From d9ea761fdd197351890418acd462c51f241014a7 Mon Sep 17 00:00:00 2001 From: "Lin, Zhenpeng" Date: Wed, 8 Sep 2021 03:40:59 +0000 Subject: dccp: don't duplicate ccid when cloning dccp sock Commit 2677d2067731 ("dccp: don't free ccid2_hc_tx_sock ...") fixed a UAF but reintroduced CVE-2017-6074. When the sock is cloned, two dccps_hc_tx_ccid will reference to the same ccid. So one can free the ccid object twice from two socks after cloning. This issue was found by "Hadar Manor" as well and assigned with CVE-2020-16119, which was fixed in Ubuntu's kernel. So here I port the patch from Ubuntu to fix it. The patch prevents cloned socks from referencing the same ccid. Fixes: 2677d2067731410 ("dccp: don't free ccid2_hc_tx_sock ...") Signed-off-by: Zhenpeng Lin Signed-off-by: David S. Miller --- net/dccp/minisocks.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'net') diff --git a/net/dccp/minisocks.c b/net/dccp/minisocks.c index c5c74a34d139..91e7a2202697 100644 --- a/net/dccp/minisocks.c +++ b/net/dccp/minisocks.c @@ -94,6 +94,8 @@ struct sock *dccp_create_openreq_child(const struct sock *sk, newdp->dccps_role = DCCP_ROLE_SERVER; newdp->dccps_hc_rx_ackvec = NULL; newdp->dccps_service_list = NULL; + newdp->dccps_hc_rx_ccid = NULL; + newdp->dccps_hc_tx_ccid = NULL; newdp->dccps_service = dreq->dreq_service; newdp->dccps_timestamp_echo = dreq->dreq_timestamp_echo; newdp->dccps_timestamp_time = dreq->dreq_timestamp_time; -- cgit v1.3-14-g43fede From 581edcd0c8a076eba2ec9e20db50921ee80f5cbc Mon Sep 17 00:00:00 2001 From: Jeremy Kerr Date: Wed, 8 Sep 2021 12:13:10 +0800 Subject: mctp: perform route destruction under RCU read lock The kernel test robot reports: [ 843.509974][ T345] ============================= [ 843.524220][ T345] WARNING: suspicious RCU usage [ 843.538791][ T345] 5.14.0-rc2-00606-g889b7da23abf #1 Not tainted [ 843.553617][ T345] ----------------------------- [ 843.567412][ T345] net/mctp/route.c:310 RCU-list traversed in non-reader section!! - we're missing the rcu read lock acquire around the destruction path. This change adds the acquire/release - the path is already atomic, and we're using the _rcu list iterators. Reported-by: kernel test robot Signed-off-by: Jeremy Kerr Signed-off-by: David S. Miller --- net/mctp/route.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'net') diff --git a/net/mctp/route.c b/net/mctp/route.c index 5265525011ad..5ca186d53cb0 100644 --- a/net/mctp/route.c +++ b/net/mctp/route.c @@ -1083,8 +1083,10 @@ static void __net_exit mctp_routes_net_exit(struct net *net) { struct mctp_route *rt; + rcu_read_lock(); list_for_each_entry_rcu(rt, &net->mctp.routes, list) mctp_route_release(rt); + rcu_read_unlock(); } static struct pernet_operations mctp_net_ops = { -- cgit v1.3-14-g43fede From 04f08eb44b5011493d77b602fdec29ff0f5c6cd5 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 8 Sep 2021 17:00:29 -0700 Subject: net/af_unix: fix a data-race in unix_dgram_poll syzbot reported another data-race in af_unix [1] Lets change __skb_insert() to use WRITE_ONCE() when changing skb head qlen. Also, change unix_dgram_poll() to use lockless version of unix_recvq_full() It is verry possible we can switch all/most unix_recvq_full() to the lockless version, this will be done in a future kernel version. [1] HEAD commit: 8596e589b787732c8346f0482919e83cc9362db1 BUG: KCSAN: data-race in skb_queue_tail / unix_dgram_poll write to 0xffff88814eeb24e0 of 4 bytes by task 25815 on cpu 0: __skb_insert include/linux/skbuff.h:1938 [inline] __skb_queue_before include/linux/skbuff.h:2043 [inline] __skb_queue_tail include/linux/skbuff.h:2076 [inline] skb_queue_tail+0x80/0xa0 net/core/skbuff.c:3264 unix_dgram_sendmsg+0xff2/0x1600 net/unix/af_unix.c:1850 sock_sendmsg_nosec net/socket.c:703 [inline] sock_sendmsg net/socket.c:723 [inline] ____sys_sendmsg+0x360/0x4d0 net/socket.c:2392 ___sys_sendmsg net/socket.c:2446 [inline] __sys_sendmmsg+0x315/0x4b0 net/socket.c:2532 __do_sys_sendmmsg net/socket.c:2561 [inline] __se_sys_sendmmsg net/socket.c:2558 [inline] __x64_sys_sendmmsg+0x53/0x60 net/socket.c:2558 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae read to 0xffff88814eeb24e0 of 4 bytes by task 25834 on cpu 1: skb_queue_len include/linux/skbuff.h:1869 [inline] unix_recvq_full net/unix/af_unix.c:194 [inline] unix_dgram_poll+0x2bc/0x3e0 net/unix/af_unix.c:2777 sock_poll+0x23e/0x260 net/socket.c:1288 vfs_poll include/linux/poll.h:90 [inline] ep_item_poll fs/eventpoll.c:846 [inline] ep_send_events fs/eventpoll.c:1683 [inline] ep_poll fs/eventpoll.c:1798 [inline] do_epoll_wait+0x6ad/0xf00 fs/eventpoll.c:2226 __do_sys_epoll_wait fs/eventpoll.c:2238 [inline] __se_sys_epoll_wait fs/eventpoll.c:2233 [inline] __x64_sys_epoll_wait+0xf6/0x120 fs/eventpoll.c:2233 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x3d/0x90 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae value changed: 0x0000001b -> 0x00000001 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 25834 Comm: syz-executor.1 Tainted: G W 5.14.0-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Fixes: 86b18aaa2b5b ("skbuff: fix a data race in skb_queue_len()") Cc: Qian Cai Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/linux/skbuff.h | 2 +- net/unix/af_unix.