From a3d1289126e7b14307074b76bf1677015ea5036f Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 21 Oct 2009 10:59:31 +0000 Subject: rtnetlink: rtnl_setlink() and rtnl_getlink() changes rtnl_getlink() & rtnl_setlink() run with RTNL held, we can use __dev_get_by_index() and __dev_get_by_name() variants and avoid dev_hold()/dev_put() Adds to rtnl_getlink() the capability to find a device by its name, not only by its index. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/core/rtnetlink.c | 38 +++++++++++++++++++------------------- 1 file changed, 19 insertions(+), 19 deletions(-) (limited to 'net/core/rtnetlink.c') diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c index eb42873f2a3a..ba13b0974a7b 100644 --- a/net/core/rtnetlink.c +++ b/net/core/rtnetlink.c @@ -910,9 +910,9 @@ static int rtnl_setlink(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg) err = -EINVAL; ifm = nlmsg_data(nlh); if (ifm->ifi_index > 0) - dev = dev_get_by_index(net, ifm->ifi_index); + dev = __dev_get_by_index(net, ifm->ifi_index); else if (tb[IFLA_IFNAME]) - dev = dev_get_by_name(net, ifname); + dev = __dev_get_by_name(net, ifname); else goto errout; @@ -922,11 +922,9 @@ static int rtnl_setlink(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg) } if ((err = validate_linkmsg(dev, tb)) < 0) - goto errout_dev; + goto errout; err = do_setlink(dev, ifm, tb, ifname, 0); -errout_dev: - dev_put(dev); errout: return err; } @@ -1154,6 +1152,7 @@ static int rtnl_getlink(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) { struct net *net = sock_net(skb->sk); struct ifinfomsg *ifm; + char ifname[IFNAMSIZ]; struct nlattr *tb[IFLA_MAX+1]; struct net_device *dev = NULL; struct sk_buff *nskb; @@ -1163,19 +1162,23 @@ static int rtnl_getlink(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) if (err < 0) return err; + if (tb[IFLA_IFNAME]) + nla_strlcpy(ifname, tb[IFLA_IFNAME], IFNAMSIZ); + ifm = nlmsg_data(nlh); - if (ifm->ifi_index > 0) { - dev = dev_get_by_index(net, ifm->ifi_index); - if (dev == NULL) - return -ENODEV; - } else + if (ifm->ifi_index > 0) + dev = __dev_get_by_index(net, ifm->ifi_index); + else if (tb[IFLA_IFNAME]) + dev = __dev_get_by_name(net, ifname); + else return -EINVAL; + if (dev == NULL) + return -ENODEV; + nskb = nlmsg_new(if_nlmsg_size(dev), GFP_KERNEL); - if (nskb == NULL) { - err = -ENOBUFS; - goto errout; - } + if (nskb == NULL) + return -ENOBUFS; err = rtnl_fill_ifinfo(nskb, dev, RTM_NEWLINK, NETLINK_CB(skb).pid, nlh->nlmsg_seq, 0, 0); @@ -1183,11 +1186,8 @@ static int rtnl_getlink(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) /* -EMSGSIZE implies BUG in if_nlmsg_size */ WARN_ON(err == -EMSGSIZE); kfree_skb(nskb); - goto errout; - } - err = rtnl_unicast(nskb, net, NETLINK_CB(skb).pid); -errout: - dev_put(dev); + } else + err = rtnl_unicast(nskb, net, NETLINK_CB(skb).pid); return err; } -- cgit v1.2.3 From 7c28bd0b8ec4d128bd7660671d1b626b0abc471f Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Sat, 24 Oct 2009 06:13:17 -0700 Subject: rtnetlink: speedup rtnl_dump_ifinfo() When handling large number of netdevice, rtnl_dump_ifinfo() is very slow because it has O(N^2) complexity. Instead of scanning one single list, we can use the 256 sub lists of the dev_index hash table. This considerably speedups "ip link" operations Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/core/rtnetlink.c | 37 ++++++++++++++++++++++++------------- 1 file changed, 24 insertions(+), 13 deletions(-) (limited