From 0e141f757b2c78c983df893e9993313e2dc21e38 Mon Sep 17 00:00:00 2001 From: Haishuang Yan Date: Fri, 27 Sep 2019 14:58:20 +0800 Subject: erspan: remove the incorrect mtu limit for erspan erspan driver calls ether_setup(), after commit 61e84623ace3 ("net: centralize net_device min/max MTU checking"), the range of mtu is [min_mtu, max_mtu], which is [68, 1500] by default. It causes the dev mtu of the erspan device to not be greater than 1500, this limit value is not correct for ipgre tap device. Tested: Before patch: # ip link set erspan0 mtu 1600 Error: mtu greater than device maximum. After patch: # ip link set erspan0 mtu 1600 # ip -d link show erspan0 21: erspan0@NONE: mtu 1600 qdisc noop state DOWN mode DEFAULT group default qlen 1000 link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff promiscuity 0 minmtu 68 maxmtu 0 Fixes: 61e84623ace3 ("net: centralize net_device min/max MTU checking") Signed-off-by: Haishuang Yan Signed-off-by: David S. Miller --- net/ipv4/ip_gre.c | 1 + 1 file changed, 1 insertion(+) (limited to 'net/ipv4') diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c index a53a543fe055..52690bb3e40f 100644 --- a/net/ipv4/ip_gre.c +++ b/net/ipv4/ip_gre.c @@ -1446,6 +1446,7 @@ static void erspan_setup(struct net_device *dev) struct ip_tunnel *t = netdev_priv(dev); ether_setup(dev); + dev->max_mtu = 0; dev->netdev_ops = &erspan_netdev_ops; dev->priv_flags &= ~IFF_TX_SKB_SHARING; dev->priv_flags |= IFF_LIVE_ADDR_CHANGE; -- cgit v1.2.3 From 895b5c9f206eb7d25dc1360a8ccfc5958895eb89 Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Sun, 29 Sep 2019 20:54:03 +0200 Subject: netfilter: drop bridge nf reset from nf_reset commit 174e23810cd31 ("sk_buff: drop all skb extensions on free and skb scrubbing") made napi recycle always drop skb extensions. The additional skb_ext_del() that is performed via nf_reset on napi skb recycle is not needed anymore. Most nf_reset() calls in the stack are there so queued skb won't block 'rmmod nf_conntrack' indefinitely. This removes the skb_ext_del from nf_reset, and renames it to a more fitting nf_reset_ct(). In a few selected places, add a call to skb_ext_reset to make sure that no active extensions remain. I am submitting this for "net", because we're still early in the release cycle. The patch applies to net-next too, but I think the rename causes needless divergence between those trees. Suggested-by: Eric Dumazet Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- net/ipv4/ip_input.c | 2 +- net/ipv4/ipmr.c | 4 ++-- net/ipv4/netfilter/nf_dup_ipv4.c | 2 +- net/ipv4/raw.c | 2 +- net/ipv4/tcp_ipv4.c | 2 +- net/ipv4/udp.c | 4 ++-- 6 files changed, 8 insertions(+), 8 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/ip_input.c b/net/ipv4/ip_input.c index 1e2392b7c64e..c59a78a267c3 100644 --- a/net/ipv4/ip_input.c +++ b/net/ipv4/ip_input.c @@ -199,7 +199,7 @@ resubmit: kfree_skb(skb); return; } - nf_reset(skb); + nf_reset_ct(skb); } ret = INDIRECT_CALL_2(ipprot->handler, tcp_v4_rcv, udp_rcv, skb); diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c index 313470f6bb14..716d5472c022 100644 --- a/net/ipv4/ipmr.c +++ b/net/ipv4/ipmr.c @@ -1794,7 +1794,7 @@ static void ip_encap(struct net *net, struct sk_buff *skb, ip_send_check(iph); memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt)); - nf_reset(skb); + nf_reset_ct(skb); } static inline int ipmr_forward_finish(struct net *net, struct sock *sk, @@ -2140,7 +2140,7 @@ int ip_mr_input(struct sk_buff *skb) mroute_sk = rcu_dereference(mrt->mroute_sk); if (mroute_sk) { - nf_reset(skb); + nf_reset_ct(skb); raw_rcv(mroute_sk, skb); return 0; } diff --git a/net/ipv4/netfilter/nf_dup_ipv4.c b/net/ipv4/netfilter/nf_dup_ipv4.c index af3fbf76dbd3..6cc5743c553a 100644 --- a/net/ipv4/netfilter/nf_dup_ipv4.c +++ b/net/ipv4/netfilter/nf_dup_ipv4.c @@ -65,7 +65,7 @@ void nf_dup_ipv4(struct net *net, struct sk_buff *skb, unsigned int hooknum, #if IS_ENABLED(CONFIG_NF_CONNTRACK) /* Avoid counting cloned packets towards the original connection. */ - nf_reset(skb); + nf_reset_ct(skb); nf_ct_set(skb, NULL, IP_CT_UNTRACKED); #endif /* diff --git a/net/ipv4/raw.c b/net/ipv4/raw.c index 80da5a66d5d7..3183413ebc6c 100644 --- a/net/ipv4/raw.c +++ b/net/ipv4/raw.c @@ -332,7 +332,7 @@ int raw_rcv(struct sock *sk, struct sk_buff *skb) kfree_skb(skb); return NET_RX_DROP; } - nf_reset(skb); + nf_reset_ct(skb); skb_push(skb, skb->data - skb_network_header(skb)); diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 2ee45e3755e9..bf124b1742df 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -1916,7 +1916,7 @@ process: if (tcp_v4_inbound_md5_hash(sk, skb)) goto discard_and_relse; - nf_reset(skb); + nf_reset_ct(skb); if (tcp_filter(sk, skb)) goto discard_and_relse; diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index cf755156a684..e8443cc5c1ab 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -1969,7 +1969,7 @@ static int udp_queue_rcv_one_skb(struct sock *sk, struct sk_buff *skb) */ if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) goto drop; - nf_reset(skb); + nf_reset_ct(skb); if (static_branch_unlikely(&udp_encap_needed_key) && up->encap_type) { int (*encap_rcv)(struct sock *sk, struct sk_buff *skb); @@ -2298,7 +2298,7 @@ int __udp4_lib_rcv(struct sk_buff *skb, struct udp_table *udptable, if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) goto drop; - nf_reset(skb); + nf_reset_ct(skb); /* No socket. Drop packet silently, if checksum is wrong */ if (udp_lib_checksum_complete(skb)) -- cgit v1.2.3 From 3256a2d6ab1f71f9a1bd2d7f6f18eb8108c48d17 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Mon, 30 Sep 2019 15:44:44 -0700 Subject: tcp: adjust rto_base in retransmits_timed_out() The cited commit exposed an old retransmits_timed_out() bug which assumed it could call tcp_model_timeout() with TCP_RTO_MIN as rto_base for all states. But flows in SYN_SENT or SYN_RECV state uses a different RTO base (1 sec instead of 200 ms, unless BPF choses another value) This caused a reduction of SYN retransmits from 6 to 4 with the default /proc/sys/net/ipv4/tcp_syn_retries value. Fixes: a41e8a88b06e ("tcp: better handle TCP_USER_TIMEOUT in SYN_SENT state") Signed-off-by: Eric Dumazet Cc: Yuchung Cheng Cc: Marek Majkowski Signed-off-by: David S. Miller --- net/ipv4/tcp_timer.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c index 40de2d2364a1..05be564414e9 100644 --- a/net/ipv4/tcp_timer.c +++ b/net/ipv4/tcp_timer.c @@ -198,8 +198,13 @@ static bool retransmits_timed_out(struct sock *sk, return false; start_ts = tcp_sk(sk)->retrans_stamp; - if (likely(timeout == 0)) - timeout = tcp_model_timeout(sk, boundary, TCP_RTO_MIN); + if (likely(timeout == 0)) { + unsigned int rto_base = TCP_RTO_MIN; + + if ((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV)) + rto_base = tcp_timeout_init(sk); + timeout = tcp_model_timeout(sk, boundary, rto_base); + } return (s32)(tcp_time_stamp(tcp_sk(sk)) - start_ts - timeout) >= 0; } -- cgit v1.2.3 From 44b321e5020d782ad6e8ae8183f09b163be6e6e2 Mon Sep 17 00:00:00 2001 From: Josh Hunt Date: Wed, 2 Oct 2019 13:29:22 -0400 Subject: udp: fix gso_segs calculations Commit dfec0ee22c0a ("udp: Record gso_segs when supporting UDP segmentation offload") added gso_segs calculation, but incorrectly got sizeof() the pointer and not the underlying data type. In addition let's fix the v6 case. Fixes: bec1f6f69736 ("udp: generate gso with UDP_SEGMENT") Fixes: dfec0ee22c0a ("udp: Record gso_segs when supporting UDP segmentation offload") Signed-off-by: Josh Hunt Reviewed-by: Alexander Duyck Acked-by: Willem de Bruijn Signed-off-by: David S. Miller --- net/ipv4/udp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/ipv4') diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index e8443cc5c1ab..3e2b90181879 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -856,7 +856,7 @@ static int udp_send_skb(struct sk_buff *skb, struct flowi4 *fl4, skb_shinfo(skb)->gso_size = cork->gso_size; skb_shinfo(skb)->gso_type = SKB_GSO_UDP_L4; - skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(len - sizeof(uh), + skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(len - sizeof(*uh), cork->gso_size); goto csum_partial; } -- cgit v1.2.3 From 4094871db1d65810acab3d57f6089aa39ef7f648 Mon Sep 17 00:00:00 2001 From: Josh Hunt Date: Wed, 2 Oct 2019 13:29:23 -0400 Subject: udp: only do GSO if # of segs > 1 Prior to this change an application sending <= 1MSS worth of data and enabling UDP GSO would fail if the system had SW GSO enabled, but the same send would succeed if HW GSO offload is enabled. In addition to this inconsistency the error in the SW GSO case does not get back to the application if sending out of a real device so the user is unaware of this failure. With this change we only perform GSO if the # of segments is > 1 even if the application has enabled segmentation. I've also updated the relevant udpgso selftests. Fixes: bec1f6f69736 ("udp: generate gso with UDP_SEGMENT") Signed-off-by: Josh Hunt Reviewed-by: Willem de Bruijn Reviewed-by: Alexander Duyck Signed-off-by: David S. Miller --- net/ipv4/udp.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index 3e2b90181879..14bc654b6842 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -821,6 +821,7 @@ static int udp_send_skb(struct sk_buff *skb, struct flowi4 *fl4, int is_udplite = IS_UDPLITE(sk); int offset = skb_transport_offset(skb); int len = skb->len - offset; + int datalen = len - sizeof(*uh); __wsum csum = 0; /* @@ -854,10 +855,12 @@ static int udp_send_skb(struct sk_buff *skb, struct flowi4 *fl4, return -EIO; } - skb_shinfo(skb)->gso_size = cork->gso_size; - skb_shinfo(skb)->gso_type = SKB_GSO_UDP_L4; - skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(len - sizeof(*uh), - cork->gso_size); + if (datalen > cork->gso_size) { + skb_shinfo(skb)->gso_size = cork->gso_size; + skb_shinfo(skb)->gso_type = SKB_GSO_UDP_L4; + skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(datalen, + cork->gso_size); + } goto csum_partial; } -- cgit v1.2.3 From 3afb0961884046c8fb4acbce65139088959681c8 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 2 Oct 2019 20:19:59 -0700 Subject: tcp: fix slab-out-of-bounds in tcp_zerocopy_receive() Apparently a refactoring patch brought a bug, that was caught by syzbot [1] Original code was correct, do not try to be smarter than the compiler :/ [1] BUG: KASAN: slab-out-of-bounds in tcp_zerocopy_receive net/ipv4/tcp.c:1807 [inline] BUG: KASAN: slab-out-of-bounds in do_tcp_getsockopt.isra.0+0x2c6c/0x3120 net/ipv4/tcp.c:3654 Read of size 4 at addr ffff8880943cf188 by task syz-executor.2/17508 CPU: 0 PID: 17508 Comm: syz-executor.2 Not tainted 5.3.0-rc7+ #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Call Trace: __dump_stack lib/dump_stack.c:77 [inline] dump_stack+0x172/0x1f0 lib/dump_stack.c:113 print_address_description.cold+0xd4/0x306 mm/kasan/report.c:351 __kasan_report.cold+0x1b/0x36 mm/kasan/report.c:482 kasan_report+0x12/0x17 mm/kasan/common.c:618 __asan_report_load4_noabort+0x14/0x20 mm/kasan/generic_report.c:131 tcp_zerocopy_receive net/ipv4/tcp.c:1807 [inline] do_tcp_getsockopt.isra.0+0x2c6c/0x3120 net/ipv4/tcp.c:3654 tcp_getsockopt+0xbf/0xe0 net/ipv4/tcp.c:3680 sock_common_getsockopt+0x94/0xd0 net/core/sock.c:3098 __sys_getsockopt+0x16d/0x310 net/socket.c:2129 __do_sys_getsockopt net/socket.c:2144 [inline] __se_sys_getsockopt net/socket.c:2141 [inline] __x64_sys_getsockopt+0xbe/0x150 net/socket.c:2141 do_syscall_64+0xfd/0x6a0 arch/x86/entry/common.c:296 Fixes: d8e18a516f8f ("net: Use skb accessors in network core") Signed-off-by: Eric Dumazet Acked-by: Soheil Hassas Yeganeh Cc: Matthew Wilcox (Oracle) Reported-by: syzbot Signed-off-by: David S. Miller --- net/ipv4/tcp.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 79c325a07ba5..f98a1882e537 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -1798,13 +1798,11 @@ static int tcp_zerocopy_receive(struct sock *sk, } if (skb_frag_size(frags) != PAGE_SIZE || skb_frag_off(frags)) { int remaining = zc->recv_skip_hint; - int size = skb_frag_size(frags); - while (remaining && (size != PAGE_SIZE || + while (remaining && (skb_frag_size(frags) != PAGE_SIZE || skb_frag_off(frags))) { - remaining -= size; + remaining -= skb_frag_size(frags); frags++; - size = skb_frag_size(frags); } zc->recv_skip_hint -= remaining; break; -- cgit v1.2.3 From b406472b5ad79ede8d10077f0c8f05505ace8b6d Mon Sep 17 00:00:00 2001 From: Paolo Abeni Date: Fri, 4 Oct 2019 15:11:17 +0200 Subject: net: ipv4: avoid mixed n_redirects and rate_tokens usage Since commit c09551c6ff7f ("net: ipv4: use a dedicated counter for icmp_v4 redirect packets") we use 'n_redirects' to account for redirect packets, but we still use 'rate_tokens' to compute the redirect packets exponential backoff. If the device sent to the relevant peer any ICMP error packet after sending a redirect, it will also update 'rate_token' according to the leaking bucket schema; typically 'rate_token' will raise above BITS_PER_LONG and the redirect packets backoff algorithm will produce undefined behavior. Fix the issue using 'n_redirects' to compute the exponential backoff in ip_rt_send_redirect(). Note that we still clear rate_tokens after a redirect silence period, to avoid changing an established behaviour. The root cause predates git history; before the mentioned commit in the critical scenario, the kernel stopped sending redirects, after the mentioned commit the behavior more randomic. Reported-by: Xiumei Mu Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Fixes: c09551c6ff7f ("net: ipv4: use a dedicated counter for icmp_v4 redirect packets") Signed-off-by: Paolo Abeni Acked-by: Lorenzo Bianconi Signed-off-by: David S. Miller --- net/ipv4/route.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/route.c b/net/ipv4/route.c index 7dcce724c78b..14654876127e 100644 --- a/net/ipv4/route.c +++ b/net/ipv4/route.c @@ -916,16 +916,15 @@ void ip_rt_send_redirect(struct sk_buff *skb) if (peer->rate_tokens == 0 || time_after(jiffies, (peer->rate_last + - (ip_rt_redirect_load << peer->rate_tokens)))) { + (ip_rt_redirect_load << peer->n_redirects)))) { __be32 gw = rt_nexthop(rt, ip_hdr(skb)->daddr); icmp_send(skb, ICMP_REDIRECT, ICMP_REDIR_HOST, gw); peer->rate_last = jiffies; - ++peer->rate_tokens; ++peer->n_redirects; #ifdef CONFIG_IP_ROUTE_VERBOSE if (log_martians && - peer->rate_tokens == ip_rt_redirect_number) + peer->n_redirects == ip_rt_redirect_number) net_warn_ratelimited("host %pI4/if%d ignores redirects for %pI4 to %pI4\n", &ip_hdr(skb)->saddr, inet_iif(skb), &ip_hdr(skb)->daddr, &gw); -- cgit v1.2.3 From 60b173ca3d1cd1782bd0096dc17298ec242f6fb1 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 9 Oct 2019 14:51:20 -0700 Subject: net: add {READ|WRITE}_ONCE() annotations on ->rskq_accept_head reqsk_queue_empty() is called from inet_csk_listen_poll() while other cpus might write ->rskq_accept_head value. Use {READ|WRITE}_ONCE() to avoid compiler tricks and potential KCSAN splats. Fixes: fff1f3001cc5 ("tcp: add a spinlock to protect struct request_sock_queue") Signed-off-by: Eric Dumazet Signed-off-by: Jakub Kicinski --- net/ipv4/inet_connection_sock.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/ipv4') diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c index a9183543ca30..dbcf34ec8dd2 100644 --- a/net/ipv4/inet_connection_sock.c +++ b/net/ipv4/inet_connection_sock.c @@ -934,7 +934,7 @@ struct sock *inet_csk_reqsk_queue_add(struct sock *sk, req->sk = child; req->dl_next = NULL; if (queue->rskq_accept_head == NULL) - queue->rskq_accept_head = req; + WRITE_ONCE(queue->rskq_accept_head, req); else queue->rskq_accept_tail->dl_next = req; queue->rskq_accept_tail = req; -- cgit v1.2.3 From 1f142c17d19a5618d5a633195a46f2c8be9bf232 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 9 Oct 2019 15:10:15 -0700 Subject: tcp: annotate lockless access to tcp_memory_pressure tcp_memory_pressure is read without holding any lock, and its value could be changed on other cpus. Use READ_ONCE() to annotate these lockless reads. The write side is already using atomic ops. Fixes: b8da51ebb1aa ("tcp: introduce tcp_under_memory_pressure()") Signed-off-by: Eric Dumazet Signed-off-by: Jakub Kicinski --- net/ipv4/tcp.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index f98a1882e537..888c92b63f5a 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -326,7 +326,7 @@ void tcp_enter_memory_pressure(struct sock *sk) { unsigned long val; - if (tcp_memory_pressure) + if (READ_ONCE(tcp_memory_pressure)) return; val = jiffies; @@ -341,7 +341,7 @@ void tcp_leave_memory_pressure(struct sock *sk) { unsigned long val; - if (!tcp_memory_pressure) + if (!READ_ONCE(tcp_memory_pressure)) return; val = xchg(&tcp_memory_pressure, 0); if (val) -- cgit v1.2.3 From 8265792bf8871acc2d00fd03883d830e2249d395 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 9 Oct 2019 15:21:13 -0700 Subject: net: silence KCSAN warnings around sk_add_backlog() calls sk_add_backlog() callers usually read sk->sk_rcvbuf without owning the socket lock. This means sk_rcvbuf value can be changed by other cpus, and KCSAN complains. Add READ_ONCE() annotations to document the lockless nature of these reads. Note that writes over sk_rcvbuf should also use WRITE_ONCE(), but this will be done in separate patches to ease stable backports (if we decide this is relevant for stable trees). BUG: KCSAN: data-race in tcp_add_backlog / tcp_recvmsg write to 0xffff88812ab369f8 of 8 bytes by interrupt on cpu 1: __sk_add_backlog include/net/sock.h:902 [inline] sk_add_backlog include/net/sock.h:933 [inline] tcp_add_backlog+0x45a/0xcc0 