item present */ + + /* All read segments in this chunk + * have the same "position". + */ + *iptr++ = cpu_to_be32(pos); + iptr = xdr_encode_rdma_segment(iptr, seg); + + dprintk("RPC: %5u %s: read segment pos %u " + "%d@0x%016llx:0x%08x (%s)\n", + rqst->rq_task->tk_pid, __func__, pos, + seg->mr_len, (unsigned long long)seg->mr_base, + seg->mr_rkey, n < nsegs ? "more" : "last"); + + r_xprt->rx_stats.read_chunk_count++; + req->rl_nchunks++; + seg += n; + nsegs -= n; + } while (nsegs); + req->rl_nextseg = seg; + + /* Finish Read list */ + *iptr++ = xdr_zero; /* Next item not present */ + return iptr; +} + +/* XDR-encode the Write list. Supports encoding a list containing + * one array of plain segments that belong to a single write chunk. + * + * Encoding key for single-list chunks (HLOO = Handle32 Length32 Offset64): + * * Write chunklist (a list of (one) counted array): * N elements: * 1 - N - HLOO - HLOO - ... - HLOO - 0 * + * Returns a pointer to the XDR word in the RDMA header following + * the end of the Write list, or an error pointer. + */ +static __be32 * +rpcrdma_encode_write_list(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req, + struct rpc_rqst *rqst, __be32 *iptr, + enum rpcrdma_chunktype wtype) +{ + struct rpcrdma_mr_seg *seg = req->rl_nextseg; + int n, nsegs, nchunks; + __be32 *segcount; + + if (wtype != rpcrdma_writech) { + *iptr++ = xdr_zero; /* no Write list present */ + return iptr; + } + + nsegs = rpcrdma_convert_iovs(&rqst->rq_rcv_buf, + rqst->rq_rcv_buf.head[0].iov_len, + wtype, seg, + RPCRDMA_MAX_SEGS - req->rl_nchunks); + if (nsegs < 0) + return ERR_PTR(nsegs); + + *iptr++ = xdr_one; /* Write list present */ + segcount = iptr++; /* save location of segment count */ + + nchunks = 0; + do { + n = r_xprt->rx_ia.ri_ops->ro_map(r_xprt, seg, nsegs, true); + if (n <= 0) + return ERR_PTR(n); + + iptr = xdr_encode_rdma_segment(iptr, seg); + + dprintk("RPC: %5u %s: write segment " + "%d@0x016%llx:0x%08x (%s)\n", + rqst->rq_task->tk_pid, __func__, + seg->mr_len, (unsigned long long)seg->mr_base, + seg->mr_rkey, n < nsegs ? "more" : "last"); + + r_xprt->rx_stats.write_chunk_count++; + r_xprt->rx_stats.total_rdma_request += seg->mr_len; + req->rl_nchunks++; + nchunks++; + seg += n; + nsegs -= n; + } while (nsegs); + req->rl_nextseg = seg; + + /* Update count of segments in this Write chunk */ + *segcount = cpu_to_be32(nchunks); + + /* Finish Write list */ + *iptr++ = xdr_zero; /* Next item not present */ + return iptr; +} + +/* XDR-encode the Reply chunk. Supports encoding an array of plain + * segments that belong to a single write (reply) chunk. + * + * Encoding key for single-list chunks (HLOO = Handle32 Length32 Offset64): + * * Reply chunk (a counted array): * N elements: * 1 - N - HLOO - HLOO - ... - HLOO * - * Returns positive RPC/RDMA header size, or negative errno. + * Returns a pointer to the XDR word in the RDMA header following + * the end of the Reply chunk, or an error pointer. */ - -static ssize_t -rpcrdma_create_chunks(struct rpc_rqst *rqst, struct xdr_buf *target, - struct rpcrdma_msg *headerp, enum rpcrdma_chunktype type) +static __be32 * +rpcrdma_encode_reply_chunk(struct rpcrdma_xprt *r_xprt, + struct rpcrdma_req *req, struct rpc_rqst *rqst, + __be32 *iptr, enum rpcrdma_chunktype wtype) { - struct rpcrdma_req *req = rpcr_to_rdmar(rqst); - struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt); - int n, nsegs, nchunks = 0; - unsigned int pos; - struct rpcrdma_mr_seg *seg = req->rl_segments; - struct rpcrdma_read_chunk *cur_rchunk = NULL; - struct rpcrdma_write_array *warray = NULL; - struct rpcrdma_write_chunk *cur_wchunk = NULL; - __be32 *iptr = headerp->rm_body.rm_chunks; - int (*map)(struct rpcrdma_xprt *, struct rpcrdma_mr_seg *, int, bool); - - if (type == rpcrdma_readch || type == rpcrdma_areadch) { - /* a read chunk - server will RDMA Read our memory */ - cur_rchunk = (struct rpcrdma_read_chunk *) iptr; - } else { - /* a write or reply chunk - server will RDMA Write our memory */ - *iptr++ = xdr_zero; /* encode a NULL read chunk list */ - if (type == rpcrdma_replych) - *iptr++ = xdr_zero; /* a NULL write chunk list */ - warray = (struct rpcrdma_write_array *) iptr; - cur_wchunk = (struct rpcrdma_write_chunk *) (warray + 1); - } + struct rpcrdma_mr_seg *seg = req->rl_nextseg; + int n, nsegs, nchunks; + __be32 *segcount; - if (type == rpcrdma_replych || type == rpcrdma_areadch) - pos = 0; - else - pos = target->head[0].iov_len; + if (wtype != rpcrdma_replych) { + *iptr++ = xdr_zero; /* no Reply chunk present */ + return iptr; + } - nsegs = rpcrdma_convert_iovs(target, pos, type, seg, RPCRDMA_MAX_SEGS); + nsegs = rpcrdma_convert_iovs(&rqst->rq_rcv_buf, 0, wtype, seg, + RPCRDMA_MAX_SEGS - req->rl_nchunks); if (nsegs < 0) - return nsegs; + return ERR_PTR(nsegs); - map = r_xprt->rx_ia.ri_ops->ro_map; + *iptr++ = xdr_one; /* Reply chunk present */ + segcount = iptr++; /* save location of segment count */ + + nchunks = 0; do { - n = map(r_xprt, seg, nsegs, cur_wchunk != NULL); + n = r_xprt->rx_ia.ri_ops->ro_map(r_xprt, seg, nsegs, true); if (n <= 0) - goto out; - if (cur_rchunk) { /* read */ - cur_rchunk->rc_discrim = xdr_one; - /* all read chunks have the same "position" */ - cur_rchunk->rc_position = cpu_to_be32(pos); - cur_rchunk->rc_target.rs_handle = - cpu_to_be32(seg->mr_rkey); - cur_rchunk->rc_target.rs_length = - cpu_to_be32(seg->mr_len); - xdr_encode_hyper( - (__be32 *)&cur_rchunk->rc_target.rs_offset, - seg->mr_base); - dprintk("RPC: %s: read chunk " - "elem %d@0x%llx:0x%x pos %u (%s)\n", __func__, - seg->mr_len, (unsigned long long)seg->mr_base, - seg->mr_rkey, pos, n < nsegs ? "more" : "last"); - cur_rchunk++; - r_xprt->rx_stats.read_chunk_count++; - } else { /* write/reply */ - cur_wchunk->wc_target.rs_handle = - cpu_to_be32(seg->mr_rkey); - cur_wchunk->wc_target.rs_length = - cpu_to_be32(seg->mr_len); - xdr_encode_hyper( - (__be32 *)&cur_wchunk->wc_target.rs_offset, - seg->mr_base); - dprintk("RPC: %s: %s chunk " - "elem %d@0x%llx:0x%x (%s)\n", __func__, - (type == rpcrdma_replych) ? "reply" : "write", - seg->mr_len, (unsigned long long)seg->mr_base, - seg->mr_rkey, n < nsegs ? "more" : "last"); - cur_wchunk++; - if (type == rpcrdma_replych) - r_xprt->rx_stats.reply_chunk_count++; - else - r_xprt->rx_stats.write_chunk_count++; - r_xprt->rx_stats.total_rdma_request += seg->mr_len; - } + return ERR_PTR(n); + + iptr = xdr_encode_rdma_segment(iptr, seg); + + dprintk("RPC: %5u %s: reply segment " + "%d@0x%016llx:0x%08x (%s)\n", + rqst->rq_task->tk_pid, __func__, + seg->mr_len, (unsigned long long)seg->mr_base, + seg->mr_rkey, n < nsegs ? "more" : "last"); + + r_xprt->rx_stats.reply_chunk_count++; + r_xprt->rx_stats.total_rdma_request += seg->mr_len; + req->rl_nchunks++; nchunks++; seg += n; nsegs -= n; } while (nsegs); + req->rl_nextseg = seg; - /* success. all failures return above */ - req->rl_nchunks = nchunks; - - /* - * finish off header. If write, marshal discrim and nchunks. - */ - if (cur_rchunk) { - iptr = (__be32 *) cur_rchunk; - *iptr++ = xdr_zero; /* finish the read chunk list */ - *iptr++ = xdr_zero; /* encode a NULL write chunk list */ - *iptr++ = xdr_zero; /* encode a NULL reply chunk */ - } else { - warray->wc_discrim = xdr_one; - warray->wc_nchunks = cpu_to_be32(nchunks); - iptr = (__be32 *) cur_wchunk; - if (type == rpcrdma_writech) { - *iptr++ = xdr_zero; /* finish the write chunk list */ - *iptr++ = xdr_zero; /* encode a NULL reply chunk */ - } - } - - /* - * Return header size. - */ - return (unsigned char *)iptr - (unsigned char *)headerp; + /* Update count of segments in the Reply chunk */ + *segcount = cpu_to_be32(nchunks); -out: - for (pos = 0; nchunks--;) - pos += r_xprt->rx_ia.ri_ops->ro_unmap(r_xprt, - &req->rl_segments[pos]); - return n; + return iptr; } /* @@ -440,13 +551,10 @@ static void rpcrdma_inline_pullup(struct rpc_rqst *rqst) * Marshal a request: the primary job of this routine is to choose * the transfer modes. See comments below. * - * Uses multiple RDMA IOVs for a request: - * [0] -- RPC RDMA header, which uses memory from the *start* of the - * preregistered buffer that already holds the RPC data in - * its middle. - * [1] -- the RPC header/data, marshaled by RPC and the NFS protocol. - * [2] -- optional padding. - * [3] -- if padded, header only in [1] and data here. + * Prepares up to two IOVs per Call message: + * + * [0] -- RPC RDMA header + * [1] -- the RPC header/data * * Returns zero on success, otherwise a negative errno. */ @@ -457,24 +565,17 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) struct rpc_xprt *xprt = rqst->rq_xprt; struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt); struct rpcrdma_req *req = rpcr_to_rdmar(rqst); - char *base; - size_t rpclen; - ssize_t hdrlen; enum rpcrdma_chunktype rtype, wtype; struct rpcrdma_msg *headerp; + ssize_t hdrlen; + size_t rpclen; + __be32 *iptr; #if defined(CONFIG_SUNRPC_BACKCHANNEL) if (test_bit(RPC_BC_PA_IN_USE, &rqst->rq_bc_pa_state)) return rpcrdma_bc_marshal_reply(rqst); #endif - /* - * rpclen gets amount of data in first buffer, which is the - * pre-registered buffer. - */ - base = rqst->rq_svec[0].iov_base; - rpclen = rqst->rq_svec[0].iov_len; - headerp = rdmab_to_msg(req->rl_rdmabuf); /* don't byte-swap XID, it's already done in request */ headerp->rm_xid = rqst->rq_xid; @@ -485,15 +586,16 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) /* * Chunks needed for results? * - * o Read ops return data as write chunk(s), header as inline. * o If the expected result is under the inline threshold, all ops * return as inline. + * o Large read ops return data as write chunk(s), header as + * inline. * o Large non-read ops return as a single reply chunk. */ - if (rqst->rq_rcv_buf.flags & XDRBUF_READ) - wtype = rpcrdma_writech; - else if (rpcrdma_results_inline(rqst)) + if (rpcrdma_results_inline(r_xprt, rqst)) wtype = rpcrdma_noch; + else if (rqst->rq_rcv_buf.flags & XDRBUF_READ) + wtype = rpcrdma_writech; else wtype = rpcrdma_replych; @@ -511,10 +613,14 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) * that both has a data payload, and whose non-data arguments * by themselves are larger than the inline threshold. */ - if (rpcrdma_args_inline(rqst)) { + if (rpcrdma_args_inline(r_xprt, rqst)) { rtype = rpcrdma_noch; + rpcrdma_inline_pullup(rqst); + rpclen = rqst->rq_svec[0].iov_len; } else if (rqst->rq_snd_buf.flags & XDRBUF_WRITE) { rtype = rpcrdma_readch; + rpclen = rqst->rq_svec[0].iov_len; + rpclen += rpcrdma_tail_pullup(&rqst->rq_snd_buf); } else { r_xprt->rx_stats.nomsg_call_count++; headerp->rm_type = htonl(RDMA_NOMSG); @@ -522,57 +628,50 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) rpclen = 0; } - /* The following simplification is not true forever */ - if (rtype != rpcrdma_noch && wtype == rpcrdma_replych) - wtype = rpcrdma_noch; - if (rtype != rpcrdma_noch && wtype != rpcrdma_noch) { - dprintk("RPC: %s: cannot marshal multiple chunk lists\n", - __func__); - return -EIO; - } - - hdrlen = RPCRDMA_HDRLEN_MIN; - - /* - * Pull up any extra send data into the preregistered buffer. - * When padding is in use and applies to the transfer, insert - * it and change the message type. + /* This implementation supports the following