From 4a85a6a3320b4a622315d2e0ea91a1d2b013bce4 Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Fri, 18 Dec 2020 12:28:41 -0500 Subject: SUNRPC: Handle TCP socket sends with kernel_sendpage() again Daire Byrne reports a ~50% aggregrate throughput regression on his Linux NFS server after commit da1661b93bf4 ("SUNRPC: Teach server to use xprt_sock_sendmsg for socket sends"), which replaced kernel_send_page() calls in NFSD's socket send path with calls to sock_sendmsg() using iov_iter. Investigation showed that tcp_sendmsg() was not using zero-copy to send the xdr_buf's bvec pages, but instead was relying on memcpy. This means copying every byte of a large NFS READ payload. It looks like TLS sockets do indeed support a ->sendpage method, so it's really not necessary to use xprt_sock_sendmsg() to support TLS fully on the server. A mechanical reversion of da1661b93bf4 is not possible at this point, but we can re-implement the server's TCP socket sendmsg path using kernel_sendpage(). Reported-by: Daire Byrne BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=209439 Signed-off-by: Chuck Lever --- net/sunrpc/svcsock.c | 86 +++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 85 insertions(+), 1 deletion(-) (limited to 'net') diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c index b248f2349437..c9766d07eb81 100644 --- a/net/sunrpc/svcsock.c +++ b/net/sunrpc/svcsock.c @@ -1062,6 +1062,90 @@ err_noclose: return 0; /* record not complete */ } +static int svc_tcp_send_kvec(struct socket *sock, const struct kvec *vec, + int flags) +{ + return kernel_sendpage(sock, virt_to_page(vec->iov_base), + offset_in_page(vec->iov_base), + vec->iov_len, flags); +} + +/* + * kernel_sendpage() is used exclusively to reduce the number of + * copy operations in this path. Therefore the caller must ensure + * that the pages backing @xdr are unchanging. + * + * In addition, the logic assumes that * .bv_len is never larger + * than PAGE_SIZE. + */ +static int svc_tcp_sendmsg(struct socket *sock, struct msghdr *msg, + struct xdr_buf *xdr, rpc_fraghdr marker, + unsigned int *sentp) +{ + const struct kvec *head = xdr->head; + const struct kvec *tail = xdr->tail; + struct kvec rm = { + .iov_base = &marker, + .iov_len = sizeof(marker), + }; + int flags, ret; + + *sentp = 0; + xdr_alloc_bvec(xdr, GFP_KERNEL); + + msg->msg_flags = MSG_MORE; + ret = kernel_sendmsg(sock, msg, &rm, 1, rm.iov_len); + if (ret < 0) + return ret; + *sentp += ret; + if (ret != rm.iov_len) + return -EAGAIN; + + flags = head->iov_len < xdr->len ? MSG_MORE | MSG_SENDPAGE_NOTLAST : 0; + ret = svc_tcp_send_kvec(sock, head, flags); + if (ret < 0) + return ret; + *sentp += ret; + if (ret != head->iov_len) + goto out; + + if (xdr->page_len) { + unsigned int offset, len, remaining; + struct bio_vec *bvec; + + bvec = xdr->bvec; + offset = xdr->page_base; + remaining = xdr->page_len; + flags = MSG_MORE | MSG_SENDPAGE_NOTLAST; + while (remaining > 0) { + if (remaining <= PAGE_SIZE && tail->iov_len == 0) + flags = 0; + len = min(remaining, bvec->bv_len); + ret = kernel_sendpage(sock, bvec->bv_page, + bvec->bv_offset + offset, + len, flags); + if (ret < 0) + return ret; + *sentp += ret; + if (ret != len) + goto out; + remaining -= len; + offset = 0; + bvec++; + } + } + + if (tail->iov_len) { + ret = svc_tcp_send_kvec(sock, tail, 0); + if (ret < 0) + return ret; + *sentp += ret; + } + +out: + return 0; +} + /** * svc_tcp_sendto - Send out a reply on a TCP socket * @rqstp: completed svc_rqst @@ -1089,7 +1173,7 @@ static int svc_tcp_sendto(struct svc_rqst *rqstp) mutex_lock(&xprt->xpt_mutex); if (svc_xprt_is_dead(xprt)) goto out_notconn; - err = xprt_sock_sendmsg(svsk->sk_sock, &msg, xdr, 0, marker, &sent); + err = svc_tcp_sendmsg(svsk->sk_sock, &msg, xdr, marker, &sent); xdr_free_bvec(xdr); trace_svcsock_tcp_send(xprt, err < 0 ? err : sent); if (err < 0 || sent != (xdr->len + sizeof(marker))) -- cgit v1.2.3 From 53475c5dd856212e91538a9501162e821cc1f791 Mon Sep 17 00:00:00 2001 From: Dongseok Yi Date: Fri, 8 Jan 2021 11:28:38 +0900 Subject: net: fix use-after-free when UDP GRO with shared fraglist skbs in fraglist could be shared by a BPF filter loaded at TC. If TC writes, it will call skb_ensure_writable -> pskb_expand_head to create a private linear section for the head_skb. And then call skb_clone_fraglist -> skb_get on each skb in the fraglist. skb_segment_list overwrites part of the skb linear section of each fragment itself. Even after skb_clone, the frag_skbs share their linear section with their clone in PF_PACKET. Both sk_receive_queue of PF_PACKET and PF_INET (or PF_INET6) can have a link for the same frag_skbs chain. If a new skb (not frags) is queued to one of the sk_receive_queue, multiple ptypes can see and release this. It causes use-after-free. [ 4443.426215] ------------[ cut here ]------------ [ 4443.426222] refcount_t: underflow; use-after-free. [ 4443.426291] WARNING: CPU: 7 PID: 28161 at lib/refcount.c:190 refcount_dec_and_test_checked+0xa4/0xc8 [ 4443.426726] pstate: 60400005 (nZCv daif +PAN -UAO) [ 4443.426732] pc : refcount_dec_and_test_checked+0xa4/0xc8 [ 4443.426737] lr : refcount_dec_and_test_checked+0xa0/0xc8 [ 4443.426808] Call trace: [ 4443.426813] refcount_dec_and_test_checked+0xa4/0xc8 [ 4443.426823] skb_release_data+0x144/0x264 [ 4443.426828] kfree_skb+0x58/0xc4 [ 4443.426832] skb_queue_purge+0x64/0x9c [ 4443.426844] packet_set_ring+0x5f0/0x820 [ 4443.426849] packet_setsockopt+0x5a4/0xcd0 [ 4443.426853] __sys_setsockopt+0x188/0x278 [ 4443.426858] __arm64_sys_setsockopt+0x28/0x38 [ 4443.426869] el0_svc_common+0xf0/0x1d0 [ 4443.426873] el0_svc_handler+0x74/0x98 [ 4443.426880] el0_svc+0x8/0xc Fixes: 3a1296a38d0c (net: Support GRO/GSO fraglist chaining.) Signed-off-by: Dongseok Yi Acked-by: Willem de Bruijn Acked-by: Daniel Borkmann Link: https://lore.kernel.org/r/1610072918-174177-1-git-send-email-dseok.yi@samsung.com Signed-off-by: Jakub Kicinski --- net/core/skbuff.c | 20 +++++++++++++++++++- 1 file changed, 19 insertions(+), 1 deletion(-) (limited to 'net') diff --git a/net/core/skbuff.c b/net/core/skbuff.c index f62cae3f75d8..b6f2b520a9b7 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -3655,7 +3655,8 @@ struct sk_buff *skb_segment_list(struct sk_buff *skb, unsigned int delta_truesize = 0; unsigned int delta_len = 0; struct sk_buff *tail = NULL; - struct sk_buff *nskb; + struct sk_buff *nskb, *tmp; + int err; skb_push(skb, -skb_network_offset(skb) + offset); @@ -3665,11 +3666,28 @@ struct sk_buff *skb_segment_list(struct sk_buff *skb, nskb = list_skb; list_skb = list_skb->next; + err = 0; + if (skb_shared(nskb)) { + tmp = skb_clone(nskb, GFP_ATOMIC); + if (tmp) { + consume_skb(nskb); + nskb = tmp; + err = skb_unclone(nskb, GFP_ATOMIC); + } else { + err = -ENOMEM; + } + } + if (!tail) skb->next = nskb; else tail->next = nskb; + if (unlikely(err)) { + nskb->next = list_skb; + goto err_linearize; + } + tail = nskb; delta_len += nskb->len; -- cgit v1.2.3 From fd2ddef043592e7de80af53f47fa46fd3573086e Mon Sep 17 00:00:00 2001 From: Baptiste Lepers Date: Thu, 7 Jan 2021 16:11:10 +1100 Subject: udp: Prevent reuseport_select_sock from reading uninitialized socks reuse->socks[] is modified concurrently by reuseport_add_sock. To prevent reading values that have not been fully initialized, only read the array up until the last known safe index instead of incorrectly re-reading the last index of the array. Fixes: acdcecc61285f ("udp: correct reuseport selection with connected sockets") Signed-off-by: Baptiste Lepers Acked-by: Willem de Bruijn Link: https://lore.kernel.org/r/20210107051110.12247-1-baptiste.lepers@gmail.com Signed-off-by: Jakub Kicinski --- net/core/sock_reuseport.