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'net') diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h index 6bdb0db3e825..841e2f0f5240 100644 --- a/include/linux/skbuff.h +++ b/include/linux/skbuff.h @@ -1940,7 +1940,7 @@ static inline void __skb_insert(struct sk_buff *newsk, WRITE_ONCE(newsk->prev, prev); WRITE_ONCE(next->prev, newsk); WRITE_ONCE(prev->next, newsk); - list->qlen++; + WRITE_ONCE(list->qlen, list->qlen + 1); } static inline void __skb_queue_splice(const struct sk_buff_head *list, diff --git a/net/unix/af_unix.c b/net/unix/af_unix.c index eb47b9de2380..92345c9bb60c 100644 --- a/net/unix/af_unix.c +++ b/net/unix/af_unix.c @@ -3073,7 +3073,7 @@ static __poll_t unix_dgram_poll(struct file *file, struct socket *sock, other = unix_peer(sk); if (other && unix_peer(other) != sk && - unix_recvq_full(other) && + unix_recvq_full_lockless(other) && unix_dgram_peer_wake_me(sk, other)) writable = 0; -- cgit v1.3-14-g43fede From 9b6ff7eb666415e1558f1ba8a742f5db6a9954de Mon Sep 17 00:00:00 2001 From: Xiyu Yang Date: Thu, 9 Sep 2021 12:32:00 +0800 Subject: net/l2tp: Fix reference count leak in l2tp_udp_recv_core The reference count leak issue may take place in an error handling path. If both conditions of tunnel->version == L2TP_HDR_VER_3 and the return value of l2tp_v3_ensure_opt_in_linear is nonzero, the function would directly jump to label invalid, without decrementing the reference count of the l2tp_session object session increased earlier by l2tp_tunnel_get_session(). This may result in refcount leaks. Fix this issue by decrease the reference count before jumping to the label invalid. Fixes: 4522a70db7aa ("l2tp: fix reading optional fields of L2TPv3") Signed-off-by: Xiyu Yang Signed-off-by: Xin Xiong Signed-off-by: Xin Tan Signed-off-by: David S. Miller --- net/l2tp/l2tp_core.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'net') diff --git a/net/l2tp/l2tp_core.c b/net/l2tp/l2tp_core.c index 53486b162f01..93271a2632b8 100644 --- a/net/l2tp/l2tp_core.c +++ b/net/l2tp/l2tp_core.c @@ -869,8 +869,10 @@ static int l2tp_udp_recv_core(struct l2tp_tunnel *tunnel, struct sk_buff *skb) } if (tunnel->version == L2TP_HDR_VER_3 && - l2tp_v3_ensure_opt_in_linear(session, skb, &ptr, &optr)) + l2tp_v3_ensure_opt_in_linear(session, skb, &ptr, &optr)) { + l2tp_session_dec_refcount(session); goto invalid; + } l2tp_recv_common(session, skb, ptr, optr, hdrflags, length); l2tp_session_dec_refcount(session); -- cgit v1.3-14-g43fede From dc41c4a98a76640e7085815f937eadd1f336ba85 Mon Sep 17 00:00:00 2001 From: Baruch Siach Date: Thu, 9 Sep 2021 20:49:47 +0300 Subject: net/packet: clarify source of pr_*() messages Add pr_fmt macro to spell out the source of messages in prefix. Before this patch: packet size is too long (1543 > 1518) With this patch: af_packet: packet size is too long (1543 > 1518) Signed-off-by: Baruch Siach Signed-off-by: David S. Miller --- net/packet/af_packet.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'net') diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c index 543365f58e97..2a2bc64f75cf 100644 --- a/net/packet/af_packet.c +++ b/net/packet/af_packet.c @@ -46,6 +46,8 @@ * Copyright (C) 2011, */ +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + #include #include #include -- cgit v1.3-14-g43fede From d7807a9adf4856171f8441f13078c33941df48ab Mon Sep 17 00:00:00 2001 From: Yajun Deng Date: Mon, 13 Sep 2021 12:04:42 +0800 Subject: Revert "ipv4: fix memory leaks in ip_cmsg_send() callers" This reverts commit 919483096bfe75dda338e98d56da91a263746a0a. There is only when ip_options_get() return zero need to free. It already called kfree() when return error. Fixes: 919483096bfe ("ipv4: fix memory leaks in ip_cmsg_send() callers") Signed-off-by: Yajun Deng Signed-off-by: David S. Miller --- net/ipv4/ip_sockglue.c | 2 +- net/ipv4/ping.c | 5 ++--- net/ipv4/raw.c | 5 ++--- net/ipv4/udp.c | 5 ++--- 4 files changed, 7 insertions(+), 10 deletions(-) (limited to 'net') diff --git a/net/ipv4/ip_sockglue.c b/net/ipv4/ip_sockglue.c index b297bb28556e..7cef9987ab4a 100644 --- a/net/ipv4/ip_sockglue.c +++ b/net/ipv4/ip_sockglue.c @@ -279,7 +279,7 @@ int ip_cmsg_send(struct sock *sk, struct msghdr *msg, struct ipcm_cookie *ipc, case IP_RETOPTS: err = cmsg->cmsg_len - sizeof(struct cmsghdr); - /* Our caller is responsible for freeing ipc->opt */ + /* Our caller is responsible for freeing ipc->opt when err = 0 */ err = ip_options_get(net, &ipc->opt, KERNEL_SOCKPTR(CMSG_DATA(cmsg)), err < 40 ? err : 40); diff --git a/net/ipv4/ping.c b/net/ipv4/ping.c index 1e44a43acfe2..c588f9f2f46c 100644 --- a/net/ipv4/ping.c +++ b/net/ipv4/ping.c @@ -727,10 +727,9 @@ static int ping_v4_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) if (msg->msg_controllen) { err = ip_cmsg_send(sk, msg, &ipc, false); - if (unlikely(err)) { - kfree(ipc.opt); + if (unlikely(err)) return err; - } + if (ipc.opt) free = 1; } diff --git a/net/ipv4/raw.c b/net/ipv4/raw.c index bb446e60cf58..1c98063a3ae8 100644 --- a/net/ipv4/raw.c +++ b/net/ipv4/raw.c @@ -562,10 +562,9 @@ static int raw_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) if (msg->msg_controllen) { err = ip_cmsg_send(sk, msg, &ipc, false); - if (unlikely(err)) { - kfree(ipc.opt); + if (unlikely(err)) goto out; - } + if (ipc.opt) free = 1; } diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index 8851c9463b4b..d5f5981d7a43 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -1122,10 +1122,9 @@ int udp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) if (err > 0) err = ip_cmsg_send(sk, msg, &ipc, sk->sk_family == AF_INET6); - if (unlikely(err < 0)) { - kfree(ipc.opt); + if (unlikely(err < 0)) return err; - } + if (ipc.opt) free = 1; connected = 0; -- cgit v1.3-14-g43fede From e50e711351bdc656a8e6ca1022b4293cae8dcd59 Mon Sep 17 00:00:00 2001 From: Aya Levin Date: Mon, 13 Sep 2021 10:53:49 +0300 Subject: udp_tunnel: Fix udp_tunnel_nic work-queue type Turn udp_tunnel_nic work-queue to an ordered work-queue. This queue holds the UDP-tunnel