combinations + * of chunk lists in one RPC-over-RDMA Call message: + * + * - Read list + * - Write list + * - Reply chunk + * - Read list + Reply chunk + * + * It might not yet support the following combinations: + * + * - Read list + Write list + * + * It does not support the following combinations: + * + * - Write list + Reply chunk + * - Read list + Write list + Reply chunk + * + * This implementation supports only a single chunk in each + * Read or Write list. Thus for example the client cannot + * send a Call message with a Position Zero Read chunk and a + * regular Read chunk at the same time. */ - if (rtype == rpcrdma_noch) { - - rpcrdma_inline_pullup(rqst); - - headerp->rm_body.rm_nochunks.rm_empty[0] = xdr_zero; - headerp->rm_body.rm_nochunks.rm_empty[1] = xdr_zero; - headerp->rm_body.rm_nochunks.rm_empty[2] = xdr_zero; - /* new length after pullup */ - rpclen = rqst->rq_svec[0].iov_len; - } else if (rtype == rpcrdma_readch) - rpclen += rpcrdma_tail_pullup(&rqst->rq_snd_buf); - if (rtype != rpcrdma_noch) { - hdrlen = rpcrdma_create_chunks(rqst, &rqst->rq_snd_buf, - headerp, rtype); - wtype = rtype; /* simplify dprintk */ - - } else if (wtype != rpcrdma_noch) { - hdrlen = rpcrdma_create_chunks(rqst, &rqst->rq_rcv_buf, - headerp, wtype); - } - if (hdrlen < 0) - return hdrlen; + req->rl_nchunks = 0; + req->rl_nextseg = req->rl_segments; + iptr = headerp->rm_body.rm_chunks; + iptr = rpcrdma_encode_read_list(r_xprt, req, rqst, iptr, rtype); + if (IS_ERR(iptr)) + goto out_unmap; + iptr = rpcrdma_encode_write_list(r_xprt, req, rqst, iptr, wtype); + if (IS_ERR(iptr)) + goto out_unmap; + iptr = rpcrdma_encode_reply_chunk(r_xprt, req, rqst, iptr, wtype); + if (IS_ERR(iptr)) + goto out_unmap; + hdrlen = (unsigned char *)iptr - (unsigned char *)headerp; + + if (hdrlen + rpclen > RPCRDMA_INLINE_WRITE_THRESHOLD(rqst)) + goto out_overflow; + + dprintk("RPC: %5u %s: %s/%s: hdrlen %zd rpclen %zd\n", + rqst->rq_task->tk_pid, __func__, + transfertypes[rtype], transfertypes[wtype], + hdrlen, rpclen); - dprintk("RPC: %s: %s: hdrlen %zd rpclen %zd" - " headerp 0x%p base 0x%p lkey 0x%x\n", - __func__, transfertypes[wtype], hdrlen, rpclen, - headerp, base, rdmab_lkey(req->rl_rdmabuf)); - - /* - * initialize send_iov's - normally only two: rdma chunk header and - * single preregistered RPC header buffer, but if padding is present, - * then use a preregistered (and zeroed) pad buffer between the RPC - * header and any write data. In all non-rdma cases, any following - * data has been copied into the RPC header buffer. - */ req->rl_send_iov[0].addr = rdmab_addr(req->rl_rdmabuf); req->rl_send_iov[0].length = hdrlen; req->rl_send_iov[0].lkey = rdmab_lkey(req->rl_rdmabuf); @@ -587,6 +686,18 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst) req->rl_niovs = 2; return 0; + +out_overflow: + pr_err("rpcrdma: send overflow: hdrlen %zd rpclen %zu %s/%s\n", + hdrlen, rpclen, transfertypes[rtype], transfertypes[wtype]); + /* Terminate this RPC. Chunks registered above will be + * released by xprt_release -> xprt_rmda_free . + */ + return -EIO; + +out_unmap: + r_xprt->rx_ia.ri_ops->ro_unmap_safe(r_xprt, req, false); + return PTR_ERR(iptr); } /* diff --git a/net/sunrpc/xprtrdma/svc_rdma_marshal.c b/net/sunrpc/xprtrdma/svc_rdma_marshal.c index 765bca4..0ba9887 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_marshal.c +++ b/net/sunrpc/xprtrdma/svc_rdma_marshal.c @@ -145,19 +145,32 @@ static __be32 *decode_reply_array(__be32 *va, __be32 *vaend) return (__be32 *)&ary->wc_array[nchunks]; } -int svc_rdma_xdr_decode_req(struct rpcrdma_msg *rmsgp, struct svc_rqst *rqstp) +/** + * svc_rdma_xdr_decode_req - Parse incoming RPC-over-RDMA header + * @rq_arg: Receive buffer + * + * On entry, xdr->head[0].iov_base points to first byte in the + * RPC-over-RDMA header. + * + * On successful exit, head[0] points to first byte past the + * RPC-over-RDMA header. For RDMA_MSG, this is the RPC message. + * The length of the RPC-over-RDMA header is returned. + */ +int svc_rdma_xdr_decode_req(struct xdr_buf *rq_arg) { + struct rpcrdma_msg *rmsgp; __be32 *va, *vaend; unsigned int len; u32 hdr_len; /* Verify that there's enough bytes for header + something */ - if (rqstp->rq_arg.len <= RPCRDMA_HDRLEN_ERR) { + if (rq_arg->len <= RPCRDMA_HDRLEN_ERR) { dprintk("svcrdma: header too short = %d\n", - rqstp->rq_arg.len); + rq_arg->len); return -EINVAL; } + rmsgp = (struct rpcrdma_msg *)rq_arg->head[0].iov_base; if (rmsgp->rm_vers != rpcrdma_version) { dprintk("%s: bad version %u\n", __func__, be32_to_cpu(rmsgp->rm_vers)); @@ -189,10 +202,10 @@ int svc_rdma_xdr_decode_req(struct rpcrdma_msg *rmsgp, struct svc_rqst *rqstp) be32_to_cpu(rmsgp->rm_body.rm_padded.rm_thresh); va = &rmsgp->rm_body.rm_padded.rm_pempty[4]; - rqstp->rq_arg.head[0].iov_base = va; + rq_arg->head[0].iov_base = va; len = (u32)((unsigned long)va - (unsigned long)rmsgp); - rqstp->rq_arg.head[0].iov_len -= len; - if (len > rqstp->rq_arg.len) + rq_arg->head[0].iov_len -= len; + if (len > rq_arg->len) return -EINVAL; return len; default: @@ -205,7 +218,7 @@ int svc_rdma_xdr_decode_req(struct rpcrdma_msg *rmsgp, struct svc_rqst *rqstp) * chunk list and a reply chunk list. */ va = &rmsgp->rm_body.rm_chunks[0]; - vaend = (__be32 *)((unsigned long)rmsgp + rqstp->rq_arg.len); + vaend = (__be32 *)((unsigned long)rmsgp + rq_arg->len); va = decode_read_list(va, vaend); if (!va) { dprintk("svcrdma: failed to decode read list\n"); @@ -222,10 +235,9 @@ int svc_rdma_xdr_decode_req(struct rpcrdma_msg *rmsgp, struct svc_rqst *rqstp) return -EINVAL; } - rqstp->rq_arg.head[0].iov_base = va; + rq_arg->head[0].iov_base = va; hdr_len = (unsigned long)va - (unsigned long)rmsgp; - rqstp->rq_arg.head[0].iov_len -= hdr_len; - + rq_arg->head[0].iov_len -= hdr_len; return hdr_len; } diff --git a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c index fbe7444..2c25606 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c +++ b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c @@ -447,10 +447,8 @@ static int rdma_read_chunks(struct svcxprt_rdma *xprt, head->arg.len = rqstp->rq_arg.len; head->arg.buflen = rqstp->rq_arg.buflen; - ch = (struct rpcrdma_read_chunk *)&rmsgp->rm_body.rm_chunks[0]; - position = be32_to_cpu(ch->rc_position); - /* RDMA_NOMSG: RDMA READ data should land just after RDMA RECV data */ + position = be32_to_cpu(ch->rc_position); if (position == 0) { head->arg.pages = &head->pages[0]; page_offset = head->byte_len; @@ -488,7 +486,7 @@ static int rdma_read_chunks(struct svcxprt_rdma *xprt, if (page_offset & 3) { u32 pad = 4 - (page_offset & 3); - head->arg.page_len += pad; + head->arg.tail[0].iov_len += pad; head->arg.len += pad; head->arg.buflen += pad; page_offset += pad; @@ -510,11 +508,10 @@ static int rdma_read_chunks(struct svcxprt_rdma *xprt, return ret; } -static int rdma_read_complete(struct svc_rqst *rqstp, - struct svc_rdma_op_ctxt *head) +static void rdma_read_complete(struct svc_rqst *rqstp, + struct svc_rdma_op_ctxt *head) { int page_no; - int ret; /* Copy RPC pages */ for (page_no = 0; page_no < head->count; page_no++) { @@ -550,23 +547,6 @@ static int rdma_read_complete(struct svc_rqst *rqstp, rqstp->rq_arg.tail[0] = head->arg.tail[0]; rqstp->rq_arg.len = head->arg.len; rqstp->rq_arg.buflen = head->arg.buflen; - - /* Free the context */ - svc_rdma_put_context(head, 0); - - /* XXX: What should this be? */ - rqstp->rq_prot = IPPROTO_MAX; - svc_xprt_copy_addrs(rqstp, rqstp->rq_xprt); - - ret = rqstp->rq_arg.head[0].iov_len - + rqstp->rq_arg.page_len - + rqstp->rq_arg.tail[0].iov_len; - dprintk("svcrdma: deferred read ret=%d, rq_arg.len=%u, " - "rq_arg.head[0].iov_base=%p, rq_arg.head[0].iov_len=%zu\n", - ret, rqstp->rq_arg.len, rqstp->rq_arg.head[0].iov_base, - rqstp->rq_arg.head[0].iov_len); - - return ret; } /* By convention, backchannel calls arrive via rdma_msg type @@ -624,7 +604,8 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp) dto_q); list_del_init(&ctxt->dto_q); spin_unlock_bh(&rdma_xprt->sc_rq_dto_lock); - return rdma_read_complete(rqstp, ctxt); + rdma_read_complete(rqstp, ctxt); + goto complete; } else if (!list_empty(&rdma_xprt->sc_rq_dto_q)) { ctxt = list_entry(rdma_xprt->sc_rq_dto_q.next, struct svc_rdma_op_ctxt, @@ -655,7 +636,7 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp) /* Decode the RDMA header. */ rmsgp = (struct rpcrdma_msg *)rqstp->rq_arg.head[0].iov_base; - ret = svc_rdma_xdr_decode_req(rmsgp, rqstp); + ret = svc_rdma_xdr_decode_req(&rqstp->rq_arg); if (ret < 0) goto out_err; if (ret == 0) @@ -682,6 +663,7 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp) return 0; } +complete: ret = rqstp->rq_arg.head[0].iov_len + rqstp->rq_arg.page_len + rqstp->rq_arg.tail[0].iov_len; diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c index 4f1b1c4..54d53330 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c @@ -463,25 +463,21 @@ static int send_reply(struct svcxprt_rdma *rdma, struct svc_rqst *rqstp, struct page *page, struct rpcrdma_msg *rdma_resp, - struct svc_rdma_op_ctxt *ctxt, struct svc_rdma_req_map *vec, int byte_count) { + struct svc_rdma_op_ctxt *ctxt; struct ib_send_wr send_wr; u32 xdr_off; int sge_no; int sge_bytes; int page_no; int pages; - int ret; - - ret = svc_rdma_repost_recv(rdma, GFP_KERNEL); - if (ret) { - svc_rdma_put_context(ctxt, 0); - return -ENOTCONN; - } + int ret = -EIO; /* Prepare the context */ + ctxt = svc_rdma_get_context(rdma); + ctxt->direction = DMA_TO_DEVICE; ctxt->pages[0] = page; ctxt->count = 1; @@ -565,8 +561,7 @@ static int send_reply(struct svcxprt_rdma *rdma, err: svc_rdma_unmap_dma(ctxt); svc_rdma_put_context(ctxt, 1); - pr_err("svcrdma: failed to send reply, rc=%d\n", ret); - return -EIO; + return ret; } void svc_rdma_prep_reply_hdr(struct svc_rqst *rqstp) @@ -585,7 +580,6 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) int ret; int inline_bytes; struct page *res_page; - struct svc_rdma_op_ctxt *ctxt; struct svc_rdma_req_map *vec; dprintk("svcrdma: sending response for rqstp=%p\n", rqstp); @@ -598,8 +592,6 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) rp_ary = svc_rdma_get_reply_array(rdma_argp, wr_ary); /* Build an req vec for the XDR */ - ctxt = svc_rdma_get_context(rdma); - ctxt->direction = DMA_TO_DEVICE; vec = svc_rdma_get_req_map(rdma); ret = svc_rdma_map_xdr(rdma, &rqstp->rq_res, vec, wr_ary != NULL); if (ret) @@ -635,7 +627,12 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) inline_bytes -= ret; } - ret = send_reply(rdma, rqstp, res_page, rdma_resp, ctxt, vec, + /* Post a fresh Receive buffer _before_ sending the reply */ + ret = svc_rdma_post_recv(rdma, GFP_KERNEL); + if (ret) + goto err1; + + ret = send_reply(rdma, rqstp, res_page, rdma_resp, vec, inline_bytes); if (ret < 0) goto err1; @@ -648,7 +645,8 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) put_page(res_page); err0: svc_rdma_put_req_map(rdma, vec); - svc_rdma_put_context(ctxt, 0); + pr_err("svcrdma: Could not send reply, err=%d. Closing transport.\n", + ret); set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags); return -ENOTCONN; } diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c index 9066896..dd94401 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_transport.c +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c @@ -789,7 +789,7 @@ static struct svc_xprt *svc_rdma_create(struct svc_serv *serv, int ret; dprintk("svcrdma: Creating RDMA socket\n"); - if (sa->sa_family != AF_INET) { + if ((sa->sa_family != AF_INET) && (sa->sa_family != AF_INET6)) { dprintk("svcrdma: Address family %d is not supported.\n", sa->sa_family); return ERR_PTR(-EAFNOSUPPORT); } @@ -805,6 +805,16 @@ static struct svc_xprt *svc_rdma_create(struct svc_serv *serv, goto err0; } + /* Allow both IPv4 and IPv6 sockets to bind a single port + * at the same time. + */ +#if IS_ENABLED(CONFIG_IPV6) + ret = rdma_set_afonly(listen_id, 1); + if (ret) { + dprintk("svcrdma: rdma_set_afonly failed = %d\n", ret); + goto err1; + } +#endif ret = rdma_bind_addr(listen_id, sa); if (ret) { dprintk("svcrdma: rdma_bind_addr failed = %d\n", ret); @@ -1073,7 +1083,7 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt) newxprt->sc_dev_caps |= SVCRDMA_DEVCAP_READ_W_INV; /* Post receive buffers */ - for (i = 0; i < newxprt->sc_rq_depth; i++) { + for (i = 0; i < newxprt->sc_max_requests; i++) { ret = svc_rdma_post_recv(newxprt, GFP_KERNEL); if (ret) { dprintk("svcrdma: failure posting receive buffers\n"); @@ -1170,6 +1180,9 @@ static void __svc_rdma_free(struct work_struct *work) dprintk("svcrdma: %s(%p)\n", __func__, rdma); + if (rdma->sc_qp && !IS_ERR(rdma->sc_qp)) + ib_drain_qp(rdma->sc_qp); + /* We should only be called from kref_put */ if (atomic_read(&xprt->xpt_ref.refcount) != 0) pr_err("svcrdma: sc_xprt still in use? (%d)\n", diff --git a/net/sunrpc/xprtrdma/transport.c b/net/sunrpc/xprtrdma/transport.c index b1b009f..99d2e5b 100644 --- a/net/sunrpc/xprtrdma/transport.c +++ b/net/sunrpc/xprtrdma/transport.c @@ -73,6 +73,8 @@ static unsigned int xprt_rdma_memreg_strategy = RPCRDMA_FRMR; static unsigned int min_slot_table_size = RPCRDMA_MIN_SLOT_TABLE; static unsigned int max_slot_table_size = RPCRDMA_MAX_SLOT_TABLE; +static unsigned int min_inline_size = RPCRDMA_MIN_INLINE; +static unsigned int max_inline_size = RPCRDMA_MAX_INLINE; static unsigned int zero; static unsigned int max_padding = PAGE_SIZE; static unsigned int min_memreg = RPCRDMA_BOUNCEBUFFERS; @@ -96,6 +98,8 @@ static struct ctl_table xr_tunables_table[] = { .maxlen = sizeof(unsigned int), .mode = 0644, .proc_handler = proc_dointvec, + .extra1 = &min_inline_size, + .extra2 = &max_inline_size, }, { .procname = "rdma_max_inline_write", @@ -103,6 +107,8 @@ static struct ctl_table xr_tunables_table[] = { .maxlen = sizeof(unsigned int), .mode = 0644, .proc_handler = proc_dointvec, + .extra1 = &min_inline_size, + .extra2 = &max_inline_size, }, { .procname = "rdma_inline_write_padding", @@ -508,6 +514,7 @@ xprt_rdma_allocate(struct rpc_task *task, size_t size) out: dprintk("RPC: %s: size %zd, request 0x%p\n", __func__, size, req); req->rl_connect_cookie = 0; /* our reserved value */ + req->rl_task = task; return req->rl_sendbuf->rg_base; out_rdmabuf: @@ -564,7 +571,6 @@ xprt_rdma_free(void *buffer) struct rpcrdma_req *req; struct rpcrdma_xprt *r_xprt; struct rpcrdma_regbuf *rb; - int i; if (buffer == NULL) return; @@ -578,11 +584,8 @@ xprt_rdma_free(void *buffer) dprintk("RPC: %s: called on 0x%p\n", __func__, req->rl_reply); - for (i = 0; req->rl_nchunks;) { - --req->rl_nchunks; - i += r_xprt->rx_ia.ri_ops->ro_unmap(r_xprt, - &req->rl_segments[i]); - } + r_xprt->rx_ia.ri_ops->ro_unmap_safe(r_xprt, req, + !RPC_IS_ASYNC(req->rl_task)); rpcrdma_buffer_put(req); } @@ -707,6 +710,7 @@ static struct rpc_xprt_ops xprt_rdma_procs = { #if defined(CONFIG_SUNRPC_BACKCHANNEL) .bc_setup = xprt_rdma_bc_setup, .bc_up = xprt_rdma_bc_up, + .bc_maxpayload = xprt_rdma_bc_maxpayload, .bc_free_rqst = xprt_rdma_bc_free_rqst, .bc_destroy = xprt_rdma_bc_destroy, #endif diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c index f5ed9f9..b044d98a 100644 --- a/net/sunrpc/xprtrdma/verbs.c +++ b/net/sunrpc/xprtrdma/verbs.c @@ -203,15 +203,6 @@ out_fail: goto out_schedule; } -static void -rpcrdma_flush_cqs(struct rpcrdma_ep *ep) -{ - struct ib_wc wc; - - while (ib_poll_cq(ep->rep_attr.recv_cq, 1, &wc) > 0) - rpcrdma_receive_wc(NULL, &wc); -} - static int rpcrdma_conn_upcall(struct rdma_cm_id *id, struct rdma_cm_event *event) { @@ -374,23 +365,6 @@ out: } /* - * Drain any cq, prior to teardown. - */ -static void -rpcrdma_clean_cq(struct ib_cq *cq) -{ - struct ib_wc wc; - int count = 0; - - while (1 == ib_poll_cq(cq, 1, &wc)) - ++count; - - if (count) - dprintk("RPC: %s: flushed %d events (last 0x%x)\n", - __func__, count, wc.opcode); -} - -/* * Exported functions. */ @@ -459,7 +433,6 @@ rpcrdma_ia_open(struct rpcrdma_xprt *xprt, struct sockaddr *addr, int memreg) dprintk("RPC: %s: memory registration strategy is '%s'\n", __func__, ia->ri_ops->ro_displayname); - rwlock_init(&ia->ri_qplock); return 0; out3: @@ -515,7 +488,7 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia, __func__); return -ENOMEM; } - max_qp_wr = ia->ri_device->attrs.max_qp_wr - RPCRDMA_BACKWARD_WRS; + max_qp_wr = ia->ri_device->attrs.max_qp_wr - RPCRDMA_BACKWARD_WRS - 1; /* check provider's send/recv wr limits */ if (cdata->max_requests > max_qp_wr) @@ -526,11 +499,13 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia, ep->rep_attr.srq = NULL; ep->rep_attr.cap.max_send_wr = cdata->max_requests; ep->rep_attr.cap.max_send_wr += RPCRDMA_BACKWARD_WRS; + ep->rep_attr.cap.max_send_wr += 1; /* drain cqe */ rc = ia->ri_ops->ro_open(ia, ep, cdata); if (rc) return rc; ep->rep_attr.cap.max_recv_wr = cdata->max_requests; ep->rep_attr.cap.max_recv_wr += RPCRDMA_BACKWARD_WRS; + ep->rep_attr.cap.max_recv_wr += 1; /* drain