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net') diff --git a/net/core/sock_reuseport.c b/net/core/sock_reuseport.c index bbdd3c7b6cb5..b065f0a103ed 100644 --- a/net/core/sock_reuseport.c +++ b/net/core/sock_reuseport.c @@ -293,7 +293,7 @@ select_by_hash: i = j = reciprocal_scale(hash, socks); while (reuse->socks[i]->sk_state == TCP_ESTABLISHED) { i++; - if (i >= reuse->num_socks) + if (i >= socks) i = 0; if (i == j) goto out; -- cgit v1.2.3 From 2b446e650b418f9a9e75f99852e2f2560cabfa17 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Wed, 6 Jan 2021 10:40:05 -0800 Subject: docs: net: explain struct net_device lifetime Explain the two basic flows of struct net_device's operation. Signed-off-by: Jakub Kicinski --- Documentation/networking/netdevices.rst | 171 ++++++++++++++++++++++++++++++-- net/core/rtnetlink.c | 2 +- 2 files changed, 166 insertions(+), 7 deletions(-) (limited to 'net') diff --git a/Documentation/networking/netdevices.rst b/Documentation/networking/netdevices.rst index e65665c5ab50..17bdcb746dcf 100644 --- a/Documentation/networking/netdevices.rst +++ b/Documentation/networking/netdevices.rst @@ -10,18 +10,177 @@ Introduction The following is a random collection of documentation regarding network devices. -struct net_device allocation rules -================================== +struct net_device lifetime rules +================================ Network device structures need to persist even after module is unloaded and must be allocated with alloc_netdev_mqs() and friends. If device has registered successfully, it will be freed on last use -by free_netdev(). This is required to handle the pathologic case cleanly -(example: rmmod mydriver needs_free_netdev = true; + } + + static void my_destructor(struct net_device *dev) + { + some_obj_destroy(priv->obj); + some_uninit(priv); + } + + int create_link() + { + struct my_device_priv *priv; + int err; + + ASSERT_RTNL(); + + dev = alloc_netdev(sizeof(*priv), "net%d", NET_NAME_UNKNOWN, my_setup); + if (!dev) + return -ENOMEM; + priv = netdev_priv(dev); + + /* Implicit constructor */ + err = some_init(priv); + if (err) + goto err_free_dev; + + priv->obj = some_obj_create(); + if (!priv->obj) { + err = -ENOMEM; + goto err_some_uninit; + } + /* End of constructor, set the destructor: */ + dev->priv_destructor = my_destructor; + + err = register_netdevice(dev); + if (err) + /* register_netdevice() calls destructor on failure */ + goto err_free_dev; + + /* If anything fails now unregister_netdevice() (or unregister_netdev()) + * will take care of calling my_destructor and free_netdev(). + */ + + return 0; + + err_some_uninit: + some_uninit(priv); + err_free_dev: + free_netdev(dev); + return err; + } + +If struct net_device.priv_destructor is set it will be called by the core +some time after unregister_netdevice(), it will also be called if +register_netdevice() fails. The callback may be invoked with or without +``rtnl_lock`` held. + +There is no explicit constructor callback, driver "constructs" the private +netdev state after allocating it and before registration. + +Setting struct net_device.needs_free_netdev makes core call free_netdevice() +automatically after unregister_netdevice() when all references to the device +are gone. It only takes effect after a successful call to register_netdevice() +so if register_netdevice() fails driver is responsible for calling +free_netdev(). + +free_netdev() is safe to call on error paths right after unregister_netdevice() +or when register_netdevice() fails. Parts of netdev (de)registration process +happen after ``rtnl_lock`` is released, therefore in those cases free_netdev() +will defer some of the processing until ``rtnl_lock`` is released. + +Devices spawned from struct rtnl_link_ops should never free the +struct net_device directly. + +.ndo_init and .ndo_uninit +~~~~~~~~~~~~~~~~~~~~~~~~~ + +``.ndo_init`` and ``.ndo_uninit`` callbacks are called during net_device +registration and de-registration, under ``rtnl_lock``. Drivers can use +those e.g. when parts of their init process need to run under ``rtnl_lock``. + +``.ndo_init`` runs before device is visible in the system, ``.ndo_uninit`` +runs during de-registering after device is closed but other subsystems +may still have outstanding references to the netdevice. MTU === diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c index bb0596c41b3e..79f514afb17d 100644 --- a/net/core/rtnetlink.c +++ b/net/core/rtnetlink.c @@ -3441,7 +3441,7 @@ replay: if (ops->newlink) { err = ops->newlink(link_net ? : net, dev, tb, data, extack); - /* Drivers should call free_netdev() in ->destructor + /* Drivers should set dev->needs_free_netdev * and unregister it on failure after registration * so that device could be finally freed in rtnl_unlock. */ -- cgit v1.2.3 From c269a24ce057abfc31130960e96ab197ef6ab196 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Wed, 6 Jan 2021 10:40:06 -0800 Subject: net: make free_netdev() more lenient with unregistering devices There are two flavors of handling netdev registration: - ones called without holding rtnl_lock: register_netdev() and unregister_netdev(); and - those called with rtnl_lock held: register_netdevice() and unregister_netdevice(). While the semantics of the former are pretty clear, the same can't be said about the latter. The netdev_todo mechanism is utilized to perform some of the device unregistering tasks and it hooks into rtnl_unlock() so the locked variants can't actually finish the work. In general free_netdev() does not mix well with locked calls. Most drivers operating under rtnl_lock set dev->needs_free_netdev to true and expect core to make the free_netdev() call some time later. The part where this becomes most problematic is error paths. There is no way to unwind the state cleanly after a call to register_netdevice(), since unreg can't be performed fully without dropping locks. Make free_netdev() more lenient, and defer the freeing if device is being unregistered. This allows error paths to simply call free_netdev() both after register_netdevice() failed, and after a call to unregister_netdevice() but before dropping rtnl_lock. Simplify the error paths which are currently doing gymnastics around free_netdev() handling. Signed-off-by: Jakub Kicinski --- net/8021q/vlan.c | 4 +--- net/core/dev.c | 11 +++++++++++ net/core/rtnetlink.c | 23 ++++++----------------- 3 files changed, 18 insertions(+), 20 deletions(-) (limited to 'net') diff --git a/net/8021q/vlan.c b/net/8021q/vlan.c index 15bbfaf943fd..8b644113715e 100644 --- a/net/8021q/vlan.c +++ b/net/8021q/vlan.c @@ -284,9 +284,7 @@ static int register_vlan_device(struct net_device *real_dev, u16 vlan_id) return 0; out_free_newdev: - if (new_dev->reg_state == NETREG_UNINITIALIZED || - new_dev->reg_state == NETREG_UNREGISTERED) - free_netdev(new_dev); + free_netdev(new_dev); return err; } diff --git a/net/core/dev.c b/net/core/dev.c index 8fa739259041..adde93cbca9f 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -10631,6 +10631,17 @@ void free_netdev(struct net_device *dev) struct napi_struct *p, *n; might_sleep(); + + /* When called immediately after register_netdevice() failed the unwind + * handling may still be dismantling the device. Handle that case by + * deferring the free. + */ + if (dev->reg_state == NETREG_UNREGISTERING) { + ASSERT_RTNL(); + dev->needs_free_netdev = true; + return; + } + netif_free_tx_queues(dev); netif_free_rx_queues(dev); diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c index 79f514afb17d..3d6ab194d0f5 100644 --- a/net/core/rtnetlink.c +++ b/net/core/rtnetlink.c @@ -3439,26 +3439,15 @@ replay: dev->ifindex = ifm->ifi_index; - if (ops->newlink) { + if (ops->newlink) err = ops->newlink(link_net ? : net, dev, tb, data, extack); - /* Drivers should set dev->needs_free_netdev - * and unregister it on failure after registration - * so that device could be finally freed in rtnl_unlock. - */ - if (err < 0) { - /* If device is not registered at all, free it now */ - if (dev->reg_state == NETREG_UNINITIALIZED || - dev->reg_state == NETREG_UNREGISTERED) - free_netdev(dev); - goto out; - } - } else { + else err = register_netdevice(dev); - if (err < 0) { - free_netdev(dev); - goto out; - } + if (err < 0) { + free_netdev(dev); + goto out; } + err = rtnl_configure_link(dev, ifm); if (err < 0) goto out_unregister; -- cgit v1.2.3 From 766b0515d5bec4b780750773ed3009b148df8c0a Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Wed, 6 Jan 2021 10:40:07 -0800 Subject: net: make sure devices go through netdev_wait_all_refs If register_netdevice() fails at the very last stage - the notifier call - some subsystems may have already seen it and grabbed a reference. struct net_device can't be freed right away without calling netdev_wait_all_refs(). Now that we have a clean interface in form of dev->needs_free_netdev and lenient free_netdev() we can undo what commit 93ee31f14f6f ("[NET]: Fix free_netdev on register_netdev failure.") has done and complete the unregistration path by bringing the net_set_todo() call back. After registration fails user is still expected to explicitly free the net_device, so make sure ->needs_free_netdev is cleared, otherwise rolling back the registration will cause the old double free for callers who release rtnl_lock before the free. This also solves the problem of priv_destructor not being called on notifier error. net_set_todo() will be moved back into unregister_netdevice_queue() in a follow up. Reported-by: Hulk Robot Reported-by: Yang Yingliang