net/ipv4/tcp_ipv4.c:1737 tcp_v4_rcv+0x1aba/0x1bf0 net/ipv4/tcp_ipv4.c:1925 ip_protocol_deliver_rcu+0x51/0x470 net/ipv4/ip_input.c:204 ip_local_deliver_finish+0x110/0x140 net/ipv4/ip_input.c:231 NF_HOOK include/linux/netfilter.h:305 [inline] NF_HOOK include/linux/netfilter.h:299 [inline] ip_local_deliver+0x133/0x210 net/ipv4/ip_input.c:252 dst_input include/net/dst.h:442 [inline] ip_rcv_finish+0x121/0x160 net/ipv4/ip_input.c:413 NF_HOOK include/linux/netfilter.h:305 [inline] NF_HOOK include/linux/netfilter.h:299 [inline] ip_rcv+0x18f/0x1a0 net/ipv4/ip_input.c:523 __netif_receive_skb_one_core+0xa7/0xe0 net/core/dev.c:5004 __netif_receive_skb+0x37/0xf0 net/core/dev.c:5118 netif_receive_skb_internal+0x59/0x190 net/core/dev.c:5208 napi_skb_finish net/core/dev.c:5671 [inline] napi_gro_receive+0x28f/0x330 net/core/dev.c:5704 receive_buf+0x284/0x30b0 drivers/net/virtio_net.c:1061 virtnet_receive drivers/net/virtio_net.c:1323 [inline] virtnet_poll+0x436/0x7d0 drivers/net/virtio_net.c:1428 napi_poll net/core/dev.c:6352 [inline] net_rx_action+0x3ae/0xa50 net/core/dev.c:6418 read to 0xffff88812ab369f8 of 8 bytes by task 7271 on cpu 0: tcp_recvmsg+0x470/0x1a30 net/ipv4/tcp.c:2047 inet_recvmsg+0xbb/0x250 net/ipv4/af_inet.c:838 sock_recvmsg_nosec net/socket.c:871 [inline] sock_recvmsg net/socket.c:889 [inline] sock_recvmsg+0x92/0xb0 net/socket.c:885 sock_read_iter+0x15f/0x1e0 net/socket.c:967 call_read_iter include/linux/fs.h:1864 [inline] new_sync_read+0x389/0x4f0 fs/read_write.c:414 __vfs_read+0xb1/0xc0 fs/read_write.c:427 vfs_read fs/read_write.c:461 [inline] vfs_read+0x143/0x2c0 fs/read_write.c:446 ksys_read+0xd5/0x1b0 fs/read_write.c:587 __do_sys_read fs/read_write.c:597 [inline] __se_sys_read fs/read_write.c:595 [inline] __x64_sys_read+0x4c/0x60 fs/read_write.c:595 do_syscall_64+0xcf/0x2f0 arch/x86/entry/common.c:296 entry_SYSCALL_64_after_hwframe+0x44/0xa9 Reported by Kernel Concurrency Sanitizer on: CPU: 0 PID: 7271 Comm: syz-fuzzer Not tainted 5.3.0+ #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Signed-off-by: Eric Dumazet Reported-by: syzbot Signed-off-by: Jakub Kicinski --- net/ipv4/tcp_ipv4.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index bf124b1742df..492bf6a6b023 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -1644,7 +1644,7 @@ int tcp_v4_early_demux(struct sk_buff *skb) bool tcp_add_backlog(struct sock *sk, struct sk_buff *skb) { - u32 limit = sk->sk_rcvbuf + sk->sk_sndbuf; + u32 limit = READ_ONCE(sk->sk_rcvbuf) + READ_ONCE(sk->sk_sndbuf); struct skb_shared_info *shinfo; const struct tcphdr *th; struct tcphdr *thtail; -- cgit v1.2.3 From eac66402d1c342f07ff38f8d631ff95eb7ad3220 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 9 Oct 2019 15:32:35 -0700 Subject: net: annotate sk->sk_rcvlowat lockless reads sock_rcvlowat() or int_sk_rcvlowat() might be called without the socket lock for example from tcp_poll(). Use READ_ONCE() to document the fact that other cpus might change sk->sk_rcvlowat under us and avoid KCSAN splats. Use WRITE_ONCE() on write sides too. Signed-off-by: Eric Dumazet Signed-off-by: Jakub Kicinski --- net/ipv4/tcp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 888c92b63f5a..8781a92ea4b6 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -1699,7 +1699,7 @@ int tcp_set_rcvlowat(struct sock *sk, int val) else cap = sock_net(sk)->ipv4.sysctl_tcp_rmem[2] >> 1; val = min(val, cap); - sk->sk_rcvlowat = val ? : 1; + WRITE_ONCE(sk->sk_rcvlowat, val ? : 1); /* Check if we need to signal EPOLLIN right now */ tcp_data_ready(sk); -- cgit v1.2.3 From d983ea6f16b835dcde2ee9a58a1e764ce68bfccc Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 10 Oct 2019 20:17:38 -0700 Subject: tcp: add rcu protection around tp->fastopen_rsk Both tcp_v4_err() and tcp_v6_err() do the following operations while they do not own the socket lock : fastopen = tp->fastopen_rsk; snd_una = fastopen ? tcp_rsk(fastopen)->snt_isn : tp->snd_una; The problem is that without appropriate barrier, the compiler might reload tp->fastopen_rsk and trigger a NULL deref. request sockets are protected by RCU, we can simply add the missing annotations and barriers to solve the issue. Fixes: 168a8f58059a ("tcp: TCP Fast Open Server - main code path") Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/inet_connection_sock.c | 4 ++-- net/ipv4/tcp.c | 11 ++++++++--- net/ipv4/tcp_fastopen.c | 2 +- net/ipv4/tcp_input.c | 13 +++++++++---- net/ipv4/tcp_ipv4.c | 4 ++-- net/ipv4/tcp_minisocks.c | 2 +- net/ipv4/tcp_output.c | 2 +- net/ipv4/tcp_timer.c | 11 ++++++----- 8 files changed, 30 insertions(+), 19 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c index dbcf34ec8dd2..eb30fc1770de 100644 --- a/net/ipv4/inet_connection_sock.c +++ b/net/ipv4/inet_connection_sock.c @@ -906,7 +906,7 @@ static void inet_child_forget(struct sock *sk, struct request_sock *req, percpu_counter_inc(sk->sk_prot->orphan_count); if (sk->sk_protocol == IPPROTO_TCP && tcp_rsk(req)->tfo_listener) { - BUG_ON(tcp_sk(child)->fastopen_rsk != req); + BUG_ON(rcu_access_pointer(tcp_sk(child)->fastopen_rsk) != req); BUG_ON(sk != req->rsk_listener); /* Paranoid, to prevent race condition if @@ -915,7 +915,7 @@ static void inet_child_forget(struct sock *sk, struct request_sock *req, * Also to satisfy an assertion in * tcp_v4_destroy_sock(). */ - tcp_sk(child)->fastopen_rsk = NULL; + RCU_INIT_POINTER(tcp_sk(child)->fastopen_rsk, NULL); } inet_csk_destroy_sock(child); } diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 8781a92ea4b6..c59d0bd29c5c 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -543,7 +543,7 @@ __poll_t tcp_poll(struct file *file, struct socket *sock, poll_table *wait) /* Connected or passive Fast Open socket? */ if (state != TCP_SYN_SENT && - (state != TCP_SYN_RECV || tp->fastopen_rsk)) { + (state != TCP_SYN_RECV || rcu_access_pointer(tp->fastopen_rsk))) { int target = sock_rcvlowat(sk, 0, INT_MAX); if (tp->urg_seq == tp->copied_seq && @@ -2487,7 +2487,10 @@ adjudge_to_death: } if (sk->sk_state == TCP_CLOSE) { - struct request_sock *req = tcp_sk(sk)->fastopen_rsk; + struct request_sock *req; + + req = rcu_dereference_protected(tcp_sk(sk)->fastopen_rsk, + lockdep_sock_is_held(sk)); /* We could get here with a non-NULL req if the socket is * aborted (e.g., closed with unread data) before 3WHS * finishes. @@ -3831,8 +3834,10 @@ EXPORT_SYMBOL(tcp_md5_hash_key); void tcp_done(struct sock *sk) { - struct request_sock *req = tcp_sk(sk)->fastopen_rsk; + struct request_sock *req; + req = rcu_dereference_protected(tcp_sk(sk)->fastopen_rsk, + lockdep_sock_is_held(sk)); if (sk->sk_state == TCP_SYN_SENT || sk->sk_state == TCP_SYN_RECV) TCP_INC_STATS(sock_net(sk), TCP_MIB_ATTEMPTFAILS); diff --git a/net/ipv4/tcp_fastopen.c b/net/ipv4/tcp_fastopen.c index 3fd451271a70..a915ade0c818 100644 --- a/net/ipv4/tcp_fastopen.c +++ b/net/ipv4/tcp_fastopen.c @@ -253,7 +253,7 @@ static struct sock *tcp_fastopen_create_child(struct sock *sk, */ tp = tcp_sk(child); - tp->fastopen_rsk = req; + rcu_assign_pointer(tp->fastopen_rsk, req); tcp_rsk(req)->tfo_listener = true; /* RFC1323: The window in SYN & SYN/ACK segments is never diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 3578357abe30..5f9b102c3b55 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -2666,7 +2666,7 @@ static void tcp_process_loss(struct sock *sk, int flag, int num_dupack, struct tcp_sock *tp = tcp_sk(sk); bool recovered = !before(tp->snd_una, tp->high_seq); - if ((flag & FLAG_SND_UNA_ADVANCED || tp->fastopen_rsk) && + if ((flag & FLAG_SND_UNA_ADVANCED || rcu_access_pointer(tp->fastopen_rsk)) && tcp_try_undo_loss(sk, false)) return; @@ -2990,7 +2990,7 @@ void tcp_rearm_rto(struct sock *sk) /* If the retrans timer is currently being used by Fast Open * for SYN-ACK retrans purpose, stay put. */ - if (tp->fastopen_rsk) + if (rcu_access_pointer(tp->fastopen_rsk)) return; if (!tp->packets_out) { @@ -6087,6 +6087,8 @@ reset_and_undo: static void tcp_rcv_synrecv_state_fastopen(struct sock *sk) { + struct request_sock *req; + tcp_try_undo_loss(sk, false); /* Reset rtx states to prevent spurious retransmits_timed_out() */ @@ -6096,7 +6098,9 @@ static void tcp_rcv_synrecv_state_fastopen(struct sock *sk) /* Once we leave TCP_SYN_RECV or TCP_FIN_WAIT_1, * we no longer need req so release it. */ - reqsk_fastopen_remove(sk, tcp_sk(sk)->fastopen_rsk, false); + req = rcu_dereference_protected(tcp_sk(sk)->fastopen_rsk, + lockdep_sock_is_held(sk)); + reqsk_fastopen_remove(sk, req, false); /* Re-arm the timer because data may have been sent out. * This is similar to the regular data transmission case @@ -6171,7 +6175,8 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb) tcp_mstamp_refresh(tp); tp->rx_opt.saw_tstamp = 0; - req = tp->fastopen_rsk; + req = rcu_dereference_protected(tp->fastopen_rsk, + lockdep_sock_is_held(sk)); if (req) { bool req_stolen; diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 492bf6a6b023..ffa366099eb2 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -478,7 +478,7 @@ int tcp_v4_err(struct sk_buff *icmp_skb, u32 info) icsk = inet_csk(sk); tp = tcp_sk(sk); /* XXX (TFO) - tp->snd_una should be ISN (tcp_create_openreq_child() */ - fastopen = tp->fastopen_rsk; + fastopen = rcu_dereference(tp->fastopen_rsk); snd_una = fastopen ? tcp_rsk(fastopen)->snt_isn : tp->snd_una; if (sk->sk_state != TCP_LISTEN && !between(seq, snd_una, tp->snd_nxt)) { @@ -2121,7 +2121,7 @@ void tcp_v4_destroy_sock(struct sock *sk) if (inet_csk(sk)->icsk_bind_hash) inet_put_port(sk); - BUG_ON(tp->fastopen_rsk); + BUG_ON(rcu_access_pointer(tp->fastopen_rsk)); /* If socket is aborted during connect operation */ tcp_free_fastopen_req(tp); diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index bb140a5db8c0..5401dbd39c8f 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -541,7 +541,7 @@ struct sock *tcp_create_openreq_child(const struct sock *sk, newtp->rx_opt.mss_clamp = req->mss; tcp_ecn_openreq_child(newtp, req); newtp->fastopen_req = NULL; - newtp->fastopen_rsk = NULL; + RCU_INIT_POINTER(newtp->fastopen_rsk, NULL); __TCP_INC_STATS(sock_net(sk), TCP_MIB_PASSIVEOPENS); diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index fec6d67bfd14..84ae4d1449ea 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -2482,7 +2482,7 @@ bool tcp_schedule_loss_probe(struct sock *sk, bool advancing_rto) /* Don't do any loss probe on a Fast Open connection before 3WHS * finishes. */ - if (tp->fastopen_rsk) + if (rcu_access_pointer(tp->fastopen_rsk)) return false; early_retrans = sock_net(sk)->ipv4.sysctl_tcp_early_retrans; diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c index 05be564414e9..dd5a6317a801 100644 --- a/net/ipv4/tcp_timer.c +++ b/net/ipv4/tcp_timer.c @@ -386,15 +386,13 @@ abort: tcp_write_err(sk); * Timer for Fast Open socket to retransmit SYNACK. Note that the * sk here is the child socket, not the parent (listener) socket. */ -static void tcp_fastopen_synack_timer(struct sock *sk) +static void tcp_fastopen_synack_timer(struct sock *sk, struct request_sock *req) { struct inet_connection_sock *icsk = inet_csk(sk); int max_retries = icsk->icsk_syn_retries ? : sock_net(sk)->ipv4.sysctl_tcp_synack_retries + 1; /* add one more retry for fastopen */ struct tcp_sock *tp = tcp_sk(sk); - struct request_sock *req; - req = tcp_sk(sk)->fastopen_rsk; req->rsk_ops->syn_ack_timeout(req); if (req->num_timeout >= max_retries) { @@ -435,11 +433,14 @@ void tcp_retransmit_timer(struct sock *sk) struct tcp_sock *tp = tcp_sk(sk); struct net *net = sock_net(sk); struct inet_connection_sock *icsk = inet_csk(sk); + struct request_sock *req; - if (tp->fastopen_rsk) { + req = rcu_dereference_protected(tp->fastopen_rsk, + lockdep_sock_is_held(sk)); + if (req) { WARN_ON_ONCE(sk->sk_state != TCP_SYN_RECV && sk->sk_state != TCP_FIN_WAIT1); - tcp_fastopen_synack_timer(sk); + tcp_fastopen_synack_timer(sk, req); /* Before we receive ACK to our SYN-ACK don't retransmit * anything else (e.g., data or FIN segments). */ -- cgit v1.2.3 From dba7d9b8c739df27ff3a234c81d6c6b23e3986fa Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 10 Oct 2019 20:17:39 -0700 Subject: tcp: annotate tp->rcv_nxt lockless reads There are few places where we fetch tp->rcv_nxt while this field can change from IRQ or other cpu. We need to add READ_ONCE() annotations, and also make sure write sides use corresponding WRITE_ONCE() to avoid store-tearing. Note that tcp_inq_hint() was already using READ_ONCE(tp->rcv_nxt) syzbot reported : BUG: KCSAN: data-race in tcp_poll / tcp_queue_rcv write to 0xffff888120425770 of 4 bytes by interrupt on cpu 0: tcp_rcv_nxt_update net/ipv4/tcp_input.c:3365 [inline] tcp_queue_rcv+0x180/0x380 net/ipv4/tcp_input.c:4638 tcp_rcv_established+0xbf1/0xf50 net/ipv4/tcp_input.c:5616 tcp_v4_do_rcv+0x381/0x4e0 net/ipv4/tcp_ipv4.c:1542 tcp_v4_rcv+0x1a03/0x1bf0 net/ipv4/tcp_ipv4.c:1923 ip_protocol_deliver_rcu+0x51/0x470 net/ipv4/ip_input.c:204 ip_local_deliver_finish+0x110/0x140 net/ipv4/ip_input.c:231 NF_HOOK include/linux/netfilter.h:305 [inline] NF_HOOK include/linux/netfilter.h:299 [inline] ip_local_deliver+0x133/0x210 net/ipv4/ip_input.c:252 dst_input include/net/dst.h:442 [inline] ip_rcv_finish+0x121/0x160 net/ipv4/ip_input.c:413 NF_HOOK include/linux/netfilter.h:305 [inline] NF_HOOK include/linux/netfilter.h:299 [inline] ip_rcv+0x18f/0x1a0 net/ipv4/ip_input.c:523 __netif_receive_skb_one_core+0xa7/0xe0 net/core/dev.c:5004 __netif_receive_skb+0x37/0xf0 net/core/dev.c:5118 netif_receive_skb_internal+0x59/0x190 net/core/dev.c:5208 napi_skb_finish net/core/dev.c:5671 [inline] napi_gro_receive+0x28f/0x330 net/core/dev.c:5704 receive_buf+0x284/0x30b0 drivers/net/virtio_net.c:1061 read to 0xffff888120425770 of 4 bytes by task 7254 on cpu 1: tcp_stream_is_readable net/ipv4/tcp.c:480 [inline] tcp_poll+0x204/0x6b0 net/ipv4/tcp.c:554 sock_poll+0xed/0x250 net/socket.c:1256 vfs_poll include/linux/poll.h:90 [inline] ep_item_poll.isra.0+0x90/0x190 fs/eventpoll.c:892 ep_send_events_proc+0x113/0x5c0 fs/eventpoll.c:1749 ep_scan_ready_list.constprop.0+0x189/0x500 fs/eventpoll.c:704 ep_send_events fs/eventpoll.c:1793 [inline] ep_poll+0xe3/0x900 fs/eventpoll.c:1930 do_epoll_wait+0x162/0x180 fs/eventpoll.c:2294 __do_sys_epoll_pwait fs/eventpoll.c:2325 [inline] __se_sys_epoll_pwait fs/eventpoll.c:2311 [inline] __x64_sys_epoll_pwait+0xcd/0x170 fs/eventpoll.c:2311 do_syscall_64+0xcf/0x2f0 arch/x86/entry/common.c:296 entry_SYSCALL_64_after_hwframe+0x44/0xa9 Reported by Kernel Concurrency Sanitizer on: CPU: 1 PID: 7254 Comm: syz-fuzzer Not tainted 5.3.0+ #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Signed-off-by: Eric Dumazet Reported-by: syzbot Signed-off-by: David S. Miller --- net/ipv4/tcp.c | 4 ++-- net/ipv4/tcp_diag.c | 2 +- net/ipv4/tcp_input.c | 6 +++--- net/ipv4/tcp_ipv4.c | 3 ++- net/ipv4/tcp_minisocks.c | 7 +++++-- 5 files changed, 13 insertions(+), 9 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index c59d0bd29c5c..883ee863db43 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -477,7 +477,7 @@ static void tcp_tx_timestamp(struct sock *sk, u16 tsflags) static inline bool tcp_stream_is_readable(const struct tcp_sock *tp, int target, struct sock *sk) { - return (tp->rcv_nxt - tp->copied_seq >= target) || + return (READ_ONCE(tp->rcv_nxt) - tp->copied_seq >= target) || (sk->sk_prot->stream_memory_read ? sk->sk_prot->stream_memory_read(sk) : false); } @@ -2935,7 +2935,7 @@ static int do_tcp_setsockopt(struct sock *sk, int level, else if (tp->repair_queue == TCP_SEND_QUEUE) tp->write_seq = val; else if (tp->repair_queue == TCP_RECV_QUEUE) - tp->rcv_nxt = val; + WRITE_ONCE(tp->rcv_nxt, val); else err = -EINVAL; break; diff --git a/net/ipv4/tcp_diag.c b/net/ipv4/tcp_diag.c index 81a8221d650a..cd219161f106 100644 --- a/net/ipv4/tcp_diag.c +++ b/net/ipv4/tcp_diag.c @@ -26,7 +26,7 @@ static void tcp_diag_get_info(struct sock *sk, struct inet_diag_msg *r, } else if (sk->sk_type == SOCK_STREAM) { const struct tcp_sock *tp = tcp_sk(sk); - r->idiag_rqueue = max_t(int, tp->rcv_nxt - tp->copied_seq, 0); + r->idiag_rqueue = max_t(int, READ_ONCE(tp->rcv_nxt) - tp->copied_seq, 0); r->idiag_wqueue = tp->write_seq - tp->snd_una; } if (info) diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 5f9b102c3b55..5b7c8768ed5f 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -3362,7 +3362,7 @@ static void tcp_rcv_nxt_update(struct tcp_sock *tp, u32 seq) sock_owned_by_me((struct sock *)tp); tp->bytes_received += delta; - tp->rcv_nxt = seq; + WRITE_ONCE(tp->rcv_nxt, seq); } /* Update our send window. @@ -5932,7 +5932,7 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb, /* Ok.. it's good. Set up sequence numbers and * move to established. */ - tp->rcv_nxt = TCP_SKB_CB(skb)->seq + 1; + WRITE_ONCE(tp->rcv_nxt, TCP_SKB_CB(skb)->seq + 1); tp->rcv_wup = TCP_SKB_CB(skb)->seq + 1; /* RFC1323: The window in SYN & SYN/ACK segments is @@ -6035,7 +6035,7 @@ discard: tp->tcp_header_len = sizeof(struct tcphdr); } - tp->rcv_nxt = TCP_SKB_CB(skb)->seq + 1; + WRITE_ONCE(tp->rcv_nxt, TCP_SKB_CB(skb)->seq + 1); tp->copied_seq = tp->rcv_nxt; tp->rcv_wup = TCP_SKB_CB(skb)->seq + 1; diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index ffa366099eb2..5089dd6bee0f 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2455,7 +2455,8 @@ static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i) /* Because we don't lock the socket, * we might find a transient negative value. */ - rx_queue = max_t(int, tp->rcv_nxt - tp->copied_seq, 0); + rx_queue = max_t(int, READ_ONCE(tp->rcv_nxt) - + tp->copied_seq, 0); seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX " "%08X %5u %8d %lu %d %pK %lu %lu %u %u %d", diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index 5401dbd39c8f..adc6ce486a38 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -462,6 +462,7 @@ struct sock *tcp_create_openreq_child(const struct sock *sk, struct tcp_request_sock *treq = tcp_rsk(req); struct inet_connection_sock *newicsk; struct tcp_sock *oldtp, *newtp; + u32 seq; if (!newsk) return NULL; @@ -475,8 +476,10 @@ struct sock *tcp_create_openreq_child(const struct sock *sk, /* Now setup tcp_sock */ newtp->pred_flags = 0; - newtp->rcv_wup = newtp->copied_seq = - newtp->rcv_nxt = treq->rcv_isn + 1; + seq = treq->rcv_isn + 1; + newtp->rcv_wup = seq; + newtp->copied_seq = seq; + WRITE_ONCE(newtp->rcv_nxt, seq); newtp->segs_in = 1; newtp->snd_sml = newtp->snd_una = -- cgit v1.2.3 From 7db48e983930285b765743ebd665aecf9850582b Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 10 Oct 2019 20:17:40 -0700 Subject: tcp: annotate tp->copied_seq lockless reads There are few places where we fetch tp->copied_seq while this field can change from IRQ or other cpu. We need to add READ_ONCE() annotations, and also make sure write sides use corresponding WRITE_ONCE() to avoid store-tearing. Note that tcp_inq_hint() was already using READ_ONCE(tp->copied_seq) Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp.c | 20 ++++++++++---------- net/ipv4/tcp_diag.c | 3 ++- net/ipv4/tcp_input.c | 6 +++--- net/ipv4/tcp_ipv4.c | 2 +- net/ipv4/tcp_minisocks.c | 2 +- net/ipv4/tcp_output.c | 2 +- 6 files changed, 18 insertions(+), 17 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 883ee863db43..c322ad071e17 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -477,7 +477,7 @@ static void tcp_tx_timestamp(struct sock *sk, u16 tsflags) static inline bool tcp_stream_is_readable(const struct tcp_sock *tp, int target, struct sock *sk) { - return (READ_ONCE(tp->rcv_nxt) - tp->copied_seq >= target) || + return (READ_ONCE(tp->rcv_nxt) - READ_ONCE(tp->copied_seq) >= target) || (sk->sk_prot->stream_memory_read ? sk->sk_prot->stream_memory_read(sk) : false); } @@ -546,7 +546,7 @@ __poll_t tcp_poll(struct file *file, struct socket *sock, poll_table *wait) (state != TCP_SYN_RECV || rcu_access_pointer(tp->fastopen_rsk))) { int target = sock_rcvlowat(sk, 0, INT_MAX); - if (tp->urg_seq == tp->copied_seq && + if (tp->urg_seq == READ_ONCE(tp->copied_seq) && !sock_flag(sk, SOCK_URGINLINE) && tp->urg_data) target++; @@ -607,7 +607,7 @@ int tcp_ioctl(struct sock *sk, int cmd, unsigned long arg) unlock_sock_fast(sk, slow); break; case SIOCATMARK: - answ = tp->urg_data && tp->urg_seq == tp->copied_seq; + answ = tp->urg_data && tp->urg_seq == READ_ONCE(tp->copied_seq); break; case SIOCOUTQ: if (sk->sk_state == TCP_LISTEN) @@ -1668,9 +1668,9 @@ int tcp_read_sock(struct sock *sk, read_descriptor_t *desc, sk_eat_skb(sk, skb); if (!desc->count) break; - tp->copied_seq = seq; + WRITE_ONCE(tp->copied_seq, seq); } - tp->copied_seq = seq; + WRITE_ONCE(tp->copied_seq, seq); tcp_rcv_space_adjust(sk); @@ -1819,7 +1819,7 @@ static int tcp_zerocopy_receive(struct sock *sk, out: up_read(¤t->mm->mmap_sem); if (length) { - tp->copied_seq = seq; + WRITE_ONCE(tp->copied_seq, seq); tcp_rcv_space_adjust(sk); /* Clean up data we have read: This will do ACK frames. */ @@ -2117,7 +2117,7 @@ found_ok_skb: if (urg_offset < used) { if (!urg_offset) { if (!sock_flag(sk, SOCK_URGINLINE)) { - ++*seq; + WRITE_ONCE(*seq, *seq + 1); urg_hole++; offset++; used--; @@ -2139,7 +2139,7 @@ found_ok_skb: } } - *seq += used; + WRITE_ONCE(*seq, *seq + used); copied += used; len -= used; @@ -2166,7 +2166,7 @@ skip_copy: found_fin_ok: /* Process the FIN. */ - ++*seq; + WRITE_ONCE(*seq, *seq + 1); if (!(flags & MSG_PEEK)) sk_eat_skb(sk, skb); break; @@ -2588,7 +2588,7 @@ int tcp_disconnect(struct sock *sk, int flags) __kfree_skb(sk->sk_rx_skb_cache); sk->sk_rx_skb_cache = NULL; } - tp->copied_seq = tp->rcv_nxt; + WRITE_ONCE(tp->copied_seq, tp->rcv_nxt); tp->urg_data = 0; tcp_write_queue_purge(sk); tcp_fastopen_active_disable_ofo_check(sk); diff --git a/net/ipv4/tcp_diag.c b/net/ipv4/tcp_diag.c index cd219161f106..66273c8a55c2 100644 --- a/net/ipv4/tcp_diag.c +++ b/net/ipv4/tcp_diag.c @@ -26,7 +26,8 @@ static void tcp_diag_get_info(struct sock *sk, struct inet_diag_msg *r, } else if (sk->sk_type == SOCK_STREAM) { const struct tcp_sock *tp = tcp_sk(sk); - r->idiag_rqueue = max_t(int, READ_ONCE(tp->rcv_nxt) - tp->copied_seq, 0); + r->idiag_rqueue = max_t(int, READ_ONCE(tp->rcv_nxt) - + READ_ONCE(tp->copied_seq), 0); r->idiag_wqueue = tp->write_seq - tp->snd_una; } if (info) diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 5b7c8768ed5f..a30aae3a6a18 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -5961,7 +5961,7 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb, /* Remember, tcp_poll() does not lock socket! * Change state from SYN-SENT only after copied_seq * is initialized. */ - tp->copied_seq = tp->rcv_nxt; + WRITE_ONCE(tp->copied_seq, tp->rcv_nxt); smc_check_reset_syn(tp); @@ -6036,7 +6036,7 @@ discard: } WRITE_ONCE(tp->rcv_nxt, TCP_SKB_CB(skb)->seq + 1); - tp->copied_seq = tp->rcv_nxt; + WRITE_ONCE(tp->copied_seq, tp->rcv_nxt); tp->rcv_wup = TCP_SKB_CB(skb)->seq + 1; /* RFC1323: The window in SYN & SYN/ACK segments is @@ -6216,7 +6216,7 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb) tcp_try_undo_spurious_syn(sk); tp->retrans_stamp = 0; tcp_init_transfer(sk, BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB); - tp->copied_seq = tp->rcv_nxt; + WRITE_ONCE(tp->copied_seq, tp->rcv_nxt); } smp_mb(); tcp_set_state(sk, TCP_ESTABLISHED); diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 5089dd6bee0f..39560f482e0b 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2456,7 +2456,7 @@ static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i) * we might find a transient negative value. */ rx_queue = max_t(int, READ_ONCE(tp->rcv_nxt) - - tp->copied_seq, 0); + READ_ONCE(tp->copied_seq), 0); seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX " "%08X %5u %8d %lu %d %pK %lu %lu %u %u %d", diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index adc6ce486a38..c4731d26ab4a 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -478,7 +478,7 @@ struct sock *tcp_create_openreq_child(const struct sock *sk, seq = treq->rcv_isn + 1; newtp->rcv_wup = seq; - newtp->copied_seq = seq; + WRITE_ONCE(newtp->copied_seq, seq); WRITE_ONCE(newtp->rcv_nxt, seq); newtp->segs_in = 1; diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index 84ae4d1449ea..7dda12720169 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -3433,7 +3433,7 @@ static void tcp_connect_init(struct sock *sk) else tp->rcv_tstamp = tcp_jiffies32; tp->rcv_wup = tp->rcv_nxt; - tp->copied_seq = tp->rcv_nxt; + WRITE_ONCE(tp->copied_seq, tp->rcv_nxt); inet_csk(sk)->icsk_rto = tcp_timeout_init(sk); inet_csk(sk)->icsk_retransmits = 0; -- cgit v1.2.3 From 0f31746452e6793ad6271337438af8f4defb8940 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 10 Oct 2019 20:17:41 -0700 Subject: tcp: annotate tp->write_seq lockless reads There are few places where we fetch tp->write_seq while this field can change from IRQ or other cpu. We need to add READ_ONCE() annotations, and also make sure write sides use corresponding WRITE_ONCE() to avoid store-tearing. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp.c | 20 ++++++++++++-------- net/ipv4/tcp_diag.c | 2 +- net/ipv4/tcp_ipv4.c | 21 ++++++++++++--------- net/ipv4/tcp_minisocks.c | 2 +- net/ipv4/tcp_output.c | 4 ++-- 5 files changed, 28 insertions(+), 21 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index c322ad071e17..96dd65cbeb85 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -616,7 +616,7 @@ int tcp_ioctl(struct sock *sk, int cmd, unsigned long arg) if ((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV)) answ = 0; else - answ = tp->write_seq - tp->snd_una; + answ = READ_ONCE(tp->write_seq) - tp->snd_una; break; case SIOCOUTQNSD: if (sk->sk_state == TCP_LISTEN) @@ -625,7 +625,7 @@ int tcp_ioctl(struct sock *sk, int cmd, unsigned long arg) if ((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV)) answ = 0; else - answ = tp->write_seq - tp->snd_nxt; + answ = READ_ONCE(tp->write_seq) - tp->snd_nxt; break; default: return -ENOIOCTLCMD; @@ -1035,7 +1035,7 @@ new_segment: sk->sk_wmem_queued += copy; sk_mem_charge(sk, copy); skb->ip_summed = CHECKSUM_PARTIAL; - tp->write_seq += copy; + WRITE_ONCE(tp->write_seq, tp->write_seq + copy); TCP_SKB_CB(skb)->end_seq += copy; tcp_skb_pcount_set(skb, 0); @@ -1362,7 +1362,7 @@ new_segment: if (!copied) TCP_SKB_CB(skb)->tcp_flags &= ~TCPHDR_PSH; - tp->write_seq += copy; + WRITE_ONCE(tp->write_seq, tp->write_seq + copy); TCP_SKB_CB(skb)->end_seq += copy; tcp_skb_pcount_set(skb, 0); @@ -2562,6 +2562,7 @@ int tcp_disconnect(struct sock *sk, int flags) struct inet_connection_sock *icsk = inet_csk(sk); struct tcp_sock *tp = tcp_sk(sk); int old_state = sk->sk_state; + u32 seq; if (old_state != TCP_CLOSE) tcp_set_state(sk, TCP_CLOSE); @@ -2604,9 +2605,12 @@ int tcp_disconnect(struct sock *sk, int flags) tp->srtt_us = 0; tp->mdev_us = jiffies_to_usecs(TCP_TIMEOUT_INIT); tp->rcv_rtt_last_tsecr = 0; - tp->write_seq += tp->max_window + 2; - if (tp->write_seq == 0) - tp->write_seq = 1; + + seq = tp->write_seq + tp->max_window + 2; + if (!seq) + seq = 1; + WRITE_ONCE(tp->write_seq, seq); + icsk->icsk_backoff = 0; tp->snd_cwnd = 2; icsk->icsk_probes_out = 0; @@ -2933,7 +2937,7 @@ static int do_tcp_setsockopt(struct sock *sk, int level, if (sk->sk_state != TCP_CLOSE) err = -EPERM; else if (tp->repair_queue == TCP_SEND_QUEUE) - tp->write_seq = val; + WRITE_ONCE(tp->write_seq, val); else if (tp->repair_queue == TCP_RECV_QUEUE) WRITE_ONCE(tp->rcv_nxt, val); else diff --git a/net/ipv4/tcp_diag.c b/net/ipv4/tcp_diag.c index 66273c8a55c2..549506162dde 100644 --- a/net/ipv4/tcp_diag.c +++ b/net/ipv4/tcp_diag.c @@ -28,7 +28,7 @@ static void tcp_diag_get_info(struct sock *sk, struct inet_diag_msg *r, r->idiag_rqueue = max_t(int, READ_ONCE(tp->rcv_nxt) - READ_ONCE(tp->copied_seq), 0); - r->idiag_wqueue = tp->write_seq - tp->snd_una; + r->idiag_wqueue = READ_ONCE(tp->write_seq) - tp->snd_una; } if (info) tcp_get_info(sk, info); diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 39560f482e0b..6be568334848 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -164,9 +164,11 @@ int tcp_twsk_unique(struct sock *sk, struct sock *sktw, void *twp) * without appearing to create any others. */ if (likely(!tp->repair)) { - tp->write_seq = tcptw->tw_snd_nxt + 65535 + 2; - if (tp->write_seq == 0) - tp->write_seq = 1; + u32 seq = tcptw->tw_snd_nxt + 65535 + 2; + + if (!seq) + seq = 1; + WRITE_ONCE(tp->write_seq, seq); tp->rx_opt.ts_recent = tcptw->tw_ts_recent; tp->rx_opt.ts_recent_stamp = tcptw->tw_ts_recent_stamp; } @@ -253,7 +255,7 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len) tp->rx_opt.ts_recent = 0; tp->rx_opt.ts_recent_stamp = 0; if (likely(!tp->repair)) - tp->write_seq = 0; + WRITE_ONCE(tp->write_seq, 0); } inet->inet_dport = usin->sin_port; @@ -291,10 +293,11 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len) if (likely(!tp->repair)) { if (!tp->write_seq) - tp->write_seq = secure_tcp_seq(inet->inet_saddr, - inet->inet_daddr, - inet->inet_sport, - usin->sin_port); + WRITE_ONCE(tp->write_seq, + secure_tcp_seq(inet->inet_saddr, + inet->inet_daddr, + inet->inet_sport, + usin->sin_port)); tp->tsoffset = secure_tcp_ts_off(sock_net(sk), inet->inet_saddr, inet->inet_daddr); @@ -2461,7 +2464,7 @@ static void get_tcp4_sock(struct sock *sk, struct seq_file *f, int i) seq_printf(f, "%4d: %08X:%04X %08X:%04X %02X %08X:%08X %02X:%08lX " "%08X %5u %8d %lu %d %pK %lu %lu %u %u %d", i, src, srcp, dest, destp, state, - tp->write_seq - tp->snd_una, + READ_ONCE(tp->write_seq) - tp->snd_una, rx_queue, timer_active, jiffies_delta_to_clock_t(timer_expires - jiffies), diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index c4731d26ab4a..339944690329 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -498,7 +498,7 @@ struct sock *tcp_create_openreq_child(const struct sock *sk, newtp->total_retrans = req->num_retrans; tcp_init_xmit_timers(newsk); - newtp->write_seq = newtp->pushed_seq = treq->snt_isn + 1; + WRITE_ONCE(newtp->write_seq, newtp->pushed_seq = treq->snt_isn + 1); if (sock_flag(newsk, SOCK_KEEPOPEN)) inet_csk_reset_keepalive_timer(newsk, diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index 7dda12720169..c17c2a78809d 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -1196,7 +1196,7 @@ static void tcp_queue_skb(struct sock *sk, struct sk_buff *skb) struct tcp_sock *tp = tcp_sk(sk); /* Advance write_seq and place onto the write_queue. */ - tp->write_seq = TCP_SKB_CB(skb)->end_seq; + WRITE_ONCE(tp->write_seq, TCP_SKB_CB(skb)->end_seq); __skb_header_release(skb); tcp_add_write_queue_tail(sk, skb); sk->sk_wmem_queued += skb->truesize; @@ -3449,7 +3449,7 @@ static void tcp_connect_queue_skb(struct sock *sk, struct sk_buff *skb) __skb_header_release(skb); sk->sk_wmem_queued += skb->truesize; sk_mem_charge(sk, skb->truesize); - tp->write_seq = tcb->end_seq; + WRITE_ONCE(tp->write_seq, tcb->end_seq); tp->packets_out += tcp_skb_pcount(skb); } -- cgit v1.2.3 From e0d694d638dba768b47be31c22e1a9b4f862f561 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 10 Oct 2019 20:17:42 -0700 Subject: tcp: annotate tp->snd_nxt lockless reads There are few places where we fetch tp->snd_nxt while this field can change from IRQ or other cpu. We need to add READ_ONCE() annotations, and also make sure write sides use corresponding WRITE_ONCE() to avoid store-tearing. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp.c | 3 ++- net/ipv4/tcp_minisocks.c | 6 ++++-- net/ipv4/tcp_output.c | 10 +++++----- 3 files changed, 11 insertions(+), 8 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 96dd65cbeb85..652568750cb1 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -625,7 +625,8 @@ int tcp_ioctl(struct sock *sk, int cmd, unsigned long arg) if ((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV)) answ = 0; else - answ = READ_ONCE(tp->write_seq) - tp->snd_nxt; + answ = READ_ONCE(tp->write_seq) - + READ_ONCE(tp->snd_nxt); break; default: return -ENOIOCTLCMD; diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index 339944690329..c802bc80c400 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -482,8 +482,10 @@ struct sock *tcp_create_openreq_child(const struct sock *sk, WRITE_ONCE(newtp->rcv_nxt, seq); newtp->segs_in = 1; - newtp->snd_sml = newtp->snd_una = - newtp->snd_nxt = newtp->snd_up = treq->snt_isn + 1; + seq = treq->snt_isn + 1; + newtp->snd_sml = newtp->snd_una = seq; + WRITE_ONCE(newtp->snd_nxt, seq); + newtp->snd_up = seq; INIT_LIST_HEAD(&newtp->tsq_node); INIT_LIST_HEAD(&newtp->tsorted_sent_queue); diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index c17c2a78809d..a115a991dfb5 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -67,7 +67,7 @@ static void tcp_event_new_data_sent(struct sock *sk, struct sk_buff *skb) struct tcp_sock *tp = tcp_sk(sk); unsigned int prior_packets = tp->packets_out; - tp->snd_nxt = TCP_SKB_CB(skb)->end_seq; + WRITE_ONCE(tp->snd_nxt, TCP_SKB_CB(skb)->end_seq); __skb_unlink(skb, &sk->sk_write_queue); tcp_rbtree_insert(&sk->tcp_rtx_queue, skb); @@ -3142,7 +3142,7 @@ void tcp_send_fin(struct sock *sk) * if FIN had been sent. This is because retransmit path * does not change tp->snd_nxt. */ - tp->snd_nxt++; + WRITE_ONCE(tp->snd_nxt, tp->snd_nxt + 1); return; } } else { @@ -3426,7 +3426,7 @@ static void tcp_connect_init(struct sock *sk) tp->snd_una = tp->write_seq; tp->snd_sml = tp->write_seq; tp->snd_up = tp->write_seq; - tp->snd_nxt = tp->write_seq; + WRITE_ONCE(tp->snd_nxt, tp->write_seq); if (likely(!tp->repair)) tp->rcv_nxt = 0; @@ -3586,11 +3586,11 @@ int tcp_connect(struct sock *sk) /* We change tp->snd_nxt after the tcp_transmit_skb() call * in order to make this packet get counted in tcpOutSegs. */ - tp->snd_nxt = tp->write_seq; + WRITE_ONCE(tp->snd_nxt, tp->write_seq); tp->pushed_seq = tp->write_seq; buff = tcp_send_head(sk); if (unlikely(buff)) { - tp->snd_nxt = TCP_SKB_CB(buff)->seq; + WRITE_ONCE(tp->snd_nxt, TCP_SKB_CB(buff)->seq); tp->pushed_seq = TCP_SKB_CB(buff)->seq; } TCP_INC_STATS(sock_net(sk), TCP_MIB_ACTIVEOPENS); -- cgit v1.2.3 From d9b55bf7b6788ec0bd1db1acefbc4feb1399144a Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 10 Oct 2019 20:17:43 -0700 Subject: tcp: annotate tp->urg_seq lockless reads There two places where we fetch tp->urg_seq while this field can change from IRQ or other cpu. We need to add READ_ONCE() annotations, and also make sure write side use corresponding WRITE_ONCE() to avoid store-tearing. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp.c | 5 +++-- net/ipv4/tcp_input.c | 2 +- 2 files changed, 4 insertions(+), 3 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 652568750cb1..577a8c6eef9f 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -546,7 +546,7 @@ __poll_t tcp_poll(struct file *file, struct socket *sock, poll_table *wait) (state != TCP_SYN_RECV || rcu_access_pointer(tp->fastopen_rsk))) { int target = sock_rcvlowat(sk, 0, INT_MAX); - if (tp->urg_seq == READ_ONCE(tp->copied_seq) && + if (READ_ONCE(tp->urg_seq) == READ_ONCE(tp->copied_seq) && !sock_flag(sk, SOCK_URGINLINE) && tp->urg_data) target++; @@ -607,7 +607,8 @@ int tcp_ioctl(struct sock *sk, int cmd, unsigned long arg) unlock_sock_fast(sk, slow); break; case SIOCATMARK: - answ = tp->urg_data && tp->urg_seq == READ_ONCE(tp->copied_seq); + answ = tp->urg_data && + READ_ONCE(tp->urg_seq) == READ_ONCE(tp->copied_seq); break; case SIOCOUTQ: if (sk->sk_state == TCP_LISTEN) diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index a30aae3a6a18..16342e043ab3 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -5356,7 +5356,7 @@ static void tcp_check_urg(struct sock *sk, const struct tcphdr *th) } tp->urg_data = TCP_URG_NOTYET; - tp->urg_seq = ptr; + WRITE_ONCE(tp->urg_seq, ptr); /* Disable header prediction. */ tp->pred_flags = 0; -- cgit v1.2.3 From ebb3b78db7bf842270a46fd4fe7cc45c78fa5ed6 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 10 Oct 2019 20:17:44 -0700 Subject: tcp: annotate sk->sk_rcvbuf lockless reads For the sake of tcp_poll(), there are few places where we fetch sk->sk_rcvbuf while this field can change from IRQ or other cpu. We need to add READ_ONCE() annotations, and also make sure write sides use corresponding WRITE_ONCE() to avoid store-tearing. Note that other transports probably need similar fixes. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp.c | 4 ++-- net/ipv4/tcp_input.c | 7 ++++--- 2 files changed, 6 insertions(+), 5 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 577a8c6eef9f..bc0481aa6633 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -451,7 +451,7 @@ void tcp_init_sock(struct sock *sk) icsk->icsk_sync_mss = tcp_sync_mss; sk->sk_sndbuf = sock_net(sk)->ipv4.sysctl_tcp_wmem[1]; - sk->sk_rcvbuf = sock_net(sk)->ipv4.sysctl_tcp_rmem[1]; + WRITE_ONCE(sk->sk_rcvbuf, sock_net(sk)->ipv4.sysctl_tcp_rmem[1]); sk_sockets_allocated_inc(sk); sk->sk_route_forced_caps = NETIF_F_GSO; @@ -1711,7 +1711,7 @@ int tcp_set_rcvlowat(struct sock *sk, int val) val <<= 1; if (val > sk->sk_rcvbuf) { - sk->sk_rcvbuf = val; + WRITE_ONCE(sk->sk_rcvbuf, val); tcp_sk(sk)->window_clamp = tcp_win_from_space(sk, val); } return 0; diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 16342e043ab3..6995df20710a 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -483,8 +483,9 @@ static void tcp_clamp_window(struct sock *sk) !(sk->sk_userlocks & SOCK_RCVBUF_LOCK) && !tcp_under_memory_pressure(sk) && sk_memory_allocated(sk) < sk_prot_mem_limits(sk, 0)) { - sk->sk_rcvbuf = min(atomic_read(&sk->sk_rmem_alloc), - net->ipv4.sysctl_tcp_rmem[2]); + WRITE_ONCE(sk->sk_rcvbuf, + min(atomic_read(&sk->sk_rmem_alloc), + net->ipv4.sysctl_tcp_rmem[2])); } if (atomic_read(&sk->sk_rmem_alloc) > sk->sk_rcvbuf) tp->rcv_ssthresh = min(tp->window_clamp, 2U * tp->advmss); @@ -648,7 +649,7 @@ void tcp_rcv_space_adjust(struct sock *sk) rcvbuf = min_t(u64, rcvwin * rcvmem, sock_net(sk)->ipv4.sysctl_tcp_rmem[2]); if (rcvbuf > sk->sk_rcvbuf) { - sk->sk_rcvbuf = rcvbuf; + WRITE_ONCE(sk->sk_rcvbuf, rcvbuf); /* Make the window clamp follow along. */ tp->window_clamp = tcp_win_from_space(sk, rcvbuf); -- cgit v1.2.3 From e292f05e0df73f9fcc93329663936e1ded97a988 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 10 Oct 2019 20:17:45 -0700 Subject: tcp: annotate sk->sk_sndbuf lockless reads For the sake of tcp_poll(), there are few places where we fetch sk->sk_sndbuf while this field can change from IRQ or other cpu. We need to add READ_ONCE() annotations, and also make sure write sides use corresponding WRITE_ONCE() to avoid store-tearing. Note that other transports probably need similar fixes. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp.c | 2 +- net/ipv4/tcp_input.c | 3 ++- 2 files changed, 3 insertions(+), 2 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index bc0481aa6633..111853262972 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -450,7 +450,7 @@ void tcp_init_sock(struct sock *sk) icsk->icsk_sync_mss = tcp_sync_mss; - sk->sk_sndbuf = sock_net(sk)->ipv4.sysctl_tcp_wmem[1]; + WRITE_ONCE(sk->sk_sndbuf, sock_net(sk)->ipv4.sysctl_tcp_wmem[1]); WRITE_ONCE(sk->sk_rcvbuf, sock_net(sk)->ipv4.sysctl_tcp_rmem[1]); sk_sockets_allocated_inc(sk); diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 6995df20710a..a2e52ad7cdab 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -359,7 +359,8 @@ static void tcp_sndbuf_expand(struct sock *sk) sndmem *= nr_segs * per_mss; if (sk->sk_sndbuf < sndmem) - sk->sk_sndbuf = min(sndmem, sock_net(sk)->ipv4.sysctl_tcp_wmem[2]); + WRITE_ONCE(sk->sk_sndbuf, + min(sndmem, sock_net(sk)->ipv4.sysctl_tcp_wmem[2])); } /* 2. Tuning advertised window (window_clamp, rcv_ssthresh) -- cgit v1.2.3 From ab4e846a82d0ae00176de19f2db3c5c64f8eb5f2 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 10 Oct 2019 20:17:46 -0700 Subject: tcp: annotate sk->sk_wmem_queued lockless reads For the sake of tcp_poll(), there are few places where we fetch sk->sk_wmem_queued while this field can change from IRQ or other cpu. We need to add READ_ONCE() annotations, and also make sure write sides use corresponding WRITE_ONCE() to avoid store-tearing. sk_wmem_queued_add() helper is added so that we can in the future convert to ADD_ONCE() or equivalent if/when available. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/inet_diag.c | 2 +- net/ipv4/tcp.c | 4 ++-- net/ipv4/tcp_output.c | 14 +++++++------- 3 files changed, 10 insertions(+), 10 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/inet_diag.c b/net/ipv4/inet_diag.c index bbb005eb5218..7dc79b973e6e 100644 --- a/net/ipv4/inet_diag.c +++ b/net/ipv4/inet_diag.c @@ -193,7 +193,7 @@ int inet_sk_diag_fill(struct sock *sk, struct inet_connection_sock *icsk, if (ext & (1 << (INET_DIAG_MEMINFO - 1))) { struct inet_diag_meminfo minfo = { .idiag_rmem = sk_rmem_alloc_get(sk), - .idiag_wmem = sk->sk_wmem_queued, + .idiag_wmem = READ_ONCE(sk->sk_wmem_queued), .idiag_fmem = sk->sk_forward_alloc, .idiag_tmem = sk_wmem_alloc_get(sk), }; diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 111853262972..b2ac4f074e2d 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -659,7 +659,7 @@ static void skb_entail(struct sock *sk, struct sk_buff *skb) tcb->sacked = 0; __skb_header_release(skb); tcp_add_write_queue_tail(sk, skb); - sk->sk_wmem_queued += skb->truesize; + sk_wmem_queued_add(sk, skb->truesize); sk_mem_charge(sk, skb->truesize); if (tp->nonagle & TCP_NAGLE_PUSH) tp->nonagle &= ~TCP_NAGLE_PUSH; @@ -1034,7 +1034,7 @@ new_segment: skb->len += copy; skb->data_len += copy; skb->truesize += copy; - sk->sk_wmem_queued += copy; + sk_wmem_queued_add(sk, copy); sk_mem_charge(sk, copy); skb->ip_summed = CHECKSUM_PARTIAL; WRITE_ONCE(tp->write_seq, tp->write_seq + copy); diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index a115a991dfb5..0488607c5cd3 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -1199,7 +1199,7 @@ static void tcp_queue_skb(struct sock *sk, struct sk_buff *skb) WRITE_ONCE(tp->write_seq, TCP_SKB_CB(skb)->end_seq); __skb_header_release(skb); tcp_add_write_queue_tail(sk, skb); - sk->sk_wmem_queued += skb->truesize; + sk_wmem_queued_add(sk, skb->truesize); sk_mem_charge(sk, skb->truesize); } @@ -1333,7 +1333,7 @@ int tcp_fragment(struct sock *sk, enum tcp_queue tcp_queue, return -ENOMEM; /* We'll just try again later. */ skb_copy_decrypted(buff, skb); - sk->sk_wmem_queued += buff->truesize; + sk_wmem_queued_add(sk, buff->truesize); sk_mem_charge(sk, buff->truesize); nlen = skb->len - len - nsize; buff->truesize += nlen; @@ -1443,7 +1443,7 @@ int tcp_trim_head(struct sock *sk, struct sk_buff *skb, u32 len) if (delta_truesize) { skb->truesize -= delta_truesize; - sk->sk_wmem_queued -= delta_truesize; + sk_wmem_queued_add(sk, -delta_truesize); sk_mem_uncharge(sk, delta_truesize); sock_set_flag(sk, SOCK_QUEUE_SHRUNK); } @@ -1888,7 +1888,7 @@ static int tso_fragment(struct sock *sk, struct sk_buff *skb, unsigned int len, return -ENOMEM; skb_copy_decrypted(buff, skb); - sk->sk_wmem_queued += buff->truesize; + sk_wmem_queued_add(sk, buff->truesize); sk_mem_charge(sk, buff->truesize); buff->truesize += nlen; skb->truesize -= nlen; @@ -2152,7 +2152,7 @@ static int tcp_mtu_probe(struct sock *sk) nskb = sk_stream_alloc_skb(sk, probe_size, GFP_ATOMIC, false); if (!nskb) return -1; - sk->sk_wmem_queued += nskb->truesize; + sk_wmem_queued_add(sk, nskb->truesize); sk_mem_charge(sk, nskb->truesize); skb = tcp_send_head(sk); @@ -3222,7 +3222,7 @@ int tcp_send_synack(struct sock *sk) tcp_rtx_queue_unlink_and_free(skb, sk); __skb_header_release(nskb); tcp_rbtree_insert(&sk->tcp_rtx_queue, nskb); - sk->sk_wmem_queued += nskb->truesize; + sk_wmem_queued_add(sk, nskb->truesize); sk_mem_charge(sk, nskb->truesize); skb = nskb; } @@ -3447,7 +3447,7 @@ static void tcp_connect_queue_skb(struct sock *sk, struct sk_buff *skb) tcb->end_seq += skb->len; __skb_header_release(skb); - sk->sk_wmem_queued += skb->truesize; + sk_wmem_queued_add(sk, skb->truesize); sk_mem_charge(sk, skb->truesize); WRITE_ONCE(tp->write_seq, tcb->end_seq); tp->packets_out += tcp_skb_pcount(skb); -- cgit v1.2.3 From cab209e571a9375f7dc6db69a6c40d2d98e57e3b Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Mon, 14 Oct 2019 06:47:57 -0700 Subject: tcp: fix a possible lockdep splat in tcp_done() syzbot found that if __inet_inherit_port() returns an error, we call tcp_done() after inet_csk_prepare_forced_close(), meaning the socket lock is no longer held. We might fix this in a different way in net-next, but for 5.4 it seems safer to relax the lockdep check. Fixes: d983ea6f16b8 ("tcp: add rcu protection around tp->fastopen_rsk") Signed-off-by: Eric Dumazet Reported-by: syzbot Signed-off-by: David S. Miller --- net/ipv4/tcp.