configuration commands of the different netdevs. When the netdevs are functions of the same NIC the order of execution may be crucial. Problem example: NIC with 2 PFs, both PFs declare offload quota of up to 3 UDP-ports. $ifconfig eth2 1.1.1.1/16 up $ip link add eth2_19503 type vxlan id 5049 remote 1.1.1.2 dev eth2 dstport 19053 $ip link set dev eth2_19503 up $ip link add eth2_19504 type vxlan id 5049 remote 1.1.1.3 dev eth2 dstport 19054 $ip link set dev eth2_19504 up $ip link add eth2_19505 type vxlan id 5049 remote 1.1.1.4 dev eth2 dstport 19055 $ip link set dev eth2_19505 up $ip link add eth2_19506 type vxlan id 5049 remote 1.1.1.5 dev eth2 dstport 19056 $ip link set dev eth2_19506 up NIC RX port offload infrastructure offloads the first 3 UDP-ports (on all devices which sets NETIF_F_RX_UDP_TUNNEL_PORT feature) and not UDP-port 19056. So both PFs gets this offload configuration. $ip link set dev eth2_19504 down This triggers udp-tunnel-core to remove the UDP-port 19504 from offload-ports-list and offload UDP-port 19056 instead. In this scenario it is important that the UDP-port of 19504 will be removed from both PFs before trying to add UDP-port 19056. The NIC can stop offloading a UDP-port only when all references are removed. Otherwise the NIC may report exceeding of the offload quota. Fixes: cc4e3835eff4 ("udp_tunnel: add central NIC RX port offload infrastructure") Signed-off-by: Aya Levin Reviewed-by: Tariq Toukan Signed-off-by: David S. Miller --- net/ipv4/udp_tunnel_nic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net') diff --git a/net/ipv4/udp_tunnel_nic.c b/net/ipv4/udp_tunnel_nic.c index 0d122edc368d..b91003538d87 100644 --- a/net/ipv4/udp_tunnel_nic.c +++ b/net/ipv4/udp_tunnel_nic.c @@ -935,7 +935,7 @@ static int __init udp_tunnel_nic_init_module(void) { int err; - udp_tunnel_nic_workqueue = alloc_workqueue("udp_tunnel_nic", 0, 0); + udp_tunnel_nic_workqueue = alloc_ordered_workqueue("udp_tunnel_nic", 0); if (!udp_tunnel_nic_workqueue) return -ENOMEM; -- cgit v1.3-14-g43fede From f4bb62e64c88c93060c051195d3bbba804e56945 Mon Sep 17 00:00:00 2001 From: Hoang Le Date: Mon, 13 Sep 2021 16:28:52 +0700 Subject: tipc: increase timeout in tipc_sk_enqueue() In tipc_sk_enqueue() we use hardcoded 2 jiffies to extract socket buffer from generic queue to particular socket. The 2 jiffies is too short in case there are other high priority tasks get CPU cycles for multiple jiffies update. As result, no buffer could be enqueued to particular socket. To solve this, we switch to use constant timeout 20msecs. Then, the function will be expired between 2 jiffies (CONFIG_100HZ) and 20 jiffies (CONFIG_1000HZ). Fixes: c637c1035534 ("tipc: resolve race problem at unicast message reception") Acked-by: Jon Maloy Signed-off-by: Hoang Le Signed-off-by: David S. Miller --- net/tipc/socket.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net') diff --git a/net/tipc/socket.c b/net/tipc/socket.c index a0a27d87f631..ad570c2450be 100644 --- a/net/tipc/socket.c +++ b/net/tipc/socket.c @@ -2423,7 +2423,7 @@ static int tipc_sk_backlog_rcv(struct sock *sk, struct sk_buff *skb) static void tipc_sk_enqueue(struct sk_buff_head *inputq, struct sock *sk, u32 dport, struct sk_buff_head *xmitq) { - unsigned long time_limit = jiffies + 2; + unsigned long time_limit = jiffies + usecs_to_jiffies(20000); struct sk_buff *skb; unsigned int lim; atomic_t *dcnt; -- cgit v1.3-14-g43fede From e87b5052271e39d62337ade531992b7e5d8c2cfa Mon Sep 17 00:00:00 2001 From: zhang kai Date: Thu, 9 Sep 2021 16:39:18 +0800 Subject: ipv6: delay fib6_sernum increase in fib6_add only increase fib6_sernum in net namespace after add fib6_info successfully. Signed-off-by: zhang kai Reviewed-by: David Ahern Signed-off-by: David S. Miller --- net/ipv6/ip6_fib.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'net') diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c index 1bec5b22f80d..0371d2c14145 100644 --- a/net/ipv6/ip6_fib.c +++ b/net/ipv6/ip6_fib.c @@ -1378,7 +1378,6 @@ int fib6_add(struct fib6_node *root, struct fib6_info *rt, int err = -ENOMEM; int allow_create = 1; int replace_required = 0; - int sernum = fib6_new_sernum(info->nl_net); if (info->nlh) { if (!(info->nlh->nlmsg_flags & NLM_F_CREATE)) @@ -1478,7 +1477,7 @@ int fib6_add(struct fib6_node *root, struct fib6_info *rt, if (!err) { if (rt->nh) list_add(&rt->nh_list, &rt->nh->f6i_list); - __fib6_update_sernum_upto_root(rt, sernum); + __fib6_update_sernum_upto_root(rt, fib6_new_sernum(info->nl_net)); fib6_start_gc(info->nl_net, rt); } -- cgit v1.3-14-g43fede From 8520e224f547cd070c7c8f97b1fc6d58cff7ccaa Mon Sep 17 00:00:00 2001 From: Daniel Borkmann Date: Tue, 14 Sep 2021 01:07:57 +0200 Subject: bpf, cgroups: Fix cgroup v2 fallback on v1/v2 mixed mode Fix cgroup v1 interference when non-root cgroup v2 BPF programs are used. Back in the days, commit bd1060a1d671 ("sock, cgroup: add sock->sk_cgroup") embedded per-socket cgroup information into sock->sk_cgrp_data and in order to save 8 bytes in struct sock made both mutually exclusive, that is, when cgroup v1 socket tagging (e.g. net_cls/net_prio) is used, then cgroup v2 falls back to the root cgroup in sock_cgroup_ptr() (&cgrp_dfl_root.cgrp). The assumption made was "there is no reason to mix the two and this is in line with how legacy and v2 compatibility is handled" as stated in bd1060a1d671. However, with Kubernetes more widely supporting cgroups v2 as well nowadays, this assumption no longer holds, and the possibility of the v1/v2 mixed mode with the v2 root fallback being hit becomes a real security issue. Many of the cgroup v2 BPF programs