From: <w15303746062@163.com>
To: luiz.dentz@gmail.com, pmenzel@molgen.mpg.de, marcel@holtmann.org,
linux-bluetooth@vger.kernel.org
Cc: linux-serial@vger.kernel.org, linux-kernel@vger.kernel.org,
greg@kroah.com, stable@vger.kernel.org,
Mingyu Wang <25181214217@stu.xidian.edu.cn>
Subject: [PATCH v9] Bluetooth: hci_uart: fix UAFs and race conditions in close and init paths
Date: Mon, 18 May 2026 10:49:49 +0800 [thread overview]
Message-ID: <20260518024949.439299-1-w15303746062@163.com> (raw)
In-Reply-To: <CABBYNZ+r3gm37FW5WqE79bRp+x9UZsaCtyvfz_FdixqEucAxGw@mail.gmail.com>
From: Mingyu Wang <25181214217@stu.xidian.edu.cn>
Vulnerabilities leading to Use-After-Free (UAF) and Null Pointer
Dereference (NPD) conditions were observed in the lifecycle management
of hci_uart.
The primary issue arises because the workqueues (init_ready and
write_work) are only flushed/cancelled if the HCI_UART_PROTO_READY
flag is set during TTY close. If a hangup occurs before setup completes,
hci_uart_tty_close() skips the teardown of these workqueues and
proceeds to free the `hu` struct. When the scheduled work executes
later, it blindly dereferences the freed `hu` struct.
Furthermore, several data races and UAFs were identified in the teardown
sequence:
1. Calling hci_uart_flush() from hci_uart_close() without effectively
disabling write_work causes a race condition where both can concurrently
double-free hu->tx_skb. This happens because protocol timers can
concurrently invoke hci_uart_tx_wakeup() and requeue write_work.
2. Calling hci_free_dev(hdev) before hu->proto->close(hu) causes a UAF
when vendor specific protocol close callbacks dereference hu->hdev.
3. In the initialization error paths, failing to take the proto_lock
write lock before clearing PROTO_READY leads to races with active
readers. Additionally, hci_uart_tty_receive() accesses hu->hdev
outside the read lock, leading to UAFs if the initialization error
path frees hdev concurrently.
Fix these synchronization and lifecycle issues by:
1. Re-ordering hci_uart_tty_close() to clear HCI_UART_PROTO_READY first,
followed immediately by a cancel_work_sync(&hu->write_work). Clearing
the flag locks out concurrent protocol timers from successfully invoking
hci_uart_tx_wakeup(), effectively rendering the cancellation permanent
and preventing the tx_skb double-free.
2. Note: Clearing PROTO_READY early causes hci_uart_close() to skip
hu->proto->flush(). This is perfectly safe in the tty_close path
because hu->proto->close() executes shortly after, which intrinsically
purges all protocol SKB queues and tears down the state.
3. Relocating hu->proto->close(hu) strictly prior to hci_free_dev(hdev)
across all close and error paths to prevent vendor-level UAFs.
4. Moving the hdev->stat.byte_rx increment in hci_uart_tty_receive()
inside the proto_lock read-side critical section to safely synchronize
with device unregistration.
5. Adding cancel_work_sync(&hu->write_work) to hci_uart_close() to safely
flush the workqueue before hci_uart_flush() is invoked via the HCI core.
6. Utilizing cancel_work_sync() instead of disable_work_sync() across
all paths to prevent permanently breaking user-space retry capabilities.
Fixes: 3b799254cf6f ("Bluetooth: hci_uart: Cancel init work before unregistering")
Cc: stable@vger.kernel.org
Signed-off-by: Mingyu Wang <25181214217@stu.xidian.edu.cn>
---
Changes in v9:
- Addressed a critical flaw identified in v8 where premature cancellation of write_work allowed active protocol timers to immediately reschedule it. The teardown sequence in hci_uart_tty_close() now strictly clears HCI_UART_PROTO_READY *before* calling cancel_work_sync(&hu->write_work). This permanently locks out hci_uart_tx_wakeup(), completely resolving the lingering UAF and double-free races.
- Documented that skipping hu->proto->flush() via early flag clearance is intrinsically safe, as hu->proto->close() executes subsequently to purge all unacked/rel queues.