to 'net/core/rtnetlink.c') diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c index ba13b0974a7b..52ea418d5302 100644 --- a/net/core/rtnetlink.c +++ b/net/core/rtnetlink.c @@ -682,22 +682,33 @@ nla_put_failure: static int rtnl_dump_ifinfo(struct sk_buff *skb, struct netlink_callback *cb) { struct net *net = sock_net(skb->sk); - int idx; - int s_idx = cb->args[0]; + int h, s_h; + int idx = 0, s_idx; struct net_device *dev; - - idx = 0; - for_each_netdev(net, dev) { - if (idx < s_idx) - goto cont; - if (rtnl_fill_ifinfo(skb, dev, RTM_NEWLINK, - NETLINK_CB(cb->skb).pid, - cb->nlh->nlmsg_seq, 0, NLM_F_MULTI) <= 0) - break; + struct hlist_head *head; + struct hlist_node *node; + + s_h = cb->args[0]; + s_idx = cb->args[1]; + + for (h = s_h; h < NETDEV_HASHENTRIES; h++, s_idx = 0) { + idx = 0; + head = &net->dev_index_head[h]; + hlist_for_each_entry(dev, node, head, index_hlist) { + if (idx < s_idx) + goto cont; + if (rtnl_fill_ifinfo(skb, dev, RTM_NEWLINK, + NETLINK_CB(cb->skb).pid, + cb->nlh->nlmsg_seq, 0, + NLM_F_MULTI) <= 0) + goto out; cont: - idx++; + idx++; + } } - cb->args[0] = idx; +out: + cb->args[1] = idx; + cb->args[0] = h; return skb->len; } -- cgit v1.2.3 From 23289a37e2b127dfc4de1313fba15bb4c9f0cd5b Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 27 Oct 2009 07:06:36 +0000 Subject: net: add a list_head parameter to dellink() method Adding a list_head parameter to rtnl_link_ops->dellink() methods allow us to queue devices on a list, in order to dismantle them all at once. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/core/rtnetlink.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'net/core/rtnetlink.c') diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c index 52ea418d5302..391a62cd9df6 100644 --- a/net/core/rtnetlink.c +++ b/net/core/rtnetlink.c @@ -248,7 +248,7 @@ static LIST_HEAD(link_ops); int __rtnl_link_register(struct rtnl_link_ops *ops) { if (!ops->dellink) - ops->dellink = unregister_netdevice; + ops->dellink = unregister_netdevice_queue; list_add_tail(&ops->list, &link_ops); return 0; @@ -277,13 +277,13 @@ EXPORT_SYMBOL_GPL(rtnl_link_register); static void __rtnl_kill_links(struct net *net, struct rtnl_link_ops *ops) { struct net_device *dev; -restart: + LIST_HEAD(list_kill); + for_each_netdev(net, dev) { - if (dev->rtnl_link_ops == ops) { - ops->dellink(dev); - goto restart; - } + if (dev->rtnl_link_ops == ops) + ops->dellink(dev, &list_kill); } + unregister_netdevice_many(&list_kill); } void rtnl_kill_links(struct net *net, struct rtnl_link_ops *ops) @@ -972,7 +972,7 @@ static int rtnl_dellink(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg) if (!ops) return -EOPNOTSUPP; - ops->dellink(dev); + ops->dellink(dev, NULL); return 0; } -- cgit v1.2.3 From e0d087af725b09358336098a6b57bb7f90f96175 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Sat, 7 Nov 2009 01:26:17 -0800 Subject: rtnetlink: Cleanups Pure cleanups patch Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/core/rtnetlink.c | 54 ++++++++++++++++++++++------------------------------ 1 file changed, 23 insertions(+), 31 deletions(-) (limited to 'net/core/rtnetlink.c') diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c index 391a62cd9df6..e2f3317f290f 100644 --- a/net/core/rtnetlink.c +++ b/net/core/rtnetlink.c @@ -38,7 +38,6 @@ #include #include -#include #include #include @@ -53,8 +52,7 @@ #include #include -struct rtnl_link -{ +struct rtnl_link { rtnl_doit_func doit; rtnl_dumpit_func