diff options
Diffstat (limited to 'net')
52 files changed, 1679 insertions, 929 deletions
diff --git a/net/Kconfig b/net/Kconfig index e134498..a8934d8 100644 --- a/net/Kconfig +++ b/net/Kconfig @@ -397,7 +397,7 @@ config LWTUNNEL with light weight tunnel state associated with fib routes. config DST_CACHE - bool "dst cache" + bool default n config NET_DEVLINK diff --git a/net/bridge/br_if.c b/net/bridge/br_if.c index a73df33..8217aec 100644 --- a/net/bridge/br_if.c +++ b/net/bridge/br_if.c @@ -437,6 +437,20 @@ int br_min_mtu(const struct net_bridge *br) return mtu; } +static void br_set_gso_limits(struct net_bridge *br) +{ + unsigned int gso_max_size = GSO_MAX_SIZE; + u16 gso_max_segs = GSO_MAX_SEGS; + const struct net_bridge_port *p; + + list_for_each_entry(p, &br->port_list, list) { + gso_max_size = min(gso_max_size, p->dev->gso_max_size); + gso_max_segs = min(gso_max_segs, p->dev->gso_max_segs); + } + br->dev->gso_max_size = gso_max_size; + br->dev->gso_max_segs = gso_max_segs; +} + /* * Recomputes features using slave's features */ @@ -564,6 +578,7 @@ int br_add_if(struct net_bridge *br, struct net_device *dev) call_netdevice_notifiers(NETDEV_CHANGEADDR, br->dev); dev_set_mtu(br->dev, br_min_mtu(br)); + br_set_gso_limits(br); kobject_uevent(&p->kobj, KOBJ_ADD); @@ -610,6 +625,7 @@ int br_del_if(struct net_bridge *br, struct net_device *dev) del_nbp(p); dev_set_mtu(br->dev, br_min_mtu(br)); + br_set_gso_limits(br); spin_lock_bh(&br->lock); changed_addr = br_stp_recalculate_bridge_id(br); diff --git a/net/ceph/pagevec.c b/net/ceph/pagevec.c index d4f5f22..10297f7 100644 --- a/net/ceph/pagevec.c +++ b/net/ceph/pagevec.c @@ -24,7 +24,7 @@ struct page **ceph_get_direct_page_vector(const void __user *data, return ERR_PTR(-ENOMEM); while (got < num_pages) { - rc = get_user_pages_unlocked(current, current->mm, + rc = get_user_pages_unlocked( (unsigned long)data + ((unsigned long)got * PAGE_SIZE), num_pages - got, write_page, 0, pages + got); if (rc < 0) diff --git a/net/core/dev.c b/net/core/dev.c index edb7179..b9bcbe7 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -4438,7 +4438,7 @@ static enum gro_result dev_gro_receive(struct napi_struct *napi, struct sk_buff NAPI_GRO_CB(skb)->same_flow = 0; NAPI_GRO_CB(skb)->flush = 0; NAPI_GRO_CB(skb)->free = 0; - NAPI_GRO_CB(skb)->udp_mark = 0; + NAPI_GRO_CB(skb)->encap_mark = 0; NAPI_GRO_CB(skb)->gro_remcsum_start = 0; /* Setup for GRO checksum validation */ @@ -6445,6 +6445,7 @@ EXPORT_SYMBOL(dev_get_phys_port_id); * dev_get_phys_port_name - Get device physical port name * @dev: device * @name: port name + * @len: limit of bytes to copy to name * * Get device physical port name */ diff --git a/net/core/gen_estimator.c b/net/core/gen_estimator.c index 92d886f..4573d81 100644 --- a/net/core/gen_estimator.c +++ b/net/core/gen_estimator.c @@ -191,6 +191,7 @@ struct gen_estimator *gen_find_node(const struct gnet_stats_basic_packed *bstats /** * gen_new_estimator - create a new rate estimator * @bstats: basic statistics + * @cpu_bstats: bstats per cpu * @rate_est: rate estimator statistics * @stats_lock: statistics lock * @opt: rate estimator configuration TLV @@ -287,6 +288,7 @@ EXPORT_SYMBOL(gen_kill_estimator); /** * gen_replace_estimator - replace rate estimator configuration * @bstats: basic statistics + * @cpu_bstats: bstats per cpu * @rate_est: rate estimator statistics * @stats_lock: statistics lock * @opt: rate estimator configuration TLV diff --git a/net/core/gen_stats.c b/net/core/gen_stats.c index 1e2f46a..e640462 100644 --- a/net/core/gen_stats.c +++ b/net/core/gen_stats.c @@ -140,6 +140,7 @@ EXPORT_SYMBOL(__gnet_stats_copy_basic); /** * gnet_stats_copy_basic - copy basic statistics into statistic TLV * @d: dumping handle + * @cpu: copy statistic per cpu * @b: basic statistics * * Appends the basic statistics to the top level TLV created by diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c index d2d9e5e..f206677 100644 --- a/net/core/rtnetlink.c +++ b/net/core/rtnetlink.c @@ -895,6 +895,8 @@ static noinline size_t if_nlmsg_size(const struct net_device *dev, + nla_total_size(4) /* IFLA_PROMISCUITY */ + nla_total_size(4) /* IFLA_NUM_TX_QUEUES */ + nla_total_size(4) /* IFLA_NUM_RX_QUEUES */ + + nla_total_size(4) /* IFLA_MAX_GSO_SEGS */ + + nla_total_size(4) /* IFLA_MAX_GSO_SIZE */ + nla_total_size(1) /* IFLA_OPERSTATE */ + nla_total_size(1) /* IFLA_LINKMODE */ + nla_total_size(4) /* IFLA_CARRIER_CHANGES */ @@ -1223,6 +1225,8 @@ static int rtnl_fill_ifinfo(struct sk_buff *skb, struct net_device *dev, nla_put_u32(skb, IFLA_GROUP, dev->group) || nla_put_u32(skb, IFLA_PROMISCUITY, dev->promiscuity) || nla_put_u32(skb, IFLA_NUM_TX_QUEUES, dev->num_tx_queues) || + nla_put_u32(skb, IFLA_GSO_MAX_SEGS, dev->gso_max_segs) || + nla_put_u32(skb, IFLA_GSO_MAX_SIZE, dev->gso_max_size) || #ifdef CONFIG_RPS nla_put_u32(skb, IFLA_NUM_RX_QUEUES, dev->num_rx_queues) || #endif @@ -1389,6 +1393,8 @@ static const struct nla_policy ifla_vf_policy[IFLA_VF_MAX+1] = { [IFLA_VF_RSS_QUERY_EN] = { .len = sizeof(struct ifla_vf_rss_query_en) }, [IFLA_VF_STATS] = { .type = NLA_NESTED }, [IFLA_VF_TRUST] = { .len = sizeof(struct ifla_vf_trust) }, + [IFLA_VF_IB_NODE_GUID] = { .len = sizeof(struct ifla_vf_guid) }, + [IFLA_VF_IB_PORT_GUID] = { .len = sizeof(struct ifla_vf_guid) }, }; static const struct nla_policy ifla_port_policy[IFLA_PORT_MAX+1] = { @@ -1593,6 +1599,22 @@ static int validate_linkmsg(struct net_device *dev, struct nlattr *tb[]) return 0; } +static int handle_infiniband_guid(struct net_device *dev, struct ifla_vf_guid *ivt, + int guid_type) +{ + const struct net_device_ops *ops = dev->netdev_ops; + + return ops->ndo_set_vf_guid(dev, ivt->vf, ivt->guid, guid_type); +} + +static int handle_vf_guid(struct net_device *dev, struct ifla_vf_guid *ivt, int guid_type) +{ + if (dev->type != ARPHRD_INFINIBAND) + return -EOPNOTSUPP; + + return handle_infiniband_guid(dev, ivt, guid_type); +} + static int do_setvfinfo(struct net_device *dev, struct nlattr **tb) { const struct net_device_ops *ops = dev->netdev_ops; @@ -1695,6 +1717,24 @@ static int do_setvfinfo(struct net_device *dev, struct nlattr **tb) return err; } + if (tb[IFLA_VF_IB_NODE_GUID]) { + struct ifla_vf_guid *ivt = nla_data(tb[IFLA_VF_IB_NODE_GUID]); + + if (!ops->ndo_set_vf_guid) + return -EOPNOTSUPP; + + return handle_vf_guid(dev, ivt, IFLA_VF_IB_NODE_GUID); + } + + if (tb[IFLA_VF_IB_PORT_GUID]) { + struct ifla_vf_guid *ivt = nla_data(tb[IFLA_VF_IB_PORT_GUID]); + + if (!ops->ndo_set_vf_guid) + return -EOPNOTSUPP; + + return handle_vf_guid(dev, ivt, IFLA_VF_IB_PORT_GUID); + } + return err; } diff --git a/net/core/skbuff.c b/net/core/skbuff.c index f044f97..d04c2d1 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -815,7 +815,7 @@ void napi_consume_skb(struct sk_buff *skb, int budget) trace_consume_skb(skb); /* if SKB is a clone, don't handle this case */ - if (unlikely(skb->fclone != SKB_FCLONE_UNAVAILABLE)) { + if (skb->fclone != SKB_FCLONE_UNAVAILABLE) { __kfree_skb(skb); return; } diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c index 0cc923f..9e48199 100644 --- a/net/ipv4/af_inet.c +++ b/net/ipv4/af_inet.c @@ -1380,6 +1380,19 @@ out: return pp; } +static struct sk_buff **ipip_gro_receive(struct sk_buff **head, + struct sk_buff *skb) +{ + if (NAPI_GRO_CB(skb)->encap_mark) { + NAPI_GRO_CB(skb)->flush = 1; + return NULL; + } + + NAPI_GRO_CB(skb)->encap_mark = 1; + + return inet_gro_receive(head, skb); +} + #define SECONDS_PER_DAY 86400 /* inet_current_timestamp - Return IP network timestamp @@ -1402,7 +1415,7 @@ __be32 inet_current_timestamp(void) msecs += (u32)ts.tv_nsec / NSEC_PER_MSEC; /* Convert to network byte order. */ - return htons(msecs); + return htonl(msecs); } EXPORT_SYMBOL(inet_current_timestamp); @@ -1448,6 +1461,13 @@ out_unlock: return err; } +static int ipip_gro_complete(struct sk_buff *skb, int nhoff) +{ + skb->encapsulation = 1; + skb_shinfo(skb)->gso_type |= SKB_GSO_IPIP; + return inet_gro_complete(skb, nhoff); +} + int inet_ctl_sock_create(struct sock **sk, unsigned short family, unsigned short type, unsigned char protocol, struct net *net) @@ -1675,8 +1695,8 @@ static struct packet_offload ip_packet_offload __read_mostly = { static const struct net_offload ipip_offload = { .callbacks = { .gso_segment = inet_gso_segment, - .gro_receive = inet_gro_receive, - .gro_complete = inet_gro_complete, + .gro_receive = ipip_gro_receive, + .gro_complete = ipip_gro_complete, }, }; diff --git a/net/ipv4/fib_frontend.c b/net/ipv4/fib_frontend.c index 21add55..8a9246d 100644 --- a/net/ipv4/fib_frontend.c +++ b/net/ipv4/fib_frontend.c @@ -280,7 +280,6 @@ __be32 fib_compute_spec_dst(struct sk_buff *skb) struct in_device *in_dev; struct fib_result res; struct rtable *rt; - struct flowi4 fl4; struct net *net; int scope; @@ -296,14 +295,13 @@ __be32 fib_compute_spec_dst(struct sk_buff *skb) scope = RT_SCOPE_UNIVERSE; if (!ipv4_is_zeronet(ip_hdr(skb)->saddr)) { - fl4.flowi4_oif = 0; - fl4.flowi4_iif = LOOPBACK_IFINDEX; - fl4.daddr = ip_hdr(skb)->saddr; - fl4.saddr = 0; - fl4.flowi4_tos = RT_TOS(ip_hdr(skb)->tos); - fl4.flowi4_scope = scope; - fl4.flowi4_mark = IN_DEV_SRC_VMARK(in_dev) ? skb->mark : 0; - fl4.flowi4_tun_key.tun_id = 0; + struct flowi4 fl4 = { + .flowi4_iif = LOOPBACK_IFINDEX, + .daddr = ip_hdr(skb)->saddr, + .flowi4_tos = RT_TOS(ip_hdr(skb)->tos), + .flowi4_scope = scope, + .flowi4_mark = IN_DEV_SRC_VMARK(in_dev) ? skb->mark : 0, + }; if (!fib_lookup(net, &fl4, &res, 0)) return FIB_RES_PREFSRC(net, res); } else { diff --git a/net/ipv4/fou.c b/net/ipv4/fou.c index 7804842..a0586b4 100644 --- a/net/ipv4/fou.c +++ b/net/ipv4/fou.c @@ -48,7 +48,7 @@ static inline struct fou *fou_from_sock(struct sock *sk) return sk->sk_user_data; } -static void fou_recv_pull(struct sk_buff *skb, size_t len) +static int fou_recv_pull(struct sk_buff *skb, size_t len) { struct iphdr *iph = ip_hdr(skb); @@ -59,6 +59,7 @@ static void fou_recv_pull(struct sk_buff *skb, size_t len) __skb_pull(skb, len); skb_postpull_rcsum(skb, udp_hdr(skb), len); skb_reset_transport_header(skb); + return iptunnel_pull_offloads(skb); } static int fou_udp_recv(struct sock *sk, struct sk_buff *skb) @@ -68,9 +69,14 @@ static int fou_udp_recv(struct sock *sk, struct sk_buff *skb) if (!fou) return 1; - fou_recv_pull(skb, sizeof(struct udphdr)); + if (fou_recv_pull(skb, sizeof(struct udphdr))) + goto drop; return -fou->protocol; + +drop: + kfree_skb(skb); + return 0; } static struct guehdr *gue_remcsum(struct sk_buff *skb, struct guehdr *guehdr, @@ -170,6 +176,9 @@ static int gue_udp_recv(struct sock *sk, struct sk_buff *skb) __skb_pull(skb, sizeof(struct udphdr) + hdrlen); skb_reset_transport_header(skb); + if (iptunnel_pull_offloads(skb)) + goto drop; + return -guehdr->proto_ctype; drop: diff --git a/net/ipv4/gre_offload.c b/net/ipv4/gre_offload.c index 540866d..c47539d 100644 --- a/net/ipv4/gre_offload.c +++ b/net/ipv4/gre_offload.c @@ -49,6 +49,7 @@ static struct sk_buff *gre_gso_segment(struct sk_buff *skb, /* setup inner skb. */ skb->encapsulation = 0; + SKB_GSO_CB(skb)->encap_level = 0; __skb_pull(skb, tnl_hlen); skb_reset_mac_header(skb); skb_set_network_header(skb, skb_inner_network_offset(skb)); @@ -126,6 +127,11 @@ static struct sk_buff **gre_gro_receive(struct sk_buff **head, struct packet_offload *ptype; __be16 type; + if (NAPI_GRO_CB(skb)->encap_mark) + goto out; + + NAPI_GRO_CB(skb)->encap_mark = 1; + off = skb_gro_offset(skb); hlen = off + sizeof(*greh); greh = skb_gro_header_fast(skb, off); diff --git a/net/ipv4/ip_tunnel_core.c b/net/ipv4/ip_tunnel_core.c index d27276f..02dd990 100644 --- a/net/ipv4/ip_tunnel_core.c +++ b/net/ipv4/ip_tunnel_core.c @@ -114,7 +114,8 @@ int iptunnel_pull_header(struct sk_buff *skb, int hdr_len, __be16 inner_proto, skb->vlan_tci = 0; skb_set_queue_mapping(skb, 0); skb_scrub_packet(skb, xnet); - return 0; + + return iptunnel_pull_offloads(skb); } EXPORT_SYMBOL_GPL(iptunnel_pull_header); diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index 836abe5..08eed5e 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -2070,10 +2070,14 @@ void udp_v4_early_demux(struct sk_buff *skb) if (!in_dev) return; - ours = ip_check_mc_rcu(in_dev, iph->daddr, iph->saddr, - iph->protocol); - if (!ours) - return; + /* we are supposed to accept bcast packets */ + if (skb->pkt_type == PACKET_MULTICAST) { + ours = ip_check_mc_rcu(in_dev, iph->daddr, iph->saddr, + iph->protocol); + if (!ours) + return; + } + sk = __udp4_lib_mcast_demux_lookup(net, uh->dest, iph->daddr, uh->source, iph->saddr, dif); } else if (skb->pkt_type == PACKET_HOST) { diff --git a/net/ipv4/udp_offload.c b/net/ipv4/udp_offload.c index 8a3405a..0ed2daf 100644 --- a/net/ipv4/udp_offload.c +++ b/net/ipv4/udp_offload.c @@ -56,6 +56,7 @@ static struct sk_buff *__skb_udp_tunnel_segment(struct sk_buff *skb, /* setup inner skb. */ skb->encapsulation = 0; + SKB_GSO_CB(skb)->encap_level = 0; __skb_pull(skb, tnl_hlen); skb_reset_mac_header(skb); skb_set_network_header(skb, skb_inner_network_offset(skb)); @@ -311,14 +312,14 @@ struct sk_buff **udp_gro_receive(struct sk_buff **head, struct sk_buff *skb, unsigned int off = skb_gro_offset(skb); int flush = 1; - if (NAPI_GRO_CB(skb)->udp_mark || + if (NAPI_GRO_CB(skb)->encap_mark || (skb->ip_summed != CHECKSUM_PARTIAL && NAPI_GRO_CB(skb)->csum_cnt == 0 && !NAPI_GRO_CB(skb)->csum_valid)) goto out; - /* mark that this skb passed once through the udp gro layer */ - NAPI_GRO_CB(skb)->udp_mark = 1; + /* mark that this skb passed once through the tunnel gro layer */ + NAPI_GRO_CB(skb)->encap_mark = 1; rcu_read_lock(); uo_priv = rcu_dereference(udp_offload_base); diff --git a/net/ipv6/ip6_offload.c b/net/ipv6/ip6_offload.c index eeca943..82e9f30 100644 --- a/net/ipv6/ip6_offload.c +++ b/net/ipv6/ip6_offload.c @@ -258,6 +258,19 @@ out: return pp; } +static struct sk_buff **sit_gro_receive(struct sk_buff **head, + struct sk_buff *skb) +{ + if (NAPI_GRO_CB(skb)->encap_mark) { + NAPI_GRO_CB(skb)->flush = 1; + return NULL; + } + + NAPI_GRO_CB(skb)->encap_mark = 1; + + return ipv6_gro_receive(head, skb); +} + static int ipv6_gro_complete(struct sk_buff *skb, int nhoff) { const struct net_offload *ops; @@ -302,7 +315,7 @@ static struct packet_offload ipv6_packet_offload __read_mostly = { static const struct net_offload sit_offload = { .callbacks = { .gso_segment = ipv6_gso_segment, - .gro_receive = ipv6_gro_receive, + .gro_receive = sit_gro_receive, .gro_complete = sit_gro_complete, }, }; diff --git a/net/ipv6/ping.c b/net/ipv6/ping.c index 263a516..c382db7 100644 --- a/net/ipv6/ping.c +++ b/net/ipv6/ping.c @@ -26,35 +26,6 @@ #include <net/transp_v6.h> #include <net/ping.h> -struct proto pingv6_prot = { - .name = "PINGv6", - .owner = THIS_MODULE, - .init = ping_init_sock, - .close = ping_close, - .connect = ip6_datagram_connect_v6_only, - .disconnect = udp_disconnect, - .setsockopt = ipv6_setsockopt, - .getsockopt = ipv6_getsockopt, - .sendmsg = ping_v6_sendmsg, - .recvmsg = ping_recvmsg, - .bind = ping_bind, - .backlog_rcv = ping_queue_rcv_skb, - .hash = ping_hash, - .unhash = ping_unhash, - .get_port = ping_get_port, - .obj_size = sizeof(struct raw6_sock), -}; -EXPORT_SYMBOL_GPL(pingv6_prot); - -static struct inet_protosw pingv6_protosw = { - .type = SOCK_DGRAM, - .protocol = IPPROTO_ICMPV6, - .prot = &pingv6_prot, - .ops = &inet6_dgram_ops, - .flags = INET_PROTOSW_REUSE, -}; - - /* Compatibility glue so we can support IPv6 when it's compiled as a module */ static int dummy_ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len) @@ -77,7 +48,7 @@ static int dummy_ipv6_chk_addr(struct net *net, const struct in6_addr *addr, return 0; } -int ping_v6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) +static int ping_v6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) { struct inet_sock *inet = inet_sk(sk); struct ipv6_pinfo *np = inet6_sk(sk); @@ -192,6 +163,34 @@ int ping_v6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) return len; } +struct proto pingv6_prot = { + .name = "PINGv6", + .owner = THIS_MODULE, + .init = ping_init_sock, + .close = ping_close, + .connect = ip6_datagram_connect_v6_only, + .disconnect = udp_disconnect, + .setsockopt = ipv6_setsockopt, + .getsockopt = ipv6_getsockopt, + .sendmsg = ping_v6_sendmsg, + .recvmsg = ping_recvmsg, + .bind = ping_bind, + .backlog_rcv = ping_queue_rcv_skb, + .hash = ping_hash, + .unhash = ping_unhash, + .get_port = ping_get_port, + .obj_size = sizeof(struct raw6_sock), +}; +EXPORT_SYMBOL_GPL(pingv6_prot); + +static