are also used for policy enforcement, just to pick _one_ example, that is, to programmatically deny socket related system calls like connect(2) or bind(2). A v2 root fallback would implicitly cause a policy bypass for the affected Pods. In production environments, we have recently seen this case due to various circumstances: i) a different 3rd party agent and/or ii) a container runtime such as [0] in the user's environment configuring legacy cgroup v1 net_cls tags, which triggered implicitly mentioned root fallback. Another case is Kubernetes projects like kind [1] which create Kubernetes nodes in a container and also add cgroup namespaces to the mix, meaning programs which are attached to the cgroup v2 root of the cgroup namespace get attached to a non-root cgroup v2 path from init namespace point of view. And the latter's root is out of reach for agents on a kind Kubernetes node to configure. Meaning, any entity on the node setting cgroup v1 net_cls tag will trigger the bypass despite cgroup v2 BPF programs attached to the namespace root. Generally, this mutual exclusiveness does not hold anymore in today's user environments and makes cgroup v2 usage from BPF side fragile and unreliable. This fix adds proper struct cgroup pointer for the cgroup v2 case to struct sock_cgroup_data in order to address these issues; this implicitly also fixes the tradeoffs being made back then with regards to races and refcount leaks as stated in bd1060a1d671, and removes the fallback, so that cgroup v2 BPF programs always operate as expected. [0] https://github.com/nestybox/sysbox/ [1] https://kind.sigs.k8s.io/ Fixes: bd1060a1d671 ("sock, cgroup: add sock->sk_cgroup") Signed-off-by: Daniel Borkmann Signed-off-by: Alexei Starovoitov Acked-by: Stanislav Fomichev Acked-by: Tejun Heo Link: https://lore.kernel.org/bpf/20210913230759.2313-1-daniel@iogearbox.net --- include/linux/cgroup-defs.h | 107 +++++++++++-------------------------------- include/linux/cgroup.h | 22 +-------- kernel/cgroup/cgroup.c | 50 ++++---------------- net/core/netclassid_cgroup.c | 7 +-- net/core/netprio_cgroup.c | 10 +--- 5 files changed, 41 insertions(+), 155 deletions(-) (limited to 'net') diff --git a/include/linux/cgroup-defs.h b/include/linux/cgroup-defs.h index e1c705fdfa7c..db2e147e069f 100644 --- a/include/linux/cgroup-defs.h +++ b/include/linux/cgroup-defs.h @@ -752,107 +752,54 @@ static inline void cgroup_threadgroup_change_end(struct task_struct *tsk) {} * sock_cgroup_data is embedded at sock->sk_cgrp_data and contains * per-socket cgroup information except for memcg association. * - * On legacy hierarchies, net_prio and net_cls controllers directly set - * attributes on each sock which can then be tested by the network layer. - * On the default hierarchy, each sock is associated with the cgroup it was - * created in and the networking layer can match the cgroup directly. - * - * To avoid carrying all three cgroup related fields separately in sock, - * sock_cgroup_data overloads (prioidx, classid) and the cgroup pointer. - * On boot, sock_cgroup_data records the cgroup that the sock was created - * in so that cgroup2 matches can be made; however, once either net_prio or - * net_cls starts being used, the area is overridden to carry prioidx and/or - * classid. The two modes are distinguished by whether the lowest bit is - * set. Clear bit indicates cgroup pointer while set bit prioidx and - * classid. - * - * While userland may start using net_prio or net_cls at any time, once - * either is used, cgroup2 matching no longer works. There is no reason to - * mix the two and this is in line with how legacy and v2 compatibility is - * handled. On mode switch, cgroup references which are already being - * pointed to by socks may be leaked. While this can be remedied by adding - * synchronization around sock_cgroup_data, given that the number of leaked - * cgroups is bound and highly unlikely to be high, this seems to be the - * better trade-off. + * On legacy hierarchies, net_prio and net_cls controllers directly + * set attributes on each sock which can then be tested by the network + * layer. On the default hierarchy, each sock is associated with the + * cgroup it was created in and the networking layer can match the + * cgroup directly. */ struct sock_cgroup_data { - union { -#ifdef __LITTLE_ENDIAN - struct { - u8 is_data : 1; - u8 no_refcnt : 1; - u8 unused : 6; - u8 padding; - u16 prioidx; - u32 classid; - } __packed; -#else - struct { - u32 classid; - u16 prioidx; - u8 padding; - u8 unused : 6; - u8 no_refcnt : 1; - u8 is_data : 1; - } __packed; + struct cgroup *cgroup; /* v2 */ +#ifdef CONFIG_CGROUP_NET_CLASSID + u32 classid; /* v1 */ +#endif +#ifdef CONFIG_CGROUP_NET_PRIO + u16 prioidx; /* v1 */ #endif - u64 val; - }; }; -/* - * There's a theoretical window where the following accessors race with - * updaters and return part of the previous pointer as the prioidx or - * classid. Such races are short-lived and the result isn't critical. - */ static inline u16 sock_cgroup_prioidx(const struct sock_cgroup_data *skcd) { - /* fallback to 1 which is always the ID of the root cgroup */ - return (skcd->is_data & 1) ? skcd->prioidx : 1; +#ifdef CONFIG_CGROUP_NET_PRIO + return READ_ONCE(skcd->prioidx); +#else + return 1; +#endif } static inline u32 sock_cgroup_classid(const struct sock_cgroup_data *skcd) { - /* fallback to 0 which is the unconfigured default classid */ - return (skcd->is_data & 1) ? skcd->classid : 0; +#ifdef CONFIG_CGROUP_NET_CLASSID + return READ_ONCE(skcd->classid); +#else + return 0; +#endif } -/* - * If invoked concurrently, the updaters may clobber each other. The - * caller is responsible for synchronization. - */ static inline void sock_cgroup_set_prioidx(struct sock_cgroup_data *skcd, u16 prioidx) { - struct sock_cgroup_data skcd_buf = {{ .val = READ_ONCE(skcd->val) }}; - - if (sock_cgroup_prioidx(&skcd_buf) == prioidx) - return; - - if (!