dumpit; }; @@ -65,6 +63,7 @@ void rtnl_lock(void) { mutex_lock(&rtnl_mutex); } +EXPORT_SYMBOL(rtnl_lock); void __rtnl_unlock(void) { @@ -76,16 +75,19 @@ void rtnl_unlock(void) /* This fellow will unlock it for us. */ netdev_run_todo(); } +EXPORT_SYMBOL(rtnl_unlock); int rtnl_trylock(void) { return mutex_trylock(&rtnl_mutex); } +EXPORT_SYMBOL(rtnl_trylock); int rtnl_is_locked(void) { return mutex_is_locked(&rtnl_mutex); } +EXPORT_SYMBOL(rtnl_is_locked); static struct rtnl_link *rtnl_msg_handlers[NPROTO]; @@ -168,7 +170,6 @@ int __rtnl_register(int protocol, int msgtype, return 0; } - EXPORT_SYMBOL_GPL(__rtnl_register); /** @@ -188,7 +189,6 @@ void rtnl_register(int protocol, int msgtype, "protocol = %d, message type = %d\n", protocol, msgtype); } - EXPORT_SYMBOL_GPL(rtnl_register); /** @@ -213,7 +213,6 @@ int rtnl_unregister(int protocol, int msgtype) return 0; } - EXPORT_SYMBOL_GPL(rtnl_unregister); /** @@ -230,7 +229,6 @@ void rtnl_unregister_all(int protocol) kfree(rtnl_msg_handlers[protocol]); rtnl_msg_handlers[protocol] = NULL; } - EXPORT_SYMBOL_GPL(rtnl_unregister_all); static LIST_HEAD(link_ops); @@ -253,7 +251,6 @@ int __rtnl_link_register(struct rtnl_link_ops *ops) list_add_tail(&ops->list, &link_ops); return 0; } - EXPORT_SYMBOL_GPL(__rtnl_link_register); /** @@ -271,7 +268,6 @@ int rtnl_link_register(struct rtnl_link_ops *ops) rtnl_unlock(); return err; } - EXPORT_SYMBOL_GPL(rtnl_link_register); static void __rtnl_kill_links(struct net *net, struct rtnl_link_ops *ops) @@ -309,7 +305,6 @@ void __rtnl_link_unregister(struct rtnl_link_ops *ops) } list_del(&ops->list); } - EXPORT_SYMBOL_GPL(__rtnl_link_unregister); /** @@ -322,7 +317,6 @@ void rtnl_link_unregister(struct rtnl_link_ops *ops) __rtnl_link_unregister(ops); rtnl_unlock(); } - EXPORT_SYMBOL_GPL(rtnl_link_unregister); static const struct rtnl_link_ops *rtnl_link_ops_get(const char *kind) @@ -427,12 +421,13 @@ void __rta_fill(struct sk_buff *skb, int attrtype, int attrlen, const void *data struct rtattr *rta; int size = RTA_LENGTH(attrlen); - rta = (struct rtattr*)skb_put(skb, RTA_ALIGN(size)); + rta = (struct rtattr *)skb_put(skb, RTA_ALIGN(size)); rta->rta_type = attrtype; rta->rta_len = size; memcpy(RTA_DATA(rta), data, attrlen); memset(RTA_DATA(rta) + attrlen, 0, RTA_ALIGN(size) - size); } +EXPORT_SYMBOL(__rta_fill); int rtnetlink_send(struct sk_buff *skb, struct net *net, u32 pid, unsigned group, int echo) { @@ -454,6 +449,7 @@ int rtnl_unicast(struct sk_buff *skb, struct net *net, u32 pid) return nlmsg_unicast(rtnl, skb, pid); } +EXPORT_SYMBOL(rtnl_unicast); void rtnl_notify(struct sk_buff *skb, struct net *net, u32 pid, u32 group, struct nlmsghdr *nlh, gfp_t flags) @@ -466,6 +462,7 @@ void rtnl_notify(struct sk_buff *skb, struct net *net, u32 pid, u32 group, nlmsg_notify(rtnl, skb, pid, group, report, flags); } +EXPORT_SYMBOL(rtnl_notify); void rtnl_set_sk_err(struct net *net, u32 group, int error) { @@ -473,6 +470,7 @@ void rtnl_set_sk_err(struct net *net, u32 group, int error) netlink_set_err(rtnl, 0, group, error); } +EXPORT_SYMBOL(rtnl_set_sk_err); int rtnetlink_put_metrics(struct sk_buff *skb, u32 *metrics) { @@ -501,6 +499,7 @@ nla_put_failure: nla_nest_cancel(skb, mx); return -EMSGSIZE; } +EXPORT_SYMBOL(rtnetlink_put_metrics); int rtnl_put_cacheinfo(struct sk_buff *skb, struct dst_entry *dst, u32 id, u32 ts, u32 tsage, long expires, u32 error) @@ -520,14 +519,13 @@ int rtnl_put_cacheinfo(struct sk_buff *skb, struct dst_entry *dst, u32 id, return nla_put(skb, RTA_CACHEINFO, sizeof(ci), &ci); } - EXPORT_SYMBOL_GPL(rtnl_put_cacheinfo); static void set_operstate(struct net_device *dev, unsigned char transition) { unsigned char operstate = dev->operstate; - switch(transition) { + switch (transition) { case IF_OPER_UP: if ((operstate == IF_OPER_DORMANT || operstate == IF_OPER_UNKNOWN) && @@ -728,6 +726,7 @@ const struct nla_policy ifla_policy[IFLA_MAX+1] = { [IFLA_NET_NS_PID] = { .type = NLA_U32 }, [IFLA_IFALIAS] = { .type = NLA_STRING, .len = IFALIASZ-1 }, }; +EXPORT_SYMBOL(ifla_policy); static const struct nla_policy ifla_info_policy[IFLA_INFO_MAX+1] = { [IFLA_INFO_KIND] = { .type = NLA_STRING }, @@ -932,7 +931,8 @@ static int rtnl_setlink(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg) goto errout; } - if ((err = validate_linkmsg(dev, tb)) < 0) + err = validate_linkmsg(dev, tb); + if (err < 0) goto errout; err = do_setlink(dev, ifm, tb, ifname, 0); @@ -985,7 +985,8 @@ struct net_device *rtnl_create_link(struct net *net, char *ifname, unsigned int real_num_queues = 1; if (ops->get_tx_queues) { - err = ops->get_tx_queues(net, tb, &num_queues, &real_num_queues); + err = ops->get_tx_queues(net, tb, &num_queues, + &real_num_queues); if (err) goto err; } @@ -1026,6 +1027,7 @@ err_free: err: return ERR_PTR(err); } +EXPORT_SYMBOL(rtnl_create_link); static int rtnl_newlink(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg) { @@ -1059,7 +1061,8 @@ replay: else dev = NULL; - if ((err = validate_linkmsg(dev, tb)) < 0) + err = validate_linkmsg(dev, tb); + if (err < 0) return err; if (tb[IFLA_LINKINFO]) { @@ -1210,7 +1213,7 @@ static int rtnl_dump_all(struct sk_buff *skb, struct netlink_callback *cb) if (s_idx == 0) s_idx = 1; - for (idx=1; idxnlh->nlmsg_type-RTM_BASE; if (idx < s_idx || idx == PF_PACKET) continue; @@ -1277,7 +1280,7 @@ static int rtnetlink_rcv_msg(struct sk_buff *skb, struct nlmsghdr *nlh) if (nlh->nlmsg_len < NLMSG_LENGTH(sizeof(struct rtgenmsg))) return 0; - family = ((struct rtgenmsg*)NLMSG_DATA(nlh))->rtgen_family; + family = ((struct rtgenmsg *)NLMSG_DATA(nlh))->rtgen_family; if (family >= NPROTO) return -EAFNOSUPPORT; @@ -1310,7 +1313,7 @@ static int rtnetlink_rcv_msg(struct sk_buff *skb, struct nlmsghdr *nlh) if (nlh->nlmsg_len > min_len) { int attrlen = nlh->nlmsg_len - NLMSG_ALIGN(min_len); - struct rtattr *attr = (void*)nlh + NLMSG_ALIGN(min_len); + struct rtattr *attr = (void *)nlh + NLMSG_ALIGN(min_len); while (RTA_OK(attr, attrlen)) { unsigned flavor = attr->rta_type; @@ -1416,14 +1419,3 @@ void __init rtnetlink_init(void) rtnl_register(PF_UNSPEC, RTM_GETROUTE, NULL, rtnl_dump_all); } -EXPORT_SYMBOL(__rta_fill); -EXPORT_SYMBOL(rtnetlink_put_metrics); -EXPORT_SYMBOL(rtnl_lock); -EXPORT_SYMBOL(rtnl_trylock); -EXPORT_SYMBOL(rtnl_unlock); -EXPORT_SYMBOL(rtnl_is_locked); -EXPORT_SYMBOL(rtnl_unicast); -EXPORT_SYMBOL(rtnl_notify); -EXPORT_SYMBOL(rtnl_set_sk_err); -EXPORT_SYMBOL(rtnl_create_link); -EXPORT_SYMBOL(ifla_policy); -- cgit v1.2.3 From 81adee47dfb608df3ad0b91d230fb3cef75f0060 Mon Sep 17 00:00:00 2001 From: "Eric W. Biederman" Date: Sun, 8 Nov 2009 00:53:51 -0800 Subject: net: Support specifying the network namespace upon device creation. There is no good reason to not support userspace specifying the network namespace during device creation, and it makes it easier to create a network device and pass it to a child network namespace with a well known name. We have to be careful to ensure that the