Changes in v8:
- Corrected the teardown sequence in hci_uart_tty_close() by unconditionally canceling write_work BEFORE hci_uart_close().
- Moved hu->hdev->stat.byte_rx increment inside the proto_lock read-side critical section in hci_uart_tty_receive() to prevent read-side UAF against concurrent registration failures.
- Added cancel_work_sync(&hu->write_work) inside hci_uart_close() to eliminate the race condition between write_work and hci_uart_flush() when the interface is brought down via the HCI core.
Changes in v7:
- Reverted disable_work_sync() back to cancel_work_sync() across all error and close paths to preserve user-space retry capabilities.
- Synchronized workqueue teardown safely by atomically clearing PROTO_READY / PROTO_INIT under proto_lock prior to calling cancel_work_sync().
- Fixed a Use-After-Free (UAF) vulnerability in the teardown sequence by relocating hu->proto->close(hu) strictly prior to hci_free_dev(hdev).
- Added cancel_work_sync(&hu->init_ready) at the very beginning of hci_uart_tty_close() to serialize teardown against active asynchronous registration.
Changes in v6:
- Fixed missing `hu->proto_lock` write lock in hci_uart_init_work() error path to prevent race with readers (reported by Sashiko).
- Added disable_work_sync() instead of cancel_work_sync() for `hu->write_work` in hci_uart_init_work() and hci_uart_register_dev() error paths.
Changes in v5:
- Relocated disable_work_sync() to the very top of hci_uart_tty_close(),
before hci_uart_close(), to ensure no new work is submitted during device teardown.
Changes in v4:
- Adopted Luiz's suggestion to use disable_work_sync() instead of
cancel_work_sync() in close path to prevent new work submissions.
Changes in v3:
- Added 'Cc: stable' tag as requested by the stable bot.
Changes in v2:
- Added KASAN/ODEBUG crash trace.
drivers/bluetooth/hci_ldisc.c | 48 +++++++++++++++++++++++++++++------
1 file changed, 40 insertions(+), 8 deletions(-)
diff --git a/drivers/bluetooth/hci_ldisc.c b/drivers/bluetooth/hci_ldisc.c
index 275ea865bc29..47f4902b40b4 100644
--- a/drivers/bluetooth/hci_ldisc.c
+++ b/drivers/bluetooth/hci_ldisc.c
@@ -194,7 +194,15 @@ void hci_uart_init_work(struct work_struct *work)
err = hci_register_dev(hu->hdev);
if (err < 0) {
BT_ERR("Can't register HCI device");
+
+ percpu_down_write(&hu->proto_lock);
clear_bit(HCI_UART_PROTO_READY, &hu->flags);
+ percpu_up_write(&hu->proto_lock);
+
+ /* Safely cancel work after clearing flags */
+ cancel_work_sync(&hu->write_work);
+
+ /* Close protocol before freeing hdev */
hu->proto->close(hu);
hdev = hu->hdev;
hu->hdev = NULL;
@@ -263,8 +271,12 @@ static int hci_uart_open(struct hci_dev *hdev)
/* Close device */
static int hci_uart_close(struct hci_dev *hdev)
{
+ struct hci_uart *hu = hci_get_drvdata(hdev);
+
BT_DBG("hdev %p", hdev);
+ cancel_work_sync(&hu->write_work);
+
hci_uart_flush(hdev);
hdev->flush = NULL;
return 0;
@@ -531,6 +543,7 @@ static void hci_uart_tty_close(struct tty_struct *tty)
{
struct hci_uart *hu = tty->disc_data;
struct hci_dev *hdev;
+ bool proto_ready;
BT_DBG("tty %p", tty);
@@ -540,24 +553,38 @@ static void hci_uart_tty_close(struct tty_struct *tty)
if (!hu)
return;
- hdev = hu->hdev;
- if (hdev)
- hci_uart_close(hdev);
+ /* Wait for init_ready to finish to prevent registration races */
+ cancel_work_sync(&hu->init_ready);
- if (test_bit(HCI_UART_PROTO_READY, &hu->flags)) {
+ proto_ready = test_bit(HCI_UART_PROTO_READY, &hu->flags);
+ if (proto_ready) {
percpu_down_write(&hu->proto_lock);
clear_bit(HCI_UART_PROTO_READY, &hu->flags);
percpu_up_write(&hu->proto_lock);
+ }
- cancel_work_sync(&hu->init_ready);
- cancel_work_sync(&hu->write_work);
+ /*
+ * Unconditionally cancel write_work AFTER clearing PROTO_READY.