cqe */ ep->rep_attr.cap.max_send_sge = RPCRDMA_MAX_IOVS; ep->rep_attr.cap.max_recv_sge = 1; ep->rep_attr.cap.max_inline_data = 0; @@ -578,6 +553,7 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia, ep->rep_attr.recv_cq = recvcq; /* Initialize cma parameters */ + memset(&ep->rep_remote_cma, 0, sizeof(ep->rep_remote_cma)); /* RPC/RDMA does not use private data */ ep->rep_remote_cma.private_data = NULL; @@ -591,7 +567,16 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia, ep->rep_remote_cma.responder_resources = ia->ri_device->attrs.max_qp_rd_atom; - ep->rep_remote_cma.retry_count = 7; + /* Limit transport retries so client can detect server + * GID changes quickly. RPC layer handles re-establishing + * transport connection and retransmission. + */ + ep->rep_remote_cma.retry_count = 6; + + /* RPC-over-RDMA handles its own flow control. In addition, + * make all RNR NAKs visible so we know that RPC-over-RDMA + * flow control is working correctly (no NAKs should be seen). + */ ep->rep_remote_cma.flow_control = 0; ep->rep_remote_cma.rnr_retry_count = 0; @@ -622,13 +607,8 @@ rpcrdma_ep_destroy(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia) cancel_delayed_work_sync(&ep->rep_connect_worker); - if (ia->ri_id->qp) - rpcrdma_ep_disconnect(ep, ia); - - rpcrdma_clean_cq(ep->rep_attr.recv_cq); - rpcrdma_clean_cq(ep->rep_attr.send_cq); - if (ia->ri_id->qp) { + rpcrdma_ep_disconnect(ep, ia); rdma_destroy_qp(ia->ri_id); ia->ri_id->qp = NULL; } @@ -659,7 +639,6 @@ retry: dprintk("RPC: %s: reconnecting...\n", __func__); rpcrdma_ep_disconnect(ep, ia); - rpcrdma_flush_cqs(ep); xprt = container_of(ia, struct rpcrdma_xprt, rx_ia); id = rpcrdma_create_id(xprt, ia, @@ -692,10 +671,8 @@ retry: goto out; } - write_lock(&ia->ri_qplock); old = ia->ri_id; ia->ri_id = id; - write_unlock(&ia->ri_qplock); rdma_destroy_qp(old); rpcrdma_destroy_id(old); @@ -785,7 +762,6 @@ rpcrdma_ep_disconnect(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia) { int rc; - rpcrdma_flush_cqs(ep); rc = rdma_disconnect(ia->ri_id); if (!rc) { /* returns without wait if not connected */ @@ -797,6 +773,8 @@ rpcrdma_ep_disconnect(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia) dprintk("RPC: %s: rdma_disconnect %i\n", __func__, rc); ep->rep_connected = rc; } + + ib_drain_qp(ia->ri_id->qp); } struct rpcrdma_req * @@ -1271,25 +1249,3 @@ out_rc: rpcrdma_recv_buffer_put(rep); return rc; } - -/* How many chunk list items fit within our inline buffers? - */ -unsigned int -rpcrdma_max_segments(struct rpcrdma_xprt *r_xprt) -{ - struct rpcrdma_create_data_internal *cdata = &r_xprt->rx_data; - int bytes, segments; - - bytes = min_t(unsigned int, cdata->inline_wsize, cdata->inline_rsize); - bytes -= RPCRDMA_HDRLEN_MIN; - if (bytes < sizeof(struct rpcrdma_segment) * 2) { - pr_warn("RPC: %s: inline threshold too small\n", - __func__); - return 0; - } - - segments = 1 << (fls(bytes / sizeof(struct rpcrdma_segment)) - 1); - dprintk("RPC: %s: max chunk list size = %d segments\n", - __func__, segments); - return segments; -} diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h index 2ebc743..95cdc66 100644 --- a/net/sunrpc/xprtrdma/xprt_rdma.h +++ b/net/sunrpc/xprtrdma/xprt_rdma.h @@ -65,7 +65,6 @@ */ struct rpcrdma_ia { const struct rpcrdma_memreg_ops *ri_ops; - rwlock_t ri_qplock; struct ib_device *ri_device; struct rdma_cm_id *ri_id; struct ib_pd *ri_pd; @@ -73,6 +72,8 @@ struct rpcrdma_ia { struct completion ri_done; int ri_async_rc; unsigned int ri_max_frmr_depth; + unsigned int ri_max_inline_write; + unsigned int ri_max_inline_read; struct ib_qp_attr ri_qp_attr; struct ib_qp_init_attr ri_qp_init_attr; }; @@ -144,6 +145,26 @@ rdmab_to_msg(struct rpcrdma_regbuf *rb) #define RPCRDMA_DEF_GFP (GFP_NOIO | __GFP_NOWARN) +/* To ensure a transport can always make forward progress, + * the number of RDMA segments allowed in header chunk lists + * is capped at 8. This prevents less-capable devices and + * memory registrations from overrunning the Send buffer + * while building chunk lists. + * + * Elements of the Read list take up more room than the + * Write list or Reply chunk. 8 read segments means the Read + * list (or Write list or Reply chunk) cannot consume more + * than + * + * ((8 + 2) * read segment size) + 1 XDR words, or 244 bytes. + * + * And the fixed part of the header is another 24 bytes. + * + * The smallest inline threshold is 1024 bytes, ensuring that + * at least 750 bytes are available for RPC messages. + */ +#define RPCRDMA_MAX_HDR_SEGS (8) + /* * struct rpcrdma_rep -- this structure encapsulates state required to recv * and complete a reply, asychronously. It needs several pieces of @@ -162,7 +183,9 @@ rdmab_to_msg(struct rpcrdma_regbuf *rb) */ #define RPCRDMA_MAX_DATA_SEGS ((1 * 1024 * 1024) / PAGE_SIZE) -#define RPCRDMA_MAX_SEGS (RPCRDMA_MAX_DATA_SEGS + 2) /* head+tail = 2 */ + +/* data segments + head/tail for Call + head/tail for Reply */ +#define RPCRDMA_MAX_SEGS (RPCRDMA_MAX_DATA_SEGS + 4) struct rpcrdma_buffer; @@ -198,14 +221,13 @@ enum rpcrdma_frmr_state { }; struct rpcrdma_frmr { - struct scatterlist *sg; - int sg_nents; + struct scatterlist *fr_sg; + int fr_nents; + enum dma_data_direction fr_dir; struct ib_mr *fr_mr; struct ib_cqe fr_cqe; enum rpcrdma_frmr_state fr_state; struct completion fr_linv_done; - struct work_struct fr_work; - struct rpcrdma_xprt *fr_xprt; union { struct ib_reg_wr fr_regwr; struct ib_send_wr fr_invwr; @@ -222,6 +244,8 @@ struct rpcrdma_mw { struct rpcrdma_fmr fmr; struct rpcrdma_frmr frmr; }; + struct work_struct mw_work; + struct rpcrdma_xprt *mw_xprt; struct list_head mw_list; struct