target network namespace for the new device exists through the life of the call. To keep that logic clear I have factored out the network namespace grabbing logic into rtnl_link_get_net. In addtion we need to continue to pass the source network namespace to the rtnl_link_ops.newlink method so that we can find the base device source network namespace. Signed-off-by: Eric W. Biederman Acked-by: Eric Dumazet --- net/core/rtnetlink.c | 38 +++++++++++++++++++++++++++----------- 1 file changed, 27 insertions(+), 11 deletions(-) (limited to 'net/core/rtnetlink.c') diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c index e2f3317f290f..33148a568199 100644 --- a/net/core/rtnetlink.c +++ b/net/core/rtnetlink.c @@ -733,6 +733,20 @@ static const struct nla_policy ifla_info_policy[IFLA_INFO_MAX+1] = { [IFLA_INFO_DATA] = { .type = NLA_NESTED }, }; +struct net *rtnl_link_get_net(struct net *src_net, struct nlattr *tb[]) +{ + struct net *net; + /* Examine the link attributes and figure out which + * network namespace we are talking about. + */ + if (tb[IFLA_NET_NS_PID]) + net = get_net_ns_by_pid(nla_get_u32(tb[IFLA_NET_NS_PID])); + else + net = get_net(src_net); + return net; +} +EXPORT_SYMBOL(rtnl_link_get_net); + static int validate_linkmsg(struct net_device *dev, struct nlattr *tb[]) { if (dev) { @@ -756,8 +770,7 @@ static int do_setlink(struct net_device *dev, struct ifinfomsg *ifm, int err; if (tb[IFLA_NET_NS_PID]) { - struct net *net; - net = get_net_ns_by_pid(nla_get_u32(tb[IFLA_NET_NS_PID])); + struct net *net = rtnl_link_get_net(dev_net(dev), tb); if (IS_ERR(net)) { err = PTR_ERR(net); goto errout; @@ -976,8 +989,8 @@ static int rtnl_dellink(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg) return 0; } -struct net_device *rtnl_create_link(struct net *net, char *ifname, - const struct rtnl_link_ops *ops, struct nlattr *tb[]) +struct net_device *rtnl_create_link(struct net *src_net, struct net *net, + char *ifname, const struct rtnl_link_ops *ops, struct nlattr *tb[]) { int err; struct net_device *dev; @@ -985,7 +998,7 @@ struct net_device *rtnl_create_link(struct net *net, char *ifname, unsigned int real_num_queues = 1; if (ops->get_tx_queues) { - err = ops->get_tx_queues(net, tb, &num_queues, + err = ops->get_tx_queues(src_net, tb, &num_queues, &real_num_queues); if (err) goto err; @@ -995,16 +1008,16 @@ struct net_device *rtnl_create_link(struct net *net, char *ifname, if (!dev) goto err; + dev_net_set(dev, net); + dev->rtnl_link_ops = ops; dev->real_num_tx_queues = real_num_queues; + if (strchr(dev->name, '%')) { err = dev_alloc_name(dev, dev->name); if (err < 0) goto err_free; } - dev_net_set(dev, net); - dev->rtnl_link_ops = ops; - if (tb[IFLA_MTU]) dev->mtu = nla_get_u32(tb[IFLA_MTU]); if (tb[IFLA_ADDRESS]) @@ -1083,6 +1096,7 @@ replay: if (1) { struct nlattr *attr[ops ? ops->maxtype + 1 : 0], **data = NULL; + struct net *dest_net; if (ops) { if (ops->maxtype && linkinfo[IFLA_INFO_DATA]) { @@ -1147,17 +1161,19 @@ replay: if (!ifname[0]) snprintf(ifname, IFNAMSIZ, "%s%%d", ops->kind); - dev = rtnl_create_link(net, ifname, ops, tb); + dest_net = rtnl_link_get_net(net, tb); + dev = rtnl_create_link(net, dest_net, ifname, ops, tb); if (IS_ERR(dev)) err = PTR_ERR(dev); else if (ops->newlink) - err = ops->newlink(dev, tb, data); + err = ops->newlink(net, dev, tb, data); else err = register_netdevice(dev); - if (err < 0 && !IS_ERR(dev)) free_netdev(dev); + + put_net(dest_net); return err; } } -- cgit v1.2.3