+ * This ensures that concurrent protocol timers cannot requeue
+ * write_work via hci_uart_tx_wakeup(), permanently preventing
+ * double-free races and UAFs.
+ */
+ cancel_work_sync(&hu->write_work);
+
+ hdev = hu->hdev;
+ if (hdev)
+ hci_uart_close(hdev); /* proto->flush is safely skipped */
+ if (proto_ready) {
if (hdev) {
if (test_bit(HCI_UART_REGISTERED, &hu->flags))
hci_unregister_dev(hdev);
- hci_free_dev(hdev);
}
+ /* Close protocol before freeing hdev (intrinsically purges queues) */
hu->proto->close(hu);
+
+ if (hdev)
+ hci_free_dev(hdev);
}
clear_bit(HCI_UART_PROTO_SET, &hu->flags);
@@ -625,11 +652,12 @@ static void hci_uart_tty_receive(struct tty_struct *tty, const u8 *data,
* tty caller
*/
hu->proto->recv(hu, data, count);
- percpu_up_read(&hu->proto_lock);
if (hu->hdev)
hu->hdev->stat.byte_rx += count;
+ percpu_up_read(&hu->proto_lock);
+
tty_unthrottle(tty);
}
@@ -695,6 +723,10 @@ static int hci_uart_register_dev(struct hci_uart *hu)
percpu_down_write(&hu->proto_lock);
clear_bit(HCI_UART_PROTO_INIT, &hu->flags);
percpu_up_write(&hu->proto_lock);
+ /* Cancel work after clearing flags */
+ cancel_work_sync(&hu->write_work);
+
+ /* Close protocol before freeing hdev */
hu->proto->close(hu);
hu->hdev = NULL;
hci_free_dev(hdev);
--
2.34.1
next prev parent reply other threads:[~2026-05-18 2:50 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-05-13 6:45 [PATCH] Bluetooth: hci_uart: fix UAF in hci_uart_tty_close() w15303746062
2026-05-13 7:54 ` bluez.test.bot
2026-05-13 9:04 ` [PATCH] " Paul Menzel
2026-05-14 15:17 ` [PATCH v2] " w15303746062
2026-05-14 17:00 ` [v2] " bluez.test.bot
2026-05-15 6:10 ` [PATCH v2] " Greg KH
2026-05-15 6:50 ` [PATCH v3] " w15303746062
2026-05-15 8:43 ` [v3] " bluez.test.bot
2026-05-15 12:37 ` [PATCH] " Luiz Augusto von Dentz
2026-05-15 13:39 ` w15303746062
2026-05-15 14:05 ` [PATCH v4] " w15303746062
2026-05-15 15:12 ` [v4] " bluez.test.bot
2026-05-15 16:08 ` [PATCH v4] " Luiz Augusto von Dentz
2026-05-16 1:41 ` w15303746062
2026-05-16 2:22 ` [PATCH v5] " w15303746062
2026-05-16 5:11 ` [v5] " bluez.test.bot
2026-05-16 5:30 ` [PATCH v6] Bluetooth: hci_uart: fix UAFs and race conditions in close and init paths w15303746062
2026-05-16 7:33 ` [v6] " bluez.test.bot
2026-05-16 8:47 ` [PATCH v7] " w15303746062
2026-05-16 9:24 ` [v7] " bluez.test.bot
2026-05-18 1:36 ` [PATCH v8] " w15303746062
2026-05-18 2:49 ` w15303746062 [this message]
2026-05-18 6:16 ` [v9] " bluez.test.bot
2026-05-19 15:00 ` [PATCH v9] " patchwork-bot+bluetooth
2026-06-13 12:58 ` [REGRESSION] " Pauli Virtanen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260518024949.439299-1-w15303746062@163.com \
--to=w15303746062@163.com \
--cc=25181214217@stu.xidian.edu.cn \
--cc=greg@kroah.com \
--cc=linux-bluetooth@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-serial@vger.kernel.org \
--cc=luiz.dentz@gmail.com \
--cc=marcel@holtmann.org \
--cc=pmenzel@molgen.mpg.de \
--cc=stable@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.