syzbot

======================================================
WARNING: possible circular locking dependency detected
syzkaller #0 Not tainted
------------------------------------------------------
kworker/u5:7/4200 is trying to acquire lock:
ffff88801fc50b98 ((work_completion)(&hdev->tx_work)){+.+.}-{0:0}, at: __flush_work+0xc1/0x1b0 kernel/workqueue.c:3090

but task is already holding lock:
ffff88801fc50ff0 (&hdev->req_lock){+.+.}-{3:3}, at: hci_dev_do_open+0xa4/0x1270 net/bluetooth/hci_core.c:1446

which lock already depends on the new lock.


the existing dependency chain (in reverse order) is:

-> #3 (&hdev->req_lock){+.+.}-{3:3}:
       __mutex_lock_common+0x1eb/0x2390 kernel/locking/mutex.c:596
       __mutex_lock kernel/locking/mutex.c:729 [inline]
       mutex_lock_nested+0x17/0x20 kernel/locking/mutex.c:743
       hci_dev_do_close+0x5f/0x1030 net/bluetooth/hci_core.c:1737
       hci_rfkill_set_block+0x10a/0x190 net/bluetooth/hci_core.c:2235
       rfkill_set_block+0x1c6/0x420 net/rfkill/core.c:345
       rfkill_epo+0x75/0x170 net/rfkill/core.c:454
       __rfkill_handle_global_op net/rfkill/input.c:60 [inline]
       rfkill_op_handler+0x76/0x220 net/rfkill/input.c:108
       process_one_work+0x863/0x1000 kernel/workqueue.c:2310
       worker_thread+0xaa8/0x12a0 kernel/workqueue.c:2457
       kthread+0x436/0x520 kernel/kthread.c:334
       ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:287

-> #2 (rfkill_global_mutex){+.+.}-{3:3}:
       __mutex_lock_common+0x1eb/0x2390 kernel/locking/mutex.c:596
       __mutex_lock kernel/locking/mutex.c:729 [inline]
       mutex_lock_nested+0x17/0x20 kernel/locking/mutex.c:743
       rfkill_register+0x33/0x8a0 net/rfkill/core.c:1045
       hci_register_dev+0x452/0x970 net/bluetooth/hci_core.c:3960
       __vhci_create_device drivers/bluetooth/hci_vhci.c:129 [inline]
       vhci_create_device+0x32c/0x5c0 drivers/bluetooth/hci_vhci.c:153
       vhci_get_user drivers/bluetooth/hci_vhci.c:210 [inline]
       vhci_write+0x391/0x450 drivers/bluetooth/hci_vhci.c:290
       call_write_iter include/linux/fs.h:2173 [inline]
       new_sync_write fs/read_write.c:507 [inline]
       vfs_write+0x712/0xd00 fs/read_write.c:594
       ksys_write+0x14d/0x250 fs/read_write.c:647
       do_syscall_x64 arch/x86/entry/common.c:50 [inline]
       do_syscall_64+0x4c/0xa0 arch/x86/entry/common.c:80
       entry_SYSCALL_64_after_hwframe+0x66/0xd0

-> #1 (&data->open_mutex){+.+.}-{3:3}:
       __mutex_lock_common+0x1eb/0x2390 kernel/locking/mutex.c:596
       __mutex_lock kernel/locking/mutex.c:729 [inline]
       mutex_lock_nested+0x17/0x20 kernel/locking/mutex.c:743
       vhci_send_frame+0x88/0x100 drivers/bluetooth/hci_vhci.c:71
       hci_send_frame+0x1a9/0x2e0 net/bluetooth/hci_core.c:4256
       hci_sched_acl_pkt net/bluetooth/hci_core.c:4781 [inline]
       hci_sched_acl net/bluetooth/hci_core.c:4866 [inline]
       hci_tx_work+0x9f9/0x1710 net/bluetooth/hci_core.c:4932
       process_one_work+0x863/0x1000 kernel/workqueue.c:2310
       worker_thread+0xaa8/0x12a0 kernel/workqueue.c:2457
       kthread+0x436/0x520 kernel/kthread.c:334
       ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:287

-> #0 ((work_completion)(&hdev->tx_work)){+.+.}-{0:0}:
       check_prev_add kernel/locking/lockdep.c:3053 [inline]
       check_prevs_add kernel/locking/lockdep.c:3172 [inline]
       validate_chain kernel/locking/lockdep.c:3788 [inline]
       __lock_acquire+0x2c33/0x7c60 kernel/locking/lockdep.c:5012
       lock_acquire+0x197/0x3f0 kernel/locking/lockdep.c:5623
       __flush_work+0xdd/0x1b0 kernel/workqueue.c:3090
       hci_dev_do_open+0xc58/0x1270 net/bluetooth/hci_core.c:1619
       hci_power_on+0x1c1/0x610 net/bluetooth/hci_core.c:2265
       process_one_work+0x863/0x1000 kernel/workqueue.c:2310
       worker_thread+0xaa8/0x12a0 kernel/workqueue.c:2457
       kthread+0x436/0x520 kernel/kthread.c:334
       ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:287

other info that might help us debug this:

Chain exists of:
  (work_completion)(&hdev->tx_work) --> rfkill_global_mutex --> &hdev->req_lock

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock(&hdev->req_lock);
                               lock(rfkill_global_mutex);
                               lock(&hdev->req_lock);
  lock((work_completion)(&hdev->tx_work));

 *** DEADLOCK ***

3 locks held by kworker/u5:7/4200:
 #0: ffff88801f9db938 ((wq_completion)hci5){+.+.}-{0:0}, at: process_one_work+0x760/0x1000 kernel/workqueue.c:-1
 #1: ffffc9000302fd00 ((work_completion)(&hdev->power_on)){+.+.}-{0:0}, at: process_one_work+0x7a3/0x1000 kernel/workqueue.c:2285
 #2: ffff88801fc50ff0 (&hdev->req_lock){+.+.}-{3:3}, at: hci_dev_do_open+0xa4/0x1270 net/bluetooth/hci_core.c:1446

stack backtrace:
CPU: 1 PID: 4200 Comm: kworker/u5:7 Not tainted syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/02/2025
Workqueue: hci5 hci_power_on
Call Trace:
 <TASK>
 dump_stack_lvl+0x168/0x230 lib/dump_stack.c:106
 check_noncircular+0x274/0x310 kernel/locking/lockdep.c:2133
 check_prev_add kernel/locking/lockdep.c:3053 [inline]
 check_prevs_add kernel/locking/lockdep.c:3172 [inline]
 validate_chain kernel/locking/lockdep.c:3788 [inline]
 __lock_acquire+0x2c33/0x7c60 kernel/locking/lockdep.c:5012
 lock_acquire+0x197/0x3f0 kernel/locking/lockdep.c:5623
 __flush_work+0xdd/0x1b0 kernel/workqueue.c:3090
 hci_dev_do_open+0xc58/0x1270 net/bluetooth/hci_core.c:1619
 hci_power_on+0x1c1/0x610 net/bluetooth/hci_core.c:2265
 process_one_work+0x863/0x1000 kernel/workqueue.c:2310
 worker_thread+0xaa8/0x12a0 kernel/workqueue.c:2457
 kthread+0x436/0x520 kernel/kthread.c:334
 ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:287
 </TASK>