(skcd_buf.is_data & 1)) { - skcd_buf.val = 0; - skcd_buf.is_data = 1; - } - - skcd_buf.prioidx = prioidx; - WRITE_ONCE(skcd->val, skcd_buf.val); /* see sock_cgroup_ptr() */ +#ifdef CONFIG_CGROUP_NET_PRIO + WRITE_ONCE(skcd->prioidx, prioidx); +#endif } static inline void sock_cgroup_set_classid(struct sock_cgroup_data *skcd, u32 classid) { - struct sock_cgroup_data skcd_buf = {{ .val = READ_ONCE(skcd->val) }}; - - if (sock_cgroup_classid(&skcd_buf) == classid) - return; - - if (!(skcd_buf.is_data & 1)) { - skcd_buf.val = 0; - skcd_buf.is_data = 1; - } - - skcd_buf.classid = classid; - WRITE_ONCE(skcd->val, skcd_buf.val); /* see sock_cgroup_ptr() */ +#ifdef CONFIG_CGROUP_NET_CLASSID + WRITE_ONCE(skcd->classid, classid); +#endif } #else /* CONFIG_SOCK_CGROUP_DATA */ diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h index 7bf60454a313..75c151413fda 100644 --- a/include/linux/cgroup.h +++ b/include/linux/cgroup.h @@ -829,33 +829,13 @@ static inline void cgroup_account_cputime_field(struct task_struct *task, */ #ifdef CONFIG_SOCK_CGROUP_DATA -#if defined(CONFIG_CGROUP_NET_PRIO) || defined(CONFIG_CGROUP_NET_CLASSID) -extern spinlock_t cgroup_sk_update_lock; -#endif - -void cgroup_sk_alloc_disable(void); void cgroup_sk_alloc(struct sock_cgroup_data *skcd); void cgroup_sk_clone(struct sock_cgroup_data *skcd); void cgroup_sk_free(struct sock_cgroup_data *skcd); static inline struct cgroup *sock_cgroup_ptr(struct sock_cgroup_data *skcd) { -#if defined(CONFIG_CGROUP_NET_PRIO) || defined(CONFIG_CGROUP_NET_CLASSID) - unsigned long v; - - /* - * @skcd->val is 64bit but the following is safe on 32bit too as we - * just need the lower ulong to be written and read atomically. - */ - v = READ_ONCE(skcd->val); - - if (v & 3) - return &cgrp_dfl_root.cgrp; - - return (struct cgroup *)(unsigned long)v ?: &cgrp_dfl_root.cgrp; -#else - return (struct cgroup *)(unsigned long)skcd->val; -#endif + return skcd->cgroup; } #else /* CONFIG_CGROUP_DATA */ diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c index 881ce1470beb..8afa8690d288 100644 --- a/kernel/cgroup/cgroup.c +++ b/kernel/cgroup/cgroup.c @@ -6572,74 +6572,44 @@ int cgroup_parse_float(const char *input, unsigned dec_shift, s64 *v) */ #ifdef CONFIG_SOCK_CGROUP_DATA -#if defined(CONFIG_CGROUP_NET_PRIO) || defined(CONFIG_CGROUP_NET_CLASSID) - -DEFINE_SPINLOCK(cgroup_sk_update_lock); -static bool cgroup_sk_alloc_disabled __read_mostly; - -void cgroup_sk_alloc_disable(void) -{ - if (cgroup_sk_alloc_disabled) - return; - pr_info("cgroup: disabling cgroup2 socket matching due to net_prio or net_cls activation\n"); - cgroup_sk_alloc_disabled = true; -} - -#else - -#define cgroup_sk_alloc_disabled false - -#endif - void cgroup_sk_alloc(struct sock_cgroup_data *skcd) { - if (cgroup_sk_alloc_disabled) { - skcd->no_refcnt = 1; - return; - } - /* Don't associate the sock with unrelated interrupted task's cgroup. */ if (in_interrupt()) return; rcu_read_lock(); - while (true) { struct css_set *cset; cset = task_css_set(current); if (likely(cgroup_tryget(cset->dfl_cgrp))) { - skcd->val = (unsigned long)cset->dfl_cgrp; + skcd->cgroup = cset->dfl_cgrp; cgroup_bpf_get(cset->dfl_cgrp); break; } cpu_relax(); } - rcu_read_unlock(); } void cgroup_sk_clone(struct sock_cgroup_data *skcd) { - if (skcd->val) { - if (skcd->no_refcnt) - return; - /* - * We might be cloning a socket which is left in an empty - * cgroup and the cgroup might have already been rmdir'd. - * Don't use cgroup_get_live(). - */ - cgroup_get(sock_cgroup_ptr(skcd)); - cgroup_bpf_get(sock_cgroup_ptr(skcd)); - } + struct cgroup *cgrp = sock_cgroup_ptr(skcd); + + /* + * We might be cloning a socket which is left in an empty + * cgroup and the cgroup might have already been rmdir'd. + * Don't use cgroup_get_live(). + */ + cgroup_get(cgrp); + cgroup_bpf_get(cgrp); } void cgroup_sk_free(struct sock_cgroup_data *skcd) { struct cgroup *cgrp = sock_cgroup_ptr(skcd); - if (skcd->no_refcnt) - return; cgroup_bpf_put(cgrp); cgroup_put(cgrp); } diff --git a/net/core/netclassid_cgroup.c b/net/core/netclassid_cgroup.c index b49c57d35a88..1a6a86693b74 100644 --- a/net/core/netclassid_cgroup.c +++ b/net/core/netclassid_cgroup.c @@ -71,11 +71,8 @@ static int update_classid_sock(const void *v, struct file *file, unsigned n) struct update_classid_context *ctx = (void *)v; struct socket *sock = sock_from_file(file); - if (sock) { - spin_lock(&cgroup_sk_update_lock); + if (sock) sock_cgroup_set_classid(&sock->sk->sk_cgrp_data, ctx->classid); - spin_unlock(&cgroup_sk_update_lock); - } if (--ctx->batch == 0) { ctx->batch = UPDATE_CLASSID_BATCH; return n + 1; @@ -121,8 +118,6 @@ static int write_classid(struct cgroup_subsys_state *css, struct cftype *cft, struct css_task_iter it; struct task_struct *p; - cgroup_sk_alloc_disable(); - cs->classid = (u32)value; css_task_iter_start(css, 0, &it); diff --git a/net/core/netprio_cgroup.c b/net/core/netprio_cgroup.c index 99a431c56f23..8456dfbe2eb4 100644 --- a/net/core/netprio_cgroup.c +++ b/net/core/netprio_cgroup.c @@ -207,8 +207,6 @@ static ssize_t write_priomap(struct kernfs_open_file *of, if (!dev) return -ENODEV; - cgroup_sk_alloc_disable(); - rtnl_lock(); ret = netprio_set_prio(of_css(of), dev, prio); @@ -221,12 +219,10 @@ static ssize_t write_priomap(struct kernfs_open_file *of, static int update_netprio(const