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index b2ac4f074e2d..42187a3b82f4 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -3842,8 +3842,12 @@ void tcp_done(struct sock *sk) { struct request_sock *req; - req = rcu_dereference_protected(tcp_sk(sk)->fastopen_rsk, - lockdep_sock_is_held(sk)); + /* We might be called with a new socket, after + * inet_csk_prepare_forced_close() has been called + * so we can not use lockdep_sock_is_held(sk) + */ + req = rcu_dereference_protected(tcp_sk(sk)->fastopen_rsk, 1); + if (sk->sk_state == TCP_SYN_SENT || sk->sk_state == TCP_SYN_RECV) TCP_INC_STATS(sock_net(sk), TCP_MIB_ATTEMPTFAILS); -- cgit v1.2.3 From 595e0651d0296bad2491a4a29a7a43eae6328b02 Mon Sep 17 00:00:00 2001 From: Stefano Brivio Date: Wed, 16 Oct 2019 20:52:09 +0200 Subject: ipv4: Return -ENETUNREACH if we can't create route but saddr is valid ...instead of -EINVAL. An issue was found with older kernel versions while unplugging a NFS client with pending RPCs, and the wrong error code here prevented it from recovering once link is back up with a configured address. Incidentally, this is not an issue anymore since commit 4f8943f80883 ("SUNRPC: Replace direct task wakeups from softirq context"), included in 5.2-rc7, had the effect of decoupling the forwarding of this error by using SO_ERROR in xs_wake_error(), as pointed out by Benjamin Coddington. To the best of my knowledge, this isn't currently causing any further issue, but the error code doesn't look appropriate anyway, and we might hit this in other paths as well. In detail, as analysed by Gonzalo Siero, once the route is deleted because the interface is down, and can't be resolved and we return -EINVAL here, this ends up, courtesy of inet_sk_rebuild_header(), as the socket error seen by tcp_write_err(), called by tcp_retransmit_timer(). In turn, tcp_write_err() indirectly calls xs_error_report(), which wakes up the RPC pending tasks with a status of -EINVAL. This is then seen by call_status() in the SUN RPC implementation, which aborts the RPC call calling rpc_exit(), instead of handling this as a potentially temporary condition, i.e. as a timeout. Return -EINVAL only if the input parameters passed to ip_route_output_key_hash_rcu() are actually invalid (this is the case if the specified source address is multicast, limited broadcast or all zeroes), but return -ENETUNREACH in all cases where, at the given moment, the given source address doesn't allow resolving the route. While at it, drop the initialisation of err to -ENETUNREACH, which was added to __ip_route_output_key() back then by commit 0315e3827048 ("net: Fix behaviour of unreachable, blackhole and prohibit routes"), but actually had no effect, as it was, and is, overwritten by the fib_lookup() return code assignment, and anyway ignored in all other branches, including the if (fl4->saddr) one: I find this rather confusing, as it would look like -ENETUNREACH is the "default" error, while that statement has no effect. Also note that after commit fc75fc8339e7 ("ipv4: dont create routes on down devices"), we would get -ENETUNREACH if the device is down, but -EINVAL if the source address is specified and we can't resolve the route, and this appears to be rather inconsistent. Reported-by: Stefan Walter Analysed-by: Benjamin Coddington Analysed-by: Gonzalo Siero Signed-off-by: Stefano Brivio Signed-off-by: David S. Miller --- net/ipv4/route.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) (limited to 'net/ipv4') diff --git a/net/ipv4/route.c b/net/ipv4/route.c index 14654876127e..5bc172abd143 100644 --- a/net/ipv4/route.c +++ b/net/ipv4/route.c @@ -2470,14 +2470,17 @@ struct rtable *ip_route_output_key_hash_rcu(struct net *net, struct flowi4 *fl4, int orig_oif = fl4->flowi4_oif; unsigned int flags = 0; struct rtable *rth; - int err = -ENETUNREACH; + int err; if (fl4->saddr) { - rth = ERR_PTR(-EINVAL); if (ipv4_is_multicast(fl4->saddr) || ipv4_is_lbcast(fl4->saddr) || - ipv4_is_zeronet(fl4->saddr)) + ipv4_is_zeronet(fl4->saddr)) { + rth = ERR_PTR(-EINVAL); goto out; + } + + rth = ERR_PTR(-ENETUNREACH); /* I removed check for oif == dev_out->oif here. It was wrong for two reasons: -- cgit v1.2.3 From 5018c59607a511cdee743b629c76206d9c9e6d7b Mon Sep 17 00:00:00 2001 From: Wei Wang Date: Wed, 16 Oct 2019 12:03:15 -0700 Subject: ipv4: fix race condition between route lookup and invalidation Jesse and Ido reported the following race condition: - Received packet A is forwarded and cached dst entry is taken from the nexthop ('nhc->nhc_rth_input'). Calls skb_dst_set() - Given Jesse has busy routers ("ingesting full BGP routing tables from multiple ISPs"), route is added / deleted and rt_cache_flush() is called - Received packet B tries to use the same cached dst entry from t0, but rt_cache_valid() is no longer true and it is replaced in rt_cache_route() by the newer one. This calls dst_dev_put() on the original dst entry which assigns the blackhole netdev to 'dst->dev' - dst_input(skb) is called on packet A and it is dropped due to 'dst->dev' being the blackhole netdev There are 2 issues in the v4 routing code: 1. A per-netns counter is used to do the validation of the route. That means whenever a route is changed in the netns, users of all routes in the netns needs to redo lookup. v6 has an implementation of only updating fn_sernum for routes that are affected. 2. When rt_cache_valid() returns false, rt_cache_route() is called to throw away the current cache, and create a new one. This seems unnecessary because as long as this route does not change, the route cache does not need to be recreated. To fully solve the above 2 issues, it probably needs quite some code changes and requires careful testing, and does not suite for net branch. So this patch only tries to add the deleted cached rt into the uncached list, so user could still be able to use it to receive packets until it's done. Fixes: 95c47f9cf5e0 ("ipv4: call dst_dev_put() properly") Signed-off-by: Wei Wang Reported-by: Ido Schimmel Reported-by: Jesse Hathaway Tested-by: Jesse Hathaway Acked-by: Martin KaFai Lau Cc: David Ahern Reviewed-by: Ido Schimmel Signed-off-by: David S. Miller --- net/ipv4/route.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/ipv4') diff --git a/net/ipv4/route.c b/net/ipv4/route.c index 5bc172abd143..621f83434b24 100644 --- a/net/ipv4/route.c +++ b/net/ipv4/route.c @@ -1482,7 +1482,7 @@ static bool rt_cache_route(struct fib_nh_common *nhc, struct rtable *rt) prev = cmpxchg(p, orig, rt); if (prev == orig) { if (orig) { - dst_dev_put(&orig->dst); + rt_add_uncached_list(orig); dst_release(&orig->dst); } } else { -- cgit v1.2.3 From 9669fffc1415bb0c30e5d2ec98a8e1c3a418cb9c Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 16 Oct 2019 18:00:56 -0700 Subject: net: ensure correct skb->tstamp in various fragmenters Thomas found that some forwarded packets would be stuck in FQ packet scheduler because their skb->tstamp contained timestamps far in the future. We thought we addressed this point in commit 8203e2d844d3 ("net: clear skb->tstamp in forwarding paths") but there is still an issue when/if a packet needs to be fragmented. In order to meet EDT requirements, we have to make sure all fragments get the original skb->tstamp. Note that this original skb->tstamp should be zero in forwarding path, but might have a non zero value in output path if user decided so. Fixes: fb420d5d91c1 ("tcp/fq: move back to CLOCK_MONOTONIC") Signed-off-by: Eric Dumazet Reported-by: Thomas Bartschies Signed-off-by: David S. Miller --- net/ipv4/ip_output.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'net/ipv4') diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c index 28fca408812c..814b9b8882a0 100644 --- a/net/ipv4/ip_output.c +++ b/net/ipv4/ip_output.c @@ -771,6 +771,7 @@ int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb, struct rtable *rt = skb_rtable(skb); unsigned int mtu, hlen, ll_rs; struct ip_fraglist_iter iter; + ktime_t tstamp = skb->tstamp; struct ip_frag_state state; int err = 0; @@ -846,6 +847,7 @@ int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb, ip_fraglist_prepare(skb, &iter); } + skb->tstamp = tstamp; err = output(net, sk, skb); if (!err) @@ -900,6 +902,7 @@ slow_path: /* * Put this fragment into the sending queue. */ + skb2->tstamp = tstamp; err = output(net, sk, skb2); if (err) goto fail; -- cgit v1.2.3