list_head mw_all; }; @@ -270,12 +294,14 @@ struct rpcrdma_req { unsigned int rl_niovs; unsigned int rl_nchunks; unsigned int rl_connect_cookie; + struct rpc_task *rl_task; struct rpcrdma_buffer *rl_buffer; struct rpcrdma_rep *rl_reply;/* holder for reply buffer */ struct ib_sge rl_send_iov[RPCRDMA_MAX_IOVS]; struct rpcrdma_regbuf *rl_rdmabuf; struct rpcrdma_regbuf *rl_sendbuf; struct rpcrdma_mr_seg rl_segments[RPCRDMA_MAX_SEGS]; + struct rpcrdma_mr_seg *rl_nextseg; struct ib_cqe rl_cqe; struct list_head rl_all; @@ -372,8 +398,8 @@ struct rpcrdma_memreg_ops { struct rpcrdma_mr_seg *, int, bool); void (*ro_unmap_sync)(struct rpcrdma_xprt *, struct rpcrdma_req *); - int (*ro_unmap)(struct rpcrdma_xprt *, - struct rpcrdma_mr_seg *); + void (*ro_unmap_safe)(struct rpcrdma_xprt *, + struct rpcrdma_req *, bool); int (*ro_open)(struct rpcrdma_ia *, struct rpcrdma_ep *, struct rpcrdma_create_data_internal *); @@ -456,7 +482,6 @@ struct rpcrdma_regbuf *rpcrdma_alloc_regbuf(struct rpcrdma_ia *, void rpcrdma_free_regbuf(struct rpcrdma_ia *, struct rpcrdma_regbuf *); -unsigned int rpcrdma_max_segments(struct rpcrdma_xprt *); int rpcrdma_ep_post_extra_recv(struct rpcrdma_xprt *, unsigned int); int frwr_alloc_recovery_wq(void); @@ -519,6 +544,9 @@ void rpcrdma_reply_handler(struct rpcrdma_rep *); * RPC/RDMA protocol calls - xprtrdma/rpc_rdma.c */ int rpcrdma_marshal_req(struct rpc_rqst *); +void rpcrdma_set_max_header_sizes(struct rpcrdma_ia *, + struct rpcrdma_create_data_internal *, + unsigned int); /* RPC/RDMA module init - xprtrdma/transport.c */ @@ -534,6 +562,7 @@ void xprt_rdma_cleanup(void); #if defined(CONFIG_SUNRPC_BACKCHANNEL) int xprt_rdma_bc_setup(struct rpc_xprt *, unsigned int); int xprt_rdma_bc_up(struct svc_serv *, struct net *); +size_t xprt_rdma_bc_maxpayload(struct rpc_xprt *); int rpcrdma_bc_post_recv(struct rpcrdma_xprt *, unsigned int); void rpcrdma_bc_receive_call(struct rpcrdma_xprt *, struct rpcrdma_rep *); int rpcrdma_bc_marshal_reply(struct rpc_rqst *); diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c index b90c539..2d3e0c4 100644 --- a/net/sunrpc/xprtsock.c +++ b/net/sunrpc/xprtsock.c @@ -1364,6 +1364,11 @@ static int xs_tcp_bc_up(struct svc_serv *serv, struct net *net) return ret; return 0; } + +static size_t xs_tcp_bc_maxpayload(struct rpc_xprt *xprt) +{ + return PAGE_SIZE; +} #else static inline int _xs_tcp_read_data(struct rpc_xprt *xprt, struct xdr_skb_reader *desc) @@ -2661,6 +2666,7 @@ static struct rpc_xprt_ops xs_tcp_ops = { #ifdef CONFIG_SUNRPC_BACKCHANNEL .bc_setup = xprt_setup_bc, .bc_up = xs_tcp_bc_up, + .bc_maxpayload = xs_tcp_bc_maxpayload, .bc_free_rqst = xprt_free_bc_rqst, .bc_destroy = xprt_destroy_bc, #endif diff --git a/net/tipc/bearer.c b/net/tipc/bearer.c index 6f11c62..bf8f05c 100644 --- a/net/tipc/bearer.c +++ b/net/tipc/bearer.c @@ -405,7 +405,7 @@ int tipc_l2_send_msg(struct net *net, struct sk_buff *skb, return 0; /* Send RESET message even if bearer is detached from device */ - tipc_ptr = rtnl_dereference(dev->tipc_ptr); + tipc_ptr = rcu_dereference_rtnl(dev->tipc_ptr); if (unlikely(!tipc_ptr && !msg_is_reset(buf_msg(skb)))) goto drop; diff --git a/net/tipc/link.c b/net/tipc/link.c index 7059c94..67b6ab9 100644 --- a/net/tipc/link.c +++ b/net/tipc/link.c @@ -704,7 +704,8 @@ static void link_profile_stats(struct tipc_link *l) */ int tipc_link_timeout(struct tipc_link *l, struct sk_buff_head *xmitq) { - int mtyp, rc = 0; + int mtyp = 0; + int rc = 0; bool state = false; bool probe = false; bool setup = false; diff --git a/net/tipc/msg.c b/net/tipc/msg.c index 8740930..17201aa 100644 --- a/net/tipc/msg.c +++ b/net/tipc/msg.c @@ -41,6 +41,8 @@ #include "name_table.h" #define MAX_FORWARD_SIZE 1024 +#define BUF_HEADROOM (LL_MAX_HEADER + 48) +#define BUF_TAILROOM 16 static unsigned int align(unsigned int i) { @@ -505,6 +507,10 @@ bool tipc_msg_reverse(u32 own_node, struct sk_buff **skb, int err) msg_set_hdr_sz(hdr, BASIC_H_SIZE); } + if (skb_cloned(_skb) && + pskb_expand_head(_skb, BUF_HEADROOM, BUF_TAILROOM, GFP_KERNEL)) + goto exit; + /* Now reverse the concerned fields */ msg_set_errcode(hdr, err); msg_set_origport(hdr, msg_destport(&ohdr)); diff --git a/net/tipc/msg.h b/net/tipc/msg.h index 024da8a..7cf52fb 100644 --- a/net/tipc/msg.h +++ b/net/tipc/msg.h @@ -94,17 +94,6 @@ struct plist; #define TIPC_MEDIA_INFO_OFFSET 5 -/** - * TIPC message buffer code - * - * TIPC message buffer headroom reserves space for the worst-case - * link-level device header (in case the message is sent off-node). - * - * Note: Headroom should be a multiple of 4 to ensure the TIPC header fields - * are word aligned for quicker access - */ -#define BUF_HEADROOM (LL_MAX_HEADER + 48) - struct tipc_skb_cb { void *handle; struct sk_buff *tail; diff --git a/net/tipc/netlink_compat.c b/net/tipc/netlink_compat.c index f795b1d..3ad9fab 100644 --- a/net/tipc/netlink_compat.c +++ b/net/tipc/netlink_compat.c @@ -604,7 +604,8 @@ static int tipc_nl_compat_link_dump(struct tipc_nl_compat_msg *msg, link_info.dest = nla_get_flag(link[TIPC_NLA_LINK_DEST]); link_info.up = htonl(nla_get_flag(link[TIPC_NLA_LINK_UP])); - strcpy(link_info.str, nla_data(link[TIPC_NLA_LINK_NAME])); + nla_strlcpy(link_info.str, nla_data(link[TIPC_NLA_LINK_NAME]), + TIPC_MAX_LINK_NAME); return tipc_add_tlv(msg->rep, TIPC_TLV_LINK_INFO, &link_info, sizeof(link_info)); diff --git a/net/tipc/socket.c b/net/tipc/socket.c index 88bfcd7..c49b8df 100644 --- a/net/tipc/socket.c +++ b/net/tipc/socket.c @@ -796,9 +796,11 @@ void tipc_sk_mcast_rcv(struct net *net, struct sk_buff_head *arrvq, * @tsk: receiving socket * @skb: pointer to