[syzbot] possible deadlock in sco_conn_del

[syzbot]

Hello,

syzbot found the following issue on:

HEAD commit: 696206280c5e Add linux-next specific files for 20220404
git tree: linux-next
console output: https://syzkaller.appspot.com/x/log.txt?x=12c57ff5700000
kernel config: https://syzkaller.appspot.com/x/.config?x=5eb81244dbc8d9b7
dashboard link: https://syzkaller.appspot.com/bug?extid=b825d87fe2d043e3e652
compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2

Unfortunately, I don't have any reproducer for this issue yet.

IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+b825d8...@syzkaller.appspotmail.com

======================================================
WARNING: possible circular locking dependency detected
5.18.0-rc1-next-20220404-syzkaller #0 Not tainted
------------------------------------------------------
syz-executor.4/21629 is trying to acquire lock:
ffff88801acb3130 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}, at: lock_sock include/net/sock.h:1682 [inline]
ffff88801acb3130 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}, at: sco_conn_del+0x131/0x2c0 net/bluetooth/sco.c:197

but task is already holding lock:
ffffffff8d76e388 (hci_cb_list_lock){+.+.}-{3:3}, at: hci_disconn_cfm include/net/bluetooth/hci_core.h:1576 [inline]
ffffffff8d76e388 (hci_cb_list_lock){+.+.}-{3:3}, at: hci_conn_hash_flush+0xda/0x260 net/bluetooth/hci_conn.c:1458

which lock already depends on the new lock.


the existing dependency chain (in reverse order) is:

-> #2 (hci_cb_list_lock){+.+.}-{3:3}:
__mutex_lock_common kernel/locking/mutex.c:600 [inline]
__mutex_lock+0x12f/0x12f0 kernel/locking/mutex.c:733
hci_connect_cfm include/net/bluetooth/hci_core.h:1561 [inline]
hci_remote_features_evt+0x62b/0xa20 net/bluetooth/hci_event.c:3736
hci_event_func net/bluetooth/hci_event.c:6890 [inline]
hci_event_packet+0x7c1/0xf60 net/bluetooth/hci_event.c:6939
hci_rx_work+0x522/0xd60 net/bluetooth/hci_core.c:3819
process_one_work+0x996/0x1610 kernel/workqueue.c:2289
worker_thread+0x665/0x1080 kernel/workqueue.c:2436
kthread+0x2e9/0x3a0 kernel/kthread.c:376
ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:298

-> #1 (&hdev->lock){+.+.}-{3:3}:
__mutex_lock_common kernel/locking/mutex.c:600 [inline]
__mutex_lock+0x12f/0x12f0 kernel/locking/mutex.c:733
sco_sock_connect+0x1eb/0xa80 net/bluetooth/sco.c:593
__sys_connect_file+0x14f/0x190 net/socket.c:1900
__sys_connect+0x161/0x190 net/socket.c:1917
__do_sys_connect net/socket.c:1927 [inline]
__se_sys_connect net/socket.c:1924 [inline]
__x64_sys_connect+0x6f/0xb0 net/socket.c:1924
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x35/0x80 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae

-> #0 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}:
check_prev_add kernel/locking/lockdep.c:3096 [inline]
check_prevs_add kernel/locking/lockdep.c:3219 [inline]
validate_chain kernel/locking/lockdep.c:3834 [inline]
__lock_acquire+0x2abe/0x5660 kernel/locking/lockdep.c:5060
lock_acquire kernel/locking/lockdep.c:5672 [inline]
lock_acquire+0x1ab/0x510 kernel/locking/lockdep.c:5637
lock_sock_nested+0x36/0xf0 net/core/sock.c:3312
lock_sock include/net/sock.h:1682 [inline]
sco_conn_del+0x131/0x2c0 net/bluetooth/sco.c:197
sco_disconn_cfm+0x71/0xb0 net/bluetooth/sco.c:1379
hci_disconn_cfm include/net/bluetooth/hci_core.h:1579 [inline]
hci_conn_hash_flush+0x127/0x260 net/bluetooth/hci_conn.c:1458
hci_dev_close_sync+0x567/0x1140 net/bluetooth/hci_sync.c:4121
hci_dev_do_close+0x32/0x70 net/bluetooth/hci_core.c:553
hci_unregister_dev+0x1d0/0x550 net/bluetooth/hci_core.c:2687
vhci_release+0x7c/0xf0 drivers/bluetooth/hci_vhci.c:568
__fput+0x277/0x9d0 fs/file_table.c:317
task_work_run+0xdd/0x1a0 kernel/task_work.c:164
exit_task_work include/linux/task_work.h:37 [inline]
do_exit+0xaff/0x2a00 kernel/exit.c:795
do_group_exit+0xd2/0x2f0 kernel/exit.c:925
__do_sys_exit_group kernel/exit.c:936 [inline]
__se_sys_exit_group kernel/exit.c:934 [inline]
__x64_sys_exit_group+0x3a/0x50 kernel/exit.c:934
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x35/0x80 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae

other info that might help us debug this:

Chain exists of:
sk_lock-AF_BLUETOOTH-BTPROTO_SCO --> &hdev->lock --> hci_cb_list_lock

Possible unsafe locking scenario:

CPU0 CPU1
---- ----
lock(hci_cb_list_lock);
lock(&hdev->lock);
lock(hci_cb_list_lock);
lock(sk_lock-AF_BLUETOOTH-BTPROTO_SCO);

*** DEADLOCK ***

3 locks held by syz-executor.4/21629:
#0: ffff88801e84d048 (&hdev->req_lock){+.+.}-{3:3}, at: hci_dev_do_close+0x2a/0x70 net/bluetooth/hci_core.c:551
#1: ffff88801e84c078 (&hdev->lock){+.+.}-{3:3}, at: hci_dev_close_sync+0x26d/0x1140 net/bluetooth/hci_sync.c:4108
#2: ffffffff8d76e388 (hci_cb_list_lock){+.+.}-{3:3}, at: hci_disconn_cfm include/net/bluetooth/hci_core.h:1576 [inline]
#2: ffffffff8d76e388 (hci_cb_list_lock){+.+.}-{3:3}, at: hci_conn_hash_flush+0xda/0x260 net/bluetooth/hci_conn.c:1458

stack backtrace:
CPU: 1 PID: 21629 Comm: syz-executor.4 Not tainted 5.18.0-rc1-next-20220404-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
<TASK>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106
check_noncircular+0x25f/0x2e0 kernel/locking/lockdep.c:2176
check_prev_add kernel/locking/lockdep.c:3096 [inline]
check_prevs_add kernel/locking/lockdep.c:3219 [inline]
validate_chain kernel/locking/lockdep.c:3834 [inline]
__lock_acquire+0x2abe/0x5660 kernel/locking/lockdep.c:5060
lock_acquire kernel/locking/lockdep.c:5672 [inline]
lock_acquire+0x1ab/0x510 kernel/locking/lockdep.c:5637
lock_sock_nested+0x36/0xf0 net/core/sock.c:3312
lock_sock include/net/sock.h:1682 [inline]
sco_conn_del+0x131/0x2c0 net/bluetooth/sco.c:197
sco_disconn_cfm+0x71/0xb0 net/bluetooth/sco.c:1379
hci_disconn_cfm include/net/bluetooth/hci_core.h:1579 [inline]
hci_conn_hash_flush+0x127/0x260 net/bluetooth/hci_conn.c:1458
hci_dev_close_sync+0x567/0x1140 net/bluetooth/hci_sync.c:4121
hci_dev_do_close+0x32/0x70 net/bluetooth/hci_core.c:553
hci_unregister_dev+0x1d0/0x550 net/bluetooth/hci_core.c:2687
vhci_release+0x7c/0xf0 drivers/bluetooth/hci_vhci.c:568
__fput+0x277/0x9d0 fs/file_table.c:317
task_work_run+0xdd/0x1a0 kernel/task_work.c:164
exit_task_work include/linux/task_work.h:37 [inline]
do_exit+0xaff/0x2a00 kernel/exit.c:795
do_group_exit+0xd2/0x2f0 kernel/exit.c:925
__do_sys_exit_group kernel/exit.c:936 [inline]
__se_sys_exit_group kernel/exit.c:934 [inline]
__x64_sys_exit_group+0x3a/0x50 kernel/exit.c:934
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x35/0x80 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x7f677aa89049
Code: Unable to access opcode bytes at RIP 0x7f677aa8901f.
RSP: 002b:00007ffc2aa139c8 EFLAGS: 00000246 ORIG_RAX: 00000000000000e7
RAX: ffffffffffffffda RBX: 0000000000000064 RCX: 00007f677aa89049
RDX: 0000000000000000 RSI: 0000000000000001 RDI: 0000000000000000
RBP: 00007f677aae225c R08: 000000000000000c R09: 00005555561573b8
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000016
R13: 00007ffc2aa14ca0 R14: 00005555561573b8 R15: 00007ffc2aa15da0
</TASK>