void *v, struct file *file, unsigned n) { struct socket *sock = sock_from_file(file); - if (sock) { - spin_lock(&cgroup_sk_update_lock); + + if (sock) sock_cgroup_set_prioidx(&sock->sk->sk_cgrp_data, (unsigned long)v); - spin_unlock(&cgroup_sk_update_lock); - } return 0; } @@ -235,8 +231,6 @@ static void net_prio_attach(struct cgroup_taskset *tset) struct task_struct *p; struct cgroup_subsys_state *css; - cgroup_sk_alloc_disable(); - cgroup_taskset_for_each(p, css, tset) { void *v = (void *)(unsigned long)css->id; -- cgit v1.3-14-g43fede From 550ac9c1aaaaf51fd42e20d461f0b1cdbd55b3d2 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Mon, 13 Sep 2021 11:08:36 -0700 Subject: net-caif: avoid user-triggerable WARN_ON(1) syszbot triggers this warning, which looks something we can easily prevent. If we initialize priv->list_field in chnl_net_init(), then always use list_del_init(), we can remove robust_list_del() completely. WARNING: CPU: 0 PID: 3233 at net/caif/chnl_net.c:67 robust_list_del net/caif/chnl_net.c:67 [inline] WARNING: CPU: 0 PID: 3233 at net/caif/chnl_net.c:67 chnl_net_uninit+0xc9/0x2e0 net/caif/chnl_net.c:375 Modules linked in: CPU: 0 PID: 3233 Comm: syz-executor.3 Not tainted 5.14.0-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:robust_list_del net/caif/chnl_net.c:67 [inline] RIP: 0010:chnl_net_uninit+0xc9/0x2e0 net/caif/chnl_net.c:375 Code: 89 eb e8 3a a3 ba f8 48 89 d8 48 c1 e8 03 42 80 3c 28 00 0f 85 bf 01 00 00 48 81 fb 00 14 4e 8d 48 8b 2b 75 d0 e8 17 a3 ba f8 <0f> 0b 5b 5d 41 5c 41 5d e9 0a a3 ba f8 4c 89 e3 e8 02 a3 ba f8 4c RSP: 0018:ffffc90009067248 EFLAGS: 00010202 RAX: 0000000000008780 RBX: ffffffff8d4e1400 RCX: ffffc9000fd34000 RDX: 0000000000040000 RSI: ffffffff88bb6e49 RDI: 0000000000000003 RBP: ffff88802cd9ee08 R08: 0000000000000000 R09: ffffffff8d0e6647 R10: ffffffff88bb6dc2 R11: 0000000000000000 R12: ffff88803791ae08 R13: dffffc0000000000 R14: 00000000e600ffce R15: ffff888073ed3480 FS: 00007fed10fa0700(0000) GS:ffff8880b9d00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000001b2c322000 CR3: 00000000164a6000 CR4: 00000000001506e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Call Trace: register_netdevice+0xadf/0x1500 net/core/dev.c:10347 ipcaif_newlink+0x4c/0x260 net/caif/chnl_net.c:468 __rtnl_newlink+0x106d/0x1750 net/core/rtnetlink.c:3458 rtnl_newlink+0x64/0xa0 net/core/rtnetlink.c:3506 rtnetlink_rcv_msg+0x413/0xb80 net/core/rtnetlink.c:5572 netlink_rcv_skb+0x153/0x420 net/netlink/af_netlink.c:2504 netlink_unicast_kernel net/netlink/af_netlink.c:1314 [inline] netlink_unicast+0x533/0x7d0 net/netlink/af_netlink.c:1340 netlink_sendmsg+0x86d/0xdb0 net/netlink/af_netlink.c:1929 sock_sendmsg_nosec net/socket.c:704 [inline] sock_sendmsg+0xcf/0x120 net/socket.c:724 __sys_sendto+0x21c/0x320 net/socket.c:2036 __do_sys_sendto net/socket.c:2048 [inline] __se_sys_sendto net/socket.c:2044 [inline] __x64_sys_sendto+0xdd/0x1b0 net/socket.c:2044 do_syscall_x64 arch/x86/entry/common.c:50 [inline] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80 entry_SYSCALL_64_after_hwframe+0x44/0xae Fixes: cc36a070b590 ("net-caif: add CAIF netdevice") Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/caif/chnl_net.c | 19 +++---------------- 1 file changed, 3 insertions(+), 16 deletions(-) (limited to 'net') diff --git a/net/caif/chnl_net.c b/net/caif/chnl_net.c index 37b67194c0df..414dc5671c45 100644 --- a/net/caif/chnl_net.c +++ b/net/caif/chnl_net.c @@ -53,20 +53,6 @@ struct chnl_net { enum caif_states state; }; -static void robust_list_del(struct list_head *delete_node) -{ - struct list_head *list_node; - struct list_head *n; - ASSERT_RTNL(); - list_for_each_safe(list_node, n, &chnl_net_list) { - if (list_node == delete_node) { - list_del(list_node); - return; - } - } - WARN_ON(1); -} - static int chnl_recv_cb(struct cflayer *layr, struct cfpkt *pkt) { struct sk_buff *skb; @@ -364,6 +350,7 @@ static int chnl_net_init(struct net_device *dev) ASSERT_RTNL(); priv = netdev_priv(dev); strncpy(priv->name, dev->name, sizeof(priv->name)); + INIT_LIST_HEAD(&priv->list_field); return 0; } @@ -372,7 +359,7 @@ static void chnl_net_uninit(struct net_device *dev) struct chnl_net *priv; ASSERT_RTNL(); priv = netdev_priv(dev); - robust_list_del(&priv->list_field); + list_del_init(&priv->list_field); } static const struct net_device_ops netdev_ops = { @@ -537,7 +524,7 @@ static void __exit chnl_exit_module(void) rtnl_lock(); list_for_each_safe(list_node, _tmp, &chnl_net_list) { dev = list_entry(list_node, struct chnl_net, list_field); - list_del(list_node); + list_del_init(list_node); delete_device(dev); } rtnl_unlock(); -- cgit v1.3-14-g43fede From 4f884f3962767877d7aabbc1ec124d2c307a4257 Mon Sep 17 00:00:00 2001 From: zhenggy Date: Tue, 14 Sep 2021 09:51:15 +0800 Subject: tcp: fix tp->undo_retrans accounting in tcp_sacktag_one() Commit 10d3be569243 ("tcp-tso: do not split TSO packets at retransmit time") may directly retrans a multiple segments TSO/GSO packet without split, Since this commit, we can no longer assume that a retransmitted packet is a single segment. This patch fixes the tp->undo_retrans accounting in tcp_sacktag_one() that use the actual segments(pcount) of the retransmitted packet. Before that commit (10d3be569243), the assumption underlying the tp->undo_retrans-- seems correct. Fixes: 10d3be569243 ("tcp-tso: do not split TSO packets at retransmit time") Signed-off-by: zhenggy Reviewed-by: Eric Dumazet Acked-by: Yuchung Cheng Acked-by: Neal Cardwell Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 3f7bd7ae7d7a..141e85e6422b 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -1346,7 +1346,7 @@ static u8 tcp_sacktag_one(struct sock *sk, if (dup_sack && (sacked & TCPCB_RETRANS)) { if (tp->undo_marker && tp->undo_retrans > 0 && after(end_seq, tp->undo_marker)) - tp->undo_retrans--; + tp->undo_retrans = max_t(int, 0, tp->undo_retrans - pcount); if ((sacked & TCPCB_SACKED_ACKED) && before(start_seq, state->reord)) state->reord = start_seq; -- cgit v1.3-14-g43fede From d198b27762644c71362e43a7533f89c92b115bcf Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Mon, 13 Sep 2021 22:18:51 -0700 Subject: Revert "Revert "ipv4: fix memory leaks in ip_cmsg_send() callers"" This reverts commit d7807a9adf4856171f8441f13078c33941df48ab. As mentioned in https://lkml.org/lkml/2021/9/13/1819 5 years old commit 919483096bfe ("ipv4: fix memory leaks in ip_cmsg_send() callers") was a correct fix. ip_cmsg_send() can loop over multiple cmsghdr() If IP_RETOPTS has been successful, but following cmsghdr generates an error, we do not free ipc.ok If IP_RETOPTS is not successful, we have freed the allocated temporary space, not the one currently in ipc.opt. Sure, code could be refactored, but let's not bring back old bugs. Fixes: d7807a9adf48 ("Revert "ipv4: fix memory leaks in ip_cmsg_send() callers"") Signed-off-by: Eric Dumazet Cc: Yajun Deng Signed-off-by: David S. Miller --- net/ipv4/ip_sockglue.c | 2 +- net/ipv4/ping.c | 5 +++-- net/ipv4/raw.c | 5 +++-- net/ipv4/udp.c | 5 +++-- 4 files changed, 10 insertions(+), 7 deletions(-) (limited to 'net') diff --git a/net/ipv4/ip_sockglue.c b/net/ipv4/ip_sockglue.c index 7cef9987ab4a..b297bb28556e 100644 --- a/net/ipv4/ip_sockglue.c +++ b/net/ipv4/ip_sockglue.c @@ -279,7 +279,7 @@ int ip_cmsg_send(struct sock *sk, struct msghdr *msg, struct ipcm_cookie *ipc, case IP_RETOPTS: err = cmsg->cmsg_len - sizeof(struct cmsghdr); - /* Our caller is responsible for freeing ipc->opt when err = 0 */ + /* Our caller is responsible for freeing ipc->opt */ err = ip_options_get(net, &ipc->opt, KERNEL_SOCKPTR(CMSG_DATA(cmsg)), err < 40 ? err : 40); diff --git a/net/ipv4/ping.c b/net/ipv4/ping.c index c588f9f2f46c..1e44a43acfe2 100644 --- a/net/ipv4/ping.c +++ b/net/ipv4/ping.c @@ -727,9 +727,10 @@ static int ping_v4_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) if (msg->msg_controllen) { err = ip_cmsg_send(sk, msg, &ipc, false); - if (unlikely(err)) + if (unlikely(err)) { + kfree(ipc.opt); return err; - + } if (ipc.opt) free = 1; } diff --git a/net/ipv4/raw.c b/net/ipv4/raw.c index 1c98063a3ae8..bb446e60cf58 100644 --- a/net/ipv4/raw.c +++ b/net/ipv4/raw.c @@ -562,9 +562,10 @@ static int raw_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) if (msg->msg_controllen) { err = ip_cmsg_send(sk, msg, &ipc, false); - if (unlikely(err)) + if (unlikely(err)) { + kfree(ipc.opt); goto out; - + } if (ipc.opt) free = 1; } diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index d5f5981d7a43..8851c9463b4b 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -1122,9 +1122,10 @@ int udp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) if (err > 0) err = ip_cmsg_send(sk, msg, &ipc, sk->sk_family == AF_INET6); - if (unlikely(err < 0)) + if (unlikely(err < 0)) { + kfree(ipc.opt); return err; - + } if (ipc.opt) free = 1; connected = 0; -- cgit v1.3-14-g43fede From 6a52e73368038f47f6618623d75061dc263b26ae Mon Sep 17 00:00:00 2001 From: Vladimir Oltean Date: Tue, 14 Sep 2021 16:43:31 +0300 Subject: net: dsa: destroy the phylink instance on any error in dsa_slave_phy_setup DSA supports connecting to a phy-handle, and has a fallback to a non-OF based method of connecting to an internal PHY on the switch's own MDIO bus, if no phy-handle and no fixed-link nodes were present. The -ENODEV error code from the first attempt (phylink_of_phy_connect) is what triggers the second attempt (phylink_connect_phy). However, when the first attempt returns a different error code than -ENODEV, this results in an unbalance of calls to phylink_create and phylink_destroy by the time we exit the function. The phylink instance has leaked. There are many other error codes that can be returned by phylink_of_phy_connect. For example, phylink_validate returns -EINVAL. So this is a practical issue too. Fixes: aab9c4067d23 ("net: dsa: Plug in PHYLINK support") Signed-off-by: Vladimir Oltean Reviewed-by: Florian Fainelli Reviewed-by: Russell King (Oracle) Link: https://lore.kernel.org/r/20210914134331.2303380-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski --- net/dsa/slave.c | 12 +++++------- 1 file changed, 5 insertions(+), 7 deletions(-) (limited to 'net') diff --git a/net/dsa/slave.c b/net/dsa/slave.c index 662ff531d4e2..a2bf2d8ac65b 100644 --- a/net/dsa/slave.c +++ b/net/dsa/slave.c @@ -1854,13 +1854,11 @@ static int dsa_slave_phy_setup(struct net_device *slave_dev) * use the switch internal MDIO bus instead */ ret = dsa_slave_phy_connect(slave_dev, dp->index, phy_flags); - if (ret) { - netdev_err(slave_dev, - "failed to connect to port %d: %d\n", - dp->index, ret); - phylink_destroy(dp->pl); - return ret; - } + } + if (ret) { + netdev_err(slave_dev, "failed to connect to PHY: %pe\n", + ERR_PTR(ret)); + phylink_destroy(dp->pl); } return ret; -- cgit v1.3-14-g43fede From a57d8c217aadac75530b8e7ffb3a3e1b7bfd0330 Mon Sep 17 00:00:00 2001 From: Vladimir Oltean Date: Tue, 14 Sep 2021 16:47:26 +0300 Subject: net: dsa: flush switchdev workqueue before tearing down CPU/DSA ports Sometimes when unbinding the mv88e6xxx driver on Turris MOX, these error messages appear: mv88e6085 d0032004.mdio-mii:12: port 1 failed to delete be:79:b4:9e:9e:96 vid 1 from fdb: -2 mv88e6085 d0032004.mdio-mii:12: port 1 failed to