message buffer. */ -static void tipc_sk_proto_rcv(struct tipc_sock *tsk, struct sk_buff *skb) +static void tipc_sk_proto_rcv(struct tipc_sock *tsk, struct sk_buff *skb, + struct sk_buff_head *xmitq) { struct sock *sk = &tsk->sk; + u32 onode = tsk_own_node(tsk); struct tipc_msg *hdr = buf_msg(skb); int mtyp = msg_type(hdr); bool conn_cong; @@ -811,7 +813,8 @@ static void tipc_sk_proto_rcv(struct tipc_sock *tsk, struct sk_buff *skb) if (mtyp == CONN_PROBE) { msg_set_type(hdr, CONN_PROBE_REPLY); - tipc_sk_respond(sk, skb, TIPC_OK); + if (tipc_msg_reverse(onode, &skb, TIPC_OK)) + __skb_queue_tail(xmitq, skb); return; } else if (mtyp == CONN_ACK) { conn_cong = tsk_conn_cong(tsk); @@ -1686,7 +1689,8 @@ static unsigned int rcvbuf_limit(struct sock *sk, struct sk_buff *skb) * * Returns true if message was added to socket receive queue, otherwise false */ -static bool filter_rcv(struct sock *sk, struct sk_buff *skb) +static bool filter_rcv(struct sock *sk, struct sk_buff *skb, + struct sk_buff_head *xmitq) { struct socket *sock = sk->sk_socket; struct tipc_sock *tsk = tipc_sk(sk); @@ -1696,7 +1700,7 @@ static bool filter_rcv(struct sock *sk, struct sk_buff *skb) int usr = msg_user(hdr); if (unlikely(msg_user(hdr) == CONN_MANAGER)) { - tipc_sk_proto_rcv(tsk, skb); + tipc_sk_proto_rcv(tsk, skb, xmitq); return false; } @@ -1739,7 +1743,8 @@ static bool filter_rcv(struct sock *sk, struct sk_buff *skb) return true; reject: - tipc_sk_respond(sk, skb, err); + if (tipc_msg_reverse(tsk_own_node(tsk), &skb, err)) + __skb_queue_tail(xmitq, skb); return false; } @@ -1755,9 +1760,24 @@ reject: static int tipc_backlog_rcv(struct sock *sk, struct sk_buff *skb) { unsigned int truesize = skb->truesize; + struct sk_buff_head xmitq; + u32 dnode, selector; - if (likely(filter_rcv(sk, skb))) + __skb_queue_head_init(&xmitq); + + if (likely(filter_rcv(sk, skb, &xmitq))) { atomic_add(truesize, &tipc_sk(sk)->dupl_rcvcnt); + return 0; + } + + if (skb_queue_empty(&xmitq)) + return 0; + + /* Send response/rejected message */ + skb = __skb_dequeue(&xmitq); + dnode = msg_destnode(buf_msg(skb)); + selector = msg_origport(buf_msg(skb)); + tipc_node_xmit_skb(sock_net(sk), skb, dnode, selector); return 0; } @@ -1771,12 +1791,13 @@ static int tipc_backlog_rcv(struct sock *sk, struct sk_buff *skb) * Caller must hold socket lock */ static void tipc_sk_enqueue(struct sk_buff_head *inputq, struct sock *sk, - u32 dport) + u32 dport, struct sk_buff_head *xmitq) { + unsigned long time_limit = jiffies + 2; + struct sk_buff *skb; unsigned int lim; atomic_t *dcnt; - struct sk_buff *skb; - unsigned long time_limit = jiffies + 2; + u32 onode; while (skb_queue_len(inputq)) { if (unlikely(time_after_eq(jiffies, time_limit))) @@ -1788,7 +1809,7 @@ static void tipc_sk_enqueue(struct sk_buff_head *inputq, struct sock *sk, /* Add message directly to receive queue if possible */ if (!sock_owned_by_user(sk)) { - filter_rcv(sk, skb); + filter_rcv(sk, skb, xmitq); continue; } @@ -1801,7 +1822,9 @@ static void tipc_sk_enqueue(struct sk_buff_head *inputq, struct sock *sk, continue; /* Overload => reject message back to sender */ - tipc_sk_respond(sk, skb, TIPC_ERR_OVERLOAD); + onode = tipc_own_addr(sock_net(sk)); + if (tipc_msg_reverse(onode, &skb, TIPC_ERR_OVERLOAD)) + __skb_queue_tail(xmitq, skb); break; } } @@ -1814,12 +1837,14 @@ static void tipc_sk_enqueue(struct sk_buff_head *inputq, struct sock *sk, */ void tipc_sk_rcv(struct net *net, struct sk_buff_head *inputq) { + struct sk_buff_head xmitq; u32 dnode, dport = 0; int err; struct tipc_sock *tsk; struct sock *sk; struct sk_buff *skb; + __skb_queue_head_init(&xmitq); while (skb_queue_len(inputq)) { dport = tipc_skb_peek_port(inputq, dport); tsk = tipc_sk_lookup(net, dport); @@ -1827,9 +1852,14 @@ void tipc_sk_rcv(struct net *net, struct sk_buff_head *inputq) if (likely(tsk)) { sk = &tsk->sk; if (likely(spin_trylock_bh(&sk->sk_lock.slock))) { - tipc_sk_enqueue(inputq, sk, dport); + tipc_sk_enqueue(inputq, sk, dport, &xmitq); spin_unlock_bh(&sk->sk_lock.slock); } + /* Send pending response/rejected messages, if any */ + while ((skb = __skb_dequeue(&xmitq))) { + dnode = msg_destnode(buf_msg(skb)); + tipc_node_xmit_skb(net, skb, dnode, dport); + } sock_put(sk); continue; } diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c index b5f1221..b96ac91 100644 --- a/net/vmw_vsock/af_vsock.c +++ b/net/vmw_vsock/af_vsock.c @@ -61,6 +61,14 @@ * function will also cleanup rejected sockets, those that reach the connected * state but leave it before they have been accepted. * + * - Lock ordering for pending or accept queue sockets is: + * + * lock_sock(listener); + * lock_sock_nested(pending, SINGLE_DEPTH_NESTING); + * + * Using explicit nested locking keeps lockdep happy since normally only one + * lock of a given class may be taken at a time. + * * - Sockets created by user action will be cleaned up when the user process * calls close(2), causing our release implementation to be called. Our release * implementation will perform some cleanup then drop the last reference so our @@ -443,7 +451,7 @@ void vsock_pending_work(struct work_struct *work) cleanup = true; lock_sock(listener); - lock_sock(sk); + lock_sock_nested(sk, SINGLE_DEPTH_NESTING); if (vsock_is_pending(sk)) { vsock_remove_pending(listener, sk); @@ -1292,7 +1300,7 @@ static int vsock_accept(struct socket *sock, struct socket *newsock, int flags) if (connected) { listener->sk_ack_backlog--; - lock_sock(connected); + lock_sock_nested(connected, SINGLE_DEPTH_NESTING); vconnected = vsock_sk(connected); /* If the listener socket has received an error, then we should |