Signed-off-by: Jakub Kicinski --- net/core/dev.c | 14 ++++---------- 1 file changed, 4 insertions(+), 10 deletions(-) (limited to 'net') diff --git a/net/core/dev.c b/net/core/dev.c index adde93cbca9f..0071a11a6dc3 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -10077,17 +10077,11 @@ int register_netdevice(struct net_device *dev) ret = call_netdevice_notifiers(NETDEV_REGISTER, dev); ret = notifier_to_errno(ret); if (ret) { + /* Expect explicit free_netdev() on failure */ + dev->needs_free_netdev = false; rollback_registered(dev); - rcu_barrier(); - - dev->reg_state = NETREG_UNREGISTERED; - /* We should put the kobject that hold in - * netdev_unregister_kobject(), otherwise - * the net device cannot be freed when - * driver calls free_netdev(), because the - * kobject is being hold. - */ - kobject_put(&dev->dev.kobj); + net_set_todo(dev); + goto out; } /* * Prevent userspace races by waiting until the network -- cgit v1.2.3 From b210de4f8c97d57de051e805686248ec4c6cfc52 Mon Sep 17 00:00:00 2001 From: Aya Levin Date: Thu, 7 Jan 2021 15:50:18 +0200 Subject: net: ipv6: Validate GSO SKB before finish IPv6 processing There are cases where GSO segment's length exceeds the egress MTU: - Forwarding of a TCP GRO skb, when DF flag is not set. - Forwarding of an skb that arrived on a virtualisation interface (virtio-net/vhost/tap) with TSO/GSO size set by other network stack. - Local GSO skb transmitted on an NETIF_F_TSO tunnel stacked over an interface with a smaller MTU. - Arriving GRO skb (or GSO skb in a virtualised environment) that is bridged to a NETIF_F_TSO tunnel stacked over an interface with an insufficient MTU. If so: - Consume the SKB and its segments. - Issue an ICMP packet with 'Packet Too Big' message containing the MTU, allowing the source host to reduce its Path MTU appropriately. Note: These cases are handled in the same manner in IPv4 output finish. This patch aligns the behavior of IPv6 and the one of IPv4. Fixes: 9e50849054a4 ("netfilter: ipv6: move POSTROUTING invocation before fragmentation") Signed-off-by: Aya Levin Reviewed-by: Tariq Toukan Link: https://lore.kernel.org/r/1610027418-30438-1-git-send-email-ayal@nvidia.com Signed-off-by: Jakub Kicinski --- net/ipv6/ip6_output.c | 41 ++++++++++++++++++++++++++++++++++++++++- 1 file changed, 40 insertions(+), 1 deletion(-) (limited to 'net') diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c index 749ad72386b2..077d43af8226 100644 --- a/net/ipv6/ip6_output.c +++ b/net/ipv6/ip6_output.c @@ -125,8 +125,43 @@ static int ip6_finish_output2(struct net *net, struct sock *sk, struct sk_buff * return -EINVAL; } +static int +ip6_finish_output_gso_slowpath_drop(struct net *net, struct sock *sk, + struct sk_buff *skb, unsigned int mtu) +{ + struct sk_buff *segs, *nskb; + netdev_features_t features; + int ret = 0; + + /* Please see corresponding comment in ip_finish_output_gso + * describing the cases where GSO segment length exceeds the + * egress MTU. + */ + features = netif_skb_features(skb); + segs = skb_gso_segment(skb, features & ~NETIF_F_GSO_MASK); + if (IS_ERR_OR_NULL(segs)) { + kfree_skb(skb); + return -ENOMEM; + } + + consume_skb(skb); + + skb_list_walk_safe(segs, segs, nskb) { + int err; + + skb_mark_not_on_list(segs); + err = ip6_fragment(net, sk, segs, ip6_finish_output2); + if (err && ret == 0) + ret = err; + } + + return ret; +} + static int __ip6_finish_output(struct net *net, struct sock *sk, struct sk_buff *skb) { + unsigned int mtu; + #if defined(CONFIG_NETFILTER) && defined(CONFIG_XFRM) /* Policy lookup after SNAT yielded a new policy */ if (skb_dst(skb)->xfrm) { @@ -135,7 +170,11 @@ static int __ip6_finish_output(struct net *net, struct sock *sk, struct sk_buff } #endif - if ((skb->len > ip6_skb_dst_mtu(skb) && !skb_is_gso(skb)) || + mtu = ip6_skb_dst_mtu(skb); + if (skb_is_gso(skb) && !skb_gso_validate_network_len(skb, mtu)) + return ip6_finish_output_gso_slowpath_drop(net, sk, skb, mtu); + + if ((skb->len > mtu && !skb_is_gso(skb)) || dst_allfrag(skb_dst(skb)) || (IP6CB(skb)->frag_max_size && skb->len > IP6CB(skb)->frag_max_size)) return ip6_fragment(net, sk, skb, ip6_finish_output2); -- cgit v1.2.3 From b77413446408fdd256599daf00d5be72b5f3e7c6 Mon Sep 17 00:00:00 2001 From: Hoang Le Date: Fri, 8 Jan 2021 14:13:37 +0700 Subject: tipc: fix NULL deref in tipc_link_xmit() The buffer list can have zero skb as following path: tipc_named_node_up()->tipc_node_xmit()->tipc_link_xmit(), so we need to check the list before casting an &sk_buff. Fault report: [] tipc: Bulk publication failure [] general protection fault, probably for non-canonical [#1] PREEMPT [...] [] KASAN: null-ptr-deref in range [0x00000000000000c8-0x00000000000000cf] [] CPU: 0 PID: 0 Comm: swapper/0 Kdump: loaded Not tainted 5.10.0-rc4+ #2 [] Hardware name: Bochs ..., BIOS Bochs 01/01/2011 [] RIP: 0010:tipc_link_xmit+0xc1/0x2180 [] Code: 24 b8 00 00 00 00 4d 39 ec 4c 0f 44 e8 e8 d7 0a 10 f9 48 [...] [] RSP: 0018:ffffc90000006ea0 EFLAGS: 00010202 [] RAX: dffffc0000000000 RBX: ffff8880224da000 RCX: 1ffff11003d3cc0d [] RDX: 0000000000000019 RSI: ffffffff886007b9 RDI: 00000000000000c8 [] RBP: ffffc90000007018 R08: 0000000000000001 R09: fffff52000000ded [] R10: 0000000000000003 R11: fffff52000000dec R12: ffffc90000007148 [] R13: 0000000000000000 R14: 0000000000000000 R15: ffffc90000007018 [] FS: 0000000000000000(0000) GS:ffff888037400000(0000) knlGS:000[...] [] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [] CR2: 00007fffd2db5000 CR3: 000000002b08f000 CR4: 00000000000006f0 Fixes: af9b028e270fd ("tipc: make media xmit call outside node spinlock context") Acked-by: Jon Maloy Signed-off-by: Hoang Le Link: https://lore.kernel.org/r/20210108071337.3598-1-hoang.h.le@dektech.com.au Signed-off-by: Jakub Kicinski --- net/tipc/link.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'net') diff --git a/net/tipc/link.c b/net/tipc/link.c index 6ae2140eb4f7..a6a694b78927 100644 --- a/net/tipc/link.c +++ b/net/tipc/link.c @@ -1030,7 +1030,6 @@ void tipc_link_reset(struct tipc_link *l) int tipc_link_xmit(struct tipc_link *l, struct sk_buff_head *list, struct sk_buff_head *xmitq) { - struct tipc_msg *hdr = buf_msg(skb_peek(list)); struct sk_buff_head *backlogq = &l->backlogq; struct sk_buff_head *transmq = &l->transmq; struct sk_buff *skb, *_skb; @@ -1038,13 +1037,18 @@ int tipc_link_xmit(struct tipc_link *l, struct sk_buff_head *list, u16 ack = l->rcv_nxt - 1; u16 seqno = l->snd_nxt; int pkt_cnt = skb_queue_len(list); - int imp = msg_importance(hdr); unsigned int mss = tipc_link_mss(l); unsigned int cwin = l->window; unsigned int mtu = l->mtu; + struct tipc_msg *hdr; bool new_bundle; int rc = 0; + int imp; + + if (pkt_cnt <= 0) + return 0; + hdr = buf_msg(skb_peek(list)); if (unlikely(msg_size(hdr) > mtu)) { pr_warn("Too large msg, purging xmit list %d %d %d %d %d!\n", skb_queue_len(list), msg_user(hdr), @@ -1053,6 +1057,7 @@ int tipc_link_xmit(struct tipc_link *l, struct sk_buff_head *list, return -EMSGSIZE; } + imp = msg_importance(hdr); /* Allow oversubscription of one data msg per source at congestion */ if (unlikely(l->backlog[imp].len >= l->backlog[imp].limit)) { if (imp == TIPC_SYSTEM_IMPORTANCE) { -- cgit v1.2.3 From f6351c3f1c27c80535d76cac2299aec44c36291e Mon Sep 17 00:00:00 2001 From: Jesper Dangaard Brouer Date: Fri, 8 Jan 2021 12:44:33 +0100 Subject: netfilter: conntrack: fix reading nf_conntrack_buckets The old way of changing the conntrack hashsize runtime was through changing the module param via file /sys/module/nf_conntrack/parameters/hashsize. This was extended to sysctl change in commit 3183ab8997a4 ("netfilter: conntrack: allow increasing bucket size via sysctl too"). The commit introduced second "user" variable nf_conntrack_htable_size_user which shadow actual variable nf_conntrack_htable_size. When hashsize is changed via module param this "user" variable isn't updated. This results in sysctl net/netfilter/nf_conntrack_buckets shows the wrong value when users update via the old way. This patch fix the issue by always updating "user" variable when reading the proc file. This will take care of changes to the actual variable without sysctl need to be aware. Fixes: 3183ab8997a4 ("netfilter: conntrack: allow increasing bucket size via sysctl too") Reported-by: Yoel Caspersen Signed-off-by: Jesper Dangaard Brouer Acked-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_conntrack_standalone.