From 5380d64f8d766576ac5c0f627418b2d0e1d2641f Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Thu, 6 Jun 2024 12:29:05 -0700 Subject: [PATCH 1/2] rtnetlink: move rtnl_lock handling out of af_netlink Now that we have an intermediate layer of code for handling rtnl-level netlink dump quirks, we can move the rtnl_lock taking there. For dump handlers with RTNL_FLAG_DUMP_SPLIT_NLM_DONE we can avoid taking rtnl_lock just to generate NLM_DONE, once again. Signed-off-by: Jakub Kicinski Reviewed-by: Kuniyuki Iwashima Reviewed-by: Eric Dumazet Signed-off-by: David S. Miller --- net/core/rtnetlink.c | 9 +++++++-- net/netlink/af_netlink.c | 2 -- 2 files changed, 7 insertions(+), 4 deletions(-) diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c index 4668d6718040..eabfc8290f5e 100644 --- a/net/core/rtnetlink.c +++ b/net/core/rtnetlink.c @@ -6486,6 +6486,7 @@ static int rtnl_mdb_del(struct sk_buff *skb, struct nlmsghdr *nlh, static int rtnl_dumpit(struct sk_buff *skb, struct netlink_callback *cb) { + const bool needs_lock = !(cb->flags & RTNL_FLAG_DUMP_UNLOCKED); rtnl_dumpit_func dumpit = cb->data; int err; @@ -6495,7 +6496,11 @@ static int rtnl_dumpit(struct sk_buff *skb, struct netlink_callback *cb) if (!dumpit) return 0; + if (needs_lock) + rtnl_lock(); err = dumpit(skb, cb); + if (needs_lock) + rtnl_unlock(); /* Old dump handlers used to send NLM_DONE as in a separate recvmsg(). * Some applications which parse netlink manually depend on this. @@ -6515,7 +6520,8 @@ static int rtnetlink_dump_start(struct sock *ssk, struct sk_buff *skb, const struct nlmsghdr *nlh, struct netlink_dump_control *control) { - if (control->flags & RTNL_FLAG_DUMP_SPLIT_NLM_DONE) { + if (control->flags & RTNL_FLAG_DUMP_SPLIT_NLM_DONE || + !(control->flags & RTNL_FLAG_DUMP_UNLOCKED)) { WARN_ON(control->data); control->data = control->dump; control->dump = rtnl_dumpit; @@ -6703,7 +6709,6 @@ static int __net_init rtnetlink_net_init(struct net *net) struct netlink_kernel_cfg cfg = { .groups = RTNLGRP_MAX, .input = rtnetlink_rcv, - .cb_mutex = &rtnl_mutex, .flags = NL_CFG_F_NONROOT_RECV, .bind = rtnetlink_bind, }; diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c index fa9c090cf629..8bbbe75e75db 100644 --- a/net/netlink/af_netlink.c +++ b/net/netlink/af_netlink.c @@ -2330,8 +2330,6 @@ static int netlink_dump(struct sock *sk, bool lock_taken) cb->extack = &extack; - if (cb->flags & RTNL_FLAG_DUMP_UNLOCKED) - extra_mutex = NULL; if (extra_mutex) mutex_lock(extra_mutex); nlk->dump_done_errno = cb->dump(skb, cb); From 5fbf57a937f418fe204f9dbb7735e91984f4ee6a Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Thu, 6 Jun 2024 12:29:06 -0700 Subject: [PATCH 2/2] net: netlink: remove the cb_mutex "injection" from netlink core Back in 2007, in commit af65bdfce98d ("[NETLINK]: Switch cb_lock spinlock to mutex and allow to override it") netlink core was extended to allow subsystems to replace the dump mutex lock with its own lock. The mechanism was used by rtnetlink to take rtnl_lock but it isn't sufficiently flexible for other users. Over the 17 years since it was added no other user appeared. Since rtnetlink needs conditional locking now, and doesn't use it either, axe this feature complete. Signed-off-by: Jakub Kicinski Reviewed-by: Kuniyuki Iwashima Signed-off-by: David S. Miller --- include/linux/netlink.h | 1 - net/netlink/af_netlink.c | 18 +++--------------- 2 files changed, 3 insertions(+), 16 deletions(-) diff --git a/include/linux/netlink.h b/include/linux/netlink.h index 5df7340d4dab..b332c2048c75 100644 --- a/include/linux/netlink.h +++ b/include/linux/netlink.h @@ -47,7 +47,6 @@ struct netlink_kernel_cfg { unsigned int groups; unsigned int flags; void (*input)(struct sk_buff *skb); - struct mutex *cb_mutex; int (*bind)(struct net *net, int group); void (*unbind)(struct net *net, int group); void (*release) (struct sock *sk, unsigned long *groups); diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c index 8bbbe75e75db..0b7a89db3ab7 100644 --- a/net/netlink/af_netlink.c +++ b/net/netlink/af_netlink.c @@ -636,8 +636,7 @@ static struct proto netlink_proto = { }; static int __netlink_create(struct net *net, struct socket *sock, - struct mutex *dump_cb_mutex, int protocol, - int kern) + int protocol, int kern) { struct sock *sk; struct netlink_sock *nlk; @@ -655,7 +654,6 @@ static int __netlink_create(struct net *net, struct socket *sock, lockdep_set_class_and_name(&nlk->nl_cb_mutex, nlk_cb_mutex_keys + protocol, nlk_cb_mutex_key_strings[protocol]); - nlk->dump_cb_mutex = dump_cb_mutex; init_waitqueue_head(&nlk->wait); sk->sk_destruct = netlink_sock_destruct; @@ -667,7 +665,6 @@ static int netlink_create(struct net *net, struct socket *sock, int protocol, int kern) { struct module *module = NULL; - struct mutex *cb_mutex; struct netlink_sock *nlk; int (*bind)(struct net *net, int group); void (*unbind)(struct net *net, int group); @@ -696,7 +693,6 @@ static int netlink_create(struct net *net, struct socket *sock, int protocol, module = nl_table[protocol].module; else err = -EPROTONOSUPPORT; - cb_mutex = nl_table[protocol].cb_mutex; bind = nl_table[protocol].bind; unbind = nl_table[protocol].unbind; release = nl_table[protocol].release; @@ -705,7 +701,7 @@ static int netlink_create(struct net *net, struct socket *sock, int protocol, if (err < 0) goto out; - err = __netlink_create(net, sock, cb_mutex, protocol, kern); + err = __netlink_create(net, sock, protocol, kern); if (err < 0) goto out_module; @@ -2016,7 +2012,6 @@ __netlink_kernel_create(struct net *net, int unit, struct module *module, struct sock *sk; struct netlink_sock *nlk; struct listeners *listeners = NULL; - struct mutex *cb_mutex = cfg ? cfg->cb_mutex : NULL; unsigned int groups; BUG_ON(!nl_table); @@ -2027,7 +2022,7 @@ __netlink_kernel_create(struct net *net, int unit, struct module *module, if (sock_create_lite(PF_NETLINK, SOCK_DGRAM, unit, &sock)) return NULL; - if (__netlink_create(net, sock, cb_mutex, unit, 1) < 0) + if (__netlink_create(net, sock, unit, 1) < 0) goto out_sock_release_nosk; sk = sock->sk; @@ -2055,7 +2050,6 @@ __netlink_kernel_create(struct net *net, int unit, struct module *module, if (!nl_table[unit].registered) { nl_table[unit].groups = groups; rcu_assign_pointer(nl_table[unit].listeners, listeners); - nl_table[unit].cb_mutex = cb_mutex; nl_table[unit].module = module; if (cfg) { nl_table[unit].bind = cfg->bind; @@ -2326,15 +2320,9 @@ static int netlink_dump(struct sock *sk, bool lock_taken) netlink_skb_set_owner_r(skb, sk); if (nlk->dump_done_errno > 0) { - struct mutex *extra_mutex = nlk->dump_cb_mutex; - cb->extack = &extack; - if (extra_mutex) - mutex_lock(extra_mutex); nlk->dump_done_errno = cb->dump(skb, cb); - if (extra_mutex) - mutex_unlock(extra_mutex); /* EMSGSIZE plus something already in the skb means * that there's more to dump but current skb has filled up.