struct inet_protosw pingv6_protosw = { + .type = SOCK_DGRAM, + .protocol = IPPROTO_ICMPV6, + .prot = &pingv6_prot, + .ops = &inet6_dgram_ops, + .flags = INET_PROTOSW_REUSE, +}; + #ifdef CONFIG_PROC_FS static void *ping_v6_seq_start(struct seq_file *seq, loff_t *pos) { diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c index f45b8ff..8338430 100644 --- a/net/ipv6/sit.c +++ b/net/ipv6/sit.c @@ -681,14 +681,16 @@ static int ipip6_rcv(struct sk_buff *skb) skb->mac_header = skb->network_header; skb_reset_network_header(skb); IPCB(skb)->flags = 0; - skb->protocol = htons(ETH_P_IPV6); + skb->dev = tunnel->dev; if (packet_is_spoofed(skb, iph, tunnel)) { tunnel->dev->stats.rx_errors++; goto out; } - __skb_tunnel_rx(skb, tunnel->dev, tunnel->net); + if (iptunnel_pull_header(skb, 0, htons(ETH_P_IPV6), + !net_eq(tunnel->net, dev_net(tunnel->dev)))) + goto out; err = IP_ECN_decapsulate(iph, skb); if (unlikely(err)) { diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c index c841679..215fc08 100644 --- a/net/netlink/af_netlink.c +++ b/net/netlink/af_netlink.c @@ -1033,6 +1033,14 @@ static int netlink_getname(struct socket *sock, struct sockaddr *addr, return 0; } +static int netlink_ioctl(struct socket *sock, unsigned int cmd, + unsigned long arg) +{ + /* try to hand this ioctl down to the NIC drivers. + */ + return -ENOIOCTLCMD; +} + static struct sock *netlink_getsockbyportid(struct sock *ssk, u32 portid) { struct sock *sock; @@ -2494,7 +2502,7 @@ static const struct proto_ops netlink_ops = { .accept = sock_no_accept, .getname = netlink_getname, .poll = datagram_poll, - .ioctl = sock_no_ioctl, + .ioctl = netlink_ioctl, .listen = sock_no_listen, .shutdown = sock_no_shutdown, .setsockopt = netlink_setsockopt, diff --git a/net/sctp/associola.c b/net/sctp/associola.c index a19b3e6..e1849f3 100644 --- a/net/sctp/associola.c +++ b/net/sctp/associola.c @@ -1406,7 +1406,8 @@ void sctp_assoc_sync_pmtu(struct sock *sk, struct sctp_association *asoc) list_for_each_entry(t, &asoc->peer.transport_addr_list, transports) { if (t->pmtu_pending && t->dst) { - sctp_transport_update_pmtu(sk, t, dst_mtu(t->dst)); + sctp_transport_update_pmtu(sk, t, + WORD_TRUNC(dst_mtu(t->dst))); t->pmtu_pending = 0; } if (!pmtu || (t->pathmtu < pmtu)) diff --git a/net/sctp/bind_addr.c b/net/sctp/bind_addr.c index 871cdf9..401c607 100644 --- a/net/sctp/bind_addr.c +++ b/net/sctp/bind_addr.c @@ -111,7 +111,8 @@ int sctp_bind_addr_dup(struct sctp_bind_addr *dest, dest->port = src->port; list_for_each_entry(addr, &src->address_list, list) { - error = sctp_add_bind_addr(dest, &addr->a, 1, gfp); + error = sctp_add_bind_addr(dest, &addr->a, sizeof(addr->a), + 1, gfp); if (error < 0) break; } @@ -150,7 +151,7 @@ void sctp_bind_addr_free(struct sctp_bind_addr *bp) /* Add an address to the bind address list in the SCTP_bind_addr structure. */ int sctp_add_bind_addr(struct sctp_bind_addr *bp, union sctp_addr *new, - __u8 addr_state, gfp_t gfp) + int new_size, __u8 addr_state, gfp_t gfp) { struct sctp_sockaddr_entry *addr; @@ -159,7 +160,7 @@ int sctp_add_bind_addr(struct sctp_bind_addr *bp, union sctp_addr *new, if (!addr) return -ENOMEM; - memcpy(&addr->a, new, sizeof(*new)); + memcpy(&addr->a, new, min_t(size_t, sizeof(*new), new_size)); /* Fix up the port if it has not yet been set. * Both v4 and v6 have the port at the same offset. @@ -291,7 +292,8 @@ int sctp_raw_to_bind_addrs(struct sctp_bind_addr *bp, __u8 *raw_addr_list, } af->from_addr_param(&addr, rawaddr, htons(port), 0); - retval = sctp_add_bind_addr(bp, &addr, SCTP_ADDR_SRC, gfp); + retval = sctp_add_bind_addr(bp, &addr, sizeof(addr), + SCTP_ADDR_SRC, gfp); if (retval) { /* Can't finish building the list, clean up. */ sctp_bind_addr_clean(bp); @@ -453,8 +455,8 @@ static int sctp_copy_one_addr(struct net *net, struct sctp_bind_addr *dest, (((AF_INET6 == addr->sa.sa_family) && (flags & SCTP_ADDR6_ALLOWED) && (flags & SCTP_ADDR6_PEERSUPP)))) - error = sctp_add_bind_addr(dest, addr, SCTP_ADDR_SRC, - gfp); + error = sctp_add_bind_addr(dest, addr, sizeof(*addr), + SCTP_ADDR_SRC, gfp); } return error; diff --git a/net/sctp/input.c b/net/sctp/input.c index db76f1a..00b84453 100644 --- a/net/sctp/input.c +++ b/net/sctp/input.c @@ -606,7 +606,8 @@ void sctp_v4_err(struct sk_buff *skb, __u32 info) /* PMTU discovery (RFC1191) */ if (ICMP_FRAG_NEEDED == code) { - sctp_icmp_frag_needed(sk, asoc, transport, info); + sctp_icmp_frag_needed(sk, asoc, transport, + WORD_TRUNC(info)); goto out_unlock; } else { if (ICMP_PROT_UNREACH == code) { diff --git a/net/sctp/outqueue.c b/net/sctp/outqueue.c index f03541d..8d3d362 100644 --- a/net/sctp/outqueue.c +++ b/net/sctp/outqueue.c @@ -978,8 +978,12 @@ static int sctp_outq_flush(struct sctp_outq *q, int rtx_timeout, gfp_t gfp) (new_transport->state == SCTP_UNCONFIRMED) || (new_transport->state == SCTP_PF))) new_transport = asoc->peer.active_path; - if (new_transport->state == SCTP_UNCONFIRMED) + if (new_transport->state == SCTP_UNCONFIRMED) { + WARN_ONCE(1, "Atempt to send packet on unconfirmed path."); + sctp_chunk_fail(chunk, 0); + sctp_chunk_free(chunk); continue; + } /* Change packets if necessary. */ if (new_transport != transport) { diff --git a/net/sctp/protocol.c b/net/sctp/protocol.c index 1099e99..d3d50da 100644 --- a/net/sctp/protocol.c +++ b/net/sctp/protocol.c @@ -216,6 +216,7 @@ int sctp_copy_local_addr_list(struct net *net, struct sctp_bind_addr *bp, (copy_flags & SCTP_ADDR6_ALLOWED) && (copy_flags & SCTP_ADDR6_PEERSUPP)))) { error = sctp_add_bind_addr(bp, &addr->a, + sizeof(addr->a), SCTP_ADDR_SRC, GFP_ATOMIC); if (error) goto end_copy; diff --git a/net/sctp/sm_make_chunk.c b/net/sctp/sm_make_chunk.c index e47abf2..7f0bf79 100644 --- a/net/sctp/sm_make_chunk.c +++ b/net/sctp/sm_make_chunk.c @@ -1849,7 +1849,8 @@ no_hmac: /* Also, add the destination address. */ if (list_empty(&retval->base.bind_addr.address_list)) { sctp_add_bind_addr(&retval->base.bind_addr, &chunk->dest, - SCTP_ADDR_SRC, GFP_ATOMIC); + sizeof(chunk->dest), SCTP_ADDR_SRC, + GFP_ATOMIC); } retval->next_tsn = retval->c.initial_tsn; diff --git a/net/sctp/sm_sideeffect.c b/net/sctp/sm_sideeffect.c index 3c22c41..7fe56d0 100644 --- a/net/sctp/sm_sideeffect.c +++ b/net/sctp/sm_sideeffect.c @@ -215,10 +215,14 @@ static int sctp_gen_sack(struct sctp_association *asoc, int force, sctp_add_cmd_sf(commands, SCTP_CMD_TIMER_RESTART, SCTP_TO(SCTP_EVENT_TIMEOUT_SACK)); } else { + __u32 old_a_rwnd = asoc->a_rwnd; + asoc->a_rwnd = asoc->rwnd; sack = sctp_make_sack(asoc); - if (!sack) + if (!sack) { + asoc->a_rwnd = old_a_rwnd; goto nomem; + } asoc->peer.sack_needed = 0; asoc->peer.sack_cnt = 0; diff --git a/net/sctp/socket.c b/net/sctp/socket.c index 96e0811..878d28e 100644 --- a/net/sctp/socket.c +++ b/net/sctp/socket.c @@ -386,7 +386,8 @@ static int sctp_do_bind(struct sock *sk, union sctp_addr *addr, int len) /* Add the address to the bind address list. * Use GFP_ATOMIC since BHs will be disabled. */ - ret = sctp_add_bind_addr(bp, addr, SCTP_ADDR_SRC, GFP_ATOMIC); + ret = sctp_add_bind_addr(bp, addr, af->sockaddr_len, + SCTP_ADDR_SRC, GFP_ATOMIC); /* Copy back into socket for getsockname() use. */ if (!ret) { @@ -577,6 +578,7 @@ static int sctp_send_asconf_add_ip(struct sock *sk, af = sctp_get_af_specific(addr->v4.sin_family); memcpy(&saveaddr, addr, af->sockaddr_len); retval = sctp_add_bind_addr(bp, &saveaddr, + sizeof(saveaddr), SCTP_ADDR_NEW, GFP_ATOMIC); addr_buf += af->sockaddr_len; } @@ -1389,7 +1391,7 @@ static int sctp_getsockopt_connectx3(struct sock *sk, int len, int err = 0; #ifdef CONFIG_COMPAT - if (is_compat_task()) { + if (in_compat_syscall()) { struct compat_sctp_getaddrs_old param32; if (len < sizeof(param32)) diff --git a/net/sctp/transport.c b/net/sctp/transport.c index d517153..9b6b48c 100644 --- a/net/sctp/transport.c +++ b/net/sctp/transport.c @@ -226,7 +226,7 @@ void sctp_transport_pmtu(struct sctp_transport *transport, struct sock *sk) } if (transport->dst) { - transport->pathmtu = dst_mtu(transport->dst); + transport->pathmtu = WORD_TRUNC(dst_mtu(transport->dst)); } else transport->pathmtu = SCTP_DEFAULT_MAXSEGMENT; } @@ -280,7 +280,7 @@ void sctp_transport_route(struct sctp_transport *transport, return; } if (transport->dst) { - transport->pathmtu = dst_mtu(transport->dst); + transport->pathmtu = WORD_TRUNC(dst_mtu(transport->dst)); /* Initialize sk->sk_rcv_saddr, if the transport is the * association's active path for getsockname(). diff --git a/net/sunrpc/Makefile b/net/sunrpc/Makefile index b512fbd..ea7ffa1 100644 --- a/net/sunrpc/Makefile +++ b/net/sunrpc/Makefile @@ -12,7 +12,8 @@ sunrpc-y := clnt.o xprt.o socklib.o xprtsock.o sched.o \ svc.o svcsock.o svcauth.o svcauth_unix.o \ addr.o rpcb_clnt.o timer.o xdr.o \ sunrpc_syms.o cache.o rpc_pipe.o \ - svc_xprt.o + svc_xprt.o \ + xprtmultipath.o sunrpc-$(CONFIG_SUNRPC_DEBUG) += debugfs.o sunrpc-$(CONFIG_SUNRPC_BACKCHANNEL) += backchannel_rqst.o sunrpc-$(CONFIG_PROC_FS) += stats.o diff --git a/net/sunrpc/auth_gss/auth_gss.c b/net/sunrpc/auth_gss/auth_gss.c index cabf586..8c6bc79 100644 --- a/net/sunrpc/auth_gss/auth_gss.c +++ b/net/sunrpc/auth_gss/auth_gss.c @@ -1181,12 +1181,12 @@ static struct rpc_auth * gss_create(struct rpc_auth_create_args *args, struct rpc_clnt *clnt) { struct gss_auth *gss_auth; - struct rpc_xprt *xprt = rcu_access_pointer(clnt->cl_xprt); + struct rpc_xprt_switch *xps = rcu_access_pointer(clnt->cl_xpi.xpi_xpswitch); while (clnt != clnt->cl_parent) { struct rpc_clnt *parent = clnt->cl_parent; /* Find the original parent for this transport */ - if (rcu_access_pointer(parent->cl_xprt) != xprt) + if (rcu_access_pointer(parent->cl_xpi.xpi_xpswitch) != xps) break; clnt = parent; } diff --git a/net/sunrpc/auth_null.c b/net/sunrpc/auth_null.c index c2a2b58..8d9eb4d 100644 --- a/net/sunrpc/auth_null.c +++ b/net/sunrpc/auth_null.c @@ -113,8 +113,8 @@ const struct rpc_authops authnull_ops = { static struct rpc_auth null_auth = { - .au_cslack = 4, - .au_rslack = 2, + .au_cslack = NUL_CALLSLACK, + .au_rslack = NUL_REPLYSLACK, .au_ops = &authnull_ops, .au_flavor = RPC_AUTH_NULL, .au_count = ATOMIC_INIT(0), diff --git a/net/sunrpc/auth_unix.c b/net/sunrpc/auth_unix.c index 548240d..0d3dd36 100644 --- a/net/sunrpc/auth_unix.c +++ b/net/sunrpc/auth_unix.c @@ -23,8 +23,6 @@ struct unx_cred { }; #define uc_uid uc_base.cr_uid -#define UNX_WRITESLACK (21 + XDR_QUADLEN(UNX_MAXNODENAME)) - #if IS_ENABLED(CONFIG_SUNRPC_DEBUG) # define RPCDBG_FACILITY RPCDBG_AUTH #endif @@ -228,8 +226,8 @@ const struct rpc_authops authunix_ops = { static struct rpc_auth unix_auth = { - .au_cslack = UNX_WRITESLACK, - .au_rslack = 2, /* assume AUTH_NULL verf */ + .au_cslack = UNX_CALLSLACK, + .au_rslack = NUL_REPLYSLACK, .au_ops = &authunix_ops, .au_flavor = RPC_AUTH_UNIX, .au_count = ATOMIC_INIT(0), diff --git a/net/sunrpc/cache.c b/net/sunrpc/cache.c index 273bc3a..008c25d 100644 --- a/net/sunrpc/cache.c +++ b/net/sunrpc/cache.c @@ -1182,14 +1182,14 @@ int sunrpc_cache_pipe_upcall(struct cache_detail *detail, struct cache_head *h) } crq->q.reader = 0; - crq->item = cache_get(h); crq->buf = buf; crq->len = 0; crq->readers = 0; spin_lock(&queue_lock); - if (test_bit(CACHE_PENDING, &h->flags)) + if (test_bit(CACHE_PENDING, &h->flags)) { + crq->item = cache_get(h); list_add_tail(&crq->q.list, &detail->queue); - else + } else /* Lost a race, no longer PENDING, so don't enqueue */ ret = -EAGAIN; spin_unlock(&queue_lock); diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c index b7f2104..7e0c9bf 100644 --- a/net/sunrpc/clnt.c +++ b/net/sunrpc/clnt.c @@ -354,6 +354,7 @@ static void rpc_free_clid(struct rpc_clnt *clnt) } static struct rpc_clnt * rpc_new_client(const struct rpc_create_args *args, + struct rpc_xprt_switch *xps, struct rpc_xprt *xprt, struct rpc_clnt *parent) { @@ -411,6 +412,8 @@ static struct rpc_clnt * rpc_new_client(const struct rpc_create_args *args, } rpc_clnt_set_transport(clnt, xprt, timeout); + xprt_iter_init(&clnt->cl_xpi, xps); + xprt_switch_put(xps); clnt->cl_rtt = &clnt->cl_rtt_default; rpc_init_rtt(&clnt->cl_rtt_default, clnt->cl_timeout->to_initval); @@ -438,6 +441,7 @@ out_no_clid: out_err: rpciod_down(); out_no_rpciod: + xprt_switch_put(xps); xprt_put(xprt); return ERR_PTR(err); } @@ -446,8 +450,13 @@ struct rpc_clnt *rpc_create_xprt(struct rpc_create_args *args, struct rpc_xprt *xprt) { struct rpc_clnt *clnt = NULL; + struct rpc_xprt_switch *xps; - clnt = rpc_new_client(args, xprt, NULL); + xps = xprt_switch_alloc(xprt, GFP_KERNEL); + if (xps == NULL) + return ERR_PTR(-ENOMEM); + + clnt = rpc_new_client(args, xps, xprt, NULL); if (IS_ERR(clnt)) return clnt; @@ -564,6 +573,7 @@ EXPORT_SYMBOL_GPL(rpc_create); static struct rpc_clnt *__rpc_clone_client(struct rpc_create_args *args, struct rpc_clnt *clnt) { + struct rpc_xprt_switch *xps; struct rpc_xprt *xprt; struct rpc_clnt *new; int err; @@ -571,13 +581,17 @@ static struct rpc_clnt *__rpc_clone_client(struct rpc_create_args *args, err = -ENOMEM; rcu_read_lock(); xprt = xprt_get(rcu_dereference(clnt->cl_xprt)); + xps = xprt_switch_get(rcu_dereference(clnt->cl_xpi.xpi_xpswitch)); rcu_read_unlock(); - if (xprt == NULL) + if (xprt == NULL || xps == NULL) { + xprt_put(xprt); + xprt_switch_put(xps); goto out_err; + } args->servername = xprt->servername; args->nodename = clnt->cl_nodename; - new = rpc_new_client(args, xprt, clnt); + new = rpc_new_client(args, xps, xprt, clnt); if (IS_ERR(new)) { err = PTR_ERR(new); goto out_err; @@ -657,6 +671,7 @@ int rpc_switch_client_transport(struct rpc_clnt *clnt, { const struct rpc_timeout *old_timeo; rpc_authflavor_t pseudoflavor; + struct rpc_xprt_switch *xps, *oldxps; struct rpc_xprt *xprt, *old; struct rpc_clnt *parent; int err; @@ -668,10 +683,17 @@ int rpc_switch_client_transport(struct rpc_clnt *clnt, return PTR_ERR(xprt); } + xps = xprt_switch_alloc(xprt, GFP_KERNEL); + if (xps == NULL) { + xprt_put(xprt); + return -ENOMEM; + } + pseudoflavor = clnt->cl_auth->au_flavor; old_timeo = clnt->cl_timeout; old = rpc_clnt_set_transport(clnt, xprt, timeout); + oldxps = xprt_iter_xchg_switch(&clnt->cl_xpi, xps); rpc_unregister_client(clnt); __rpc_clnt_remove_pipedir(clnt); @@ -697,20 +719,74 @@ int rpc_switch_client_transport(struct rpc_clnt *clnt, synchronize_rcu(); if (parent != clnt) rpc_release_client(parent); + xprt_switch_put(oldxps); xprt_put(old); dprintk("RPC: replaced xprt for clnt %p\n", clnt); return 0; out_revert: + xps = xprt_iter_xchg_switch(&clnt->cl_xpi, oldxps); rpc_clnt_set_transport(clnt, old, old_timeo); clnt->cl_parent = parent; rpc_client_register(clnt, pseudoflavor, NULL); + xprt_switch_put(xps); xprt_put(xprt); dprintk("RPC: failed to switch xprt for clnt %p\n", clnt); return err; } EXPORT_SYMBOL_GPL(rpc_switch_client_transport); +static +int rpc_clnt_xprt_iter_init(struct rpc_clnt *clnt, struct rpc_xprt_iter *xpi) +{ + struct rpc_xprt_switch *xps; + + rcu_read_lock(); + xps = xprt_switch_get(rcu_dereference(clnt->cl_xpi.xpi_xpswitch)); + rcu_read_unlock(); + if (xps == NULL) + return -EAGAIN; + xprt_iter_init_listall(xpi, xps); + xprt_switch_put(xps); + return 0; +} + +/** + * rpc_clnt_iterate_for_each_xprt - Apply a function to all transports + * @clnt: pointer to client + * @fn: function to apply + * @data: void pointer to function data + * + * Iterates through the list of RPC transports currently attached to the + * client and applies the function fn(clnt, xprt, data). + * + * On error, the iteration stops, and the function returns the error value. + */ +int rpc_clnt_iterate_for_each_xprt(struct rpc_clnt *clnt, + int (*fn)(struct rpc_clnt *, struct rpc_xprt *, void *), + void *data) +{ + struct rpc_xprt_iter xpi; + int ret; + + ret = rpc_clnt_xprt_iter_init(clnt, &xpi); + if (ret) + return ret; + for (;;) { + struct rpc_xprt *xprt = xprt_iter_get_next(&xpi); + + if (!xprt) + break; + ret = fn(clnt, xprt, data); + xprt_put(xprt); + if (ret < 0) + break; + } + xprt_iter_destroy(&xpi); + return ret; +} +EXPORT_SYMBOL_GPL(rpc_clnt_iterate_for_each_xprt); + /* * Kill all tasks for the given client. * XXX: kill their descendants as well? @@ -783,6 +859,7 @@ rpc_free_client(struct rpc_clnt *clnt) rpc_free_iostats(clnt->cl_metrics); clnt->cl_metrics = NULL; xprt_put(rcu_dereference_raw(clnt->cl_xprt)); + xprt_iter_destroy(&clnt->cl_xpi); rpciod_down(); rpc_free_clid(clnt); kfree(clnt); @@ -868,6 +945,7 @@ EXPORT_SYMBOL_GPL(rpc_bind_new_program); void rpc_task_release_client(struct rpc_task *task) { struct rpc_clnt *clnt = task->tk_client; + struct rpc_xprt *xprt = task->tk_xprt; if (clnt != NULL) { /* Remove from client task list */ @@ -878,13 +956,22 @@ void rpc_task_release_client(struct rpc_task *task) rpc_release_client(clnt); } + + if (xprt != NULL) { + task->tk_xprt = NULL; + + xprt_put(xprt); + } } static void rpc_task_set_client(struct rpc_task *task, struct rpc_clnt *clnt) { + if (clnt != NULL) { rpc_task_release_client(task); + if (task->tk_xprt == NULL) + task->tk_xprt = xprt_iter_get_next(&clnt->cl_xpi); task->tk_client = clnt; atomic_inc(&clnt->cl_count); if (clnt->cl_softrtry) @@ -900,14 +987,6 @@ void rpc_task_set_client(struct rpc_task *task, struct rpc_clnt *clnt) } } -void rpc_task_reset_client(struct rpc_task *task, struct rpc_clnt *clnt) -{ - rpc_task_release_client(task); - rpc_task_set_client(task, clnt); -} -EXPORT_SYMBOL_GPL(rpc_task_reset_client); - - static void rpc_task_set_rpc_message(struct rpc_task *task, const struct rpc_message *msg) { @@ -2104,11 +2183,9 @@ call_timeout(struct rpc_task *task) } if (RPC_IS_SOFT(task)) { if (clnt->cl_chatty) { - rcu_read_lock(); printk(KERN_NOTICE "%s: server %s not responding, timed out\n", clnt->cl_program->name, - rcu_dereference(clnt->cl_xprt)->servername); - rcu_read_unlock(); + task->tk_xprt->servername); } if (task->tk_flags & RPC_TASK_TIMEOUT) rpc_exit(task, -ETIMEDOUT); @@ -2120,11 +2197,9 @@ call_timeout(struct rpc_task *task) if (!(task->tk_flags & RPC_CALL_MAJORSEEN)) { task->tk_flags |= RPC_CALL_MAJORSEEN; if (clnt->cl_chatty) { - rcu_read_lock(); printk(KERN_NOTICE "%s: server %s not responding, still trying\n", clnt->cl_program->name, - rcu_dereference(clnt->cl_xprt)->servername); - rcu_read_unlock(); + task->tk_xprt->servername); } } rpc_force_rebind(clnt); @@ -2154,11 +2229,9 @@ call_decode(struct rpc_task *task) if (task->tk_flags & RPC_CALL_MAJORSEEN) { if (clnt->cl_chatty) { - rcu_read_lock(); printk(KERN_NOTICE "%s: server %s OK\n", clnt->cl_program->name, - rcu_dereference(clnt->cl_xprt)->servername); - rcu_read_unlock(); + task->tk_xprt->servername); } task->tk_flags &= ~RPC_CALL_MAJORSEEN; } @@ -2312,11 +2385,9 @@ rpc_verify_header(struct rpc_task *task) task->tk_action = call_bind; goto out_retry; case RPC_AUTH_TOOWEAK: - rcu_read_lock(); printk(KERN_NOTICE "RPC: server %s requires stronger " "authentication.\n", - rcu_dereference(clnt->cl_xprt)->servername); - rcu_read_unlock(); + task->tk_xprt->servername); break; default: dprintk("RPC: %5u %s: unknown auth error: %x\n", @@ -2341,27 +2412,27 @@ rpc_verify_header(struct rpc_task *task) case RPC_SUCCESS: return p; case RPC_PROG_UNAVAIL: - dprintk_rcu("RPC: %5u %s: program %u is unsupported " + dprintk("RPC: %5u %s: program %u is unsupported " "by server %s\n", task->tk_pid, __func__, (unsigned int)clnt->cl_prog, - rcu_dereference(clnt->cl_xprt)->servername); + task->tk_xprt->servername); error = -EPFNOSUPPORT; goto out_err; case RPC_PROG_MISMATCH: - dprintk_rcu("RPC: %5u %s: program %u, version %u unsupported " + dprintk("RPC: %5u %s: program %u, version %u unsupported " "by server %s\n", task->tk_pid, __func__, (unsigned int)clnt->cl_prog, (unsigned int)clnt->cl_vers, - rcu_dereference(clnt->cl_xprt)->servername); + task->tk_xprt->servername); error = -EPROTONOSUPPORT; goto out_err; case RPC_PROC_UNAVAIL: - dprintk_rcu("RPC: %5u %s: proc %s unsupported by program %u, " + dprintk("RPC: %5u %s: proc %s unsupported by program %u, " "version %u on server %s\n", task->tk_pid, __func__, rpc_proc_name(task), clnt->cl_prog, clnt->cl_vers, - rcu_dereference(clnt->cl_xprt)->servername); + task->tk_xprt->servername); error = -EOPNOTSUPP; goto out_err; case RPC_GARBAGE_ARGS: @@ -2421,7 +2492,10 @@ static int rpc_ping(struct rpc_clnt *clnt) return err; } -struct rpc_task *rpc_call_null(struct rpc_clnt *clnt, struct rpc_cred *cred, int flags) +static +struct rpc_task *rpc_call_null_helper(struct rpc_clnt *clnt, + struct rpc_xprt *xprt, struct rpc_cred *cred, int flags, + const struct rpc_call_ops *ops, void *data) { struct rpc_message msg = { .rpc_proc = &rpcproc_null, @@ -2429,14 +2503,140 @@ struct rpc_task *rpc_call_null(struct rpc_clnt *clnt, struct rpc_cred *cred, int }; struct rpc_task_setup task_setup_data = { .rpc_client = clnt, + .rpc_xprt = xprt, .rpc_message = &msg, - .callback_ops = &rpc_default_ops, + .callback_ops = (ops != NULL) ? ops : &rpc_default_ops, + .callback_data = data, .flags = flags, }; + return rpc_run_task(&task_setup_data); } + +struct rpc_task *rpc_call_null(struct rpc_clnt *clnt, struct rpc_cred *cred, int flags) +{ + return rpc_call_null_helper(clnt, NULL, cred, flags, NULL, NULL); +} EXPORT_SYMBOL_GPL(rpc_call_null); +struct rpc_cb_add_xprt_calldata { + struct rpc_xprt_switch *xps; + struct rpc_xprt *xprt; +}; + +static void rpc_cb_add_xprt_done(struct rpc_task *task, void *calldata) +{ + struct rpc_cb_add_xprt_calldata *data = calldata; + + if (task->tk_status == 0) + rpc_xprt_switch_add_xprt(data->xps, data->xprt); +} + +static void rpc_cb_add_xprt_release(void *calldata) +{ + struct rpc_cb_add_xprt_calldata *data = calldata; + + xprt_put(data->xprt); + xprt_switch_put(data->xps); + kfree(data); +} + +const static struct rpc_call_ops rpc_cb_add_xprt_call_ops = { + .rpc_call_done = rpc_cb_add_xprt_done, + .rpc_release = rpc_cb_add_xprt_release, +}; + +/** + * rpc_clnt_test_and_add_xprt - Test and add a new transport to a rpc_clnt + * @clnt: pointer to struct rpc_clnt + * @xps: pointer to struct rpc_xprt_switch, + * @xprt: pointer struct rpc_xprt + * @dummy: unused + */ +int rpc_clnt_test_and_add_xprt(struct rpc_clnt *clnt, + struct rpc_xprt_switch *xps, struct rpc_xprt *xprt, + void *dummy) +{ + struct rpc_cb_add_xprt_calldata *data; + struct rpc_cred *cred; + struct rpc_task *task; + + data = kmalloc(sizeof(*data), GFP_NOFS); + if (!data) + return -ENOMEM; + data->xps = xprt_switch_get(xps); + data->xprt = xprt_get(xprt); + + cred = authnull_ops.lookup_cred(NULL, NULL, 0); + task = rpc_call_null_helper(clnt, xprt, cred, + RPC_TASK_SOFT|RPC_TASK_SOFTCONN|RPC_TASK_ASYNC, + &rpc_cb_add_xprt_call_ops, data); + put_rpccred(cred); + if (IS_ERR(task)) + return PTR_ERR(task); + rpc_put_task(task); + return 1; +} +EXPORT_SYMBOL_GPL(rpc_clnt_test_and_add_xprt); + +/** + * rpc_clnt_add_xprt - Add a new transport to a rpc_clnt + * @clnt: pointer to struct rpc_clnt + * @xprtargs: pointer to struct xprt_create + * @setup: callback to test and/or set up the connection + * @data: pointer to setup function data + * + * Creates a new transport using the parameters set in args and + * adds it to clnt. + * If ping is set, then test that connectivity succeeds before + * adding the new transport. + * + */ +int rpc_clnt_add_xprt(struct rpc_clnt *clnt, + struct xprt_create *xprtargs, + int (*setup)(struct rpc_clnt *, + struct rpc_xprt_switch *, + struct rpc_xprt *, + void *), + void *data) +{ + struct rpc_xprt_switch *xps; + struct rpc_xprt *xprt; + unsigned char resvport; + int ret = 0; + + rcu_read_lock(); + xps = xprt_switch_get(rcu_dereference(clnt->cl_xpi.xpi_xpswitch)); + xprt = xprt_iter_xprt(&clnt->cl_xpi); + if (xps == NULL || xprt == NULL) { + rcu_read_unlock(); + return -EAGAIN; + } + resvport = xprt->resvport; + rcu_read_unlock(); + + xprt = xprt_create_transport(xprtargs); + if (IS_ERR(xprt)) { + ret = PTR_ERR(xprt); + goto out_put_switch; + } + xprt->resvport = resvport; + + rpc_xprt_switch_set_roundrobin(xps); + if (setup) { + ret = setup(clnt, xps, xprt, data); + if (ret != 0) + goto out_put_xprt; + } + rpc_xprt_switch_add_xprt(xps, xprt); +out_put_xprt: + xprt_put(xprt); +out_put_switch: + xprt_switch_put(xps); + return ret; +} +EXPORT_SYMBOL_GPL(rpc_clnt_add_xprt); + #if IS_ENABLED(CONFIG_SUNRPC_DEBUG) static void rpc_show_header(void) { @@ -2483,57 +2683,39 @@ void rpc_show_tasks(struct net *net) #endif #if IS_ENABLED(CONFIG_SUNRPC_SWAP) +static int +rpc_clnt_swap_activate_callback(struct rpc_clnt *clnt, + struct rpc_xprt *xprt, + void *dummy) +{ + return xprt_enable_swap(xprt); +} + int rpc_clnt_swap_activate(struct rpc_clnt *clnt) { - int ret = 0; - struct rpc_xprt *xprt; - - if (atomic_inc_return(&clnt->cl_swapper) == 1) { -retry: - rcu_read_lock(); - xprt = xprt_get(rcu_dereference(clnt->cl_xprt)); - rcu_read_unlock(); - if (!xprt) { - /* - * If we didn't get a reference, then we likely are - * racing with a migration event. Wait for a grace - * period and try again. - */ - synchronize_rcu(); - goto retry; - } - - ret = xprt_enable_swap(xprt); - xprt_put(xprt); - } - return ret; + if (atomic_inc_return(&clnt->cl_swapper) == 1) + return rpc_clnt_iterate_for_each_xprt(clnt, + rpc_clnt_swap_activate_callback, NULL); + return 0; } EXPORT_SYMBOL_GPL(rpc_clnt_swap_activate); +static int +rpc_clnt_swap_deactivate_callback(struct rpc_clnt *clnt, + struct rpc_xprt *xprt, + void *dummy) +{ + xprt_disable_swap(xprt); + return 0; +} + void rpc_clnt_swap_deactivate(struct rpc_clnt *clnt) { - struct rpc_xprt *xprt; - - if (atomic_dec_if_positive(&clnt->cl_swapper) == 0) { -retry: - rcu_read_lock(); - xprt = xprt_get(rcu_dereference(clnt->cl_xprt)); - rcu_read_unlock(); - if (!xprt) { - /* - * If we didn't get a reference, then we likely are - * racing with a migration event. Wait for a grace - * period and try again. - */ - synchronize_rcu(); - goto retry; - } - - xprt_disable_swap(xprt); - xprt_put(xprt); - } + if (atomic_dec_if_positive(&clnt->cl_swapper) == 0) + rpc_clnt_iterate_for_each_xprt(clnt, + rpc_clnt_swap_deactivate_callback, NULL); } EXPORT_SYMBOL_GPL(rpc_clnt_swap_deactivate); #endif /* CONFIG_SUNRPC_SWAP */ diff --git a/net/sunrpc/rpcb_clnt.c b/net/sunrpc/rpcb_clnt.c index cf5770d..5b30603 100644 --- a/net/sunrpc/rpcb_clnt.c +++ b/net/sunrpc/rpcb_clnt.c @@ -648,10 +648,10 @@ static struct rpc_task *rpcb_call_async(struct rpc_clnt *rpcb_clnt, struct rpcbi static struct rpc_clnt *rpcb_find_transport_owner(struct rpc_clnt *clnt) { struct rpc_clnt *parent = clnt->cl_parent; - struct rpc_xprt *xprt = rcu_dereference(clnt->cl_xprt); + struct rpc_xprt_switch *xps = rcu_access_pointer(clnt->cl_xpi.xpi_xpswitch); while (parent != clnt) { - if (rcu_dereference(parent->cl_xprt) != xprt) + if (rcu_access_pointer(parent->cl_xpi.xpi_xpswitch) != xps) break; if (clnt->cl_autobind) break; @@ -683,11 +683,9 @@ void rpcb_getport_async(struct rpc_task *task) int status; rcu_read_lock(); - do { - clnt = rpcb_find_transport_owner(task->tk_client); - xprt = xprt_get(rcu_dereference(clnt->cl_xprt)); - } while (xprt == NULL); + clnt = rpcb_find_transport_owner(task->tk_client); rcu_read_unlock(); + xprt = xprt_get(task->tk_xprt); dprintk("RPC: %5u %s(%s, %u, %u, %d)\n", task->tk_pid, __func__, diff --git a/net/sunrpc/sched.c b/net/sunrpc/sched.c index 73ad57a..fcfd48d 100644 --- a/net/sunrpc/sched.c +++ b/net/sunrpc/sched.c @@ -909,6 +909,8 @@ static void rpc_init_task(struct rpc_task *task, const struct rpc_task_setup *ta /* Initialize workqueue for async tasks */ task->tk_workqueue = task_setup_data->workqueue; + task->tk_xprt = xprt_get(task_setup_data->rpc_xprt); + if (task->tk_ops->rpc_call_prepare != NULL) task->tk_action = rpc_prepare_task; diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c index 37edea6..216a138 100644 --- a/net/sunrpc/xprt.c +++ b/net/sunrpc/xprt.c @@ -48,6 +48,7 @@ #include <linux/sunrpc/clnt.h> #include <linux/sunrpc/metrics.h> #include <linux/sunrpc/bc_xprt.h> +#include <linux/rcupdate.h> #include <trace/events/sunrpc.h> @@ -1166,7 +1167,7 @@ void xprt_free(struct rpc_xprt *xprt) { put_net(xprt->xprt_net); xprt_free_all_slots(xprt); - kfree(xprt); + kfree_rcu(xprt, rcu); } EXPORT_SYMBOL_GPL(xprt_free); @@ -1180,7 +1181,7 @@ EXPORT_SYMBOL_GPL(xprt_free); */ void xprt_reserve(struct rpc_task *task) { - struct rpc_xprt *xprt; + struct rpc_xprt *xprt = task->tk_xprt; task->tk_status = 0; if (task->tk_rqstp != NULL) @@ -1188,11 +1189,8 @@ void xprt_reserve(struct rpc_task *task) task->tk_timeout = 0; task->tk_status = -EAGAIN; - rcu_read_lock(); - xprt = rcu_dereference(task->tk_client->cl_xprt); if (!xprt_throttle_congested(xprt, task)) xprt->ops->alloc_slot(xprt, task); - rcu_read_unlock(); } /** @@ -1206,7 +1204,7 @@ void xprt_reserve(struct rpc_task *task) */ void xprt_retry_reserve(struct rpc_task *task) { - struct rpc_xprt *xprt; + struct rpc_xprt *xprt = task->tk_xprt; task->tk_status = 0; if (task->tk_rqstp != NULL) @@ -1214,10 +1212,7 @@ void xprt_retry_reserve(struct rpc_task *task) task->tk_timeout = 0; task->tk_status = -EAGAIN; - rcu_read_lock(); - xprt = rcu_dereference(task->tk_client->cl_xprt); xprt->ops->alloc_slot(xprt, task); - rcu_read_unlock(); } static inline __be32 xprt_alloc_xid(struct rpc_xprt *xprt) @@ -1264,11 +1259,9 @@ void xprt_release(struct rpc_task *task) if (req == NULL) { if (task->tk_client) { - rcu_read_lock(); - xprt = rcu_dereference(task->tk_client->cl_xprt); + xprt = task->tk_xprt; if (xprt->snd_task == task) xprt_release_write(xprt, task); - rcu_read_unlock(); } return; } @@ -1307,7 +1300,7 @@ void xprt_release(struct rpc_task *task) static void xprt_init(struct rpc_xprt *xprt, struct net *net) { - atomic_set(&xprt->count, 1); + kref_init(&xprt->kref); spin_lock_init(&xprt->transport_lock); spin_lock_init(&xprt->reserve_lock); @@ -1318,6 +1311,7 @@ static void xprt_init(struct rpc_xprt *xprt, struct net *net) spin_lock_init(&xprt->bc_pa_lock); INIT_LIST_HEAD(&xprt->bc_pa_list); #endif /* CONFIG_SUNRPC_BACKCHANNEL */ + INIT_LIST_HEAD(&xprt->xprt_switch); xprt->last_used = jiffies; xprt->cwnd = RPC_INITCWND; @@ -1415,6 +1409,24 @@ static void xprt_destroy(struct rpc_xprt *xprt) xprt->ops->destroy(xprt); } +static void xprt_destroy_kref(struct kref *kref) +{ + xprt_destroy(container_of(kref, struct rpc_xprt, kref)); +} + +/** + * xprt_get - return a reference to an RPC transport. + * @xprt: pointer to the transport + * + */ +struct rpc_xprt *xprt_get(struct rpc_xprt *xprt) +{ + if (xprt != NULL && kref_get_unless_zero(&xprt->kref)) + return xprt; + return NULL; +} +EXPORT_SYMBOL_GPL(xprt_get); + /** * xprt_put - release a reference to an RPC transport. * @xprt: pointer to the transport @@ -1422,7 +1434,7 @@ static void xprt_destroy(struct rpc_xprt *xprt) */ void xprt_put(struct rpc_xprt *xprt) { - if (atomic_dec_and_test(&xprt->count)) - xprt_destroy(xprt); + if (xprt != NULL) + kref_put(&xprt->kref, xprt_destroy_kref); } EXPORT_SYMBOL_GPL(xprt_put); diff --git a/net/sunrpc/xprtmultipath.c b/net/sunrpc/xprtmultipath.c new file mode 100644 index 0000000..e7fd769 --- /dev/null +++ b/net/sunrpc/xprtmultipath.c @@ -0,0 +1,475 @@ +/* + * Multipath support for RPC + * + * Copyright (c) 2015, 2016, Primary Data, Inc. All rights reserved. + * + * Trond Myklebust <trond.myklebust@primarydata.com> + * + */ +#include <linux/types.h> +#include <linux/kref.h> +#include <linux/list.h> +#include <linux/rcupdate.h> +#include <linux/rculist.h> +#include <linux/slab.h> +#include <asm/cmpxchg.h> +#include <linux/spinlock.h> +#include <linux/sunrpc/xprt.h> +#include <linux/sunrpc/xprtmultipath.h> + +typedef struct rpc_xprt *(*xprt_switch_find_xprt_t)(struct list_head *head, + const struct rpc_xprt *cur); + +static const struct rpc_xprt_iter_ops rpc_xprt_iter_singular; +static const struct rpc_xprt_iter_ops rpc_xprt_iter_roundrobin; +static const struct rpc_xprt_iter_ops rpc_xprt_iter_listall; + +static void xprt_switch_add_xprt_locked(struct rpc_xprt_switch *xps, + struct rpc_xprt *xprt) +{ + if (unlikely(xprt_get(xprt) == NULL)) + return; + list_add_tail_rcu(&xprt->xprt_switch, &xps->xps_xprt_list); + smp_wmb(); + if (xps->xps_nxprts == 0) + xps->xps_net = xprt->xprt_net; + xps->xps_nxprts++; +} + +/** + * rpc_xprt_switch_add_xprt - Add a new rpc_xprt to an rpc_xprt_switch + * @xps: pointer to struct rpc_xprt_switch + * @xprt: pointer to struct rpc_xprt + * + * Adds xprt to the end of the list of struct rpc_xprt in xps. + */ +void rpc_xprt_switch_add_xprt(struct rpc_xprt_switch *xps, + struct rpc_xprt *xprt) +{ + if (xprt == NULL) + return; + spin_lock(&xps->xps_lock); + if (xps->xps_net == xprt->xprt_net || xps->xps_net == NULL) + xprt_switch_add_xprt_locked(xps, xprt); + spin_unlock(&xps->xps_lock); +} + +static void xprt_switch_remove_xprt_locked(struct rpc_xprt_switch *xps, + struct rpc_xprt *xprt) +{ + if (unlikely(xprt == NULL)) + return; + xps->xps_nxprts--; + if (xps->xps_nxprts == 0) + xps->xps_net = NULL; + smp_wmb(); + list_del_rcu(&xprt->xprt_switch); +} + +/** + * rpc_xprt_switch_remove_xprt - Removes an rpc_xprt from a rpc_xprt_switch + * @xps: pointer to struct rpc_xprt_switch + * @xprt: pointer to struct rpc_xprt + * + * Removes xprt from the list of struct rpc_xprt in xps. + */ +void rpc_xprt_switch_remove_xprt(struct rpc_xprt_switch *xps, + struct rpc_xprt *xprt) +{ + spin_lock(&xps->xps_lock); + xprt_switch_remove_xprt_locked(xps, xprt); + spin_unlock(&xps->xps_lock); + xprt_put(xprt); +} + +/** + * xprt_switch_alloc - Allocate a new struct rpc_xprt_switch + * @xprt: pointer to struct rpc_xprt + * @gfp_flags: allocation flags + * + * On success, returns an initialised struct rpc_xprt_switch, containing + * the entry xprt. Returns NULL on failure. + */ +struct rpc_xprt_switch *xprt_switch_alloc(struct rpc_xprt *xprt, + gfp_t gfp_flags) +{ + struct rpc_xprt_switch *xps; + + xps = kmalloc(sizeof(*xps), gfp_flags); + if (xps != NULL) { + spin_lock_init(&xps->xps_lock); + kref_init(&xps->xps_kref); + xps->xps_nxprts = 0; + INIT_LIST_HEAD(&xps->xps_xprt_list); + xps->xps_iter_ops = &rpc_xprt_iter_singular; + xprt_switch_add_xprt_locked(xps, xprt); + } + + return xps; +} + +static void xprt_switch_free_entries(struct rpc_xprt_switch *xps) +{ + spin_lock(&xps->xps_lock); + while (!list_empty(&xps->xps_xprt_list)) { + struct rpc_xprt *xprt; + + xprt = list_first_entry(&xps->xps_xprt_list, + struct rpc_xprt, xprt_switch); + xprt_switch_remove_xprt_locked(xps, xprt); + spin_unlock(&xps->xps_lock); + xprt_put(xprt); + spin_lock(&xps->xps_lock); + } + spin_unlock(&xps->xps_lock); +} + +static void xprt_switch_free(struct kref *kref) +{ + struct rpc_xprt_switch *xps = container_of(kref, + struct rpc_xprt_switch, xps_kref); + + xprt_switch_free_entries(xps); + kfree_rcu(xps, xps_rcu); +} + +/** + * xprt_switch_get - Return a reference to a rpc_xprt_switch + * @xps: pointer to struct rpc_xprt_switch + * + * Returns a reference to xps unless the refcount is already zero. + */ +struct rpc_xprt_switch *xprt_switch_get(struct rpc_xprt_switch *xps) +{ + if (xps != NULL && kref_get_unless_zero(&xps->xps_kref)) + return xps; + return NULL; +} + +/** + * xprt_switch_put - Release a reference to a rpc_xprt_switch + * @xps: pointer to struct rpc_xprt_switch + * + * Release the reference to xps, and free it once the refcount is zero. + */ +void xprt_switch_put(struct rpc_xprt_switch *xps) +{ + if (xps != NULL) + kref_put(&xps->xps_kref, xprt_switch_free); +} + +/** + * rpc_xprt_switch_set_roundrobin - Set a round-robin policy on rpc_xprt_switch + * @xps: pointer to struct rpc_xprt_switch + * + * Sets a round-robin default policy for iterators acting on xps. + */ +void rpc_xprt_switch_set_roundrobin(struct rpc_xprt_switch *xps) +{ + if (READ_ONCE(xps->xps_iter_ops) != &rpc_xprt_iter_roundrobin) + WRITE_ONCE(xps->xps_iter_ops, &rpc_xprt_iter_roundrobin); +} + +static +const struct rpc_xprt_iter_ops *xprt_iter_ops(const struct rpc_xprt_iter *xpi) +{ + if (xpi->xpi_ops != NULL) + return xpi->xpi_ops; + return rcu_dereference(xpi->xpi_xpswitch)->xps_iter_ops; +} + +static +void xprt_iter_no_rewind(struct rpc_xprt_iter *xpi) +{ +} + +static +void xprt_iter_default_rewind(struct rpc_xprt_iter *xpi) +{ + WRITE_ONCE(xpi->xpi_cursor, NULL); +} + +static +struct rpc_xprt *xprt_switch_find_first_entry(struct list_head *head) +{ + return list_first_or_null_rcu(head, struct rpc_xprt, xprt_switch); +} + +static +struct rpc_xprt *xprt_iter_first_entry(struct rpc_xprt_iter *xpi) +{ + struct rpc_xprt_switch *xps = rcu_dereference(xpi->xpi_xpswitch); + + if (xps == NULL) + return NULL; + return xprt_switch_find_first_entry(&xps->xps_xprt_list); +} + +static +struct rpc_xprt *xprt_switch_find_current_entry(struct list_head *head, + const struct rpc_xprt *cur) +{ + struct rpc_xprt *pos; + + list_for_each_entry_rcu(pos, head, xprt_switch) { + if (cur == pos) + return pos; + } + return NULL; +} + +static +struct rpc_xprt *xprt_iter_current_entry(struct rpc_xprt_iter *xpi) +{ + struct rpc_xprt_switch *xps = rcu_dereference(xpi->xpi_xpswitch); + struct list_head *head; + + if (xps == NULL) + return NULL; + head = &xps->xps_xprt_list; + if (xpi->xpi_cursor == NULL || xps->xps_nxprts < 2) + return xprt_switch_find_first_entry(head); + return xprt_switch_find_current_entry(head, xpi->xpi_cursor); +} + +static +struct rpc_xprt *xprt_switch_find_next_entry(struct list_head *head, + const struct rpc_xprt *cur) +{ + struct rpc_xprt *pos, *prev = NULL; + + list_for_each_entry_rcu(pos, head, xprt_switch) { + if (cur == prev) + return pos; + prev = pos; + } + return NULL; +} + +static +struct rpc_xprt *xprt_switch_set_next_cursor(struct list_head *head, + struct rpc_xprt **cursor, + xprt_switch_find_xprt_t find_next) +{ + struct rpc_xprt *cur, *pos, *old; + + cur = READ_ONCE(*cursor); + for (;;) { + old = cur; + pos = find_next(head, old); + if (pos == NULL) + break; + cur = cmpxchg_relaxed(cursor, old, pos); + if (cur == old) + break; + } + return pos; +} + +static +struct rpc_xprt *xprt_iter_next_entry_multiple(struct rpc_xprt_iter *xpi, + xprt_switch_find_xprt_t find_next) +{ + struct rpc_xprt_switch *xps = rcu_dereference(xpi->xpi_xpswitch); + struct list_head *head; + + if (xps == NULL) + return NULL; + head = &xps->xps_xprt_list; + if (xps->xps_nxprts < 2) + return xprt_switch_find_first_entry(head); + return xprt_switch_set_next_cursor(head, &xpi->xpi_cursor, find_next); +} + +static +struct rpc_xprt *xprt_switch_find_next_entry_roundrobin(struct list_head *head, + const struct rpc_xprt *cur) +{ + struct rpc_xprt *ret; + + ret = xprt_switch_find_next_entry(head, cur); + if (ret != NULL) + return ret; + return xprt_switch_find_first_entry(head); +} + +static +struct rpc_xprt *xprt_iter_next_entry_roundrobin(struct rpc_xprt_iter *xpi) +{ + return xprt_iter_next_entry_multiple(xpi, + xprt_switch_find_next_entry_roundrobin); +} + +static +struct rpc_xprt *xprt_iter_next_entry_all(struct rpc_xprt_iter *xpi) +{ + return xprt_iter_next_entry_multiple(xpi, xprt_switch_find_next_entry); +} + +/* + * xprt_iter_rewind - Resets the xprt iterator + * @xpi: pointer to rpc_xprt_iter + * + * Resets xpi to ensure that it points to the first entry in the list + * of transports. + */ +static +void xprt_iter_rewind(struct rpc_xprt_iter *xpi) +{ + rcu_read_lock(); + xprt_iter_ops(xpi)->xpi_rewind(xpi); + rcu_read_unlock(); +} + +static void __xprt_iter_init(struct rpc_xprt_iter *xpi, + struct rpc_xprt_switch *xps, + const struct rpc_xprt_iter_ops *ops) +{ + rcu_assign_pointer(xpi->xpi_xpswitch, xprt_switch_get(xps)); + xpi->xpi_cursor = NULL; + xpi->xpi_ops = ops; +} + +/** + * xprt_iter_init - Initialise an xprt iterator + * @xpi: pointer to rpc_xprt_iter + * @xps: pointer to rpc_xprt_switch + * + * Initialises the iterator to use the default iterator ops + * as set in xps. This function is mainly intended for internal + * use in the rpc_client. + */ +void xprt_iter_init(struct rpc_xprt_iter *xpi, + struct rpc_xprt_switch *xps) +{ + __xprt_iter_init(xpi, xps, NULL); +} + +/** + * xprt_iter_init_listall - Initialise an xprt iterator + * @xpi: pointer to rpc_xprt_iter + * @xps: pointer to rpc_xprt_switch + * + * Initialises the iterator to iterate once through the entire list + * of entries in xps. + */ +void xprt_iter_init_listall(struct rpc_xprt_iter *xpi, + struct rpc_xprt_switch *xps) +{ + __xprt_iter_init(xpi, xps, &rpc_xprt_iter_listall); +} + +/** + * xprt_iter_xchg_switch - Atomically swap out the rpc_xprt_switch + * @xpi: pointer to rpc_xprt_iter + * @xps: pointer to a new rpc_xprt_switch or NULL + * + * Swaps out the existing xpi->xpi_xpswitch with a new value. + */ +struct rpc_xprt_switch *xprt_iter_xchg_switch(struct rpc_xprt_iter *xpi, + struct rpc_xprt_switch *newswitch) +{ + struct rpc_xprt_switch __rcu *oldswitch; + + /* Atomically swap out the old xpswitch */ + oldswitch = xchg(&xpi->xpi_xpswitch, RCU_INITIALIZER(newswitch)); + if (newswitch != NULL) + xprt_iter_rewind(xpi); + return rcu_dereference_protected(oldswitch, true); +} + +/** + * xprt_iter_destroy - Destroys the xprt iterator + * @xpi pointer to rpc_xprt_iter + */ +void xprt_iter_destroy(struct rpc_xprt_iter *xpi) +{ + xprt_switch_put(xprt_iter_xchg_switch(xpi, NULL)); +} + +/** + * xprt_iter_xprt - Returns the rpc_xprt pointed to by the cursor + * @xpi: pointer to rpc_xprt_iter + * + * Returns a pointer to the struct rpc_xprt that is currently + * pointed to by the cursor. + * Caller must be holding rcu_read_lock(). + */ +struct rpc_xprt *xprt_iter_xprt(struct rpc_xprt_iter *xpi) +{ + WARN_ON_ONCE(!rcu_read_lock_held()); + return xprt_iter_ops(xpi)->xpi_xprt(xpi); +} + +static +struct rpc_xprt *xprt_iter_get_helper(struct rpc_xprt_iter *xpi, + struct rpc_xprt *(*fn)(struct rpc_xprt_iter *)) +{ + struct rpc_xprt *ret; + + do { + ret = fn(xpi); + if (ret == NULL) + break; + ret = xprt_get(ret); + } while (ret == NULL); + return ret; +} + +/** + * xprt_iter_get_xprt - Returns the rpc_xprt pointed to by the cursor + * @xpi: pointer to rpc_xprt_iter + * + * Returns a reference to the struct rpc_xprt that is currently + * pointed to by the cursor. + */ +struct rpc_xprt *xprt_iter_get_xprt(struct rpc_xprt_iter *xpi) +{ + struct rpc_xprt *xprt; + + rcu_read_lock(); + xprt = xprt_iter_get_helper(xpi, xprt_iter_ops(xpi)->xpi_xprt); + rcu_read_unlock(); + return xprt; +} + +/** + * xprt_iter_get_next - Returns the next rpc_xprt following the cursor + * @xpi: pointer to rpc_xprt_iter + * + * Returns a reference to the struct rpc_xprt that immediately follows the + * entry pointed to by the cursor. + */ +struct rpc_xprt *xprt_iter_get_next(struct rpc_xprt_iter *xpi) +{ + struct rpc_xprt *xprt; + + rcu_read_lock(); + xprt = xprt_iter_get_helper(xpi, xprt_iter_ops(xpi)->xpi_next); + rcu_read_unlock(); + return xprt; +} + +/* Policy for always returning the first entry in the rpc_xprt_switch */ +static +const struct rpc_xprt_iter_ops rpc_xprt_iter_singular = { + .xpi_rewind = xprt_iter_no_rewind, + .xpi_xprt = xprt_iter_first_entry, + .xpi_next = xprt_iter_first_entry, +}; + +/* Policy for round-robin iteration of entries in the rpc_xprt_switch */ +static +const struct rpc_xprt_iter_ops rpc_xprt_iter_roundrobin = { + .xpi_rewind = xprt_iter_default_rewind, + .xpi_xprt = xprt_iter_current_entry, + .xpi_next = xprt_iter_next_entry_roundrobin, +}; + +/* Policy for once-through iteration of entries in the rpc_xprt_switch */ +static +const struct rpc_xprt_iter_ops rpc_xprt_iter_listall = { + .xpi_rewind = xprt_iter_default_rewind, + .xpi_xprt = xprt_iter_current_entry, + .xpi_next = xprt_iter_next_entry_all, +}; diff --git a/net/sunrpc/xprtrdma/fmr_ops.c b/net/sunrpc/xprtrdma/fmr_ops.c index c14f3a4..b289e10 100644 --- a/net/sunrpc/xprtrdma/fmr_ops.c +++ b/net/sunrpc/xprtrdma/fmr_ops.c @@ -80,13 +80,13 @@ fmr_op_init(struct rpcrdma_xprt *r_xprt) if (!r) goto out; - r->r.fmr.physaddrs = kmalloc(RPCRDMA_MAX_FMR_SGES * - sizeof(u64), GFP_KERNEL); - if (!r->r.fmr.physaddrs) + r->fmr.physaddrs = kmalloc(RPCRDMA_MAX_FMR_SGES * + sizeof(u64), GFP_KERNEL); + if (!r->fmr.physaddrs) goto out_free; - r->r.fmr.fmr = ib_alloc_fmr(pd, mr_access_flags, &fmr_attr); - if (IS_ERR(r->r.fmr.fmr)) + r->fmr.fmr = ib_alloc_fmr(pd, mr_access_flags, &fmr_attr); + if (IS_ERR(r->fmr.fmr)) goto out_fmr_err; list_add(&r->mw_list, &buf->rb_mws); @@ -95,9 +95,9 @@ fmr_op_init(struct rpcrdma_xprt *r_xprt) return 0; out_fmr_err: - rc = PTR_ERR(r->r.fmr.fmr); + rc = PTR_ERR(r->fmr.fmr); dprintk("RPC: %s: ib_alloc_fmr status %i\n", __func__, rc); - kfree(r->r.fmr.physaddrs); + kfree(r->fmr.physaddrs); out_free: kfree(r); out: @@ -109,7 +109,7 @@ __fmr_unmap(struct rpcrdma_mw *r) { LIST_HEAD(l); - list_add(&r->r.fmr.fmr->list, &l); + list_add(&r->fmr.fmr->list, &l); return ib_unmap_fmr(&l); } @@ -148,7 +148,7 @@ fmr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg, nsegs = RPCRDMA_MAX_FMR_SGES; for (i = 0; i < nsegs;) { rpcrdma_map_one(device, seg, direction); - mw->r.fmr.physaddrs[i] = seg->mr_dma; + mw->fmr.physaddrs[i] = seg->mr_dma; len += seg->mr_len; ++seg; ++i; @@ -158,13 +158,13 @@ fmr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg, break; } - rc = ib_map_phys_fmr(mw->r.fmr.fmr, mw->r.fmr.physaddrs, + rc = ib_map_phys_fmr(mw->fmr.fmr, mw->fmr.physaddrs, i, seg1->mr_dma); if (rc) goto out_maperr; seg1->rl_mw = mw; - seg1->mr_rkey = mw->r.fmr.fmr->rkey; + seg1->mr_rkey = mw->fmr.fmr->rkey; seg1->mr_base = seg1->mr_dma + pageoff; seg1->mr_nsegs = i; seg1->mr_len = len; @@ -219,7 +219,7 @@ fmr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) seg = &req->rl_segments[i]; mw = seg->rl_mw; - list_add(&mw->r.fmr.fmr->list, &unmap_list); + list_add(&mw->fmr.fmr->list, &unmap_list); i += seg->mr_nsegs; } @@ -281,9 +281,9 @@ fmr_op_destroy(struct rpcrdma_buffer *buf) while (!list_empty(&buf->rb_all)) { r = list_entry(buf->rb_all.next, struct rpcrdma_mw, mw_all); list_del(&r->mw_all); - kfree(r->r.fmr.physaddrs); + kfree(r->fmr.physaddrs); - rc = ib_dealloc_fmr(r->r.fmr.fmr); + rc = ib_dealloc_fmr(r->fmr.fmr); if (rc) dprintk("RPC: %s: ib_dealloc_fmr failed %i\n", __func__, rc); diff --git a/net/sunrpc/xprtrdma/frwr_ops.c b/net/sunrpc/xprtrdma/frwr_ops.c index e165673..c250924 100644 --- a/net/sunrpc/xprtrdma/frwr_ops.c +++ b/net/sunrpc/xprtrdma/frwr_ops.c @@ -109,20 +109,20 @@ static void __frwr_recovery_worker(struct work_struct *work) { struct rpcrdma_mw *r = container_of(work, struct rpcrdma_mw, - r.frmr.fr_work); - struct rpcrdma_xprt *r_xprt = r->r.frmr.fr_xprt; + 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->r.frmr.fr_mr)) + if (ib_dereg_mr(r->frmr.fr_mr)) goto out_fail; - r->r.frmr.fr_mr = ib_alloc_mr(pd, IB_MR_TYPE_MEM_REG, depth); - if (IS_ERR(r->r.frmr.fr_mr)) + 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->r.frmr.fr_state = FRMR_IS_INVALID; + r->frmr.fr_state = FRMR_IS_INVALID; rpcrdma_put_mw(r_xprt, r); return; @@ -137,15 +137,15 @@ out_fail: static void __frwr_queue_recovery(struct rpcrdma_mw *r) { - INIT_WORK(&r->r.frmr.fr_work, __frwr_recovery_worker); - queue_work(frwr_recovery_wq, &r->r.frmr.fr_work); + INIT_WORK(&r->frmr.fr_work, __frwr_recovery_worker); + queue_work(frwr_recovery_wq, &r->frmr.fr_work); } static int __frwr_init(struct rpcrdma_mw *r, struct ib_pd *pd, struct ib_device *device, unsigned int depth) { - struct rpcrdma_frmr *f = &r->r.frmr; + struct rpcrdma_frmr *f = &r->frmr; int rc; f->fr_mr = ib_alloc_mr(pd, IB_MR_TYPE_MEM_REG, depth); @@ -158,6 +158,8 @@ __frwr_init(struct rpcrdma_mw *r, struct ib_pd *pd, struct ib_device *device, sg_init_table(f->sg, depth); + init_completion(&f->fr_linv_done); + return 0; out_mr_err: @@ -179,11 +181,11 @@ __frwr_release(struct rpcrdma_mw *r) { int rc; - rc = ib_dereg_mr(r->r.frmr.fr_mr); + rc = ib_dereg_mr(r->frmr.fr_mr); if (rc) dprintk("RPC: %s: ib_dereg_mr status %i\n", __func__, rc); - kfree(r->r.frmr.sg); + kfree(r->frmr.sg); } static int @@ -244,39 +246,76 @@ frwr_op_maxpages(struct rpcrdma_xprt *r_xprt) rpcrdma_max_segments(r_xprt) * ia->ri_max_frmr_depth); } -/* If FAST_REG or LOCAL_INV failed, indicate the frmr needs - * to be reset. +static void +__frwr_sendcompletion_flush(struct ib_wc *wc, struct rpcrdma_frmr *frmr, + const char *wr) +{ + frmr->fr_state = FRMR_IS_STALE; + if (wc->status != IB_WC_WR_FLUSH_ERR) + pr_err("rpcrdma: %s: %s (%u/0x%x)\n", + wr, ib_wc_status_msg(wc->status), + wc->status, wc->vendor_err); +} + +/** + * frwr_wc_fastreg - Invoked by RDMA provider for each polled FastReg WC + * @cq: completion queue (ignored) + * @wc: completed WR * - * WARNING: Only wr_id and status are reliable at this point */ static void -__frwr_sendcompletion_flush(struct ib_wc *wc, struct rpcrdma_mw *r) +frwr_wc_fastreg(struct ib_cq *cq, struct ib_wc *wc) { - if (likely(wc->status == IB_WC_SUCCESS)) - return; - - /* WARNING: Only wr_id and status are reliable at this point */ - r = (struct rpcrdma_mw *)(unsigned long)wc->wr_id; - if (wc->status == IB_WC_WR_FLUSH_ERR) - dprintk("RPC: %s: frmr %p flushed\n", __func__, r); - else - pr_warn("RPC: %s: frmr %p error, status %s (%d)\n", - __func__, r, ib_wc_status_msg(wc->status), wc->status); + struct rpcrdma_frmr *frmr; + struct ib_cqe *cqe; - r->r.frmr.fr_state = FRMR_IS_STALE; + /* WARNING: Only wr_cqe and status are reliable at this point */ + if (wc->status != IB_WC_SUCCESS) { + cqe = wc->wr_cqe; + frmr = container_of(cqe, struct rpcrdma_frmr, fr_cqe); + __frwr_sendcompletion_flush(wc, frmr, "fastreg"); + } } +/** + * frwr_wc_localinv - Invoked by RDMA provider for each polled LocalInv WC + * @cq: completion queue (ignored) + * @wc: completed WR + * + */ static void -frwr_sendcompletion(struct ib_wc *wc) +frwr_wc_localinv(struct ib_cq *cq, struct ib_wc *wc) { - struct