This is a note to let you know that I've just added the patch titled
Bluetooth: SCO: Fix possible circular locking dependency sco_sock_getsockopt
to the 6.2-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary
The filename of the patch is:
bluetooth-sco-fix-possible-circular-locking-dependen.patch
and it can be found in the queue-6.2 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable@xxxxxxxxxxxxxxx> know about it.
commit 84491013e7f9617b31d31dd872d57d28f0bfd23c
Author: Luiz Augusto von Dentz <luiz.von.dentz@xxxxxxxxx>
Date: Thu Mar 30 14:45:03 2023 -0700
Bluetooth: SCO: Fix possible circular locking dependency sco_sock_getsockopt
[ Upstream commit 975abc0c90fc485ff9b4a6afa475c3b1398d5d47 ]
This attempts to fix the following trace:
======================================================
WARNING: possible circular locking dependency detected
6.3.0-rc2-g68fcb3a7bf97 #4706 Not tainted
------------------------------------------------------
sco-tester/31 is trying to acquire lock:
ffff8880025b8070 (&hdev->lock){+.+.}-{3:3}, at:
sco_sock_getsockopt+0x1fc/0xa90
but task is already holding lock:
ffff888001eeb130 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}, at:
sco_sock_getsockopt+0x104/0xa90
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #2 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0}:
lock_sock_nested+0x32/0x80
sco_connect_cfm+0x118/0x4a0
hci_sync_conn_complete_evt+0x1e6/0x3d0
hci_event_packet+0x55c/0x7c0
hci_rx_work+0x34c/0xa00
process_one_work+0x575/0x910
worker_thread+0x89/0x6f0
kthread+0x14e/0x180
ret_from_fork+0x2b/0x50
-> #1 (hci_cb_list_lock){+.+.}-{3:3}:
__mutex_lock+0x13b/0xcc0
hci_sync_conn_complete_evt+0x1ad/0x3d0
hci_event_packet+0x55c/0x7c0
hci_rx_work+0x34c/0xa00
process_one_work+0x575/0x910
worker_thread+0x89/0x6f0
kthread+0x14e/0x180
ret_from_fork+0x2b/0x50
-> #0 (&hdev->lock){+.+.}-{3:3}:
__lock_acquire+0x18cc/0x3740
lock_acquire+0x151/0x3a0
__mutex_lock+0x13b/0xcc0
sco_sock_getsockopt+0x1fc/0xa90
__sys_getsockopt+0xe9/0x190
__x64_sys_getsockopt+0x5b/0x70
do_syscall_64+0x42/0x90
entry_SYSCALL_64_after_hwframe+0x70/0xda
other info that might help us debug this:
Chain exists of:
&hdev->lock --> hci_cb_list_lock --> sk_lock-AF_BLUETOOTH-BTPROTO_SCO
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(sk_lock-AF_BLUETOOTH-BTPROTO_SCO);
lock(hci_cb_list_lock);
lock(sk_lock-AF_BLUETOOTH-BTPROTO_SCO);
lock(&hdev->lock);
*** DEADLOCK ***
1 lock held by sco-tester/31:
#0: ffff888001eeb130 (sk_lock-AF_BLUETOOTH-BTPROTO_SCO){+.+.}-{0:0},
at: sco_sock_getsockopt+0x104/0xa90
Fixes: 248733e87d50 ("Bluetooth: Allow querying of supported offload codecs over SCO socket")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@xxxxxxxxx>
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
diff --git a/net/bluetooth/sco.c b/net/bluetooth/sco.c
index 1111da4e2f2bd..1755f91a66f6a 100644
--- a/net/bluetooth/sco.c
+++ b/net/bluetooth/sco.c
@@ -1129,6 +1129,8 @@ static int sco_sock_getsockopt(struct socket *sock, int level, int optname,
break;
}
+ release_sock(sk);
+
/* find total buffer size required to copy codec + caps */
hci_dev_lock(hdev);
list_for_each_entry(c, &hdev->local_codecs, list) {
@@ -1146,15 +1148,13 @@ static int sco_sock_getsockopt(struct socket *sock, int level, int optname,
buf_len += sizeof(struct bt_codecs);
if (buf_len > len) {
hci_dev_put(hdev);
- err = -ENOBUFS;
- break;
+ return -ENOBUFS;
}
ptr = optval;
if (put_user(num_codecs, ptr)) {
hci_dev_put(hdev);
- err = -EFAULT;
- break;
+ return -EFAULT;
}
ptr += sizeof(num_codecs);
@@ -1194,12 +1194,14 @@ static int sco_sock_getsockopt(struct socket *sock, int level, int optname,
ptr += len;
}
- if (!err && put_user(buf_len, optlen))
- err = -EFAULT;
-
hci_dev_unlock(hdev);
hci_dev_put(hdev);
+ lock_sock(sk);
+
+ if (!err && put_user(buf_len, optlen))
+ err = -EFAULT;
+
break;
default:
