From 84602761ca4495dd409be936dfa93ed20c946684 Mon Sep 17 00:00:00 2001 From: Ying Xue Date: Fri, 27 Dec 2013 10:18:28 +0800 Subject: tipc: fix deadlock during socket release A deadlock might occur if name table is withdrawn in socket release routine, and while packets are still being received from bearer. CPU0 CPU1 T0: recv_msg() release() T1: tipc_recv_msg() tipc_withdraw() T2: [grab node lock] [grab port lock] T3: tipc_link_wakeup_ports() tipc_nametbl_withdraw() T4: [grab port lock]* named_cluster_distribute() T5: wakeupdispatch() tipc_link_send() T6: [grab node lock]* The opposite order of holding port lock and node lock on above two different paths may result in a deadlock. If socket lock instead of port lock is used to protect port instance in tipc_withdraw(), the reverse order of holding port lock and node lock will be eliminated, as a result, the deadlock is killed as well. Reported-by: Lars Everbrand Reviewed-by: Erik Hugne Signed-off-by: Ying Xue Signed-off-by: David S. Miller --- net/tipc/port.c | 45 +++++++++++++++------------------------------ net/tipc/port.h | 6 +++--- net/tipc/socket.c | 46 +++++++++++++++++++++++++++++++--------------- 3 files changed, 49 insertions(+), 48 deletions(-) (limited to 'net/tipc') diff --git a/net/tipc/port.c b/net/tipc/port.c index c081a763230..d43f3182b1d 100644 --- a/net/tipc/port.c +++ b/net/tipc/port.c @@ -251,18 +251,15 @@ struct tipc_port *tipc_createport(struct sock *sk, return p_ptr; } -int tipc_deleteport(u32 ref) +int tipc_deleteport(struct tipc_port *p_ptr) { - struct tipc_port *p_ptr; struct sk_buff *buf = NULL; - tipc_withdraw(ref, 0, NULL); - p_ptr = tipc_port_lock(ref); - if (!p_ptr) - return -EINVAL; + tipc_withdraw(p_ptr, 0, NULL); - tipc_ref_discard(ref); - tipc_port_unlock(p_ptr); + spin_lock_bh(p_ptr->lock); + tipc_ref_discard(p_ptr->ref); + spin_unlock_bh(p_ptr->lock); k_cancel_timer(&p_ptr->timer); if (p_ptr->connected) { @@ -704,47 +701,36 @@ int tipc_set_portimportance(u32 ref, unsigned int imp) } -int tipc_publish(u32 ref, unsigned int scope, struct tipc_name_seq const *seq) +int tipc_publish(struct tipc_port *p_ptr, unsigned int scope, + struct tipc_name_seq const *seq) { - struct tipc_port *p_ptr; struct publication *publ; u32 key; - int res = -EINVAL; - p_ptr = tipc_port_lock(ref); - if (!p_ptr) + if (p_ptr->connected) return -EINVAL; + key = p_ptr->ref + p_ptr->pub_count + 1; + if (key == p_ptr->ref) + return -EADDRINUSE; - if (p_ptr->connected) - goto exit; - key = ref + p_ptr->pub_count + 1; - if (key == ref) { - res = -EADDRINUSE; - goto exit; - } publ = tipc_nametbl_publish(seq->type, seq->lower, seq->upper, scope, p_ptr->ref, key); if (publ) { list_add(&publ->pport_list, &p_ptr->publications); p_ptr->pub_count++; p_ptr->published = 1; - res = 0; + return 0; } -exit: - tipc_port_unlock(p_ptr); - return res; + return -EINVAL; } -int tipc_withdraw(u32 ref, unsigned int scope, struct tipc_name_seq const *seq) +int tipc_withdraw(struct tipc_port *p_ptr, unsigned int scope, + struct tipc_name_seq const *seq) { - struct tipc_port *p_ptr; struct publication *publ; struct publication *tpubl; int res = -EINVAL; - p_ptr = tipc_port_lock(ref); - if (!p_ptr) - return -EINVAL; if (!seq) { list_for_each_entry_safe(publ, tpubl, &p_ptr->publications, pport_list) { @@ -771,7 +757,6 @@ int tipc_withdraw(u32 ref, unsigned int scope, struct tipc_name_seq const *seq) } if (list_empty(&p_ptr->publications)) p_ptr->published = 0; - tipc_port_unlock(p_ptr); return res; } diff --git a/net/tipc/port.h b/net/tipc/port.h index 91225359734..34f12bd4074 100644 --- a/net/tipc/port.h +++ b/net/tipc/port.h @@ -116,7 +116,7 @@ int tipc_reject_msg(struct sk_buff *buf, u32 err); void tipc_acknowledge(u32 port_ref, u32 ack); -int tipc_deleteport(u32 portref); +int tipc_deleteport(struct tipc_port *p_ptr); int tipc_portimportance(u32 portref, unsigned int *importance); int tipc_set_portimportance(u32 portref, unsigned int importance); @@ -127,9 +127,9 @@ int tipc_set_portunreliable(u32 portref, unsigned int isunreliable); int tipc_portunreturnable(u32 portref, unsigned int *isunreturnable); int tipc_set_portunreturnable(u32 portref, unsigned int isunreturnable); -int tipc_publish(u32 portref, unsigned int scope, +int tipc_publish(struct tipc_port *p_ptr, unsigned int scope, struct tipc_name_seq const *name_seq); -int tipc_withdraw(u32 portref, unsigned int scope, +int tipc_withdraw(struct tipc_port *p_ptr, unsigned int scope, struct tipc_name_seq const *name_seq); int tipc_connect(u32 portref, struct tipc_portid const *port); diff --git a/net/tipc/socket.c b/net/tipc/socket.c index 3b61851bb92..e741416d1d2 100644 --- a/net/tipc/socket.c +++ b/net/tipc/socket.c @@ -354,7 +354,7 @@ static int release(struct socket *sock) * Delete TIPC port; this ensures no more messages are queued * (also disconnects an active connection & sends a 'FIN-' to peer) */ - res = tipc_deleteport(tport->ref); + res = tipc_deleteport(tport); /* Discard any remaining (connection-based) messages in receive queue */ __skb_queue_purge(&sk->sk_receive_queue); @@ -386,30 +386,46 @@ static int release(struct socket *sock) */ static int bind(struct socket *sock, struct sockaddr *uaddr, int uaddr_len) { + struct sock *sk = sock->sk; struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr; - u32 portref = tipc_sk_port(sock->sk)->ref; + struct tipc_port *tport = tipc_sk_port(sock->sk); + int res = -EINVAL; - if (unlikely(!uaddr_len)) - return tipc_withdraw(portref, 0, NULL); + lock_sock(sk); + if (unlikely(!uaddr_len)) { + res = tipc_withdraw(tport, 0, NULL); + goto exit; + } - if (uaddr_len < sizeof(struct sockaddr_tipc)) - return -EINVAL; - if (addr->family != AF_TIPC) - return -EAFNOSUPPORT; + if (uaddr_len < sizeof(struct sockaddr_tipc)) { + res = -EINVAL; + goto exit; + } + if (addr->family != AF_TIPC) { + res = -EAFNOSUPPORT; + goto exit; + } if (addr->addrtype == TIPC_ADDR_NAME) addr->addr.nameseq.upper = addr->addr.nameseq.lower; - else if (addr->addrtype != TIPC_ADDR_NAMESEQ) - return -EAFNOSUPPORT; + else if (addr->addrtype != TIPC_ADDR_NAMESEQ) { + res = -EAFNOSUPPORT; + goto exit; + } if ((addr->addr.nameseq.type < TIPC_RESERVED_TYPES) && (addr->addr.nameseq.type != TIPC_TOP_SRV) && - (addr->addr.nameseq.type != TIPC_CFG_SRV)) - return -EACCES; + (addr->addr.nameseq.type != TIPC_CFG_SRV)) { + res = -EACCES; + goto exit; + } - return (addr->scope > 0) ? - tipc_publish(portref, addr->scope, &addr->addr.nameseq) : - tipc_withdraw(portref, -addr->scope, &addr->addr.nameseq); + res = (addr->scope > 0) ? + tipc_publish(tport, addr->scope, &addr->addr.nameseq) : + tipc_withdraw(tport, -addr->scope, &addr->addr.nameseq); +exit: + release_sock(sk); + return res; } /** -- cgit v1.2.3-70-g09d2 From 732256b9335f8456623bb772d86c2a24e3cafca2 Mon Sep 17 00:00:00 2001 From: Erik Hugne Date: Tue, 7 Jan 2014 15:51:36 -0500 Subject: tipc: correctly unlink packets from deferred packet queue When we pull a received packet from a link's 'deferred packets' queue for processing, its 'next' pointer is not cleared, and still refers to the next packet in that queue, if any. This is incorrect, but caused no harm before commit 40ba3cdf542a469aaa9083fa041656e59b109b90 ("tipc: message reassembly using fragment chain") was introduced. After that commit, it may sometimes lead to the following oops: general protection fault: 0000 [#1] SMP DEBUG_PAGEALLOC Modules linked in: tipc CPU: 4 PID: 0 Comm: swapper/4 Tainted: G W 3.13.0-rc2+ #6 Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007 task: ffff880017af4880 ti: ffff880017aee000 task.ti: ffff880017aee000 RIP: 0010:[] [] skb_try_coalesce+0x44/0x3d0 RSP: 0018:ffff880016603a78 EFLAGS: 00010212 RAX: 6b6b6b6bd6d6d6d6 RBX: ffff880013106ac0 RCX: ffff880016603ad0 RDX: ffff880016603ad7 RSI: ffff88001223ed00 RDI: ffff880013106ac0 RBP: ffff880016603ab8 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000000 R12: ffff88001223ed00 R13: ffff880016603ad0 R14: 000000000000058c R15: ffff880012297650 FS: 0000000000000000(0000) GS:ffff880016600000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 000000000805b000 CR3: 0000000011f5d000 CR4: 00000000000006e0 Stack: ffff880016603a88 ffffffff810a38ed ffff880016603aa8 ffff88001223ed00 0000000000000001 ffff880012297648 ffff880016603b68 ffff880012297650 ffff880016603b08 ffffffffa0006c51 ffff880016603b08 00ffffffa00005fc Call Trace: [] ? trace_hardirqs_on+0xd/0x10 [] tipc_link_recv_fragment+0xd1/0x1b0 [tipc] [] tipc_recv_msg+0x4e4/0x920 [tipc] [] ? tipc_l2_rcv_msg+0x40/0x250 [tipc] [] tipc_l2_rcv_msg+0xcc/0x250 [tipc] [] ? tipc_l2_rcv_msg+0x40/0x250 [tipc] [] __netif_receive_skb_core+0x80b/0xd00 [] ? __netif_receive_skb_core+0x144/0xd00 [] __netif_receive_skb+0x26/0x70 [] netif_receive_skb+0x2d/0x200 [] napi_gro_receive+0xb0/0x130 [] e1000_clean_rx_irq+0x2c2/0x530 [] e1000_clean+0x266/0x9c0 [] ? notifier_call_chain+0x2b/0x160 [] net_rx_action+0x141/0x310 [] __do_softirq+0xeb/0x480 [] ? _raw_spin_unlock+0x2b/0x40 [] ? handle_fasteoi_irq+0x72/0x100 [] irq_exit+0x96/0xc0 [] do_IRQ+0x63/0xe0 [] common_interrupt+0x6f/0x6f This happens when the last fragment of a message has passed through the the receiving link's 'deferred packets' queue, and at least one other packet was added to that queue while it was there. After the fragment chain with the complete message has been successfully delivered to the receiving socket, it is released. Since 'next' pointer of the last fragment in the released chain now is non-NULL, we get the crash shown above. We fix this by clearing the 'next' pointer of all received packets, including those being pulled from the 'deferred' queue, before they undergo any further processing. Fixes: 40ba3cdf542a4 ("tipc: message reassembly using fragment chain") Signed-off-by: Erik Hugne Reported-by: Ying Xue Reviewed-by: Paul Gortmaker Signed-off-by: Jon Maloy Signed-off-by: David S. Miller --- net/tipc/link.c | 1 + 1 file changed, 1 insertion(+) (limited to 'net/tipc') diff --git a/net/tipc/link.c b/net/tipc/link.c index 69cd9bf3f56..13b98774582 100644 --- a/net/tipc/link.c +++ b/net/tipc/link.c @@ -1498,6 +1498,7 @@ void tipc_recv_msg(struct sk_buff *head, struct tipc_bearer *b_ptr) int type; head = head->next; + buf->next = NULL; /* Ensure bearer is still enabled */ if (unlikely(!b_ptr->active)) -- cgit v1.2.3-70-g09d2