rpcrdma_mw *r = (struct rpcrdma_mw *)(unsigned long)wc->wr_id; - struct rpcrdma_frmr *f = &r->r.frmr; + struct rpcrdma_frmr *frmr; + struct ib_cqe *cqe; - if (unlikely(wc->status != IB_WC_SUCCESS)) - __frwr_sendcompletion_flush(wc, r); + /* WARNING: Only wr_cqe and status are reliable at this point */ + if (wc->status != IB_WC_SUCCESS) { + cqe = wc->wr_cqe; + frmr = container_of(cqe, struct rpcrdma_frmr, fr_cqe); + __frwr_sendcompletion_flush(wc, frmr, "localinv"); + } +} - if (f->fr_waiter) - complete(&f->fr_linv_done); +/** + * frwr_wc_localinv - Invoked by RDMA provider for each polled LocalInv WC + * @cq: completion queue (ignored) + * @wc: completed WR + * + * Awaken anyone waiting for an MR to finish being fenced. + */ +static void +frwr_wc_localinv_wake(struct ib_cq *cq, struct ib_wc *wc) +{ + struct rpcrdma_frmr *frmr; + struct ib_cqe *cqe; + + /* WARNING: Only wr_cqe and status are reliable at this point */ + cqe = wc->wr_cqe; + frmr = container_of(cqe, struct rpcrdma_frmr, fr_cqe); + if (wc->status != IB_WC_SUCCESS) + __frwr_sendcompletion_flush(wc, frmr, "localinv"); + complete_all(&frmr->fr_linv_done); } static int @@ -313,8 +352,7 @@ frwr_op_init(struct rpcrdma_xprt *r_xprt) list_add(&r->mw_list, &buf->rb_mws); list_add(&r->mw_all, &buf->rb_all); - r->mw_sendcompletion = frwr_sendcompletion; - r->r.frmr.fr_xprt = r_xprt; + r->frmr.fr_xprt = r_xprt; } return 0; @@ -347,10 +385,9 @@ frwr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg, mw = rpcrdma_get_mw(r_xprt); if (!mw) return -ENOMEM; - } while (mw->r.frmr.fr_state != FRMR_IS_INVALID); - frmr = &mw->r.frmr; + } while (mw->frmr.fr_state != FRMR_IS_INVALID); + frmr = &mw->frmr; frmr->fr_state = FRMR_IS_VALID; - frmr->fr_waiter = false; mr = frmr->fr_mr; reg_wr = &frmr->fr_regwr; @@ -400,7 +437,8 @@ frwr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg, reg_wr->wr.next = NULL; reg_wr->wr.opcode = IB_WR_REG_MR; - reg_wr->wr.wr_id = (uintptr_t)mw; + frmr->fr_cqe.done = frwr_wc_fastreg; + reg_wr->wr.wr_cqe = &frmr->fr_cqe; reg_wr->wr.num_sge = 0; reg_wr->wr.send_flags = 0; reg_wr->mr = mr; @@ -434,15 +472,15 @@ static struct ib_send_wr * __frwr_prepare_linv_wr(struct rpcrdma_mr_seg *seg) { struct rpcrdma_mw *mw = seg->rl_mw; - struct rpcrdma_frmr *f = &mw->r.frmr; + struct rpcrdma_frmr *f = &mw->frmr; struct ib_send_wr *invalidate_wr; - f->fr_waiter = false; f->fr_state = FRMR_IS_INVALID; invalidate_wr = &f->fr_invwr; memset(invalidate_wr, 0, sizeof(*invalidate_wr)); - invalidate_wr->wr_id = (unsigned long)(void *)mw; + f->fr_cqe.done = frwr_wc_localinv; + invalidate_wr->wr_cqe = &f->fr_cqe; invalidate_wr->opcode = IB_WR_LOCAL_INV; invalidate_wr->ex.invalidate_rkey = f->fr_mr->rkey; @@ -455,7 +493,7 @@ __frwr_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; - struct rpcrdma_frmr *f = &mw->r.frmr; + struct rpcrdma_frmr *f = &mw->frmr; seg->rl_mw = NULL; @@ -504,15 +542,15 @@ frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) i += seg->mr_nsegs; } - f = &seg->rl_mw->r.frmr; + f = &seg->rl_mw->frmr; /* Strong send queue ordering guarantees that when the * last WR in the chain completes, all WRs in the chain * are complete. */ f->fr_invwr.send_flags = IB_SEND_SIGNALED; - f->fr_waiter = true; - init_completion(&f->fr_linv_done); + f->fr_cqe.done = frwr_wc_localinv_wake; + reinit_completion(&f->fr_linv_done); INIT_CQCOUNT(&r_xprt->rx_ep); /* Transport disconnect drains the receive CQ before it @@ -520,14 +558,18 @@ 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) + if (rc) { pr_warn("%s: ib_post_send failed %i\n", __func__, rc); + rdma_disconnect(ia->ri_id); + goto unmap; + } wait_for_completion(&f->fr_linv_done); /* ORDER: Now DMA unmap all of the req's MRs, and return * them to the free MW list. */ +unmap: for (i = 0, nchunks = req->rl_nchunks; nchunks; nchunks--) { seg = &req->rl_segments[i]; @@ -549,7 +591,7 @@ 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->r.frmr; + struct rpcrdma_frmr *frmr = &mw->frmr; struct ib_send_wr *invalidate_wr, *bad_wr; int rc, nsegs = seg->mr_nsegs; @@ -557,10 +599,11 @@ frwr_op_unmap(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg) seg1->rl_mw = NULL; frmr->fr_state = FRMR_IS_INVALID; - invalidate_wr = &mw->r.frmr.fr_invwr; + invalidate_wr = &mw->frmr.fr_invwr; memset(invalidate_wr, 0, sizeof(*invalidate_wr)); - invalidate_wr->wr_id = (uintptr_t)mw; + 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); diff --git a/net/sunrpc/xprtrdma/physical_ops.c b/net/sunrpc/xprtrdma/physical_ops.c index dbb302e..481b9b6 100644 --- a/net/sunrpc/xprtrdma/physical_ops.c +++ b/net/sunrpc/xprtrdma/physical_ops.c @@ -68,7 +68,6 @@ physical_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg, rpcrdma_map_one(ia->ri_device, seg, rpcrdma_data_dir(writing)); seg->mr_rkey = ia->ri_dma_mr->rkey; seg->mr_base = seg->mr_dma; - seg->mr_nsegs = 1; return 1; } diff --git a/net/sunrpc/xprtrdma/rpc_rdma.c b/net/sunrpc/xprtrdma/rpc_rdma.c index 0f28f2d..888823b 100644 --- a/net/sunrpc/xprtrdma/rpc_rdma.c +++ b/net/sunrpc/xprtrdma/rpc_rdma.c @@ -132,6 +132,33 @@ rpcrdma_tail_pullup(struct xdr_buf *buf) return tlen; } +/* Split "vec" on page boundaries into segments. FMR registers pages, + * not a byte range. Other modes coalesce these segments into a single + * MR when they can. + */ +static int +rpcrdma_convert_kvec(struct kvec *vec, struct rpcrdma_mr_seg *seg, + int n, int nsegs) +{ + size_t page_offset; + u32 remaining; + char *base; + + base = vec->iov_base; + page_offset = offset_in_page(base); + remaining = vec->iov_len; + while (remaining && n < nsegs) { + seg[n].mr_page = NULL; + seg[n].mr_offset = base; + seg[n].mr_len = min_t(u32, PAGE_SIZE - page_offset, remaining); + remaining -= seg[n].mr_len; + base += seg[n].mr_len; + ++n; + page_offset = 0; + } + return n; +} + /* * Chunk assembly from upper layer xdr_buf. * @@ -150,11 +177,10 @@ rpcrdma_convert_iovs(struct xdr_buf *xdrbuf, unsigned int pos, int page_base; struct page **ppages; - if (pos == 0 && xdrbuf->head[0].iov_len) { - seg[n].mr_page = NULL; - seg[n].mr_offset = xdrbuf->head[0].iov_base; - seg[n].mr_len = xdrbuf->head[0].iov_len; - ++n; + if (pos == 0) { + n = rpcrdma_convert_kvec(&xdrbuf->head[0], seg, n, nsegs); + if (n == nsegs) + return -EIO; } len = xdrbuf->page_len; @@ -192,13 +218,9 @@ rpcrdma_convert_iovs(struct xdr_buf *xdrbuf, unsigned int pos, * xdr pad bytes, saving the server an RDMA operation. */ if (xdrbuf->tail[0].iov_len < 4 && xprt_rdma_pad_optimize) return n; + n = rpcrdma_convert_kvec(&xdrbuf->tail[0], seg, n, nsegs); if (n == nsegs) - /* Tail remains, but we're out of segments */ return -EIO; - seg[n].mr_page = NULL; - seg[n].mr_offset = xdrbuf->tail[0].iov_base; - seg[n].mr_len = xdrbuf->tail[0].iov_len; - ++n; } return n; @@ -773,20 +795,17 @@ rpcrdma_reply_handler(struct rpcrdma_rep *rep) struct rpcrdma_xprt *r_xprt = rep->rr_rxprt; struct rpc_xprt *xprt = &r_xprt->rx_xprt; __be32 *iptr; - int rdmalen, status; + int rdmalen, status, rmerr; unsigned long cwnd; - u32 credits; dprintk("RPC: %s: incoming rep %p\n", __func__, rep); if (rep->rr_len == RPCRDMA_BAD_LEN) goto out_badstatus; - if (rep->rr_len < RPCRDMA_HDRLEN_MIN) + if (rep->rr_len < RPCRDMA_HDRLEN_ERR) goto out_shortreply; headerp = rdmab_to_msg(rep->rr_rdmabuf); - if (headerp->rm_vers != rpcrdma_version) - goto out_badversion; #if defined(CONFIG_SUNRPC_BACKCHANNEL) if (rpcrdma_is_bcall(headerp)) goto out_bcall; @@ -809,15 +828,16 @@ rpcrdma_reply_handler(struct rpcrdma_rep *rep) */ list_del_init(&rqst->rq_list); spin_unlock_bh(&xprt->transport_lock); - dprintk("RPC: %s: reply 0x%p completes request 0x%p\n" - " RPC request 0x%p xid 0x%08x\n", - __func__, rep, req, rqst, - be32_to_cpu(headerp->rm_xid)); + dprintk("RPC: %s: reply %p completes request %p (xid 0x%08x)\n", + __func__, rep, req, be32_to_cpu(headerp->rm_xid)); /* from here on, the reply is no longer an orphan */ req->rl_reply = rep; xprt->reestablish_timeout = 0; + if (headerp->rm_vers != rpcrdma_version) + goto out_badversion; + /* check for expected message types */ /* The order of some of these tests is important. */ switch (headerp->rm_type) { @@ -878,6 +898,9 @@ rpcrdma_reply_handler(struct rpcrdma_rep *rep) status = rdmalen; break; + case rdma_error: + goto out_rdmaerr; + badheader: default: dprintk("%s: invalid rpcrdma reply header (type %d):" @@ -893,6 +916,7 @@ badheader: break; } +out: /* Invalidate and flush the data payloads before waking the * waiting application. This guarantees the memory region is * properly fenced from the server before the application @@ -903,15 +927,9 @@ badheader: if (req->rl_nchunks) r_xprt->rx_ia.ri_ops->ro_unmap_sync(r_xprt, req); - credits = be32_to_cpu(headerp->rm_credit); - if (credits == 0) - credits = 1; /* don't deadlock */ - else if (credits > r_xprt->rx_buf.rb_max_requests) - credits = r_xprt->rx_buf.rb_max_requests; - spin_lock_bh(&xprt->transport_lock); cwnd = xprt->cwnd; - xprt->cwnd = credits << RPC_CWNDSHIFT; + xprt->cwnd = atomic_read(&r_xprt->rx_buf.rb_credits) << RPC_CWNDSHIFT; if (xprt->cwnd > cwnd) xprt_release_rqst_cong(rqst->rq_task); @@ -935,13 +953,43 @@ out_bcall: return; #endif -out_shortreply: - dprintk("RPC: %s: short/invalid reply\n", __func__); - goto repost; - +/* If the incoming reply terminated a pending RPC, the next + * RPC call will post a replacement receive buffer as it is + * being marshaled. + */ out_badversion: dprintk("RPC: %s: invalid version %d\n", __func__, be32_to_cpu(headerp->rm_vers)); + status = -EIO; + r_xprt->rx_stats.bad_reply_count++; + goto out; + +out_rdmaerr: + rmerr = be32_to_cpu(headerp->rm_body.rm_error.rm_err); + switch (rmerr) { + case ERR_VERS: + pr_err("%s: server reports header version error (%u-%u)\n", + __func__, + be32_to_cpu(headerp->rm_body.rm_error.rm_vers_low), + be32_to_cpu(headerp->rm_body.rm_error.rm_vers_high)); + break; + case ERR_CHUNK: + pr_err("%s: server reports header decoding error\n", + __func__); + break; + default: + pr_err("%s: server reports unknown error %d\n", + __func__, rmerr); + } + status = -EREMOTEIO; + r_xprt->rx_stats.bad_reply_count++; + goto out; + +/* If no pending RPC transaction was matched, post a replacement + * receive buffer before returning. + */ +out_shortreply: + dprintk("RPC: %s: short/invalid reply\n", __func__); goto repost; out_nomatch: diff --git a/net/sunrpc/xprtrdma/svc_rdma_backchannel.c b/net/sunrpc/xprtrdma/svc_rdma_backchannel.c index 65a7c23..a2a7519 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_backchannel.c +++ b/net/sunrpc/xprtrdma/svc_rdma_backchannel.c @@ -107,26 +107,18 @@ static int svc_rdma_bc_sendto(struct svcxprt_rdma *rdma, int ret; vec = svc_rdma_get_req_map(rdma); - ret = svc_rdma_map_xdr(rdma, sndbuf, vec); + ret = svc_rdma_map_xdr(rdma, sndbuf, vec, false); if (ret) goto out_err; - /* Post a recv buffer to handle the reply for this request. */ - ret = svc_rdma_post_recv(rdma, GFP_NOIO); - if (ret) { - pr_err("svcrdma: Failed to post bc receive buffer, err=%d.\n", - ret); - pr_err("svcrdma: closing transport %p.\n", rdma); - set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags); - ret = -ENOTCONN; + ret = svc_rdma_repost_recv(rdma, GFP_NOIO); + if (ret) goto out_err; - } ctxt = svc_rdma_get_context(rdma); ctxt->pages[0] = virt_to_page(rqst->rq_buffer); ctxt->count = 1; - ctxt->wr_op = IB_WR_SEND; ctxt->direction = DMA_TO_DEVICE; ctxt->sge[0].lkey = rdma->sc_pd->local_dma_lkey; ctxt->sge[0].length = sndbuf->len; @@ -140,7 +132,8 @@ static int svc_rdma_bc_sendto(struct svcxprt_rdma *rdma, atomic_inc(&rdma->sc_dma_used); memset(&send_wr, 0, sizeof(send_wr)); - send_wr.wr_id = (unsigned long)ctxt; + ctxt->cqe.done = svc_rdma_wc_send; + send_wr.wr_cqe = &ctxt->cqe; send_wr.sg_list = ctxt->sge; send_wr.num_sge = 1; send_wr.opcode = IB_WR_SEND; diff --git a/net/sunrpc/xprtrdma/svc_rdma_marshal.c b/net/sunrpc/xprtrdma/svc_rdma_marshal.c index e2fca76..765bca4 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_marshal.c +++ b/net/sunrpc/xprtrdma/svc_rdma_marshal.c @@ -145,29 +145,44 @@ static __be32 *decode_reply_array(__be32 *va, __be32 *vaend) return (__be32 *)&ary->wc_array[nchunks]; } -int svc_rdma_xdr_decode_req(struct rpcrdma_msg **rdma_req, - struct svc_rqst *rqstp) +int svc_rdma_xdr_decode_req(struct rpcrdma_msg *rmsgp, struct svc_rqst *rqstp) { - struct rpcrdma_msg *rmsgp = NULL; __be32 *va, *vaend; + unsigned int len; u32 hdr_len; - rmsgp = (struct rpcrdma_msg *)rqstp->rq_arg.head[0].iov_base; - /* Verify that there's enough bytes for header + something */ - if (rqstp->rq_arg.len <= RPCRDMA_HDRLEN_MIN) { + if (rqstp->rq_arg.len <= RPCRDMA_HDRLEN_ERR) { dprintk("svcrdma: header too short = %d\n", rqstp->rq_arg.len); return -EINVAL; } - if (rmsgp->rm_vers != rpcrdma_version) - return -ENOSYS; - - /* Pull in the extra for the padded case and bump our pointer */ - if (rmsgp->rm_type == rdma_msgp) { - int hdrlen; + if (rmsgp->rm_vers != rpcrdma_version) { + dprintk("%s: bad version %u\n", __func__, + be32_to_cpu(rmsgp->rm_vers)); + return -EPROTONOSUPPORT; + } + switch (be32_to_cpu(rmsgp->rm_type)) { + case RDMA_MSG: + case RDMA_NOMSG: + break; + + case RDMA_DONE: + /* Just drop it */ + dprintk("svcrdma: dropping RDMA_DONE message\n"); + return 0; + + case RDMA_ERROR: + /* Possible if this is a backchannel reply. + * XXX: We should cancel this XID, though. + */ + dprintk("svcrdma: dropping RDMA_ERROR message\n"); + return 0; + + case RDMA_MSGP: + /* Pull in the extra for the padded case, bump our pointer */ rmsgp->rm_body.rm_padded.rm_align = be32_to_cpu(rmsgp->rm_body.rm_padded.rm_align); rmsgp->rm_body.rm_padded.rm_thresh = @@ -175,11 +190,15 @@ int svc_rdma_xdr_decode_req(struct rpcrdma_msg **rdma_req, va = &rmsgp->rm_body.rm_padded.rm_pempty[4]; rqstp->rq_arg.head[0].iov_base = va; - hdrlen = (u32)((unsigned long)va - (unsigned long)rmsgp); - rqstp->rq_arg.head[0].iov_len -= hdrlen; - if (hdrlen > rqstp->rq_arg.len) + len = (u32)((unsigned long)va - (unsigned long)rmsgp); + rqstp->rq_arg.head[0].iov_len -= len; + if (len > rqstp->rq_arg.len) return -EINVAL; - return hdrlen; + return len; + default: + dprintk("svcrdma: bad rdma procedure (%u)\n", + be32_to_cpu(rmsgp->rm_type)); + return -EINVAL; } /* The chunk list may contain either a read chunk list or a write @@ -188,20 +207,25 @@ int svc_rdma_xdr_decode_req(struct rpcrdma_msg **rdma_req, va = &rmsgp->rm_body.rm_chunks[0]; vaend = (__be32 *)((unsigned long)rmsgp + rqstp->rq_arg.len); va = decode_read_list(va, vaend); - if (!va) + if (!va) { + dprintk("svcrdma: failed to decode read list\n"); return -EINVAL; + } va = decode_write_list(va, vaend); - if (!va) + if (!va) { + dprintk("svcrdma: failed to decode write list\n"); return -EINVAL; + } va = decode_reply_array(va, vaend); - if (!va) + if (!va) { + dprintk("svcrdma: failed to decode reply chunk\n"); return -EINVAL; + } rqstp->rq_arg.head[0].iov_base = va; hdr_len = (unsigned long)va - (unsigned long)rmsgp; rqstp->rq_arg.head[0].iov_len -= hdr_len; - *rdma_req = rmsgp; return hdr_len; } diff --git a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c index c8b8a8b..3b24a64 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c +++ b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c @@ -180,9 +180,9 @@ int rdma_read_chunk_lcl(struct svcxprt_rdma *xprt, clear_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags); memset(&read_wr, 0, sizeof(read_wr)); - read_wr.wr.wr_id = (unsigned long)ctxt; + ctxt->cqe.done = svc_rdma_wc_read; + read_wr.wr.wr_cqe = &ctxt->cqe; read_wr.wr.opcode = IB_WR_RDMA_READ; - ctxt->wr_op = read_wr.wr.opcode; read_wr.wr.send_flags = IB_SEND_SIGNALED; read_wr.rkey = rs_handle; read_wr.remote_addr = rs_offset; @@ -299,8 +299,9 @@ int rdma_read_chunk_frmr(struct svcxprt_rdma *xprt, ctxt->read_hdr = head; /* Prepare REG WR */ + ctxt->reg_cqe.done = svc_rdma_wc_reg; + reg_wr.wr.wr_cqe = &ctxt->reg_cqe; reg_wr.wr.opcode = IB_WR_REG_MR; - reg_wr.wr.wr_id = 0; reg_wr.wr.send_flags = IB_SEND_SIGNALED; reg_wr.wr.num_sge = 0; reg_wr.mr = frmr->mr; @@ -310,6 +311,8 @@ int rdma_read_chunk_frmr(struct svcxprt_rdma *xprt, /* Prepare RDMA_READ */ memset(&read_wr, 0, sizeof(read_wr)); + ctxt->cqe.done = svc_rdma_wc_read; + read_wr.wr.wr_cqe = &ctxt->cqe; read_wr.wr.send_flags = IB_SEND_SIGNALED; read_wr.rkey = rs_handle; read_wr.remote_addr = rs_offset; @@ -317,19 +320,18 @@ int rdma_read_chunk_frmr(struct svcxprt_rdma *xprt, read_wr.wr.num_sge = 1; if (xprt->sc_dev_caps & SVCRDMA_DEVCAP_READ_W_INV) { read_wr.wr.opcode = IB_WR_RDMA_READ_WITH_INV; - read_wr.wr.wr_id = (unsigned long)ctxt; read_wr.wr.ex.invalidate_rkey = ctxt->frmr->mr->lkey; } else { read_wr.wr.opcode = IB_WR_RDMA_READ; read_wr.wr.next = &inv_wr; /* Prepare invalidate */ memset(&inv_wr, 0, sizeof(inv_wr)); - inv_wr.wr_id = (unsigned long)ctxt; + ctxt->inv_cqe.done = svc_rdma_wc_inv; + inv_wr.wr_cqe = &ctxt->inv_cqe; inv_wr.opcode = IB_WR_LOCAL_INV; inv_wr.send_flags = IB_SEND_SIGNALED | IB_SEND_FENCE; inv_wr.ex.invalidate_rkey = frmr->mr->lkey; } - ctxt->wr_op = read_wr.wr.opcode; /* Post the chain */ ret = svc_rdma_send(xprt, ®_wr.wr); @@ -612,7 +614,6 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp) struct svc_rdma_op_ctxt *ctxt = NULL; struct rpcrdma_msg *rmsgp; int ret = 0; - int len; dprintk("svcrdma: rqstp=%p\n", rqstp); @@ -642,8 +643,7 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp) * transport list */ if (test_bit(XPT_CLOSE, &xprt->xpt_flags)) - goto close_out; - + goto defer; goto out; } dprintk("svcrdma: processing ctxt=%p on xprt=%p, rqstp=%p, status=%d\n", @@ -654,15 +654,13 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp) rdma_build_arg_xdr(rqstp, ctxt, ctxt->byte_len); /* Decode the RDMA header. */ - len = svc_rdma_xdr_decode_req(&rmsgp, rqstp); - rqstp->rq_xprt_hlen = len; - - /* If the request is invalid, reply with an error */ - if (len < 0) { - if (len == -ENOSYS) - svc_rdma_send_error(rdma_xprt, rmsgp, ERR_VERS); - goto close_out; - } + rmsgp = (struct rpcrdma_msg *)rqstp->rq_arg.head[0].iov_base; + ret = svc_rdma_xdr_decode_req(rmsgp, rqstp); + if (ret < 0) + goto out_err; + if (ret == 0) + goto out_drop; + rqstp->rq_xprt_hlen = ret; if (svc_rdma_is_backchannel_reply(xprt, rmsgp)) { ret = svc_rdma_handle_bc_reply(xprt->xpt_bc_xprt, rmsgp, @@ -698,26 +696,16 @@ int svc_rdma_recvfrom(struct svc_rqst *rqstp) svc_xprt_copy_addrs(rqstp, xprt); return ret; - close_out: - if (ctxt) - svc_rdma_put_context(ctxt, 1); - dprintk("svcrdma: transport %p is closing\n", xprt); - /* - * Set the close bit and enqueue it. svc_recv will see the - * close bit and call svc_xprt_delete - */ - set_bit(XPT_CLOSE, &xprt->xpt_flags); +out_err: + svc_rdma_send_error(rdma_xprt, rmsgp, ret); + svc_rdma_put_context(ctxt, 0); + return 0; + defer: return 0; +out_drop: + svc_rdma_put_context(ctxt, 1); repost: - ret = svc_rdma_post_recv(rdma_xprt, GFP_KERNEL); - if (ret) { - pr_err("svcrdma: could not post a receive buffer, err=%d.