From a51cd2be864a3cc0272359b1995e213341dfc7e7 Mon Sep 17 00:00:00 2001 From: Andre Guedes Date: Fri, 27 Jan 2012 19:42:02 -0300 Subject: Bluetooth: Fix potential deadlock We don't need to use the _sync variant in hci_conn_hold and hci_conn_put to cancel conn->disc_work delayed work. This way we avoid potential deadlocks like this one reported by lockdep. ====================================================== [ INFO: possible circular locking dependency detected ] 3.2.0+ #1 Not tainted ------------------------------------------------------- kworker/u:1/17 is trying to acquire lock: (&hdev->lock){+.+.+.}, at: [] hci_conn_timeout+0x62/0x158 [bluetooth] but task is already holding lock: ((&(&conn->disc_work)->work)){+.+...}, at: [] process_one_work+0x11a/0x2bf which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #2 ((&(&conn->disc_work)->work)){+.+...}: [] lock_acquire+0x8a/0xa7 [] wait_on_work+0x3d/0xaa [] __cancel_work_timer+0xac/0xef [] cancel_delayed_work_sync+0xd/0xf [] smp_chan_create+0xde/0xe6 [bluetooth] [] smp_conn_security+0xa3/0x12d [bluetooth] [] l2cap_connect_cfm+0x237/0x2e8 [bluetooth] [] hci_proto_connect_cfm+0x2d/0x6f [bluetooth] [] hci_event_packet+0x29d1/0x2d60 [bluetooth] [] hci_rx_work+0xd0/0x2e1 [bluetooth] [] process_one_work+0x178/0x2bf [] worker_thread+0xce/0x152 [] kthread+0x95/0x9d [] kernel_thread_helper+0x4/0x10 -> #1 (slock-AF_BLUETOOTH-BTPROTO_L2CAP){+.+...}: [] lock_acquire+0x8a/0xa7 [] _raw_spin_lock_bh+0x36/0x6a [] lock_sock_nested+0x24/0x7f [] lock_sock+0xb/0xd [bluetooth] [] l2cap_chan_connect+0xa9/0x26f [bluetooth] [] l2cap_sock_connect+0xb3/0xff [bluetooth] [] sys_connect+0x69/0x8a [] system_call_fastpath+0x16/0x1b -> #0 (&hdev->lock){+.+.+.}: [] __lock_acquire+0xa80/0xd74 [] lock_acquire+0x8a/0xa7 [] __mutex_lock_common+0x48/0x38e [] mutex_lock_nested+0x2a/0x31 [] hci_conn_timeout+0x62/0x158 [bluetooth] [] process_one_work+0x178/0x2bf [] worker_thread+0xce/0x152 [] kthread+0x95/0x9d [] kernel_thread_helper+0x4/0x10 other info that might help us debug this: Chain exists of: &hdev->lock --> slock-AF_BLUETOOTH-BTPROTO_L2CAP --> (&(&conn->disc_work)->work) Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock((&(&conn->disc_work)->work)); lock(slock-AF_BLUETOOTH-BTPROTO_L2CAP); lock((&(&conn->disc_work)->work)); lock(&hdev->lock); *** DEADLOCK *** 2 locks held by kworker/u:1/17: #0: (hdev->name){.+.+.+}, at: [] process_one_work+0x11a/0x2bf #1: ((&(&conn->disc_work)->work)){+.+...}, at: [] process_one_work+0x11a/0x2bf stack backtrace: Pid: 17, comm: kworker/u:1 Not tainted 3.2.0+ #1 Call Trace: [] print_circular_bug+0x1f8/0x209 [] __lock_acquire+0xa80/0xd74 [] ? arch_local_irq_restore+0x6/0xd [] ? vprintk+0x3f9/0x41e [] lock_acquire+0x8a/0xa7 [] ? hci_conn_timeout+0x62/0x158 [bluetooth] [] __mutex_lock_common+0x48/0x38e [] ? hci_conn_timeout+0x62/0x158 [bluetooth] [] ? __dynamic_pr_debug+0x6d/0x6f [] ? hci_conn_timeout+0x62/0x158 [bluetooth] [] ? trace_hardirqs_off+0xd/0xf [] mutex_lock_nested+0x2a/0x31 [] hci_conn_timeout+0x62/0x158 [bluetooth] [] process_one_work+0x178/0x2bf [] ? process_one_work+0x11a/0x2bf [] ? lock_acquired+0x1d0/0x1df [] ? hci_acl_disconn+0x65/0x65 [bluetooth] [] worker_thread+0xce/0x152 [] ? finish_task_switch+0x45/0xc5 [] ? manage_workers.isra.25+0x16a/0x16a [] kthread+0x95/0x9d [] kernel_thread_helper+0x4/0x10 [] ? retint_restore_args+0x13/0x13 [] ? __init_kthread_worker+0x55/0x55 [] ? gs_change+0x13/0x13 Signed-off-by: Andre Guedes Signed-off-by: Vinicius Costa Gomes Reviewed-by: Ulisses Furquim Acked-by: Marcel Holtmann Signed-off-by: Johan Hedberg --- include/net/bluetooth/hci_core.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'include/net') diff --git a/include/net/bluetooth/hci_core.h b/include/net/bluetooth/hci_core.h index 92b8fea553d..453893b3120 100644 --- a/include/net/bluetooth/hci_core.h +++ b/include/net/bluetooth/hci_core.h @@ -540,7 +540,7 @@ void hci_conn_put_device(struct hci_conn *conn); static inline void hci_conn_hold(struct hci_conn *conn) { atomic_inc(&conn->refcnt); - cancel_delayed_work_sync(&conn->disc_work); + cancel_delayed_work(&conn->disc_work); } static inline void hci_conn_put(struct hci_conn *conn) @@ -559,7 +559,7 @@ static inline void hci_conn_put(struct hci_conn *conn) } else { timeo = msecs_to_jiffies(10); } - cancel_delayed_work_sync(&conn->disc_work); + cancel_delayed_work(&conn->disc_work); queue_delayed_work(conn->hdev->workqueue, &conn->disc_work, timeo); } -- cgit v1.2.3-70-g09d2