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'net') diff --git a/net/netfilter/nf_conntrack_standalone.c b/net/netfilter/nf_conntrack_standalone.c index 46c5557c1fec..0ee702d374b0 100644 --- a/net/netfilter/nf_conntrack_standalone.c +++ b/net/netfilter/nf_conntrack_standalone.c @@ -523,6 +523,9 @@ nf_conntrack_hash_sysctl(struct ctl_table *table, int write, { int ret; + /* module_param hashsize could have changed value */ + nf_conntrack_htable_size_user = nf_conntrack_htable_size; + ret = proc_dointvec(table, write, buffer, lenp, ppos); if (ret < 0 || !write) return ret; -- cgit v1.2.3 From 86b53fbf08f48d353a86a06aef537e78e82ba721 Mon Sep 17 00:00:00 2001 From: "j.nixdorf@avm.de" Date: Tue, 5 Jan 2021 15:17:01 +0100 Subject: net: sunrpc: interpret the return value of kstrtou32 correctly A return value of 0 means success. This is documented in lib/kstrtox.c. This was found by trying to mount an NFS share from a link-local IPv6 address with the interface specified by its index: mount("[fe80::1%1]:/srv/nfs", "/mnt", "nfs", 0, "nolock,addr=fe80::1%1") Before this commit this failed with EINVAL and also caused the following message in dmesg: [...] NFS: bad IP address specified: addr=fe80::1%1 The syscall using the same address based on the interface name instead of its index succeeds. Credits for this patch go to my colleague Christian Speich, who traced the origin of this bug to this line of code. Signed-off-by: Johannes Nixdorf Fixes: 00cfaa943ec3 ("replace strict_strto calls") Signed-off-by: Trond Myklebust --- net/sunrpc/addr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net') diff --git a/net/sunrpc/addr.c b/net/sunrpc/addr.c index 010dcb876f9d..6e4dbd577a39 100644 --- a/net/sunrpc/addr.c +++ b/net/sunrpc/addr.c @@ -185,7 +185,7 @@ static int rpc_parse_scope_id(struct net *net, const char *buf, scope_id = dev->ifindex; dev_put(dev); } else { - if (kstrtou32(p, 10, &scope_id) == 0) { + if (kstrtou32(p, 10, &scope_id) != 0) { kfree(p); return 0; } -- cgit v1.2.3 From 869f4fdaf4ca7bb6e0d05caf6fa1108dddc346a7 Mon Sep 17 00:00:00 2001 From: Dinghao Liu Date: Sat, 9 Jan 2021 20:01:21 +0800 Subject: netfilter: nf_nat: Fix memleak in nf_nat_init When register_pernet_subsys() fails, nf_nat_bysource should be freed just like when nf_ct_extend_register() fails. Fixes: 1cd472bf036ca ("netfilter: nf_nat: add nat hook register functions to nf_nat") Signed-off-by: Dinghao Liu Acked-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_nat_core.c | 1 + 1 file changed, 1 insertion(+) (limited to 'net') diff --git a/net/netfilter/nf_nat_core.c b/net/netfilter/nf_nat_core.c index ea923f8cf9c4..b7c3c902290f 100644 --- a/net/netfilter/nf_nat_core.c +++ b/net/netfilter/nf_nat_core.c @@ -1174,6 +1174,7 @@ static int __init nf_nat_init(void) ret = register_pernet_subsys(&nat_net_ops); if (ret < 0) { nf_ct_extend_unregister(&nat_extend); + kvfree(nf_nat_bysource); return ret; } -- cgit v1.2.3 From 97550f6fa59254435d864b92603de3ca4b5a99f8 Mon Sep 17 00:00:00 2001 From: Willem de Bruijn Date: Sat, 9 Jan 2021 17:18:33 -0500 Subject: net: compound page support in skb_seq_read skb_seq_read iterates over an skb, returning pointer and length of the next data range with each call. It relies on kmap_atomic to access highmem pages when needed. An skb frag may be backed by a compound page, but kmap_atomic maps only a single page. There are not enough kmap slots to always map all pages concurrently. Instead, if kmap_atomic is needed, iterate over each page. As this increases the number of calls, avoid this unless needed. The necessary condition is captured in skb_frag_must_loop. I tried to make the change as obvious as possible. It should be easy to verify that nothing changes if skb_frag_must_loop returns false. Tested: On an x86 platform with CONFIG_HIGHMEM=y CONFIG_DEBUG_KMAP_LOCAL_FORCE_MAP=y CONFIG_NETFILTER_XT_MATCH_STRING=y Run ip link set dev lo mtu 1500 iptables -A OUTPUT -m string --string 'badstring' -algo bm -j ACCEPT dd if=/dev/urandom of=in bs=1M count=20 nc -l -p 8000 > /dev/null & nc -w 1 -q 0 localhost 8000 < in Signed-off-by: Willem de Bruijn Signed-off-by: Jakub Kicinski --- include/linux/skbuff.h | 1 + net/core/skbuff.c | 28 +++++++++++++++++++++++----- 2 files changed, 24 insertions(+), 5 deletions(-) (limited to 'net') diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h index c858adfb5a82..5f60c9e907c9 100644 --- a/include/linux/skbuff.h +++ b/include/linux/skbuff.h @@ -1203,6 +1203,7 @@ struct skb_seq_state { struct sk_buff *root_skb; struct sk_buff *cur_skb; __u8 *frag_data; + __u32 frag_off; }; void skb_prepare_seq_read(struct sk_buff *skb, unsigned int from, diff --git a/net/core/skbuff.c b/net/core/skbuff.c index b6f2b520a9b7..0da035c1e53f 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -3442,6 +3442,7 @@ void skb_prepare_seq_read(struct sk_buff *skb, unsigned int from, st->root_skb = st->cur_skb = skb; st->frag_idx = st->stepped_offset = 0; st->frag_data = NULL; + st->frag_off = 0; } EXPORT_SYMBOL(skb_prepare_seq_read); @@ -3496,14 +3497,27 @@ next_skb: st->stepped_offset += skb_headlen(st->cur_skb); while (st->frag_idx < skb_shinfo(st->cur_skb)->nr_frags) { + unsigned int pg_idx, pg_off, pg_sz; + frag = &skb_shinfo(st->cur_skb)->frags[st->frag_idx]; - block_limit = skb_frag_size(frag) + st->stepped_offset; + pg_idx = 0; + pg_off = skb_frag_off(frag); + pg_sz = skb_frag_size(frag); + + if (skb_frag_must_loop(skb_frag_page(frag))) { + pg_idx = (pg_off + st->frag_off) >> PAGE_SHIFT; + pg_off = offset_in_page(pg_off + st->frag_off); + pg_sz = min_t(unsigned int, pg_sz - st->frag_off, + PAGE_SIZE - pg_off); + } + + block_limit = pg_sz + st->stepped_offset; if (abs_offset < block_limit) { if (!st->frag_data) - st->frag_data = kmap_atomic(skb_frag_page(frag)); + st->frag_data = kmap_atomic(skb_frag_page(frag) + pg_idx); - *data = (u8 *) st->frag_data + skb_frag_off(frag) + + *data = (u8 *)st->frag_data + pg_off + (abs_offset - st->stepped_offset); return block_limit - abs_offset; @@ -3514,8 +3528,12 @@ next_skb: st->frag_data = NULL; } - st->frag_idx++; - st->stepped_offset += skb_frag_size(frag); + st->stepped_offset += pg_sz; + st->frag_off += pg_sz; + if (st->frag_off == skb_frag_size(frag)) { + st->frag_off = 0; + st->frag_idx++; + } } if (st->frag_data) { -- cgit v1.2.3 From 9bd6b629c39e3fa9e14243a6d8820492be1a5b2e Mon Sep 17 00:00:00 2001 From: Willem de Bruijn Date: Sat, 9 Jan 2021 17:18:34 -0500 Subject: esp: avoid unneeded kmap_atomic call esp(6)_output_head uses skb_page_frag_refill to allocate a buffer for the esp trailer. It accesses the page with kmap_atomic to handle highmem. But skb_page_frag_refill can return compound pages, of which kmap_atomic only maps the first underlying page. skb_page_frag_refill does not return highmem, because flag __GFP_HIGHMEM is not set. ESP uses it in the same manner as TCP. That also does not call kmap_atomic, but directly uses page_address, in skb_copy_to_page_nocache. Do the same for ESP. This issue has become easier to trigger with recent kmap local debugging feature CONFIG_DEBUG_KMAP_LOCAL_FORCE_MAP. Fixes: cac2661c53f3 ("esp4: Avoid skb_cow_data whenever possible") Fixes: 03e2a30f6a27 ("esp6: Avoid skb_cow_data whenever possible") Signed-off-by: Willem de Bruijn Acked-by: Steffen Klassert Signed-off-by: Jakub Kicinski --- net/ipv4/esp4.c | 7 +------ net/ipv6/esp6.c | 7 +------ 2 files changed, 2 insertions(+), 12 deletions(-) (limited to 'net') diff --git a/net/ipv4/esp4.c b/net/ipv4/esp4.c index 8b07f3a4f2db..a3271ec3e162 100644 --- a/net/ipv4/esp4.c +++ b/net/ipv4/esp4.c @@ -443,7 +443,6 @@ static int esp_output_encap(struct xfrm_state *x, struct sk_buff *skb, int esp_output_head(struct xfrm_state *x, struct sk_buff *skb, struct esp_info *esp) { u8 *tail; - u8 *vaddr; int nfrags; int esph_offset; struct page *page; @@ -485,14 +484,10 @@ int esp_output_head(struct xfrm_state *x, struct sk_buff *skb, struct esp_info * page = pfrag->page; get_page(page); - vaddr = kmap_atomic(page); - - tail = vaddr + pfrag->offset; + tail = page_address(page) + pfrag->offset; esp_output_fill_trailer(tail, esp->tfclen, esp->plen, esp->proto); - kunmap_atomic(vaddr); - nfrags = skb_shinfo(skb)->nr_frags; __skb_fill_page_desc(skb, nfrags, page, pfrag->offset, diff --git a/net/ipv6/esp6.c b/net/ipv6/esp6.c index 52c2f063529f..2b804fcebcc6 100644 --- a/net/ipv6/esp6.c +++ b/net/ipv6/esp6.c @@ -478,7 +478,6 @@ static int esp6_output_encap(struct xfrm_state *x, struct sk_buff *skb, int esp6_output_head(struct xfrm_state *x, struct sk_buff *skb, struct esp_info *esp) { u8 *tail; - u8 *vaddr; int nfrags; int esph_offset; struct page *page; @@ -519,14 +518,10 @@ int esp6_output_head(struct xfrm_state *x, struct sk_buff *skb, struct esp_info page = pfrag->page; get_page(page); - vaddr = kmap_atomic(page); - - tail = vaddr + pfrag->offset; + tail = page_address(page) + pfrag->offset; esp_output_fill_trailer(tail, esp->tfclen, esp->plen, esp->proto); - kunmap_atomic(vaddr); - nfrags = skb_shinfo(skb)->nr_frags; __skb_fill_page_desc(skb, nfrags, page, pfrag->offset, -- cgit v1.2.3 From df85bc140a4d6cbaa78d8e9c35154e1a2f0622c7 Mon Sep 17 00:00:00 2001 From: Petr Machata Date: Mon, 11 Jan 2021 18:07:07 +0100 Subject: net: dcb: Accept RTM_GETDCB messages carrying set-like DCB commands In commit 826f328e2b7e ("net: dcb: Validate netlink message in DCB handler"), Linux started rejecting RTM_GETDCB netlink messages if they contained a set-like DCB_CMD_ command. The reason was that privileges were only verified for RTM_SETDCB messages, but the value that determined the action to be taken is the command, not the message type. And validation of message type against the DCB command was the obvious missing piece. Unfortunately it turns out that mlnx_qos, a somewhat widely deployed tool for configuration of DCB, accesses the DCB set-like APIs through RTM_GETDCB. Therefore do not bounce the discrepancy between message type and command. Instead, in addition to validating privileges based on the actual message type, validate them also based on the expected message type. This closes the loophole of allowing DCB configuration on non-admin accounts, while maintaining backward compatibility. Fixes: 2f90b8657ec9 ("ixgbe: this patch adds support for DCB to the kernel and ixgbe driver") Fixes: 826f328e2b7e ("net: dcb: Validate netlink message in DCB handler") Signed-off-by: Petr Machata Link: https://lore.kernel.org/r/a3edcfda0825f2aa2591801c5232f2bbf2d8a554.1610384801.git.me@pmachata.org Signed-off-by: Jakub Kicinski --- net/dcb/dcbnl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net') diff --git a/net/dcb/dcbnl.c b/net/dcb/dcbnl.c index 7d49b6fd6cef..653e3bc9c87b 100644 --- a/net/dcb/dcbnl.c +++ b/net/dcb/dcbnl.c @@ -1765,7 +1765,7 @@ static int dcb_doit(struct sk_buff *skb, struct nlmsghdr *nlh, fn = &reply_funcs[dcb->cmd]; if (!fn->cb) return -EOPNOTSUPP; - if (fn->type != nlh->nlmsg_type) + if (fn->type == RTM_SETDCB && !netlink_capable(skb, CAP_NET_ADMIN)) return -EPERM; if (!tb[DCB_ATTR_IFNAME]) -- cgit v1.2.3 From 07b90056cb15ff9877dca0d8f1b6583d1051f724 Mon Sep 17 00:00:00 2001 From: Vladimir Oltean Date: Tue, 12 Jan 2021 01:09:43 +0200 Subject: net: dsa: unbind all switches from tree when DSA master unbinds Currently the following happens when a DSA master driver unbinds while there are DSA switches attached to it: $ echo 0000:00:00.5 > /sys/bus/pci/drivers/mscc_felix/unbind ------------[ cut here ]------------ WARNING: CPU: 0 PID: 392 at net/core/dev.c:9507 Call trace: rollback_registered_many+0x5fc/0x688 unregister_netdevice_queue+0x98/0x120 dsa_slave_destroy+0x4c/0x88 dsa_port_teardown.part.16+0x78/0xb0 dsa_tree_teardown_switches+0x58/0xc0 dsa_unregister_switch+0x104/0x1b8 felix_pci_remove+0x24/0x48 pci_device_remove+0x48/0xf0 device_release_driver_internal+0x118/0x1e8 device_driver_detach+0x28/0x38 unbind_store+0xd0/0x100 Located at the above location is this WARN_ON: /* Notifier chain MUST detach us all upper devices. */ WARN_ON(netdev_has_any_upper_dev(dev)); Other stacked interfaces, like VLAN, do indeed listen for NETDEV_UNREGISTER on the real_dev and also unregister themselves at that time, which is clearly the behavior that rollback_registered_many expects. But DSA interfaces are not VLAN. They have backing hardware (platform devices, PCI devices, MDIO, SPI etc) which have a life cycle of their own and we can't just trigger an unregister from the DSA framework when we receive a netdev notifier that the master unregisters. Luckily, there is something we can do, and that is to inform the driver core that we have a runtime dependency to the DSA master interface's device, and create a device link where that is the supplier and we are the consumer. Having this device link will make the DSA switch unbind before the DSA master unbinds, which is enough to avoid the WARN_ON from rollback_registered_many. Note that even before the blamed commit, DSA did nothing intelligent when the master interface got unregistered either. See the discussion here: https://lore.kernel.org/netdev/20200505210253.20311-1-f.fainelli@gmail.com/ But this time, at least the WARN_ON is loud enough that the upper_dev_link commit can be blamed. The advantage with this approach vs dev_hold(master) in the attached link is that the latter is not meant for long term reference counting. With dev_hold, the only thing that will happen is that when the user attempts an unbind of the DSA master, netdev_wait_allrefs will keep waiting and waiting, due to DSA keeping the refcount forever. DSA would not access freed memory corresponding to the master interface, but the unbind would still result in a freeze. Whereas with device links, graceful teardown is ensured. It even works with cascaded DSA trees. $ echo 0000:00:00.2 > /sys/bus/pci/drivers/fsl_enetc/unbind [ 1818.797546] device swp0 left promiscuous mode [ 1819.301112] sja1105 spi2.0: Link is Down [ 1819.307981] DSA: tree 1 torn down [ 1819.312408] device eno2 left promiscuous mode [ 1819.656803] mscc_felix 0000:00:00.5: Link is Down [ 1819.667194] DSA: tree 0 torn down [ 1819.711557] fsl_enetc 0000:00:00.2 eno2: Link is Down This approach allows us to keep the DSA framework absolutely unchanged, and the driver core will just know to unbind us first when the master goes away - as opposed to the large (and probably impossible) rework required if attempting to listen for NETDEV_UNREGISTER. As per the documentation at Documentation/driver-api/device_link.rst, specifying the DL_FLAG_AUTOREMOVE_CONSUMER flag causes the device link to be automatically purged when the consumer fails to probe or later unbinds. So we don't need to keep the consumer_link variable in struct dsa_switch. Fixes: 2f1e8ea726e9 ("net: dsa: link interfaces with the DSA master to get rid of lockdep warnings") Signed-off-by: Vladimir Oltean Reviewed-by: Florian Fainelli Tested-by: Florian Fainelli Link: https://lore.kernel.org/r/20210111230943.3701806-1-olteanv@gmail.com Signed-off-by: Jakub Kicinski --- net/dsa/master.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'net') diff --git a/net/dsa/master.c b/net/dsa/master.c index 5a0f6fec4271..cb3a5cf99b25 100644 --- a/net/dsa/master.c +++ b/net/dsa/master.c @@ -309,8 +309,18 @@ static struct lock_class_key dsa_master_addr_list_lock_key; int dsa_master_setup(struct net_device *dev, struct dsa_port *cpu_dp) { int mtu = ETH_DATA_LEN + cpu_dp->tag_ops->overhead; + struct dsa_switch *ds = cpu_dp->ds; + struct device_link *consumer_link; int ret; + /* The DSA master must use SET_NETDEV_DEV for this to work. */ + consumer_link = device_link_add(ds->dev, dev->dev.parent, + DL_FLAG_AUTOREMOVE_CONSUMER); + if (!consumer_link) + netdev_err(dev, + "Failed to create a device link to DSA switch %s\n", + dev_name(ds->dev)); + rtnl_lock(); ret = dev_set_mtu(dev, mtu); rtnl_unlock(); -- cgit v1.2.3 From 91158e1680b164c8d101144ca916a3dca10c3e17 Mon Sep 17 00:00:00 2001 From: Vladimir Oltean Date: Tue, 12 Jan 2021 02:48:31 +0200 Subject: net: dsa: clear devlink port type before unregistering slave netdevs Florian reported a use-after-free bug in devlink_nl_port_fill found with KASAN: (devlink_nl_port_fill) (devlink_port_notify) (devlink_port_unregister) (dsa_switch_teardown.part.3) (dsa_tree_teardown_switches) (dsa_unregister_switch) (bcm_sf2_sw_remove) (platform_remove) (device_release_driver_internal) (device_links_unbind_consumers) (device_release_driver_internal) (device_driver_detach) (unbind_store) Allocated by task 31: alloc_netdev_mqs+0x5c/0x50c dsa_slave_create+0x110/0x9c8 dsa_register_switch+0xdb0/0x13a4 b53_switch_register+0x47c/0x6dc bcm_sf2_sw_probe+0xaa4/0xc98 platform_probe+0x90/0xf4 really_probe+0x184/0x728 driver_probe_device+0xa4/0x278 __device_attach_driver+0xe8/0x148 bus_for_each_drv+0x108/0x158 Freed by task 249: free_netdev+0x170/0x194 dsa_slave_destroy+0xac/0xb0 dsa_port_teardown.part.2+0xa0/0xb4 dsa_tree_teardown_switches+0x50/0xc4 dsa_unregister_switch+0x124/0x250 bcm_sf2_sw_remove+0x98/0x13c platform_remove+0x44/0x5c device_release_driver_internal+0x150/0x254 device_links_unbind_consumers+0xf8/0x12c device_release_driver_internal+0x84/0x254 device_driver_detach+0x30/0x34 unbind_store+0x90/0x134 What happens is that devlink_port_unregister emits a netlink DEVLINK_CMD_PORT_DEL message which associates the devlink port that is getting unregistered with the ifindex of its corresponding net_device. Only trouble is, the net_device has already been unregistered. It looks like we can stub out the search for a corresponding net_device if we clear the devlink_port's type. This looks like a bit of a hack, but also seems to be the reason why the devlink_port_type_clear function exists in the first place. Fixes: 3122433eb533 ("net: dsa: Register devlink ports before calling DSA driver setup()") Signed-off-by: Vladimir Oltean Reviewed-by: Florian Fainelli Tested-by: Florian fainelli Reported-by: Florian Fainelli Link: https://lore.kernel.org/r/20210112004831.3778323-1-olteanv@gmail.com Signed-off-by: Jakub Kicinski --- net/dsa/dsa2.