---
This report is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzk...@googlegroups.com.

syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.

[syzbot]

syzbot has found a reproducer for the following issue on:

HEAD commit: d12d7e1cfe38 Add linux-next specific files for 20220411
git tree: linux-next
console output: https://syzkaller.appspot.com/x/log.txt?x=162e29a8f00000
kernel config: https://syzkaller.appspot.com/x/.config?x=58fcaf7d8df169a6

dashboard link: https://syzkaller.appspot.com/bug?extid=b825d87fe2d043e3e652
compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2

syz repro: https://syzkaller.appspot.com/x/repro.syz?x=16a2ff0f700000
C reproducer: https://syzkaller.appspot.com/x/repro.c?x=149fd2df700000

IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+b825d8...@syzkaller.appspotmail.com

======================================================
WARNING: possible circular locking dependency detected

5.18.0-rc1-next-20220411-syzkaller #0 Not tainted
------------------------------------------------------
syz-executor104/3623 is trying to acquire lock:
ffff88807bd9b130 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}, at: lock_sock include/net/sock.h:1682 [inline]
ffff88807bd9b130 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}, at: sco_conn_del+0x131/0x2c0 net/bluetooth/sco.c:197

but task is already holding lock:

ffffffff8d777e08 (hci_cb_list_lock){+.+.}-{3:3}, at: hci_disconn_cfm include/net/bluetooth/hci_core.h:1580 [inline]
ffffffff8d777e08 (hci_cb_list_lock){+.+.}-{3:3}, at: hci_conn_hash_flush+0xda/0x260 net/bluetooth/hci_conn.c:1458

which lock already depends on the new lock.


the existing dependency chain (in reverse order) is:

-> #2 (hci_cb_list_lock){+.+.}-{3:3}:

__mutex_lock_common kernel/locking/mutex.c:603 [inline]
__mutex_lock+0x12f/0x1350 kernel/locking/mutex.c:747
hci_connect_cfm include/net/bluetooth/hci_core.h:1565 [inline]

hci_remote_features_evt+0x62b/0xa20 net/bluetooth/hci_event.c:3736
hci_event_func net/bluetooth/hci_event.c:6890 [inline]
hci_event_packet+0x7c1/0xf60 net/bluetooth/hci_event.c:6939
hci_rx_work+0x522/0xd60 net/bluetooth/hci_core.c:3819
process_one_work+0x996/0x1610 kernel/workqueue.c:2289
worker_thread+0x665/0x1080 kernel/workqueue.c:2436
kthread+0x2e9/0x3a0 kernel/kthread.c:376
ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:298

-> #1 (&hdev->lock){+.+.}-{3:3}:

__mutex_lock_common kernel/locking/mutex.c:603 [inline]
__mutex_lock+0x12f/0x1350 kernel/locking/mutex.c:747

sco_sock_connect+0x1eb/0xa80 net/bluetooth/sco.c:593
__sys_connect_file+0x14f/0x190 net/socket.c:1900
__sys_connect+0x161/0x190 net/socket.c:1917
__do_sys_connect net/socket.c:1927 [inline]
__se_sys_connect net/socket.c:1924 [inline]
__x64_sys_connect+0x6f/0xb0 net/socket.c:1924
do_syscall_x64 arch/x86/entry/common.c:50 [inline]

do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80

entry_SYSCALL_64_after_hwframe+0x44/0xae

-> #0 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}:

check_prev_add kernel/locking/lockdep.c:3095 [inline]
check_prevs_add kernel/locking/lockdep.c:3214 [inline]
validate_chain kernel/locking/lockdep.c:3829 [inline]
__lock_acquire+0x2abe/0x5660 kernel/locking/lockdep.c:5053
lock_acquire kernel/locking/lockdep.c:5665 [inline]
lock_acquire+0x1ab/0x570 kernel/locking/lockdep.c:5630
lock_sock_nested+0x36/0xf0 net/core/sock.c:3314

lock_sock include/net/sock.h:1682 [inline]
sco_conn_del+0x131/0x2c0 net/bluetooth/sco.c:197
sco_disconn_cfm+0x71/0xb0 net/bluetooth/sco.c:1379

hci_disconn_cfm include/net/bluetooth/hci_core.h:1583 [inline]
hci_conn_hash_flush+0x127/0x260 net/bluetooth/hci_conn.c:1458
hci_dev_close_sync+0x567/0x1140 net/bluetooth/hci_sync.c:4152
hci_dev_do_close+0x32/0x70 net/bluetooth/hci_core.c:553
hci_rfkill_set_block+0x19c/0x1d0 net/bluetooth/hci_core.c:935
rfkill_set_block+0x1f9/0x540 net/rfkill/core.c:345
rfkill_fop_write+0x2c3/0x570 net/rfkill/core.c:1286
vfs_write+0x269/0xac0 fs/read_write.c:589
ksys_write+0x1e8/0x250 fs/read_write.c:644
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80

entry_SYSCALL_64_after_hwframe+0x44/0xae

other info that might help us debug this:

Chain exists of:
sk_lock-AF_BLUETOOTH-BTPROTO_SCO --> &hdev->lock --> hci_cb_list_lock

Possible unsafe locking scenario:

CPU0 CPU1
---- ----
lock(hci_cb_list_lock);
lock(&hdev->lock);
lock(hci_cb_list_lock);
lock(sk_lock-AF_BLUETOOTH-BTPROTO_SCO);

*** DEADLOCK ***

4 locks held by syz-executor104/3623:
#0: ffffffff8d9583a8 (rfkill_global_mutex){+.+.}-{3:3}, at: rfkill_fop_write+0x15c/0x570 net/rfkill/core.c:1278
#1: ffff888022e89048 (&hdev->req_lock){+.+.}-{3:3}, at: hci_dev_do_close+0x2a/0x70 net/bluetooth/hci_core.c:551
#2: ffff888022e88078 (&hdev->lock){+.+.}-{3:3}, at: hci_dev_close_sync+0x26d/0x1140 net/bluetooth/hci_sync.c:4139
#3: ffffffff8d777e08 (hci_cb_list_lock){+.+.}-{3:3}, at: hci_disconn_cfm include/net/bluetooth/hci_core.h:1580 [inline]
#3: ffffffff8d777e08 (hci_cb_list_lock){+.+.}-{3:3}, at: hci_conn_hash_flush+0xda/0x260 net/bluetooth/hci_conn.c:1458

stack backtrace:
CPU: 0 PID: 3623 Comm: syz-executor104 Not tainted 5.18.0-rc1-next-20220411-syzkaller #0

Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
<TASK>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106

check_noncircular+0x25f/0x2e0 kernel/locking/lockdep.c:2175
check_prev_add kernel/locking/lockdep.c:3095 [inline]
check_prevs_add kernel/locking/lockdep.c:3214 [inline]
validate_chain kernel/locking/lockdep.c:3829 [inline]
__lock_acquire+0x2abe/0x5660 kernel/locking/lockdep.c:5053
lock_acquire kernel/locking/lockdep.c:5665 [inline]
lock_acquire+0x1ab/0x570 kernel/locking/lockdep.c:5630
lock_sock_nested+0x36/0xf0 net/core/sock.c:3314

lock_sock include/net/sock.h:1682 [inline]
sco_conn_del+0x131/0x2c0 net/bluetooth/sco.c:197
sco_disconn_cfm+0x71/0xb0 net/bluetooth/sco.c:1379

hci_disconn_cfm include/net/bluetooth/hci_core.h:1583 [inline]
hci_conn_hash_flush+0x127/0x260 net/bluetooth/hci_conn.c:1458
hci_dev_close_sync+0x567/0x1140 net/bluetooth/hci_sync.c:4152
hci_dev_do_close+0x32/0x70 net/bluetooth/hci_core.c:553
hci_rfkill_set_block+0x19c/0x1d0 net/bluetooth/hci_core.c:935
rfkill_set_block+0x1f9/0x540 net/rfkill/core.c:345
rfkill_fop_write+0x2c3/0x570 net/rfkill/core.c:1286
vfs_write+0x269/0xac0 fs/read_write.c:589
ksys_write+0x1e8/0x250 fs/read_write.c:644
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x7f841a998629
Code: 28 00 00 00 75 05 48 83 c4 28 c3 e8 d1 15 00 00 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f841a1472f8 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
RAX: ffffffffffffffda RBX: 00007f841aa214c0 RCX: 00007f841a998629
RDX: 0000000000000008 RSI: 0000000020000080 RDI: 0000000000000003
RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 00007fff20c0bcce
R13: 00007fff20c0bccf R14: 00007f841aa214c8 R15: 0000000000022000
</TASK>

[syzbot]

syzbot has bisected this issue to:

commit 92b8aa6d18f7a9ae36a0f71d31742aeef201207a
Author: Ying Hsu <yin...@chromium.org>
Date: Sat Mar 26 07:09:28 2022 +0000

Bluetooth: fix dangling sco_conn and use-after-free in sco_sock_timeout

bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=10fada04f00000
start commit: d12d7e1cfe38 Add linux-next specific files for 20220411
git tree: linux-next
final oops: https://syzkaller.appspot.com/x/report.txt?x=12fada04f00000
console output: https://syzkaller.appspot.com/x/log.txt?x=14fada04f00000

kernel config: https://syzkaller.appspot.com/x/.config?x=58fcaf7d8df169a6
dashboard link: https://syzkaller.appspot.com/bug?extid=b825d87fe2d043e3e652

syz repro: https://syzkaller.appspot.com/x/repro.syz?x=16a2ff0f700000
C reproducer: https://syzkaller.appspot.com/x/repro.c?x=149fd2df700000

Reported-by: syzbot+b825d8...@syzkaller.appspotmail.com
Fixes: 92b8aa6d18f7 ("Bluetooth: fix dangling sco_conn and use-after-free in sco_sock_timeout")

For information about bisection process see: https://goo.gl/tpsmEJ#bisection

[Hillf Danton]

On Mon, 11 Apr 2022 11:51:20 -0700

In bid to ix deadlock, move hci_connect_cfm() out of the section under
hdev lock in the path handling hci event.