delete be:79:b4:9e:9e:96 vid 0 from fdb: -2 mv88e6085 d0032004.mdio-mii:12: port 1 failed to delete d8:58:d7:00:ca:6d vid 100 from fdb: -2 mv88e6085 d0032004.mdio-mii:12: port 1 failed to delete d8:58:d7:00:ca:6d vid 1 from fdb: -2 mv88e6085 d0032004.mdio-mii:12: port 1 failed to delete d8:58:d7:00:ca:6d vid 0 from fdb: -2 (and similarly for other ports) What happens is that DSA has a policy "even if there are bugs, let's at least not leak memory" and dsa_port_teardown() clears the dp->fdbs and dp->mdbs lists, which are supposed to be empty. But deleting that cleanup code, the warnings go away. => the FDB and MDB lists (used for refcounting on shared ports, aka CPU and DSA ports) will eventually be empty, but are not empty by the time we tear down those ports. Aka we are deleting them too soon. The addresses that DSA complains about are host-trapped addresses: the local addresses of the ports, and the MAC address of the bridge device. The problem is that offloading those entries happens from a deferred work item scheduled by the SWITCHDEV_FDB_DEL_TO_DEVICE handler, and this races with the teardown of the CPU and DSA ports where the refcounting is kept. In fact, not only it races, but fundamentally speaking, if we iterate through the port list linearly, we might end up tearing down the shared ports even before we delete a DSA user port which has a bridge upper. So as it turns out, we need to first tear down the user ports (and the unused ones, for no better place of doing that), then the shared ports (the CPU and DSA ports). In between, we need to ensure that all work items scheduled by our switchdev handlers (which only run for user ports, hence the reason why we tear them down first) have finished. Fixes: 161ca59d39e9 ("net: dsa: reference count the MDB entries at the cross-chip notifier level") Signed-off-by: Vladimir Oltean Reviewed-by: Florian Fainelli Link: https://lore.kernel.org/r/20210914134726.2305133-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski --- include/net/dsa.h | 5 +++++ net/dsa/dsa.c | 5 +++++ net/dsa/dsa2.c | 46 +++++++++++++++++++++++++++++++--------------- net/dsa/dsa_priv.h | 1 + 4 files changed, 42 insertions(+), 15 deletions(-) (limited to 'net') diff --git a/include/net/dsa.h b/include/net/dsa.h index f9a17145255a..258867eff230 100644 --- a/include/net/dsa.h +++ b/include/net/dsa.h @@ -447,6 +447,11 @@ static inline bool dsa_port_is_user(struct dsa_port *dp) return dp->type == DSA_PORT_TYPE_USER; } +static inline bool dsa_port_is_unused(struct dsa_port *dp) +{ + return dp->type == DSA_PORT_TYPE_UNUSED; +} + static inline bool dsa_is_unused_port(struct dsa_switch *ds, int p) { return dsa_to_port(ds, p)->type == DSA_PORT_TYPE_UNUSED; diff --git a/net/dsa/dsa.c b/net/dsa/dsa.c index 1dc45e40f961..41f36ad8b0ec 100644 --- a/net/dsa/dsa.c +++ b/net/dsa/dsa.c @@ -345,6 +345,11 @@ bool dsa_schedule_work(struct work_struct *work) return queue_work(dsa_owq, work); } +void dsa_flush_workqueue(void) +{ + flush_workqueue(dsa_owq); +} + int dsa_devlink_param_get(struct devlink *dl, u32 id, struct devlink_param_gset_ctx *ctx) { diff --git a/net/dsa/dsa2.c b/net/dsa/dsa2.c index 1b2b25d7bd02..eef13cd20f19 100644 --- a/net/dsa/dsa2.c +++ b/net/dsa/dsa2.c @@ -897,6 +897,33 @@ static void dsa_switch_teardown(struct dsa_switch *ds) ds->setup = false; } +/* First tear down the non-shared, then the shared ports. This ensures that + * all work items scheduled by our switchdev handlers for user ports have + * completed before we destroy the refcounting kept on the shared ports. + */ +static void dsa_tree_teardown_ports(struct dsa_switch_tree *dst) +{ + struct dsa_port *dp; + + list_for_each_entry(dp, &dst->ports, list) + if (dsa_port_is_user(dp) || dsa_port_is_unused(dp)) + dsa_port_teardown(dp); + + dsa_flush_workqueue(); + + list_for_each_entry(dp, &dst->ports, list) + if (dsa_port_is_dsa(dp) || dsa_port_is_cpu(dp)) + dsa_port_teardown(dp); +} + +static void dsa_tree_teardown_switches(struct dsa_switch_tree *dst) +{ + struct dsa_port *dp; + + list_for_each_entry(dp, &dst->ports, list) + dsa_switch_teardown(dp->ds); +} + static int dsa_tree_setup_switches(struct dsa_switch_tree *dst) { struct dsa_port *dp; @@ -923,26 +950,13 @@ static int dsa_tree_setup_switches(struct dsa_switch_tree *dst) return 0; teardown: - list_for_each_entry(dp, &dst->ports, list) - dsa_port_teardown(dp); + dsa_tree_teardown_ports(dst); - list_for_each_entry(dp, &dst->ports, list) - dsa_switch_teardown(dp->ds); + dsa_tree_teardown_switches(dst); return err; } -static void dsa_tree_teardown_switches(struct dsa_switch_tree *dst) -{ - struct dsa_port *dp; - - list_for_each_entry(dp, &dst->ports, list) - dsa_port_teardown(dp); - - list_for_each_entry(dp, &dst->ports, list) - dsa_switch_teardown(dp->ds); -} - static int dsa_tree_setup_master(struct dsa_switch_tree *dst) { struct dsa_port *dp; @@ -1052,6 +1066,8 @@ static void dsa_tree_teardown(struct dsa_switch_tree *dst) dsa_tree_teardown_master(dst); + dsa_tree_teardown_ports(dst); + dsa_tree_teardown_switches(dst); dsa_tree_teardown_cpu_ports(dst); diff --git a/net/dsa/dsa_priv.h b/net/dsa/dsa_priv.h index 33ab7d7af9eb..a5c9bc7b66c6 100644 --- a/net/dsa/dsa_priv.h +++ b/net/dsa/dsa_priv.h @@ -170,6 +170,7 @@ void dsa_tag_driver_put(const struct dsa_device_ops *ops); const struct dsa_device_ops *dsa_find_tagger_by_name(const char *buf); bool dsa_schedule_work(struct work_struct *work); +void dsa_flush_workqueue(void); const char *dsa_tag_protocol_to_str(const struct dsa_device_ops *ops); static inline int dsa_tag_protocol_overhead(const struct dsa_device_ops *ops) -- cgit v1.3-14-g43fede