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'net') diff --git a/net/dsa/dsa2.c b/net/dsa/dsa2.c index 183003e45762..a47e0f9b20d0 100644 --- a/net/dsa/dsa2.c +++ b/net/dsa/dsa2.c @@ -353,9 +353,13 @@ static int dsa_port_devlink_setup(struct dsa_port *dp) static void dsa_port_teardown(struct dsa_port *dp) { + struct devlink_port *dlp = &dp->devlink_port; + if (!dp->setup) return; + devlink_port_type_clear(dlp); + switch (dp->type) { case DSA_PORT_TYPE_UNUSED: break; -- cgit v1.2.3 From 20bc80b6f582ad1151c52ca09ab66b472768c9c8 Mon Sep 17 00:00:00 2001 From: Paolo Abeni Date: Tue, 12 Jan 2021 18:25:23 +0100 Subject: mptcp: more strict state checking for acks Syzkaller found a way to trigger division by zero in mptcp_subflow_cleanup_rbuf(). The current checks implemented into tcp_can_send_ack() are too week, let's be more accurate. Reported-by: Christoph Paasch Fixes: ea4ca586b16f ("mptcp: refine MPTCP-level ack scheduling") Fixes: fd8976790a6c ("mptcp: be careful on MPTCP-level ack.") Signed-off-by: Paolo Abeni Reviewed-by: Mat Martineau Signed-off-by: Jakub Kicinski --- net/mptcp/protocol.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net') diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index 6628d8d74203..2ff8c7caf74f 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -427,7 +427,7 @@ static bool mptcp_subflow_active(struct mptcp_subflow_context *subflow) static bool tcp_can_send_ack(const struct sock *ssk) { return !((1 << inet_sk_state_load(ssk)) & - (TCPF_SYN_SENT | TCPF_SYN_RECV | TCPF_TIME_WAIT | TCPF_CLOSE)); + (TCPF_SYN_SENT | TCPF_SYN_RECV | TCPF_TIME_WAIT | TCPF_CLOSE | TCPF_LISTEN)); } static void mptcp_send_ack(struct mptcp_sock *msk) -- cgit v1.2.3 From 76e2a55d16259b51116767b28b19d759bff43f72 Mon Sep 17 00:00:00 2001 From: Paolo Abeni Date: Tue, 12 Jan 2021 18:25:24 +0100 Subject: mptcp: better msk-level shutdown. Instead of re-implementing most of inet_shutdown, re-use such helper, and implement the MPTCP-specific bits at the 'proto' level. The msk-level disconnect() can now be invoked, lets provide a suitable implementation. As a side effect, this fixes bad state management for listener sockets. The latter could lead to division by 0 oops since commit ea4ca586b16f ("mptcp: refine MPTCP-level ack scheduling"). Fixes: 43b54c6ee382 ("mptcp: Use full MPTCP-level disconnect state machine") Fixes: ea4ca586b16f ("mptcp: refine MPTCP-level ack scheduling") Signed-off-by: Paolo Abeni Reviewed-by: Mat Martineau Signed-off-by: Jakub Kicinski --- net/mptcp/protocol.c | 62 ++++++++++++++-------------------------------------- 1 file changed, 17 insertions(+), 45 deletions(-) (limited to 'net') diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index 2ff8c7caf74f..81faeff8f3bb 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -2642,11 +2642,12 @@ static void mptcp_copy_inaddrs(struct sock *msk, const struct sock *ssk) static int mptcp_disconnect(struct sock *sk, int flags) { - /* Should never be called. - * inet_stream_connect() calls ->disconnect, but that - * refers to the subflow socket, not the mptcp one. - */ - WARN_ON_ONCE(1); + struct mptcp_subflow_context *subflow; + struct mptcp_sock *msk = mptcp_sk(sk); + + __mptcp_flush_join_list(msk); + mptcp_for_each_subflow(msk, subflow) + tcp_disconnect(mptcp_subflow_tcp_sock(subflow), flags); return 0; } @@ -3089,6 +3090,14 @@ bool mptcp_finish_join(struct sock *ssk) return true; } +static void mptcp_shutdown(struct sock *sk, int how) +{ + pr_debug("sk=%p, how=%d", sk, how); + + if ((how & SEND_SHUTDOWN) && mptcp_close_state(sk)) + __mptcp_wr_shutdown(sk); +} + static struct proto mptcp_prot = { .name = "MPTCP", .owner = THIS_MODULE, @@ -3098,7 +3107,7 @@ static struct proto mptcp_prot = { .accept = mptcp_accept, .setsockopt = mptcp_setsockopt, .getsockopt = mptcp_getsockopt, - .shutdown = tcp_shutdown, + .shutdown = mptcp_shutdown, .destroy = mptcp_destroy, .sendmsg = mptcp_sendmsg, .recvmsg = mptcp_recvmsg, @@ -3344,43 +3353,6 @@ static __poll_t mptcp_poll(struct file *file, struct socket *sock, return mask; } -static int mptcp_shutdown(struct socket *sock, int how) -{ - struct mptcp_sock *msk = mptcp_sk(sock->sk); - struct sock *sk = sock->sk; - int ret = 0; - - pr_debug("sk=%p, how=%d", msk, how); - - lock_sock(sk); - - how++; - if ((how & ~SHUTDOWN_MASK) || !how) { - ret = -EINVAL; - goto out_unlock; - } - - if (sock->state == SS_CONNECTING) { - if ((1 << sk->sk_state) & - (TCPF_SYN_SENT | TCPF_SYN_RECV | TCPF_CLOSE)) - sock->state = SS_DISCONNECTING; - else - sock->state = SS_CONNECTED; - } - - sk->sk_shutdown |= how; - if ((how & SEND_SHUTDOWN) && mptcp_close_state(sk)) - __mptcp_wr_shutdown(sk); - - /* Wake up anyone sleeping in poll. */ - sk->sk_state_change(sk); - -out_unlock: - release_sock(sk); - - return ret; -} - static const struct proto_ops mptcp_stream_ops = { .family = PF_INET, .owner = THIS_MODULE, @@ -3394,7 +3366,7 @@ static const struct proto_ops mptcp_stream_ops = { .ioctl = inet_ioctl, .gettstamp = sock_gettstamp, .listen = mptcp_listen, - .shutdown = mptcp_shutdown, + .shutdown = inet_shutdown, .setsockopt = sock_common_setsockopt, .getsockopt = sock_common_getsockopt, .sendmsg = inet_sendmsg, @@ -3444,7 +3416,7 @@ static const struct proto_ops mptcp_v6_stream_ops = { .ioctl = inet6_ioctl, .gettstamp = sock_gettstamp, .listen = mptcp_listen, - .shutdown = mptcp_shutdown, + .shutdown = inet_shutdown, .setsockopt = sock_common_setsockopt, .getsockopt = sock_common_getsockopt, .sendmsg = inet6_sendmsg, -- cgit v1.2.3 From 25fe2c9c4cd2e97c5f5b69f3aefe69aad3057936 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Tue, 12 Jan 2021 17:21:21 +0100 Subject: smc: fix out of bound access in smc_nl_get_sys_info() smc_clc_get_hostname() sets the host pointer to a buffer which is not NULL-terminated (see smc_clc_init()). Reported-by: syzbot+f4708c391121cfc58396@syzkaller.appspotmail.com Fixes: 099b990bd11a ("net/smc: Add support for obtaining system information") Signed-off-by: Karsten Graul Signed-off-by: Jakub Kicinski --- net/smc/smc_core.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'net') diff --git a/net/smc/smc_core.c b/net/smc/smc_core.c index 59342b519e34..8d866b4ed8f6 100644 --- a/net/smc/smc_core.c +++ b/net/smc/smc_core.c @@ -246,7 +246,8 @@ int smc_nl_get_sys_info(struct sk_buff *skb, struct netlink_callback *cb) goto errattr; smc_clc_get_hostname(&host); if (host) { - snprintf(hostname, sizeof(hostname), "%s", host); + memcpy(hostname, host, SMC_MAX_HOSTNAME_LEN); + hostname[SMC_MAX_HOSTNAME_LEN] = 0; if (nla_put_string(skb, SMC_NLA_SYS_LOCAL_HOST, hostname)) goto errattr; } -- cgit v1.2.3 From 8a4465368964b4fbaf084760c94c7aabf61059fb Mon Sep 17 00:00:00 2001 From: Guvenc Gulce Date: Tue, 12 Jan 2021 17:21:22 +0100 Subject: net/smc: use memcpy instead of snprintf to avoid out of bounds read Using snprintf() to convert not null-terminated strings to null terminated strings may cause out of bounds read in the source string. Therefore use memcpy() and terminate the target string with a null afterwards. Fixes: a3db10efcc4c ("net/smc: Add support for obtaining SMCR device list") Signed-off-by: Guvenc Gulce Signed-off-by: Karsten Graul Signed-off-by: Jakub Kicinski --- net/smc/smc_core.c | 17 +++++++++++------ net/smc/smc_ib.c | 6 +++--- net/smc/smc_ism.c | 3 ++- 3 files changed, 16 insertions(+), 10 deletions(-) (limited to 'net') diff --git a/net/smc/smc_core.c b/net/smc/smc_core.c index 8d866b4ed8f6..0df85a12651e 100644 --- a/net/smc/smc_core.c +++ b/net/smc/smc_core.c @@ -258,7 +258,8 @@ int smc_nl_get_sys_info(struct sk_buff *skb, struct netlink_callback *cb) smc_ism_get_system_eid(smcd_dev, &seid); mutex_unlock(&smcd_dev_list.mutex); if (seid && smc_ism_is_v2_capable()) { - snprintf(smc_seid, sizeof(smc_seid), "%s", seid); + memcpy(smc_seid, seid, SMC_MAX_EID_LEN); + smc_seid[SMC_MAX_EID_LEN] = 0; if (nla_put_string(skb, SMC_NLA_SYS_SEID, smc_seid)) goto errattr; } @@ -296,7 +297,8 @@ static int smc_nl_fill_lgr(struct smc_link_group *lgr, goto errattr; if (nla_put_u8(skb, SMC_NLA_LGR_R_VLAN_ID, lgr->vlan_id)) goto errattr; - snprintf(smc_target, sizeof(smc_target), "%s", lgr->pnet_id); + memcpy(smc_target, lgr->pnet_id, SMC_MAX_PNETID_LEN); + smc_target[SMC_MAX_PNETID_LEN] = 0; if (nla_put_string(skb, SMC_NLA_LGR_R_PNETID, smc_target)) goto errattr; @@ -313,7 +315,7 @@ static int smc_nl_fill_lgr_link(struct smc_link_group *lgr, struct sk_buff *skb, struct netlink_callback *cb) { - char smc_ibname[IB_DEVICE_NAME_MAX + 1]; + char smc_ibname[IB_DEVICE_NAME_MAX]; u8 smc_gid_target[41]; struct nlattr *attrs; u32 link_uid = 0; @@ -462,7 +464,8 @@ static int smc_nl_fill_smcd_lgr(struct smc_link_group *lgr, goto errattr; if (nla_put_u32(skb, SMC_NLA_LGR_D_CHID, smc_ism_get_chid(lgr->smcd))) goto errattr; - snprintf(smc_pnet, sizeof(smc_pnet), "%s", lgr->smcd->pnetid); + memcpy(smc_pnet, lgr->smcd->pnetid, SMC_MAX_PNETID_LEN); + smc_pnet[SMC_MAX_PNETID_LEN] = 0; if (nla_put_string(skb, SMC_NLA_LGR_D_PNETID, smc_pnet)) goto errattr; @@ -475,10 +478,12 @@ static int smc_nl_fill_smcd_lgr(struct smc_link_group *lgr, goto errv2attr; if (nla_put_u8(skb, SMC_NLA_LGR_V2_OS, lgr->peer_os)) goto errv2attr; - snprintf(smc_host, sizeof(smc_host), "%s", lgr->peer_hostname); + memcpy(smc_host, lgr->peer_hostname, SMC_MAX_HOSTNAME_LEN); + smc_host[SMC_MAX_HOSTNAME_LEN] = 0; if (nla_put_string(skb, SMC_NLA_LGR_V2_PEER_HOST, smc_host)) goto errv2attr; - snprintf(smc_eid, sizeof(smc_eid), "%s", lgr->negotiated_eid); + memcpy(smc_eid, lgr->negotiated_eid, SMC_MAX_EID_LEN); + smc_eid[SMC_MAX_EID_LEN] = 0; if (nla_put_string(skb, SMC_NLA_LGR_V2_NEG_EID, smc_eid)) goto errv2attr; diff --git a/net/smc/smc_ib.c b/net/smc/smc_ib.c index ddd7fac98b1d..7d7ba0320d5a 100644 --- a/net/smc/smc_ib.c +++ b/net/smc/smc_ib.c @@ -371,8 +371,8 @@ static int smc_nl_handle_dev_port(struct sk_buff *skb, if (nla_put_u8(skb, SMC_NLA_DEV_PORT_PNET_USR, smcibdev->pnetid_by_user[port])) goto errattr; - snprintf(smc_pnet, sizeof(smc_pnet), "%s", - (char *)&smcibdev->pnetid[port]); + memcpy(smc_pnet, &smcibdev->pnetid[port], SMC_MAX_PNETID_LEN); + smc_pnet[SMC_MAX_PNETID_LEN] = 0; if (nla_put_string(skb, SMC_NLA_DEV_PORT_PNETID, smc_pnet)) goto errattr; if (nla_put_u32(skb, SMC_NLA_DEV_PORT_NETDEV, @@ -414,7 +414,7 @@ static int smc_nl_handle_smcr_dev(struct smc_ib_device *smcibdev, struct sk_buff *skb, struct netlink_callback *cb) { - char smc_ibname[IB_DEVICE_NAME_MAX + 1]; + char smc_ibname[IB_DEVICE_NAME_MAX]; struct smc_pci_dev smc_pci_dev; struct pci_dev *pci_dev; unsigned char is_crit; diff --git a/net/smc/smc_ism.c b/net/smc/smc_ism.c index 524ef64a191a..9c6e95882553 100644 --- a/net/smc/smc_ism.c +++ b/net/smc/smc_ism.c @@ -250,7 +250,8 @@ static int smc_nl_handle_smcd_dev(struct smcd_dev *smcd, goto errattr; if (nla_put_u8(skb, SMC_NLA_DEV_PORT_PNET_USR, smcd->pnetid_by_user)) goto errportattr; - snprintf(smc_pnet, sizeof(smc_pnet), "%s", smcd->pnetid); + memcpy(smc_pnet, smcd->pnetid, SMC_MAX_PNETID_LEN); + smc_pnet[SMC_MAX_PNETID_LEN] = 0; if (nla_put_string(skb, SMC_NLA_DEV_PORT_PNETID, smc_pnet)) goto errportattr; -- cgit v1.2.3 From d52e419ac8b50c8bef41b398ed13528e75d7ad48 Mon Sep 17 00:00:00 2001 From: David Howells Date: Tue, 12 Jan 2021 15:23:51 +0000 Subject: rxrpc: Fix handling of an unsupported token type in rxrpc_read() Clang static analysis reports the following: net/rxrpc/key.c:657:11: warning: Assigned value is garbage or undefined toksize = toksizes[tok++]; ^ ~~~~~~~~~~~~~~~ rxrpc_read() contains two consecutive loops. The first loop calculates the token sizes and stores the results in toksizes[] and the second one uses the array. When there is an error in identifying the token in the first loop, the token is skipped, no change is made to the toksizes[] array. When the same error happens in the second loop, the token is not skipped. This will cause the toksizes[] array to be out of step and will overrun past the calculated sizes. Fix this by making both loops log a message and return an error in this case. This should only happen if a new token type is incompletely implemented, so it should normally be impossible to trigger this. Fixes: 9a059cd5ca7d ("rxrpc: Downgrade the BUG() for unsupported token type in rxrpc_read()") Reported-by: Tom Rix Signed-off-by: David Howells Reviewed-by: Tom Rix Link: https://lore.kernel.org/r/161046503122.2445787.16714129930607546635.stgit@warthog.procyon.org.uk Signed-off-by: Jakub Kicinski --- net/rxrpc/key.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'net') diff --git a/net/rxrpc/key.c b/net/rxrpc/key.c index 9631aa8543b5..8d2073e0e3da 100644 --- a/net/rxrpc/key.c +++ b/net/rxrpc/key.c @@ -598,7 +598,7 @@ static long rxrpc_read(const struct key *key, default: /* we have a ticket we can't encode */ pr_err("Unsupported key token type (%u)\n", token->security_index); - continue; + return -ENOPKG; } _debug("token[%u]: toksize=%u", ntoks, toksize); @@ -674,7 +674,9 @@ static long rxrpc_read(const struct key *key, break; default: - break; + pr_err("Unsupported key token type (%u)\n", + token->security_index); + return -ENOPKG; } ASSERTCMP((unsigned long)xdr - (unsigned long)oldxdr, ==, -- cgit v1.2.3 From a95d25dd7b94a5ba18246da09b4218f132fed60e Mon Sep 17 00:00:00 2001 From: Baptiste Lepers Date: Tue, 12 Jan 2021 15:59:15 +0000 Subject: rxrpc: Call state should be read with READ_ONCE() under some circumstances The call state may be changed at any time by the data-ready routine in response to received packets, so if the call state is to be read and acted upon several times in a function, READ_ONCE() must be used unless the call state lock is held. As it happens, we used READ_ONCE() to read the state a few lines above the unmarked read in rxrpc_input_data(), so use that value rather than re-reading it. Fixes: a158bdd3247b ("rxrpc: Fix call timeouts") Signed-off-by: Baptiste Lepers Signed-off-by: David Howells Link: https://lore.kernel.org/r/161046715522.2450566.488819910256264150.stgit@warthog.procyon.org.uk Signed-off-by: Jakub Kicinski --- net/rxrpc/input.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net') diff --git a/net/rxrpc/input.c b/net/rxrpc/input.c index 667c44aa5a63..dc201363f2c4 100644 --- a/net/rxrpc/input.c +++ b/net/rxrpc/input.c @@ -430,7 +430,7 @@ static void rxrpc_input_data(struct rxrpc_call *call, struct sk_buff *skb) return; } - if (call->state == RXRPC_CALL_SERVER_RECV_REQUEST) { + if (state == RXRPC_CALL_SERVER_RECV_REQUEST) { unsigned long timo = READ_ONCE(call->next_req_timo); unsigned long now, expect_req_by; -- cgit v1.2.3 From b42b3a2744b3e8f427de79896720c72823af91ad Mon Sep 17 00:00:00 2001 From: Oliver Hartkopp Date: Tue, 12 Jan 2021 10:16:43 +0100 Subject: can: isotp: isotp_getname(): fix kernel information leak Initialize the sockaddr_can structure to prevent a data leak to user space. Suggested-by: Cong Wang Reported-by: syzbot+057884e2f453e8afebc8@syzkaller.appspotmail.com Fixes: e057dd3fc20f ("can: add ISO 15765-2:2016 transport protocol") Signed-off-by: Oliver Hartkopp Link: https://lore.kernel.org/r/20210112091643.11789-1-socketcan@hartkopp.net Signed-off-by: Marc Kleine-Budde --- net/can/isotp.c | 1 + 1 file changed, 1 insertion(+) (limited to 'net') diff --git a/net/can/isotp.c b/net/can/isotp.c index 7839c3b9e5be..3ef7f78e553b 100644 --- a/net/can/isotp.c +++ b/net/can/isotp.c @@ -1155,6 +1155,7 @@ static int isotp_getname(struct socket *sock, struct sockaddr *uaddr, int peer) if (peer) return -EOPNOTSUPP; + memset(addr, 0, sizeof(*addr)); addr->can_family = AF_CAN; addr->can_ifindex = so->ifindex; addr->can_addr.tp.rx_id = so->rxid; -- cgit v1.2.3 From 47e4bb147a96f1c9b4e7691e7e994e53838bfff8 Mon Sep 17 00:00:00 2001 From: Jakub Kicinski Date: Wed, 13 Jan 2021 17:29:47 -0800 Subject: net: sit: unregister_netdevice on newlink's error path We need to unregister the netdevice if config failed. .ndo_uninit takes care of most of the heavy lifting. This was uncovered by recent commit c269a24ce057 ("net: make free_netdev() more lenient with unregistering devices"). Previously the partially-initialized device would be left in the system. Reported-and-tested-by: syzbot+2393580080a2da190f04@syzkaller.appspotmail.com Fixes: e2f1f072db8d ("sit: allow to configure 6rd tunnels via netlink") Acked-by: Nicolas Dichtel Link: https://lore.kernel.org/r/20210114012947.2515313-1-kuba@kernel.org Signed-off-by: Jakub Kicinski --- net/ipv6/sit.