#syz test https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/ d12d7e1cfe38

--- y/net/bluetooth/hci_event.c
+++ x/net/bluetooth/hci_event.c
@@ -3733,8 +3733,10 @@ static void hci_remote_features_evt(stru

if (!hci_outgoing_auth_needed(hdev, conn)) {
conn->state = BT_CONNECTED;
+ hci_dev_unlock(hdev);
hci_connect_cfm(conn, ev->status);
hci_conn_drop(conn);
+ return;
}

unlock:
--

[syzbot]

Hello,

syzbot has tested the proposed patch but the reproducer is still triggering an issue:
possible deadlock in sco_conn_del

======================================================
WARNING: possible circular locking dependency detected

5.18.0-rc1-next-20220411-syzkaller-dirty #0 Not tainted
------------------------------------------------------
syz-executor129/4082 is trying to acquire lock:
ffff88807c11d130 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}, at: lock_sock include/net/sock.h:1682 [inline]
ffff88807c11d130 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}, at: sco_conn_del+0x131/0x2c0 net/bluetooth/sco.c:197

but task is already holding lock:
ffffffff8d777e08 (hci_cb_list_lock){+.+.}-{3:3}, at: hci_disconn_cfm include/net/bluetooth/hci_core.h:1580 [inline]
ffffffff8d777e08 (hci_cb_list_lock){+.+.}-{3:3}, at: hci_conn_hash_flush+0xda/0x260 net/bluetooth/hci_conn.c:1458

which lock already depends on the new lock.


the existing dependency chain (in reverse order) is:

-> #2 (hci_cb_list_lock){+.+.}-{3:3}:
__mutex_lock_common kernel/locking/mutex.c:603 [inline]
__mutex_lock+0x12f/0x1350 kernel/locking/mutex.c:747

hci_connect_cfm+0x26/0x140 include/net/bluetooth/hci_core.h:1565
le_conn_complete_evt.part.0+0x145e/0x17e0 net/bluetooth/hci_event.c:5666
le_conn_complete_evt net/bluetooth/hci_event.c:5692 [inline]
hci_le_conn_complete_evt+0x2fd/0x390 net/bluetooth/hci_event.c:5692
hci_le_meta_evt+0x2bd/0x520 net/bluetooth/hci_event.c:6559
hci_event_func net/bluetooth/hci_event.c:6889 [inline]
hci_event_packet+0x5da/0xf60 net/bluetooth/hci_event.c:6941

4 locks held by syz-executor129/4082:

#0: ffffffff8d9583a8 (rfkill_global_mutex){+.+.}-{3:3}, at: rfkill_fop_write+0x15c/0x570 net/rfkill/core.c:1278

#1: ffff88801f891048 (&hdev->req_lock){+.+.}-{3:3}, at: hci_dev_do_close+0x2a/0x70 net/bluetooth/hci_core.c:551
#2: ffff88801f890078 (&hdev->lock){+.+.}-{3:3}, at: hci_dev_close_sync+0x26d/0x1140 net/bluetooth/hci_sync.c:4139

#3: ffffffff8d777e08 (hci_cb_list_lock){+.+.}-{3:3}, at: hci_disconn_cfm include/net/bluetooth/hci_core.h:1580 [inline]
#3: ffffffff8d777e08 (hci_cb_list_lock){+.+.}-{3:3}, at: hci_conn_hash_flush+0xda/0x260 net/bluetooth/hci_conn.c:1458

stack backtrace:

CPU: 1 PID: 4082 Comm: syz-executor129 Not tainted 5.18.0-rc1-next-20220411-syzkaller-dirty #0

RIP: 0033:0x7fab5d4df609

Code: 28 00 00 00 75 05 48 83 c4 28 c3 e8 d1 15 00 00 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48

RSP: 002b:00007fab5cc8e2f8 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
RAX: ffffffffffffffda RBX: 00007fab5d5684c0 RCX: 00007fab5d4df609

RDX: 0000000000000008 RSI: 0000000020000080 RDI: 0000000000000003
RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000

R10: 0000000000000000 R11: 0000000000000246 R12: 00007ffeea3e1ffe
R13: 00007ffeea3e1fff R14: 00007fab5d5684c8 R15: 0000000000022000
</TASK>


Tested on:

commit: d12d7e1c Add linux-next specific files for 20220411
git tree: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/
console output: https://syzkaller.appspot.com/x/log.txt?x=14908904f00000

kernel config: https://syzkaller.appspot.com/x/.config?x=58fcaf7d8df169a6
dashboard link: https://syzkaller.appspot.com/bug?extid=b825d87fe2d043e3e652
compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2

patch: https://syzkaller.appspot.com/x/patch.diff?x=16223d2f700000

[Hillf Danton]

On Mon, 11 Apr 2022 16:37:07 -0700

In bid to fix deadlock, kill sock in workqueu to avoid lock_sock() with
hci_cb_list_lock held.

#syz test https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/ d12d7e1cfe38

--- y/net/bluetooth/sco.c
+++ x/net/bluetooth/sco.c
@@ -49,6 +49,8 @@ struct sco_conn {
struct sock *sk;

struct delayed_work timeout_work;
+ struct work_struct free_work;
+ int err;

unsigned int mtu;
};
@@ -176,6 +178,23 @@ static void sco_chan_del(struct sock *sk
sock_set_flag(sk, SOCK_ZAPPED);
}

+static void sco_conn_free_workfn(struct work_struct *work)
+{
+ struct sco_conn *conn = container_of(work, struct sco_conn, free_work);
+ struct sock *sk = conn->sk;
+
+ lock_sock(sk);
+ sco_sock_clear_timer(sk);
+ sco_chan_del(sk, conn->err);
+ release_sock(sk);
+ /* pair with sock_hold(sk) in sco_conn_del() */
+ sock_put(sk);
+
+ /* Ensure no more work items will run before freeing conn */
+ cancel_delayed_work_sync(&conn->timeout_work);
+ kfree(conn);
+}
+
static void sco_conn_del(struct hci_conn *hcon, int err)
{
struct sco_conn *conn = hcon->sco_data;
@@ -194,11 +213,12 @@ static void sco_conn_del(struct hci_conn
sco_conn_unlock(conn);

if (sk) {
- lock_sock(sk);
- sco_sock_clear_timer(sk);
- sco_chan_del(sk, err);
- release_sock(sk);
- sock_put(sk);
+ /* kill sk in workqueue just to avoid lock_sock() with hci_cb_list_lock held */
+ conn->err = err;
+ INIT_WORK(&conn->free_work, sco_conn_free_workfn);
+ queue_work(system_unbound_wq, &conn->free_work);
+ hcon->sco_data = NULL;
+ return;
}

/* Ensure no more work items will run before freeing conn. */
--

[syzbot]

Hello,

syzbot has tested the proposed patch and the reproducer did not trigger any issue:

Reported-and-tested-by: syzbot+b825d8...@syzkaller.appspotmail.com

Tested on:

commit: d12d7e1c Add linux-next specific files for 20220411
git tree: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/

kernel config: https://syzkaller.appspot.com/x/.config?x=58fcaf7d8df169a6
dashboard link: https://syzkaller.appspot.com/bug?extid=b825d87fe2d043e3e652
compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2

patch: https://syzkaller.appspot.com/x/patch.diff?x=13879aa8f00000

Note: testing is done by a robot and is best-effort only.

[syzbot]

syzbot suspects this issue was fixed by commit:

commit a2ac591cb4d83e1f2d4b4adb3c14b2c79764650a
Author: Ruihan Li <lrh...@pku.edu.cn>
Date: Wed May 3 13:39:36 2023 +0000

Bluetooth: Fix UAF in hci_conn_hash_flush again

bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=13755717280000
start commit: e4cf7c25bae5 Merge tag 'kbuild-fixes-v6.2' of git://git.ke..
git tree: upstream
kernel config: https://syzkaller.appspot.com/x/.config?x=555d27e379d75ff1
dashboard link: https://syzkaller.appspot.com/bug?extid=b825d87fe2d043e3e652
syz repro: https://syzkaller.appspot.com/x/repro.syz?x=10052058480000
C reproducer: https://syzkaller.appspot.com/x/repro.c?x=1190687c480000

If the result looks correct, please mark the issue as fixed by replying with:

#syz fix: Bluetooth: Fix UAF in hci_conn_hash_flush again

[Dmitry Vyukov]

Looks reasonable: