Received-SPF: pass (google.com: domain of linux-bluetooth-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Luiz Augusto von Dentz <luiz.dentz@gmail.com>
To:     linux-bluetooth@vger.kernel.org
Subject: [PATCH] Bluetooth: ISO: Fix possible circular locking dependency
Date:   Tue, 10 Jan 2023 17:22:54 -0800
Message-Id: <20230111012254.3724082-2-luiz.dentz@gmail.com>
In-Reply-To: <20230111012254.3724082-1-luiz.dentz@gmail.com>
References: <20230111012254.3724082-1-luiz.dentz@gmail.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

From: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>

This attempts to fix the following trace:

iso-tester/52 is trying to acquire lock:
ffff8880024e0070 (&hdev->lock){+.+.}-{3:3}, at:
iso_sock_listen+0x29e/0x440

but task is already holding lock:
ffff888001978130 (sk_lock-AF_BLUETOOTH-BTPROTO_ISO){+.+.}-{0:0}, at:
iso_sock_listen+0x8b/0x440

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #2 (sk_lock-AF_BLUETOOTH-BTPROTO_ISO){+.+.}-{0:0}:
       lock_acquire+0x176/0x3d0
       lock_sock_nested+0x32/0x80
       iso_connect_cfm+0x1a3/0x630
       hci_cc_le_setup_iso_path+0x195/0x340
       hci_cmd_complete_evt+0x1ae/0x500
       hci_event_packet+0x38e/0x7c0
       hci_rx_work+0x34c/0x980
       process_one_work+0x5a5/0x9a0
       worker_thread+0x89/0x6f0
       kthread+0x14e/0x180
       ret_from_fork+0x22/0x30

-> #1 (hci_cb_list_lock){+.+.}-{3:3}:
       lock_acquire+0x176/0x3d0
       __mutex_lock+0x13b/0xf50
       hci_le_remote_feat_complete_evt+0x17e/0x320
       hci_event_packet+0x38e/0x7c0
       hci_rx_work+0x34c/0x980
       process_one_work+0x5a5/0x9a0
       worker_thread+0x89/0x6f0
       kthread+0x14e/0x180
       ret_from_fork+0x22/0x30

-> #0 (&hdev->lock){+.+.}-{3:3}:
       check_prev_add+0xfc/0x1190
       __lock_acquire+0x1e27/0x2750
       lock_acquire+0x176/0x3d0
       __mutex_lock+0x13b/0xf50
       iso_sock_listen+0x29e/0x440
       __sys_listen+0xe6/0x160
       __x64_sys_listen+0x25/0x30
       do_syscall_64+0x42/0x90
       entry_SYSCALL_64_after_hwframe+0x62/0xcc

other info that might help us debug this:

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

 Possible unsafe locking scenario:

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

 *** DEADLOCK ***

1 lock held by iso-tester/52:
 #0: ffff888001978130 (sk_lock-AF_BLUETOOTH-BTPROTO_ISO){+.+.}-{0:0}, at:
 iso_sock_listen+0x8b/0x440

Fixes: f764a6c2c1e4 ("Bluetooth: ISO: Add broadcast support")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
---
 net/bluetooth/iso.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/net/bluetooth/iso.c b/net/bluetooth/iso.c
index 6157bc12b373..24444b502e58 100644
--- a/net/bluetooth/iso.c
+++ b/net/bluetooth/iso.c
@@ -893,13 +893,10 @@ static int iso_listen_bis(struct sock *sk)
 	if (!hdev)
 		return -EHOSTUNREACH;
 
-	hci_dev_lock(hdev);
-
 	err = hci_pa_create_sync(hdev, &iso_pi(sk)->dst,
 				 le_addr_type(iso_pi(sk)->dst_type),
 				 iso_pi(sk)->bc_sid);
 
-	hci_dev_unlock(hdev);
 	hci_dev_put(hdev);
 
 	return err;
-- 
2.37.3