From: syzbot <syzbot+bca09f5d8b843bbf7571@syzkaller.appspotmail.com>
To: linux-kernel@vger.kernel.org, syzkaller-bugs@googlegroups.com
Subject: Forwarded: [PATCH] usb: gadget: u_serial: fix use-after-free of gs_port on concurrent open
Date: Sat, 29 Aug 2026 05:34:09 -0700 [thread overview]
Message-ID: <6a92d1c1.1d9ded08.62e62.0108.GAE@google.com> (raw)
In-Reply-To: <6a7a82f8.01d0871a.3a0d52.00b9.GAE@google.com>
For archival purposes, forwarding an incoming command email to
linux-kernel@vger.kernel.org, syzkaller-bugs@googlegroups.com.
***
Subject: [PATCH] usb: gadget: u_serial: fix use-after-free of gs_port on concurrent open
Author: lingrain580@gmail.com
#syz test: git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
gserial_free_line() frees the gs_port while the tty layer may still be
about to use the tty_port embedded in it. gs_tty_driver->ports[] keeps
pointing at the freed memory, because tty_unregister_device() only clears
cdevs[] and never touches ports[].
A concurrent open() then picks up that stale pointer in tty_init_dev():
if (!tty->port)
tty->port = driver->ports[idx];
...
tty->port->itty = tty;
gs_open() does check ports[port_num].port and returns -ENODEV, but that
check runs after tty->port has already been assigned, so even the failed
open dereferences freed memory on its release path:
BUG: KASAN: slab-use-after-free in release_tty+0x589/0x5d0
Write of size 8 at addr ffff88804b450120 by task repro/4930
Allocated by task 4929:
gs_port_alloc
gserial_alloc_line_no_console
acm_alloc_instance
configfs_mkdir
Freed by task 4929:
gserial_free_port
gserial_free_line
acm_free_instance
config_item_cleanup
The existing wait_event(port->close_wait, gs_closed(port)) does not help
here: a failed open never increments port.count, so gs_closed() reports
the port as closed and the wait returns immediately.
Give gs_port proper reference counting, the way cdc-acm.c already does on
the host side. Add a tty_port_operations::destruct callback that performs
the kfree(), acquire a reference in a new ->install callback while holding
ports[].lock, and release it in ->cleanup. gserial_free_port() now calls
tty_port_put() instead of freeing directly, so the memory is released only
once the last user is gone.
Reproduced with the syzbot C reproducer on v7.2-rc6 with KASAN and
PREEMPT_RT enabled; the report is gone after this change and the
WARN_ON(port->itty) in tty_port_destructor() does not trigger, confirming
the references are balanced.
Reported-by: syzbot+bca09f5d8b843bbf7571@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=bca09f5d8b843bbf7571
Cc: stable@vger.kernel.org
Signed-off-by: Tianqi Dong <lingrain580@gmail.com>
---
drivers/usb/gadget/function/u_serial.c | 48 ++++++++++++++++++++++++--
1 file changed, 46 insertions(+), 2 deletions(-)
diff --git a/drivers/usb/gadget/function/u_serial.c b/drivers/usb/gadget/function/u_serial.c
index cdd1dfc66..6eabccf9b 100644
--- a/drivers/usb/gadget/function/u_serial.c
+++ b/drivers/usb/gadget/function/u_serial.c
@@ -908,7 +908,51 @@ static int gs_get_icount(struct tty_struct *tty,
return 0;
}
+static void gs_port_destruct(struct tty_port *port)
+{
+ struct gs_port *gs = container_of(port, struct gs_port, port);
+
+ kfree(gs);
+}
+
+static const struct tty_port_operations gs_port_ops = {
+ .destruct = gs_port_destruct,
+};
+
+/*
+ * Take a reference to the port before the tty core stores it in tty->port.
+ * Otherwise gserial_free_line() may free the port while a concurrent open()
+ * is about to dereference the stale pointer left in gs_tty_driver->ports[].
+ */
+static int gs_install(struct tty_driver *driver, struct tty_struct *tty)
+{
+ struct gs_port *port;
+ int status;
+
+ mutex_lock(&ports[tty->index].lock);
+ port = ports[tty->index].port;
+ if (!port) {
+ mutex_unlock(&ports[tty->index].lock);
+ return -ENODEV;
+ }
+ tty_port_get(&port->port);
+ mutex_unlock(&ports[tty->index].lock);
+
+ status = tty_port_install(&port->port, driver, tty);
+ if (status)
+ tty_port_put(&port->port);
+
+ return status;
+}
+
+static void gs_cleanup(struct tty_struct *tty)
+{
+ tty_port_put(tty->port);
+}
+
static const struct tty_operations gs_tty_ops = {
+ .install = gs_install,
+ .cleanup = gs_cleanup,
.open = gs_open,
.close = gs_close,
.write = gs_write,
@@ -1222,6 +1266,7 @@ gs_port_alloc(unsigned port_num, struct usb_cdc_line_coding *coding)
}
tty_port_init(&port->port);
+ port->port.ops = &gs_port_ops;
spin_lock_init(&port->port_lock);
init_waitqueue_head(&port->drain_wait);
init_waitqueue_head(&port->close_wait);
@@ -1258,8 +1303,7 @@ static void gserial_free_port(struct gs_port *port)
/* wait for old opens to finish */
wait_event(port->close_wait, gs_closed(port));
WARN_ON(port->port_usb != NULL);
- tty_port_destroy(&port->port);
- kfree(port);
+ tty_port_put(&port->port);
}
void gserial_free_line(unsigned char port_num)
--
2.43.0
prev parent reply other threads:[~2026-08-29 12:34 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-08-11 2:03 [syzbot] [serial?] KASAN: slab-use-after-free Write in release_tty syzbot
2026-08-14 3:07 ` syzbot
2026-08-29 12:34 ` syzbot [this message]
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=6a92d1c1.1d9ded08.62e62.0108.GAE@google.com \
--to=syzbot+bca09f5d8b843bbf7571@syzkaller.appspotmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=syzkaller-bugs@googlegroups.com \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®