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'net') diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c index 2da0ee703779..93636867aee2 100644 --- a/net/ipv6/sit.c +++ b/net/ipv6/sit.c @@ -1645,8 +1645,11 @@ static int ipip6_newlink(struct net *src_net, struct net_device *dev, } #ifdef CONFIG_IPV6_SIT_6RD - if (ipip6_netlink_6rd_parms(data, &ip6rd)) + if (ipip6_netlink_6rd_parms(data, &ip6rd)) { err = ipip6_tunnel_update_6rd(nt, &ip6rd); + if (err < 0) + unregister_netdevice_queue(dev, NULL); + } #endif return err; -- cgit v1.2.3 From 25764779298f23a659f3daf39f9e2b5975a7a89d Mon Sep 17 00:00:00 2001 From: Mauro Carvalho Chehab Date: Thu, 14 Jan 2021 09:04:48 +0100 Subject: net: tip: fix a couple kernel-doc markups A function has a different name between their prototype and its kernel-doc markup: ../net/tipc/link.c:2551: warning: expecting prototype for link_reset_stats(). Prototype was for tipc_link_reset_stats() instead ../net/tipc/node.c:1678: warning: expecting prototype for is the general link level function for message sending(). Prototype was for tipc_node_xmit() instead Signed-off-by: Mauro Carvalho Chehab Acked-by: Jon Maloy Signed-off-by: Jakub Kicinski --- net/tipc/link.c | 2 +- net/tipc/node.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'net') diff --git a/net/tipc/link.c b/net/tipc/link.c index a6a694b78927..115109259430 100644 --- a/net/tipc/link.c +++ b/net/tipc/link.c @@ -2544,7 +2544,7 @@ void tipc_link_set_queue_limits(struct tipc_link *l, u32 min_win, u32 max_win) } /** - * link_reset_stats - reset link statistics + * tipc_link_reset_stats - reset link statistics * @l: pointer to link */ void tipc_link_reset_stats(struct tipc_link *l) diff --git a/net/tipc/node.c b/net/tipc/node.c index 83d9eb830592..008670d1f43e 100644 --- a/net/tipc/node.c +++ b/net/tipc/node.c @@ -1665,7 +1665,7 @@ static void tipc_lxc_xmit(struct net *peer_net, struct sk_buff_head *list) } /** - * tipc_node_xmit() is the general link level function for message sending + * tipc_node_xmit() - general link level function for message sending * @net: the applicable net namespace * @list: chain of buffers containing message * @dnode: address of destination node -- cgit v1.2.3 From 3226b158e67cfaa677fd180152bfb28989cb2fac Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 13 Jan 2021 08:18:19 -0800 Subject: net: avoid 32 x truesize under-estimation for tiny skbs Both virtio net and napi_get_frags() allocate skbs with a very small skb->head While using page fragments instead of a kmalloc backed skb->head might give a small performance improvement in some cases, there is a huge risk of under estimating memory usage. For both GOOD_COPY_LEN and GRO_MAX_HEAD, we can fit at least 32 allocations per page (order-3 page in x86), or even 64 on PowerPC We have been tracking OOM issues on GKE hosts hitting tcp_mem limits but consuming far more memory for TCP buffers than instructed in tcp_mem[2] Even if we force napi_alloc_skb() to only use order-0 pages, the issue would still be there on arches with PAGE_SIZE >= 32768 This patch makes sure that small skb head are kmalloc backed, so that other objects in the slab page can be reused instead of being held as long as skbs are sitting in socket queues. Note that we might in the future use the sk_buff napi cache, instead of going through a more expensive __alloc_skb() Another idea would be to use separate page sizes depending on the allocated length (to never have more than 4 frags per page) I would like to thank Greg Thelen for his precious help on this matter, analysing crash dumps is always a time consuming task. Fixes: fd11a83dd363 ("net: Pull out core bits of __netdev_alloc_skb and add __napi_alloc_skb") Signed-off-by: Eric Dumazet Cc: Paolo Abeni Cc: Greg Thelen Reviewed-by: Alexander Duyck Acked-by: Michael S. Tsirkin Link: https://lore.kernel.org/r/20210113161819.1155526-1-eric.dumazet@gmail.com Signed-off-by: Jakub Kicinski --- net/core/skbuff.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'net') diff --git a/net/core/skbuff.c b/net/core/skbuff.c index 0da035c1e53f..c1a6f262636a 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -501,13 +501,17 @@ EXPORT_SYMBOL(__netdev_alloc_skb); struct sk_buff *__napi_alloc_skb(struct napi_struct *napi, unsigned int len, gfp_t gfp_mask) { - struct napi_alloc_cache *nc = this_cpu_ptr(&napi_alloc_cache); + struct napi_alloc_cache *nc; struct sk_buff *skb; void *data; len += NET_SKB_PAD + NET_IP_ALIGN; - if ((len > SKB_WITH_OVERHEAD(PAGE_SIZE)) || + /* If requested length is either too small or too big, + * we use kmalloc() for skb->head allocation. + */ + if (len <= SKB_WITH_OVERHEAD(1024) || + len > SKB_WITH_OVERHEAD(PAGE_SIZE) || (gfp_mask & (__GFP_DIRECT_RECLAIM | GFP_DMA))) { skb = __alloc_skb(len, gfp_mask, SKB_ALLOC_RX, NUMA_NO_NODE); if (!skb) @@ -515,6 +519,7 @@ struct sk_buff *__napi_alloc_skb(struct napi_struct *napi, unsigned int len, goto skb_success; } + nc = this_cpu_ptr(&napi_alloc_cache); len += SKB_DATA_ALIGN(sizeof(struct skb_shared_info)); len = SKB_DATA_ALIGN(len); -- cgit v1.2.3 From 25537d71e2d007faf42a244a75e5a2bb7c356234 Mon Sep 17 00:00:00 2001 From: Tariq Toukan Date: Thu, 14 Jan 2021 17:12:15 +0200 Subject: net: Allow NETIF_F_HW_TLS_TX if IP_CSUM && IPV6_CSUM Cited patch below blocked the TLS TX device offload unless HW_CSUM is set. This broke devices that use IP_CSUM && IP6_CSUM. Here we fix it. Note that the single HW_TLS_TX feature flag indicates support for both IPv4/6, hence it should still be disabled in case only one of (IP_CSUM | IPV6_CSUM) is set. Fixes: ae0b04b238e2 ("net: Disable NETIF_F_HW_TLS_TX when HW_CSUM is disabled") Signed-off-by: Tariq Toukan Reported-by: Rohit Maheshwari Reviewed-by: Maxim Mikityanskiy Link: https://lore.kernel.org/r/20210114151215.7061-1-tariqt@nvidia.com Signed-off-by: Jakub Kicinski --- Documentation/networking/tls-offload.rst | 2 +- net/core/dev.c | 12 +++++++++--- 2 files changed, 10 insertions(+), 4 deletions(-) (limited to 'net') diff --git a/Documentation/networking/tls-offload.rst b/Documentation/networking/tls-offload.rst index 0f55c6d540f9..9af3334d9ad0 100644 --- a/Documentation/networking/tls-offload.rst +++ b/Documentation/networking/tls-offload.rst @@ -530,7 +530,7 @@ TLS device feature flags only control adding of new TLS connection offloads, old connections will remain active after flags are cleared. TLS encryption cannot be offloaded to devices without checksum calculation -offload. Hence, TLS TX device feature flag requires NETIF_F_HW_CSUM being set. +offload. Hence, TLS TX device feature flag requires TX csum offload being set. Disabling the latter implies clearing the former. Disabling TX checksum offload should not affect old connections, and drivers should make sure checksum calculation does not break for them. diff --git a/net/core/dev.c b/net/core/dev.c index 0071a11a6dc3..c360bb5367e2 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -9661,9 +9661,15 @@ static netdev_features_t netdev_fix_features(struct net_device *dev, } } - if ((features & NETIF_F_HW_TLS_TX) && !(features & NETIF_F_HW_CSUM)) { - netdev_dbg(dev, "Dropping TLS TX HW offload feature since no CSUM feature.\n"); - features &= ~NETIF_F_HW_TLS_TX; + if (features & NETIF_F_HW_TLS_TX) { + bool ip_csum = (features & (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM)) == + (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM); + bool hw_csum = features & NETIF_F_HW_CSUM; + + if (!ip_csum && !hw_csum) { + netdev_dbg(dev, "Dropping TLS TX HW offload feature since no CSUM feature.\n"); + features &= ~NETIF_F_HW_TLS_TX; + } } return features; -- cgit v1.2.3 From 13a9499e833387fcc7a53915bbe5cddf3c336b59 Mon Sep 17 00:00:00 2001 From: Paolo Abeni Date: Thu, 14 Jan 2021 16:37:37 +0100 Subject: mptcp: fix locking in mptcp_disconnect() tcp_disconnect() expects the caller acquires the sock lock, but mptcp_disconnect() is not doing that. Add the missing required lock. Reported-by: Eric Dumazet Fixes: 76e2a55d1625 ("mptcp: better msk-level shutdown.") Signed-off-by: Paolo Abeni Link: https://lore.kernel.org/r/f818e82b58a556feeb71dcccc8bf1c87aafc6175.1610638176.git.pabeni@redhat.com Signed-off-by: Jakub Kicinski --- net/mptcp/protocol.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'net') diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index 81faeff8f3bb..f998a077c7dd 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -2646,8 +2646,13 @@ static int mptcp_disconnect(struct sock *sk, int flags) struct mptcp_sock *msk = mptcp_sk(sk); __mptcp_flush_join_list(msk); - mptcp_for_each_subflow(msk, subflow) - tcp_disconnect(mptcp_subflow_tcp_sock(subflow), flags); + mptcp_for_each_subflow(msk, subflow) { + struct sock *ssk = mptcp_subflow_tcp_sock(subflow); + + lock_sock(ssk); + tcp_disconnect(ssk, flags); + release_sock(ssk); + } return 0; } -- cgit v1.2.3