\n", - ret); - pr_err("svcrdma: closing transport %p.\n", rdma_xprt); - set_bit(XPT_CLOSE, &rdma_xprt->sc_xprt.xpt_flags); - ret = -ENOTCONN; - } - return ret; + return svc_rdma_repost_recv(rdma_xprt, GFP_KERNEL); } diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c index df57f3c..4f1b1c4 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c +++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c @@ -50,9 +50,15 @@ #define RPCDBG_FACILITY RPCDBG_SVCXPRT +static u32 xdr_padsize(u32 len) +{ + return (len & 3) ? (4 - (len & 3)) : 0; +} + int svc_rdma_map_xdr(struct svcxprt_rdma *xprt, struct xdr_buf *xdr, - struct svc_rdma_req_map *vec) + struct svc_rdma_req_map *vec, + bool write_chunk_present) { int sge_no; u32 sge_bytes; @@ -92,9 +98,20 @@ int svc_rdma_map_xdr(struct svcxprt_rdma *xprt, /* Tail SGE */ if (xdr->tail[0].iov_len) { - vec->sge[sge_no].iov_base = xdr->tail[0].iov_base; - vec->sge[sge_no].iov_len = xdr->tail[0].iov_len; - sge_no++; + unsigned char *base = xdr->tail[0].iov_base; + size_t len = xdr->tail[0].iov_len; + u32 xdr_pad = xdr_padsize(xdr->page_len); + + if (write_chunk_present && xdr_pad) { + base += xdr_pad; + len -= xdr_pad; + } + + if (len) { + vec->sge[sge_no].iov_base = base; + vec->sge[sge_no].iov_len = len; + sge_no++; + } } dprintk("svcrdma: %s: sge_no %d page_no %d " @@ -166,10 +183,10 @@ svc_rdma_get_write_array(struct rpcrdma_msg *rmsgp) * reply array is present */ static struct rpcrdma_write_array * -svc_rdma_get_reply_array(struct rpcrdma_msg *rmsgp) +svc_rdma_get_reply_array(struct rpcrdma_msg *rmsgp, + struct rpcrdma_write_array *wr_ary) { struct rpcrdma_read_chunk *rch; - struct rpcrdma_write_array *wr_ary; struct rpcrdma_write_array *rp_ary; /* XXX: Need to fix when reply chunk may occur with read list @@ -191,7 +208,6 @@ svc_rdma_get_reply_array(struct rpcrdma_msg *rmsgp) goto found_it; } - wr_ary = svc_rdma_get_write_array(rmsgp); if (wr_ary) { int chunk = be32_to_cpu(wr_ary->wc_nchunks); @@ -281,8 +297,8 @@ static int send_write(struct svcxprt_rdma *xprt, struct svc_rqst *rqstp, /* Prepare WRITE WR */ memset(&write_wr, 0, sizeof write_wr); - ctxt->wr_op = IB_WR_RDMA_WRITE; - write_wr.wr.wr_id = (unsigned long)ctxt; + ctxt->cqe.done = svc_rdma_wc_write; + write_wr.wr.wr_cqe = &ctxt->cqe; write_wr.wr.sg_list = &sge[0]; write_wr.wr.num_sge = sge_no; write_wr.wr.opcode = IB_WR_RDMA_WRITE; @@ -298,41 +314,37 @@ static int send_write(struct svcxprt_rdma *xprt, struct svc_rqst *rqstp, err: svc_rdma_unmap_dma(ctxt); svc_rdma_put_context(ctxt, 0); - /* Fatal error, close transport */ return -EIO; } +noinline static int send_write_chunks(struct svcxprt_rdma *xprt, - struct rpcrdma_msg *rdma_argp, + struct rpcrdma_write_array *wr_ary, struct rpcrdma_msg *rdma_resp, struct svc_rqst *rqstp, struct svc_rdma_req_map *vec) { - u32 xfer_len = rqstp->rq_res.page_len + rqstp->rq_res.tail[0].iov_len; + u32 xfer_len = rqstp->rq_res.page_len; int write_len; u32 xdr_off; int chunk_off; int chunk_no; int nchunks; - struct rpcrdma_write_array *arg_ary; struct rpcrdma_write_array *res_ary; int ret; - arg_ary = svc_rdma_get_write_array(rdma_argp); - if (!arg_ary) - return 0; res_ary = (struct rpcrdma_write_array *) &rdma_resp->rm_body.rm_chunks[1]; /* Write chunks start at the pagelist */ - nchunks = be32_to_cpu(arg_ary->wc_nchunks); + nchunks = be32_to_cpu(wr_ary->wc_nchunks); for (xdr_off = rqstp->rq_res.head[0].iov_len, chunk_no = 0; xfer_len && chunk_no < nchunks; chunk_no++) { struct rpcrdma_segment *arg_ch; u64 rs_offset; - arg_ch = &arg_ary->wc_array[chunk_no].wc_target; + arg_ch = &wr_ary->wc_array[chunk_no].wc_target; write_len = min(xfer_len, be32_to_cpu(arg_ch->rs_length)); /* Prepare the response chunk given the length actually @@ -350,11 +362,8 @@ static int send_write_chunks(struct svcxprt_rdma *xprt, xdr_off, write_len, vec); - if (ret <= 0) { - dprintk("svcrdma: RDMA_WRITE failed, ret=%d\n", - ret); - return -EIO; - } + if (ret <= 0) + goto out_err; chunk_off += ret; xdr_off += ret; xfer_len -= ret; @@ -364,11 +373,16 @@ static int send_write_chunks(struct svcxprt_rdma *xprt, /* Update the req with the number of chunks actually used */ svc_rdma_xdr_encode_write_list(rdma_resp, chunk_no); - return rqstp->rq_res.page_len + rqstp->rq_res.tail[0].iov_len; + return rqstp->rq_res.page_len; + +out_err: + pr_err("svcrdma: failed to send write chunks, rc=%d\n", ret); + return -EIO; } +noinline static int send_reply_chunks(struct svcxprt_rdma *xprt, - struct rpcrdma_msg *rdma_argp, + struct rpcrdma_write_array *rp_ary, struct rpcrdma_msg *rdma_resp, struct svc_rqst *rqstp, struct svc_rdma_req_map *vec) @@ -380,25 +394,21 @@ static int send_reply_chunks(struct svcxprt_rdma *xprt, int chunk_off; int nchunks; struct rpcrdma_segment *ch; - struct rpcrdma_write_array *arg_ary; struct rpcrdma_write_array *res_ary; int ret; - arg_ary = svc_rdma_get_reply_array(rdma_argp); - if (!arg_ary) - return 0; /* XXX: need to fix when reply lists occur with read-list and or * write-list */ res_ary = (struct rpcrdma_write_array *) &rdma_resp->rm_body.rm_chunks[2]; /* xdr offset starts at RPC message */ - nchunks = be32_to_cpu(arg_ary->wc_nchunks); + nchunks = be32_to_cpu(rp_ary->wc_nchunks); for (xdr_off = 0, chunk_no = 0; xfer_len && chunk_no < nchunks; chunk_no++) { u64 rs_offset; - ch = &arg_ary->wc_array[chunk_no].wc_target; + ch = &rp_ary->wc_array[chunk_no].wc_target; write_len = min(xfer_len, be32_to_cpu(ch->rs_length)); /* Prepare the reply chunk given the length actually @@ -415,11 +425,8 @@ static int send_reply_chunks(struct svcxprt_rdma *xprt, xdr_off, write_len, vec); - if (ret <= 0) { - dprintk("svcrdma: RDMA_WRITE failed, ret=%d\n", - ret); - return -EIO; - } + if (ret <= 0) + goto out_err; chunk_off += ret; xdr_off += ret; xfer_len -= ret; @@ -430,6 +437,10 @@ static int send_reply_chunks(struct svcxprt_rdma *xprt, svc_rdma_xdr_encode_reply_array(res_ary, chunk_no); return rqstp->rq_res.len; + +out_err: + pr_err("svcrdma: failed to send reply chunks, rc=%d\n", ret); + return -EIO; } /* This function prepares the portion of the RPCRDMA message to be @@ -464,13 +475,8 @@ static int send_reply(struct svcxprt_rdma *rdma, int pages; int ret; - /* Post a recv buffer to handle another request. */ - ret = svc_rdma_post_recv(rdma, GFP_KERNEL); + ret = svc_rdma_repost_recv(rdma, GFP_KERNEL); if (ret) { - printk(KERN_INFO - "svcrdma: could not post a receive buffer, err=%d." - "Closing transport %p.\n", ret, rdma); - set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags); svc_rdma_put_context(ctxt, 0); return -ENOTCONN; } @@ -543,8 +549,8 @@ static int send_reply(struct svcxprt_rdma *rdma, goto err; } memset(&send_wr, 0, sizeof send_wr); - ctxt->wr_op = IB_WR_SEND; - send_wr.wr_id = (unsigned long)ctxt; + ctxt->cqe.done = svc_rdma_wc_send; + send_wr.wr_cqe = &ctxt->cqe; send_wr.sg_list = ctxt->sge; send_wr.num_sge = sge_no; send_wr.opcode = IB_WR_SEND; @@ -559,6 +565,7 @@ 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; } @@ -573,7 +580,7 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) container_of(xprt, struct svcxprt_rdma, sc_xprt); struct rpcrdma_msg *rdma_argp; struct rpcrdma_msg *rdma_resp; - struct rpcrdma_write_array *reply_ary; + struct rpcrdma_write_array *wr_ary, *rp_ary; enum rpcrdma_proc reply_type; int ret; int inline_bytes; @@ -587,12 +594,14 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) * places this at the start of page 0. */ rdma_argp = page_address(rqstp->rq_pages[0]); + wr_ary = svc_rdma_get_write_array(rdma_argp); + 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); + ret = svc_rdma_map_xdr(rdma, &rqstp->rq_res, vec, wr_ary != NULL); if (ret) goto err0; inline_bytes = rqstp->rq_res.len; @@ -603,8 +612,7 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) if (!res_page) goto err0; rdma_resp = page_address(res_page); - reply_ary = svc_rdma_get_reply_array(rdma_argp); - if (reply_ary) + if (rp_ary) reply_type = RDMA_NOMSG; else reply_type = RDMA_MSG; @@ -612,27 +620,26 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) rdma_resp, reply_type); /* Send any write-chunk data and build resp write-list */ - ret = send_write_chunks(rdma, rdma_argp, rdma_resp, - rqstp, vec); - if (ret < 0) { - printk(KERN_ERR "svcrdma: failed to send write chunks, rc=%d\n", - ret); - goto err1; + if (wr_ary) { + ret = send_write_chunks(rdma, wr_ary, rdma_resp, rqstp, vec); + if (ret < 0) + goto err1; + inline_bytes -= ret + xdr_padsize(ret); } - inline_bytes -= ret; /* Send any reply-list data and update resp reply-list */ - ret = send_reply_chunks(rdma, rdma_argp, rdma_resp, - rqstp, vec); - if (ret < 0) { - printk(KERN_ERR "svcrdma: failed to send reply chunks, rc=%d\n", - ret); - goto err1; + if (rp_ary) { + ret = send_reply_chunks(rdma, rp_ary, rdma_resp, rqstp, vec); + if (ret < 0) + goto err1; + inline_bytes -= ret; } - inline_bytes -= ret; ret = send_reply(rdma, rqstp, res_page, rdma_resp, ctxt, vec, inline_bytes); + if (ret < 0) + goto err1; + svc_rdma_put_req_map(rdma, vec); dprintk("svcrdma: send_reply returns %d\n", ret); return ret; @@ -642,5 +649,68 @@ int svc_rdma_sendto(struct svc_rqst *rqstp) err0: svc_rdma_put_req_map(rdma, vec); svc_rdma_put_context(ctxt, 0); - return ret; + set_bit(XPT_CLOSE, &rdma->sc_xprt.xpt_flags); + return -ENOTCONN; +} + +void svc_rdma_send_error(struct svcxprt_rdma *xprt, struct rpcrdma_msg *rmsgp, + int status) +{ + struct ib_send_wr err_wr; + struct page *p; + struct svc_rdma_op_ctxt *ctxt; + enum rpcrdma_errcode err; + __be32 *va; + int length; + int ret; + + ret = svc_rdma_repost_recv(xprt, GFP_KERNEL); + if (ret) + return; + + p = alloc_page(GFP_KERNEL); + if (!p) + return; + va = page_address(p); + + /* XDR encode an error reply */ + err = ERR_CHUNK; + if (status == -EPROTONOSUPPORT) + err = ERR_VERS; + length = svc_rdma_xdr_encode_error(xprt, rmsgp, err, va); + + ctxt = svc_rdma_get_context(xprt); + ctxt->direction = DMA_TO_DEVICE; + ctxt->count = 1; + ctxt->pages[0] = p; + + /* Prepare SGE for local address */ + ctxt->sge[0].lkey = xprt->sc_pd->local_dma_lkey; + ctxt->sge[0].length = length; + ctxt->sge[0].addr = ib_dma_map_page(xprt->sc_cm_id->device, + p, 0, length, DMA_TO_DEVICE); + if (ib_dma_mapping_error(xprt->sc_cm_id->device, ctxt->sge[0].addr)) { + dprintk("svcrdma: Error mapping buffer for protocol error\n"); + svc_rdma_put_context(ctxt, 1); + return; + } + atomic_inc(&xprt->sc_dma_used); + + /* Prepare SEND WR */ + memset(&err_wr, 0, sizeof(err_wr)); + ctxt->cqe.done = svc_rdma_wc_send; + err_wr.wr_cqe = &ctxt->cqe; + err_wr.sg_list = ctxt->sge; + err_wr.num_sge = 1; + err_wr.opcode = IB_WR_SEND; + err_wr.send_flags = IB_SEND_SIGNALED; + + /* Post It */ + ret = svc_rdma_send(xprt, &err_wr); + if (ret) { + dprintk("svcrdma: Error %d posting send for protocol error\n", + ret); + svc_rdma_unmap_dma(ctxt); + svc_rdma_put_context(ctxt, 1); + } } diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c index 5763825..9066896 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_transport.c +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c @@ -63,17 +63,10 @@ static struct svc_xprt *svc_rdma_create(struct svc_serv *serv, int flags); static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt); static void svc_rdma_release_rqst(struct svc_rqst *); -static void dto_tasklet_func(unsigned long data); static void svc_rdma_detach(struct svc_xprt *xprt); static void svc_rdma_free(struct svc_xprt *xprt); static int svc_rdma_has_wspace(struct svc_xprt *xprt); static int svc_rdma_secure_port(struct svc_rqst *); -static void rq_cq_reap(struct svcxprt_rdma *xprt); -static void sq_cq_reap(struct svcxprt_rdma *xprt); - -static DECLARE_TASKLET(dto_tasklet, dto_tasklet_func, 0UL); -static DEFINE_SPINLOCK(dto_lock); -static LIST_HEAD(dto_xprt_q); static struct svc_xprt_ops svc_rdma_ops = { .xpo_create = svc_rdma_create, @@ -352,15 +345,6 @@ static void svc_rdma_destroy_maps(struct svcxprt_rdma *xprt) } } -/* ib_cq event handler */ -static void cq_event_handler(struct ib_event *event, void *context) -{ - struct svc_xprt *xprt = context; - dprintk("svcrdma: received CQ event %s (%d), context=%p\n", - ib_event_msg(event->event), event->event, context); - set_bit(XPT_CLOSE, &xprt->xpt_flags); -} - /* QP event handler */ static void qp_event_handler(struct ib_event *event, void *context) { @@ -392,251 +376,171 @@ static void qp_event_handler(struct ib_event *event, void *context) } } -/* - * Data Transfer Operation Tasklet +/** + * svc_rdma_wc_receive - Invoked by RDMA provider for each polled Receive WC + * @cq: completion queue + * @wc: completed WR * - * Walks a list of transports with I/O pending, removing entries as - * they are added to the server's I/O pending list. Two bits indicate - * if SQ, RQ, or both have I/O pending. The dto_lock is an irqsave - * spinlock that serializes access to the transport list with the RQ - * and SQ interrupt handlers. */ -static void dto_tasklet_func(unsigned long data) +static void svc_rdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc) { - struct svcxprt_rdma *xprt; - unsigned long flags; + struct svcxprt_rdma *xprt = cq->cq_context; + struct ib_cqe *cqe = wc->wr_cqe; + struct svc_rdma_op_ctxt *ctxt; - spin_lock_irqsave(&dto_lock, flags); - while (!list_empty(&dto_xprt_q)) { - xprt = list_entry(dto_xprt_q.next, - struct svcxprt_rdma, sc_dto_q); - list_del_init(&xprt->sc_dto_q); - spin_unlock_irqrestore(&dto_lock, flags); + /* WARNING: Only wc->wr_cqe and wc->status are reliable */ + ctxt = container_of(cqe, struct svc_rdma_op_ctxt, cqe); + ctxt->wc_status = wc->status; + svc_rdma_unmap_dma(ctxt); - rq_cq_reap(xprt); - sq_cq_reap(xprt); + if (wc->status != IB_WC_SUCCESS) + goto flushed; - svc_xprt_put(&xprt->sc_xprt); - spin_lock_irqsave(&dto_lock, flags); - } - spin_unlock_irqrestore(&dto_lock, flags); + /* All wc fields are now known to be valid */ + ctxt->byte_len = wc->byte_len; + spin_lock(&xprt->sc_rq_dto_lock); + list_add_tail(&ctxt->dto_q, &xprt->sc_rq_dto_q); + spin_unlock(&xprt->sc_rq_dto_lock); + + set_bit(XPT_DATA, &xprt->sc_xprt.xpt_flags); + if (test_bit(RDMAXPRT_CONN_PENDING, &xprt->sc_flags)) + goto out; + svc_xprt_enqueue(&xprt->sc_xprt); + goto out; + +flushed: + if (wc->status != IB_WC_WR_FLUSH_ERR) + pr_warn("svcrdma: receive: %s (%u/0x%x)\n", + ib_wc_status_msg(wc->status), + wc->status, wc->vendor_err); + set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags); + svc_rdma_put_context(ctxt, 1); + +out: + svc_xprt_put(&xprt->sc_xprt); } -/* - * Receive Queue Completion Handler - * - * Since an RQ completion handler is called on interrupt context, we - * need to defer the handling of the I/O to a tasklet - */ -static void rq_comp_handler(struct ib_cq *cq, void *cq_context) +static void svc_rdma_send_wc_common(struct svcxprt_rdma *xprt, + struct ib_wc *wc, + const char *opname) { - struct svcxprt_rdma *xprt = cq_context; - unsigned long flags; - - /* Guard against unconditional flush call for destroyed QP */ - if (atomic_read(&xprt->sc_xprt.xpt_ref.refcount)==0) - return; + if (wc->status != IB_WC_SUCCESS) + goto err; - /* - * Set the bit regardless of whether or not it's on the list - * because it may be on the list already due to an SQ - * completion. - */ - set_bit(RDMAXPRT_RQ_PENDING, &xprt->sc_flags); +out: + atomic_dec(&xprt->sc_sq_count); + wake_up(&xprt->sc_send_wait); + return; + +err: + set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags); + if (wc->status != IB_WC_WR_FLUSH_ERR) + pr_err("svcrdma: %s: %s (%u/0x%x)\n", + opname, ib_wc_status_msg(wc->status), + wc->status, wc->vendor_err); + goto out; +} - /* - * If this transport is not already on the DTO transport queue, - * add it - */ - spin_lock_irqsave(&dto_lock, flags); - if (list_empty(&xprt->sc_dto_q)) { - svc_xprt_get(&xprt->sc_xprt); - list_add_tail(&xprt->sc_dto_q, &dto_xprt_q); - } - spin_unlock_irqrestore(&dto_lock, flags); +static void svc_rdma_send_wc_common_put(struct ib_cq *cq, struct ib_wc *wc, + const char *opname) +{ + struct svcxprt_rdma *xprt = cq->cq_context; - /* Tasklet does all the work to avoid irqsave locks. */ - tasklet_schedule(&dto_tasklet); + svc_rdma_send_wc_common(xprt, wc, opname); + svc_xprt_put(&xprt->sc_xprt); } -/* - * rq_cq_reap - Process the RQ CQ. - * - * Take all completing WC off the CQE and enqueue the associated DTO - * context on the dto_q for the transport. +/** + * svc_rdma_wc_send - Invoked by RDMA provider for each polled Send WC + * @cq: completion queue + * @wc: completed WR * - * Note that caller must hold a transport reference. */ -static void rq_cq_reap(struct svcxprt_rdma *xprt) +void svc_rdma_wc_send(struct ib_cq *cq, struct ib_wc *wc) { - int ret; - struct ib_wc wc; - struct svc_rdma_op_ctxt *ctxt = NULL; + struct ib_cqe *cqe = wc->wr_cqe; + struct svc_rdma_op_ctxt *ctxt; - if (!test_and_clear_bit(RDMAXPRT_RQ_PENDING, &xprt->sc_flags)) - return; + svc_rdma_send_wc_common_put(cq, wc, "send"); - ib_req_notify_cq(xprt->sc_rq_cq, IB_CQ_NEXT_COMP); - atomic_inc(&rdma_stat_rq_poll); + ctxt = container_of(cqe, struct svc_rdma_op_ctxt, cqe); + svc_rdma_unmap_dma(ctxt); + svc_rdma_put_context(ctxt, 1); +} - while ((ret = ib_poll_cq(xprt->sc_rq_cq, 1, &wc)) > 0) { - ctxt = (struct svc_rdma_op_ctxt *)(unsigned long)wc.wr_id; - ctxt->wc_status = wc.status; - ctxt->byte_len = wc.byte_len; - svc_rdma_unmap_dma(ctxt); - if (wc.status != IB_WC_SUCCESS) { - /* Close the transport */ - dprintk("svcrdma: transport closing putting ctxt %p\n", ctxt); - set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags); - svc_rdma_put_context(ctxt, 1); - svc_xprt_put(&xprt->sc_xprt); - continue; - } - spin_lock_bh(&xprt->sc_rq_dto_lock); - list_add_tail(&ctxt->dto_q, &xprt->sc_rq_dto_q); - spin_unlock_bh(&xprt->sc_rq_dto_lock); - svc_xprt_put(&xprt->sc_xprt); - } +/** + * svc_rdma_wc_write - Invoked by RDMA provider for each polled Write WC + * @cq: completion queue + * @wc: completed WR + * + */ +void svc_rdma_wc_write(struct ib_cq *cq, struct ib_wc *wc) +{ + struct ib_cqe *cqe = wc->wr_cqe; + struct svc_rdma_op_ctxt *ctxt; - if (ctxt) - atomic_inc(&rdma_stat_rq_prod); + svc_rdma_send_wc_common_put(cq, wc, "write"); - set_bit(XPT_DATA, &xprt->sc_xprt.xpt_flags); - /* - * If data arrived before established event, - * don't enqueue. This defers RPC I/O until the - * RDMA connection is complete. - */ - if (!test_bit(RDMAXPRT_CONN_PENDING, &xprt->sc_flags)) - svc_xprt_enqueue(&xprt->sc_xprt); + ctxt = container_of(cqe, struct svc_rdma_op_ctxt, cqe); + svc_rdma_unmap_dma(ctxt); + svc_rdma_put_context(ctxt, 0); } -/* - * Process a completion context +/** + * svc_rdma_wc_reg - Invoked by RDMA provider for each polled FASTREG WC + * @cq: completion queue + * @wc: completed WR + * */ -static void process_context(struct svcxprt_rdma *xprt, - struct svc_rdma_op_ctxt *ctxt) +void svc_rdma_wc_reg(struct ib_cq *cq, struct ib_wc *wc) { - struct svc_rdma_op_ctxt *read_hdr; - int free_pages = 0; - - svc_rdma_unmap_dma(ctxt); + svc_rdma_send_wc_common_put(cq, wc, "fastreg"); +} - switch (ctxt->wr_op) { - case IB_WR_SEND: - free_pages = 1; - break; +/** + * svc_rdma_wc_read - Invoked by RDMA provider for each polled Read WC + * @cq: completion queue + * @wc: completed WR + * + */ +void svc_rdma_wc_read(struct ib_cq *cq, struct ib_wc *wc) +{ + struct svcxprt_rdma *xprt = cq->cq_context; + struct ib_cqe *cqe = wc->wr_cqe; + struct svc_rdma_op_ctxt *ctxt; - case IB_WR_RDMA_WRITE: - break; + svc_rdma_send_wc_common(xprt, wc, "read"); - case IB_WR_RDMA_READ: - case IB_WR_RDMA_READ_WITH_INV: - svc_rdma_put_frmr(xprt, ctxt->frmr); + ctxt = container_of(cqe, struct svc_rdma_op_ctxt, cqe); + svc_rdma_unmap_dma(ctxt); + svc_rdma_put_frmr(xprt, ctxt->frmr); - if (!test_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags)) - break; + if (test_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags)) { + struct svc_rdma_op_ctxt *read_hdr; read_hdr = ctxt->read_hdr; - svc_rdma_put_context(ctxt, 0); - - spin_lock_bh(&xprt->sc_rq_dto_lock); - set_bit(XPT_DATA, &xprt->sc_xprt.xpt_flags); + spin_lock(&xprt->sc_rq_dto_lock); list_add_tail(&read_hdr->dto_q, &xprt->sc_read_complete_q); - spin_unlock_bh(&xprt->sc_rq_dto_lock); - svc_xprt_enqueue(&xprt->sc_xprt); - return; + spin_unlock(&xprt->sc_rq_dto_lock); - default: - dprintk("svcrdma: unexpected completion opcode=%d\n", - ctxt->wr_op); - break; + set_bit(XPT_DATA, &xprt->sc_xprt.xpt_flags); + svc_xprt_enqueue(&xprt->sc_xprt); } - svc_rdma_put_context(ctxt, free_pages); + svc_rdma_put_context(ctxt, 0); + svc_xprt_put(&xprt->sc_xprt); } -/* - * Send Queue Completion Handler - potentially called on interrupt context. +/** + * svc_rdma_wc_inv - Invoked by RDMA provider for each polled LOCAL_INV WC + * @cq: completion queue + * @wc: completed WR * - * Note that caller must hold a transport reference. */ -static void sq_cq_reap(struct svcxprt_rdma *xprt) -{ - struct svc_rdma_op_ctxt *ctxt = NULL; - struct ib_wc wc_a[6]; - struct ib_wc *wc; - struct ib_cq *cq = xprt->sc_sq_cq; - int ret; - - memset(wc_a, 0, sizeof(wc_a)); - - if (!test_and_clear_bit(RDMAXPRT_SQ_PENDING, &xprt->sc_flags)) - return; - - ib_req_notify_cq(xprt->sc_sq_cq, IB_CQ_NEXT_COMP); - atomic_inc(&rdma_stat_sq_poll); - while ((ret = ib_poll_cq(cq, ARRAY_SIZE(wc_a), wc_a)) > 0) { - int i; - - for (i = 0; i < ret; i++) { - wc = &wc_a[i]; - if (wc->status != IB_WC_SUCCESS) { - dprintk("svcrdma: sq wc err status %s (%d)\n", - ib_wc_status_msg(wc->status), - wc->status); - - /* Close the transport */ - set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags); - } - - /* Decrement used SQ WR count */ - atomic_dec(&xprt->sc_sq_count); - wake_up(&xprt->sc_send_wait); - - ctxt = (struct svc_rdma_op_ctxt *) - (unsigned long)wc->wr_id; - if (ctxt) - process_context(xprt, ctxt); - - svc_xprt_put(&xprt->sc_xprt); - } - } - - if (ctxt) - atomic_inc(&rdma_stat_sq_prod); -} - -static void sq_comp_handler(struct ib_cq *cq, void *cq_context) +void svc_rdma_wc_inv(struct ib_cq *cq, struct ib_wc *wc) { - struct svcxprt_rdma *xprt = cq_context; - unsigned long flags; - - /* Guard against unconditional flush call for destroyed QP */ - if (atomic_read(&xprt->sc_xprt.xpt_ref.refcount)==0) - return; - - /* - * Set the bit regardless of whether or not it's on the list - * because it may be on the list already due to an RQ - * completion. - */ - set_bit(RDMAXPRT_SQ_PENDING, &xprt->sc_flags); - - /* - * If this transport is not already on the DTO transport queue, - * add it - */ - spin_lock_irqsave(&dto_lock, flags); - if (list_empty(&xprt->sc_dto_q)) { - svc_xprt_get(&xprt->sc_xprt); - list_add_tail(&xprt->sc_dto_q, &dto_xprt_q); - } - spin_unlock_irqrestore(&dto_lock, flags); - - /* Tasklet does all the work to avoid irqsave locks. */ - tasklet_schedule(&dto_tasklet); + svc_rdma_send_wc_common_put(cq, wc, "localInv"); } static struct svcxprt_rdma *rdma_create_xprt(struct svc_serv *serv, @@ -681,6 +585,7 @@ int svc_rdma_post_recv(struct svcxprt_rdma *xprt, gfp_t flags) ctxt = svc_rdma_get_context(xprt); buflen = 0; ctxt->direction = DMA_FROM_DEVICE; + ctxt->cqe.done = svc_rdma_wc_receive; for (sge_no = 0; buflen < xprt->sc_max_req_size; sge_no++) { if (sge_no >= xprt->sc_max_sge) { pr_err("svcrdma: Too many sges (%d)\n", sge_no); @@ -705,7 +610,7 @@ int svc_rdma_post_recv(struct svcxprt_rdma *xprt, gfp_t flags) recv_wr.next = NULL; recv_wr.sg_list = &ctxt->sge[0]; recv_wr.num_sge = ctxt->count; - recv_wr.wr_id = (u64)(unsigned long)ctxt; + recv_wr.wr_cqe = &ctxt->cqe; svc_xprt_get(&xprt->sc_xprt); ret = ib_post_recv(xprt->sc_qp, &recv_wr, &bad_recv_wr); @@ -722,6 +627,21 @@ int svc_rdma_post_recv(struct svcxprt_rdma *xprt, gfp_t flags) return -ENOMEM; } +int svc_rdma_repost_recv(struct svcxprt_rdma *xprt, gfp_t flags) +{ + int ret = 0; + + ret = svc_rdma_post_recv(xprt, flags); + if (ret) { + pr_err("svcrdma: could not post a receive buffer, err=%d.\n", + ret); + pr_err("svcrdma: closing transport %p.\n", xprt); + set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags); + ret = -ENOTCONN; + } + return ret; +} + /* * This function handles the CONNECT_REQUEST event on a listening * endpoint. It is passed the cma_id for the _new_ connection. The context in @@ -1011,7 +931,6 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt) struct svcxprt_rdma *listen_rdma; struct svcxprt_rdma *newxprt = NULL; struct rdma_conn_param conn_param; - struct ib_cq_init_attr cq_attr = {}; struct ib_qp_init_attr qp_attr; struct ib_device *dev; unsigned int i; @@ -1069,22 +988,14 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt) dprintk("svcrdma: error creating PD for connect request\n"); goto errout; } - cq_attr.cqe = newxprt->sc_sq_depth; - newxprt->sc_sq_cq = ib_create_cq(dev, - sq_comp_handler, - cq_event_handler, - newxprt, - &cq_attr); + newxprt->sc_sq_cq = ib_alloc_cq(dev, newxprt, newxprt->sc_sq_depth, + 0, IB_POLL_SOFTIRQ); if (IS_ERR(newxprt->sc_sq_cq)) { dprintk("svcrdma: error creating SQ CQ for connect request\n"); goto errout; } - cq_attr.cqe = newxprt->sc_rq_depth; - newxprt->sc_rq_cq = ib_create_cq(dev, - rq_comp_handler, - cq_event_handler, - newxprt, - &cq_attr); + newxprt->sc_rq_cq = ib_alloc_cq(dev, newxprt, newxprt->sc_rq_depth, + 0, IB_POLL_SOFTIRQ); if (IS_ERR(newxprt->sc_rq_cq)) { dprintk("svcrdma: error creating RQ CQ for connect request\n"); goto errout; @@ -1173,13 +1084,6 @@ static struct svc_xprt *svc_rdma_accept(struct svc_xprt *xprt) /* Swap out the handler */ newxprt->sc_cm_id->event_handler = rdma_cma_handler; - /* - * Arm the CQs for the SQ and RQ before accepting so we can't - * miss the first message - */ - ib_req_notify_cq(newxprt->sc_sq_cq, IB_CQ_NEXT_COMP); - ib_req_notify_cq(newxprt->sc_rq_cq, IB_CQ_NEXT_COMP); - /* Accept Connection */ set_bit(RDMAXPRT_CONN_PENDING, &newxprt->sc_flags); memset(&conn_param, 0, sizeof conn_param); @@ -1319,10 +1223,10 @@ static void __svc_rdma_free(struct work_struct *work) ib_destroy_qp(rdma->sc_qp); if (rdma->sc_sq_cq && !IS_ERR(rdma->sc_sq_cq)) - ib_destroy_cq(rdma->sc_sq_cq); + ib_free_cq(rdma->sc_sq_cq); if (rdma->sc_rq_cq && !IS_ERR(rdma->sc_rq_cq)) - ib_destroy_cq(rdma->sc_rq_cq); + ib_free_cq(rdma->sc_rq_cq); if (rdma->sc_pd && !IS_ERR(rdma->sc_pd)) ib_dealloc_pd(rdma->sc_pd); @@ -1383,9 +1287,6 @@ int svc_rdma_send(struct svcxprt_rdma *xprt, struct ib_send_wr *wr) spin_unlock_bh(&xprt->sc_lock); atomic_inc(&rdma_stat_sq_starve); - /* See if we can opportunistically reap SQ WR to make room */ - sq_cq_reap(xprt); - /* Wait until SQ WR available if SQ still full */ wait_event(xprt->sc_send_wait, atomic_read(&xprt->sc_sq_count) < @@ -1418,57 +1319,3 @@ int svc_rdma_send(struct svcxprt_rdma *xprt, struct ib_send_wr *wr) } return ret; } - -void svc_rdma_send_error(struct svcxprt_rdma *xprt, struct rpcrdma_msg *rmsgp, - enum rpcrdma_errcode err) -{ - struct ib_send_wr err_wr; - struct page *p; - struct svc_rdma_op_ctxt *ctxt; - __be32 *va; - int length; - int ret; - - p = alloc_page(GFP_KERNEL); - if (!p) - return; - va = page_address(p); - - /* XDR encode error */ - length = svc_rdma_xdr_encode_error(xprt, rmsgp, err, va); - - ctxt = svc_rdma_get_context(xprt); - ctxt->direction = DMA_FROM_DEVICE; - ctxt->count = 1; - ctxt->pages[0] = p; - - /* Prepare SGE for local address */ - ctxt->sge[0].addr = ib_dma_map_page(xprt->sc_cm_id->device, - p, 0, length, DMA_FROM_DEVICE); - if (ib_dma_mapping_error(xprt->sc_cm_id->device, ctxt->sge[0].addr)) { - put_page(p); - svc_rdma_put_context(ctxt, 1); - return; - } - atomic_inc(&xprt->sc_dma_used); - ctxt->sge[0].lkey = xprt->sc_pd->local_dma_lkey; - ctxt->sge[0].length = length; - - /* Prepare SEND WR */ - memset(&err_wr, 0, sizeof err_wr); - ctxt->wr_op = IB_WR_SEND; - err_wr.wr_id = (unsigned long)ctxt; - err_wr.sg_list = ctxt->sge; - err_wr.num_sge = 1; - err_wr.opcode = IB_WR_SEND; - err_wr.send_flags = IB_SEND_SIGNALED; - - /* Post It */ - ret = svc_rdma_send(xprt, &err_wr); - if (ret) { - dprintk("svcrdma: Error %d posting send for protocol error\n", - ret); - svc_rdma_unmap_dma(ctxt); - svc_rdma_put_context(ctxt, 1); - } -} diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c index 878f1bf..f5ed9f9 100644 --- a/net/sunrpc/xprtrdma/verbs.c +++ b/net/sunrpc/xprtrdma/verbs.c @@ -112,89 +112,65 @@ rpcrdma_qp_async_error_upcall(struct ib_event *event, void *context) } } +/** + * rpcrdma_wc_send - Invoked by RDMA provider for each polled Send WC + * @cq: completion queue (ignored) + * @wc: completed WR + * + */ static void -rpcrdma_cq_async_error_upcall(struct ib_event *event, void *context) +rpcrdma_wc_send(struct ib_cq *cq, struct ib_wc *wc) { - struct rpcrdma_ep *ep = context; - - pr_err("RPC: %s: %s on device %s ep %p\n", - __func__, ib_event_msg(event->event), - event->device->name, context); - if (ep->rep_connected == 1) { - ep->rep_connected = -EIO; - rpcrdma_conn_func(ep); - wake_up_all(&ep->rep_connect_wait); - } + /* WARNING: Only wr_cqe and status are reliable at this point */ + if (wc->status != IB_WC_SUCCESS && wc->status != IB_WC_WR_FLUSH_ERR) + pr_err("rpcrdma: Send: %s (%u/0x%x)\n", + ib_wc_status_msg(wc->status), + wc->status, wc->vendor_err); } static void -rpcrdma_sendcq_process_wc(struct ib_wc *wc) +rpcrdma_receive_worker(struct work_struct *work) { - /* WARNING: Only wr_id and status are reliable at this point */ - if (wc->wr_id == RPCRDMA_IGNORE_COMPLETION) { - if (wc->status != IB_WC_SUCCESS && - wc->status != IB_WC_WR_FLUSH_ERR) - pr_err("RPC: %s: SEND: %s\n", - __func__, ib_wc_status_msg(wc->status)); - } else { - struct rpcrdma_mw *r; + struct rpcrdma_rep *rep = + container_of(work, struct rpcrdma_rep, rr_work); - r = (struct rpcrdma_mw *)(unsigned long)wc->wr_id; - r->mw_sendcompletion(wc); - } + rpcrdma_reply_handler(rep); } -/* The common case is a single send completion is waiting. By - * passing two WC entries to ib_poll_cq, a return code of 1 - * means there is exactly one WC waiting and no more. We don't - * have to invoke ib_poll_cq again to know that the CQ has been - * properly drained. +/* Perform basic sanity checking to avoid using garbage + * to update the credit grant value. */ static void -rpcrdma_sendcq_poll(struct ib_cq *cq) +rpcrdma_update_granted_credits(struct rpcrdma_rep *rep) { - struct ib_wc *pos, wcs[2]; - int count, rc; + struct rpcrdma_msg *rmsgp = rdmab_to_msg(rep->rr_rdmabuf); + struct rpcrdma_buffer *buffer = &rep->rr_rxprt->rx_buf; + u32 credits; - do { - pos = wcs; + if (rep->rr_len < RPCRDMA_HDRLEN_ERR) + return; - rc = ib_poll_cq(cq, ARRAY_SIZE(wcs), pos); - if (rc < 0) - break; + credits = be32_to_cpu(rmsgp->rm_credit); + if (credits == 0) + credits = 1; /* don't deadlock */ + else if (credits > buffer->rb_max_requests) + credits = buffer->rb_max_requests; - count = rc; - while (count-- > 0) - rpcrdma_sendcq_process_wc(pos++); - } while (rc == ARRAY_SIZE(wcs)); - return; + atomic_set(&buffer->rb_credits, credits); } -/* Handle provider send completion upcalls. +/** + * rpcrdma_receive_wc - Invoked by RDMA provider for each polled Receive WC + * @cq: completion queue (ignored) + * @wc: completed WR + * */ static void -rpcrdma_sendcq_upcall(struct ib_cq *cq, void *cq_context) +rpcrdma_receive_wc(struct ib_cq *cq, struct ib_wc *wc) { - do { - rpcrdma_sendcq_poll(cq); - } while (ib_req_notify_cq(cq, IB_CQ_NEXT_COMP | - IB_CQ_REPORT_MISSED_EVENTS) > 0); -} - -static void -rpcrdma_receive_worker(struct work_struct *work) -{ - struct rpcrdma_rep *rep = - container_of(work, struct rpcrdma_rep, rr_work); - - rpcrdma_reply_handler(rep); -} - -static void -rpcrdma_recvcq_process_wc(struct ib_wc *wc) -{ - struct rpcrdma_rep *rep = - (struct rpcrdma_rep *)(unsigned long)wc->wr_id; + struct ib_cqe *cqe = wc->wr_cqe; + struct rpcrdma_rep *rep = container_of(cqe, struct rpcrdma_rep, + rr_cqe); /* WARNING: Only wr_id and status are reliable at this point */ if (wc->status != IB_WC_SUCCESS) @@ -211,7 +187,8 @@ rpcrdma_recvcq_process_wc(struct ib_wc *wc) ib_dma_sync_single_for_cpu(rep->rr_device, rdmab_addr(rep->rr_rdmabuf), rep->rr_len, DMA_FROM_DEVICE); - prefetch(rdmab_to_msg(rep->rr_rdmabuf)); + + rpcrdma_update_granted_credits(rep); out_schedule: queue_work(rpcrdma_receive_wq, &rep->rr_work); @@ -219,57 +196,20 @@ out_schedule: out_fail: if (wc->status != IB_WC_WR_FLUSH_ERR) - pr_err("RPC: %s: rep %p: %s\n", - __func__, rep, ib_wc_status_msg(wc->status)); + pr_err("rpcrdma: Recv: %s (%u/0x%x)\n", + ib_wc_status_msg(wc->status), + wc->status, wc->vendor_err); rep->rr_len = RPCRDMA_BAD_LEN; goto out_schedule; } -/* The wc array is on stack: automatic memory is always CPU-local. - * - * struct ib_wc is 64 bytes, making the poll array potentially - * large. But this is at the bottom of the call chain. Further - * substantial work is done in another thread. - */ -static void -rpcrdma_recvcq_poll(struct ib_cq *cq) -{ - struct ib_wc *pos, wcs[4]; - int count, rc; - - do { - pos = wcs; - - rc = ib_poll_cq(cq, ARRAY_SIZE(wcs), pos); - if (rc < 0) - break; - - count = rc; - while (count-- > 0) - rpcrdma_recvcq_process_wc(pos++); - } while (rc == ARRAY_SIZE(wcs)); -} - -/* Handle provider receive completion upcalls. - */ -static void -rpcrdma_recvcq_upcall(struct ib_cq *cq, void *cq_context) -{ - do { - rpcrdma_recvcq_poll(cq); - } while (ib_req_notify_cq(cq, IB_CQ_NEXT_COMP | - IB_CQ_REPORT_MISSED_EVENTS) > 0); -} - 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_recvcq_process_wc(&wc); - while (ib_poll_cq(ep->rep_attr.send_cq, 1, &wc) > 0) - rpcrdma_sendcq_process_wc(&wc); + rpcrdma_receive_wc(NULL, &wc); } static int @@ -330,6 +270,7 @@ rpcrdma_conn_upcall(struct rdma_cm_id *id, struct rdma_cm_event *event) connected: dprintk("RPC: %s: %sconnected\n", __func__, connstate > 0 ? "" : "dis"); + atomic_set(&xprt->rx_buf.rb_credits, 1); ep->rep_connected = connstate; rpcrdma_conn_func(ep); wake_up_all(&ep->rep_connect_wait); @@ -560,9 +501,8 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia, struct rpcrdma_create_data_internal *cdata) { struct ib_cq *sendcq, *recvcq; - struct ib_cq_init_attr cq_attr = {}; unsigned int max_qp_wr; - int rc, err; + int rc; if (ia->ri_device->attrs.max_sge < RPCRDMA_MAX_IOVS) { dprintk("RPC: %s: insufficient sge's available\n", @@ -614,9 +554,9 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia, init_waitqueue_head(&ep->rep_connect_wait); INIT_DELAYED_WORK(&ep->rep_connect_worker, rpcrdma_connect_worker); - cq_attr.cqe = ep->rep_attr.cap.max_send_wr + 1; - sendcq = ib_create_cq(ia->ri_device, rpcrdma_sendcq_upcall, - rpcrdma_cq_async_error_upcall, NULL, &cq_attr); + sendcq = ib_alloc_cq(ia->ri_device, NULL, + ep->rep_attr.cap.max_send_wr + 1, + 0, IB_POLL_SOFTIRQ); if (IS_ERR(sendcq)) { rc = PTR_ERR(sendcq); dprintk("RPC: %s: failed to create send CQ: %i\n", @@ -624,16 +564,9 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia, goto out1; } - rc = ib_req_notify_cq(sendcq, IB_CQ_NEXT_COMP); - if (rc) { - dprintk("RPC: %s: ib_req_notify_cq failed: %i\n", - __func__, rc); - goto out2; - } - - cq_attr.cqe = ep->rep_attr.cap.max_recv_wr + 1; - recvcq = ib_create_cq(ia->ri_device, rpcrdma_recvcq_upcall, - rpcrdma_cq_async_error_upcall, NULL, &cq_attr); + recvcq = ib_alloc_cq(ia->ri_device, NULL, + ep->rep_attr.cap.max_recv_wr + 1, + 0, IB_POLL_SOFTIRQ); if (IS_ERR(recvcq)) { rc = PTR_ERR(recvcq); dprintk("RPC: %s: failed to create recv CQ: %i\n", @@ -641,14 +574,6 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia, goto out2; } - rc = ib_req_notify_cq(recvcq, IB_CQ_NEXT_COMP); - if (rc) { - dprintk("RPC: %s: ib_req_notify_cq failed: %i\n", - __func__, rc); - ib_destroy_cq(recvcq); - goto out2; - } - ep->rep_attr.send_cq = sendcq; ep->rep_attr.recv_cq = recvcq; @@ -673,10 +598,7 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia, return 0; out2: - err = ib_destroy_cq(sendcq); - if (err) - dprintk("RPC: %s: ib_destroy_cq returned %i\n", - __func__, err); + ib_free_cq(sendcq); out1: if (ia->ri_dma_mr) ib_dereg_mr(ia->ri_dma_mr); @@ -711,15 +633,8 @@ rpcrdma_ep_destroy(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia) ia->ri_id->qp = NULL; } - rc = ib_destroy_cq(ep->rep_attr.recv_cq); - if (rc) - dprintk("RPC: %s: ib_destroy_cq returned %i\n", - __func__, rc); - - rc = ib_destroy_cq(ep->rep_attr.send_cq); - if (rc) - dprintk("RPC: %s: ib_destroy_cq returned %i\n", - __func__, rc); + ib_free_cq(ep->rep_attr.recv_cq); + ib_free_cq(ep->rep_attr.send_cq); if (ia->ri_dma_mr) { rc = ib_dereg_mr(ia->ri_dma_mr); @@ -898,6 +813,7 @@ rpcrdma_create_req(struct rpcrdma_xprt *r_xprt) spin_lock(&buffer->rb_reqslock); list_add(&req->rl_all, &buffer->rb_allreqs); spin_unlock(&buffer->rb_reqslock); + req->rl_cqe.done = rpcrdma_wc_send; req->rl_buffer = &r_xprt->rx_buf; return req; } @@ -923,6 +839,7 @@ rpcrdma_create_rep(struct rpcrdma_xprt *r_xprt) } rep->rr_device = ia->ri_device; + rep->rr_cqe.done = rpcrdma_receive_wc; rep->rr_rxprt = r_xprt; INIT_WORK(&rep->rr_work, rpcrdma_receive_worker); return rep; @@ -943,6 +860,7 @@ rpcrdma_buffer_create(struct rpcrdma_xprt *r_xprt) buf->rb_max_requests = r_xprt->rx_data.max_requests; buf->rb_bc_srv_max_requests = 0; spin_lock_init(&buf->rb_lock); + atomic_set(&buf->rb_credits, 1); rc = ia->ri_ops->ro_init(r_xprt); if (rc) @@ -1259,7 +1177,7 @@ rpcrdma_ep_post(struct rpcrdma_ia *ia, } send_wr.next = NULL; - send_wr.wr_id = RPCRDMA_IGNORE_COMPLETION; + send_wr.wr_cqe = &req->rl_cqe; send_wr.sg_list = iov; send_wr.num_sge = req->rl_niovs; send_wr.opcode = IB_WR_SEND; @@ -1297,7 +1215,7 @@ rpcrdma_ep_post_recv(struct rpcrdma_ia *ia, int rc; recv_wr.next = NULL; - recv_wr.wr_id = (u64) (unsigned long) rep; + recv_wr.wr_cqe = &rep->rr_cqe; recv_wr.sg_list = &rep->rr_rdmabuf->rg_iov; recv_wr.num_sge = 1; diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h index 38fe11b..2ebc743 100644 --- a/net/sunrpc/xprtrdma/xprt_rdma.h +++ b/net/sunrpc/xprtrdma/xprt_rdma.h @@ -95,10 +95,6 @@ struct rpcrdma_ep { #define INIT_CQCOUNT(ep) atomic_set(&(ep)->rep_cqcount, (ep)->rep_cqinit) #define DECR_CQCOUNT(ep) atomic_sub_return(1, &(ep)->rep_cqcount) -/* Force completion handler to ignore the signal - */ -#define RPCRDMA_IGNORE_COMPLETION (0ULL) - /* Pre-allocate extra Work Requests for handling backward receives * and sends. This is a fixed value because the Work Queues are * allocated when the forward channel is set up. @@ -171,6 +167,7 @@ rdmab_to_msg(struct rpcrdma_regbuf *rb) struct rpcrdma_buffer; struct rpcrdma_rep { + struct ib_cqe rr_cqe; unsigned int rr_len; struct ib_device *rr_device; struct rpcrdma_xprt *rr_rxprt; @@ -204,11 +201,11 @@ struct rpcrdma_frmr { struct scatterlist *sg; int sg_nents; 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; - bool fr_waiter; - struct completion fr_linv_done;; union { struct ib_reg_wr fr_regwr; struct ib_send_wr fr_invwr; @@ -224,8 +221,7 @@ struct rpcrdma_mw { union { struct rpcrdma_fmr fmr; struct rpcrdma_frmr frmr; - } r; - void (*mw_sendcompletion)(struct ib_wc *); + }; struct list_head mw_list; struct list_head mw_all; }; @@ -281,6 +277,7 @@ struct rpcrdma_req { struct rpcrdma_regbuf *rl_sendbuf; struct rpcrdma_mr_seg rl_segments[RPCRDMA_MAX_SEGS]; + struct ib_cqe rl_cqe; struct list_head rl_all; bool rl_backchannel; }; @@ -311,6 +308,7 @@ struct rpcrdma_buffer { struct list_head rb_send_bufs; struct list_head rb_recv_bufs; u32 rb_max_requests; + atomic_t rb_credits; /* most recent credit grant */ u32 rb_bc_srv_max_requests; spinlock_t rb_reqslock; /* protect rb_allreqs */ diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c index fde2138..65e7595 100644 --- a/net/sunrpc/xprtsock.c +++ b/net/sunrpc/xprtsock.c @@ -1844,9 +1844,7 @@ static int xs_bind(struct sock_xprt *transport, struct socket *sock) */ static void xs_local_rpcbind(struct rpc_task *task) { - rcu_read_lock(); - xprt_set_bound(rcu_dereference(task->tk_client->cl_xprt)); - rcu_read_unlock(); + xprt_set_bound(task->tk_xprt); } static void xs_local_set_port(struct rpc_xprt *xprt, unsigned short port) diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c index bbe65dc..3dce53e 100644 --- a/net/vmw_vsock/af_vsock.c +++ b/net/vmw_vsock/af_vsock.c @@ -1209,10 +1209,14 @@ static int vsock_stream_connect(struct socket *sock, struct sockaddr *addr, if (signal_pending(current)) { err = sock_intr_errno(timeout); - goto out_wait_error; + sk->sk_state = SS_UNCONNECTED; + sock->state = SS_UNCONNECTED; + goto out_wait; } else if (timeout == 0) { err = -ETIMEDOUT; - goto out_wait_error; + sk->sk_state = SS_UNCONNECTED; + sock->state = SS_UNCONNECTED; + goto out_wait; } prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); @@ -1220,20 +1224,17 @@ static int vsock_stream_connect(struct socket *sock, struct sockaddr *addr, if (sk->sk_err) { err = -sk->sk_err; - goto out_wait_error; - } else + sk->sk_state = SS_UNCONNECTED; + sock->state = SS_UNCONNECTED; + } else { err = 0; + } out_wait: finish_wait(sk_sleep(sk), &wait); out: release_sock(sk); return err; - -out_wait_error: - sk->sk_state = SS_UNCONNECTED; - sock->state = SS_UNCONNECTED; - goto out_wait; } static int vsock_accept(struct socket *sock, struct socket *newsock, int flags) @@ -1270,18 +1271,20 @@ static int vsock_accept(struct socket *sock, struct socket *newsock, int flags) listener->sk_err == 0) { release_sock(listener); timeout = schedule_timeout(timeout); + finish_wait(sk_sleep(listener), &wait); lock_sock(listener); if (signal_pending(current)) { err = sock_intr_errno(timeout); - goto out_wait; + goto out; } else if (timeout == 0) { err = -EAGAIN; - goto out_wait; + goto out; } prepare_to_wait(sk_sleep(listener), &wait, TASK_INTERRUPTIBLE); } + finish_wait(sk_sleep(listener), &wait); if (listener->sk_err) err = -listener->sk_err; @@ -1301,19 +1304,15 @@ static int vsock_accept(struct socket *sock, struct socket *newsock, int flags) */ if (err) { vconnected->rejected = true; - release_sock(connected); - sock_put(connected); - goto out_wait; + } else { + newsock->state = SS_CONNECTED; + sock_graft(connected, newsock); } - newsock->state = SS_CONNECTED; - sock_graft(connected, newsock); release_sock(connected); sock_put(connected); } -out_wait: - finish_wait(sk_sleep(listener), &wait); out: release_sock(listener); return err; @@ -1557,9 +1556,11 @@ static int vsock_stream_sendmsg(struct socket *sock, struct msghdr *msg, if (err < 0) goto out; + while (total_written < len) { ssize_t written; + prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); while (vsock_stream_has_space(vsk) == 0 && sk->sk_err == 0 && !(sk->sk_shutdown & SEND_SHUTDOWN) && @@ -1568,27 +1569,33 @@ static int vsock_stream_sendmsg(struct socket *sock, struct msghdr *msg, /* Don't wait for non-blocking sockets. */ if (timeout == 0) { err = -EAGAIN; - goto out_wait; + finish_wait(sk_sleep(sk), &wait); + goto out_err; } err = transport->notify_send_pre_block(vsk, &send_data); - if (err < 0) - goto out_wait; + if (err < 0) { + finish_wait(sk_sleep(sk), &wait); + goto out_err; + } release_sock(sk); - prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); timeout = schedule_timeout(timeout); - finish_wait(sk_sleep(sk), &wait); lock_sock(sk); if (signal_pending(current)) { err = sock_intr_errno(timeout); - goto out_wait; + finish_wait(sk_sleep(sk), &wait); + goto out_err; } else if (timeout == 0) { err = -EAGAIN; - goto out_wait; + finish_wait(sk_sleep(sk), &wait); + goto out_err; } + prepare_to_wait(sk_sleep(sk), &wait, + TASK_INTERRUPTIBLE); } + finish_wait(sk_sleep(sk), &wait); /* These checks occur both as part of and after the loop * conditional since we need to check before and after @@ -1596,16 +1603,16 @@ static int vsock_stream_sendmsg(struct socket *sock, struct msghdr *msg, */ if (sk->sk_err) { err = -sk->sk_err; - goto out_wait; + goto out_err; } else if ((sk->sk_shutdown & SEND_SHUTDOWN) || (vsk->peer_shutdown & RCV_SHUTDOWN)) { err = -EPIPE; - goto out_wait; + goto out_err; } err = transport->notify_send_pre_enqueue(vsk, &send_data); if (err < 0) - goto out_wait; + goto out_err; /* Note that enqueue will only write as many bytes as are free * in the produce queue, so we don't need to ensure len is @@ -1618,7 +1625,7 @@ static int vsock_stream_sendmsg(struct socket *sock, struct msghdr *msg, len - total_written); if (written < 0) { err = -ENOMEM; - goto out_wait; + goto out_err; } total_written += written; @@ -1626,11 +1633,11 @@ static int vsock_stream_sendmsg(struct socket *sock, struct msghdr *msg, err = transport->notify_send_post_enqueue( vsk, written, &send_data); if (err < 0) - goto out_wait; + goto out_err; } -out_wait: +out_err: if (total_written > 0) err = total_written; out: @@ -1715,18 +1722,59 @@ vsock_stream_recvmsg(struct socket *sock, struct msghdr *msg, size_t len, while (1) { - s64 ready = vsock_stream_has_data(vsk); + s64 ready; - if (ready < 0) { - /* Invalid queue pair content. XXX This should be - * changed to a connection reset in a later change. - */ + prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); + ready = vsock_stream_has_data(vsk); - err = -ENOMEM; - goto out; - } else if (ready > 0) { + if (ready == 0) { + if (sk->sk_err != 0 || + (sk->sk_shutdown & RCV_SHUTDOWN) || + (vsk->peer_shutdown & SEND_SHUTDOWN)) { + finish_wait(sk_sleep(sk), &wait); + break; + } + /* Don't wait for non-blocking sockets. */ + if (timeout == 0) { + err = -EAGAIN; + finish_wait(sk_sleep(sk), &wait); + break; + } + + err = transport->notify_recv_pre_block( + vsk, target, &recv_data); + if (err < 0) { + finish_wait(sk_sleep(sk), &wait); + break; + } + release_sock(sk); + timeout = schedule_timeout(timeout); + lock_sock(sk); + + if (signal_pending(current)) { + err = sock_intr_errno(timeout); + finish_wait(sk_sleep(sk), &wait); + break; + } else if (timeout == 0) { + err = -EAGAIN; + finish_wait(sk_sleep(sk), &wait); + break; + } + } else { ssize_t read; + finish_wait(sk_sleep(sk), &wait); + + if (ready < 0) { + /* Invalid queue pair content. XXX This should + * be changed to a connection reset in a later + * change. + */ + + err = -ENOMEM; + goto out; + } + err = transport->notify_recv_pre_dequeue( vsk, target, &recv_data); if (err < 0) @@ -1752,35 +1800,6 @@ vsock_stream_recvmsg(struct socket *sock, struct msghdr *msg, size_t len, break; target -= read; - } else { - if (sk->sk_err != 0 || (sk->sk_shutdown & RCV_SHUTDOWN) - || (vsk->peer_shutdown & SEND_SHUTDOWN)) { - break; - } - /* Don't wait for non-blocking sockets. */ - if (timeout == 0) { - err = -EAGAIN; - break; - } - - err = transport->notify_recv_pre_block( - vsk, target, &recv_data); - if (err < 0) - break; - - release_sock(sk); - prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); - timeout = schedule_timeout(timeout); - finish_wait(sk_sleep(sk), &wait); - lock_sock(sk); - - if (signal_pending(current)) { - err = sock_intr_errno(timeout); - break; - } else if (timeout == 0) { - err = -EAGAIN; - break; - } } } diff --git a/net/xfrm/xfrm_user.c b/net/xfrm/xfrm_user.c index 805681a..2cc7af8 100644 --- a/net/xfrm/xfrm_user.c +++ b/net/xfrm/xfrm_user.c @@ -2449,7 +2449,7 @@ static int xfrm_user_rcv_msg(struct sk_buff *skb, struct nlmsghdr *nlh) int type, err; #ifdef CONFIG_COMPAT - if (is_compat_task()) + if (in_compat_syscall()) return -ENOTSUPP; #endif |
