[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20220414094225.4527-17-daniel.starke@siemens.com>
Date: Thu, 14 Apr 2022 02:42:22 -0700
From: "D. Starke" <daniel.starke@...mens.com>
To: linux-serial@...r.kernel.org, gregkh@...uxfoundation.org,
jirislaby@...nel.org
Cc: linux-kernel@...r.kernel.org,
Daniel Starke <daniel.starke@...mens.com>
Subject: [PATCH 17/20] tty: n_gsm: fix reset fifo race condition
From: Daniel Starke <daniel.starke@...mens.com>
gsmtty_write() and gsm_dlci_data_output() properly guard the fifo access.
However, gsm_dlci_close() and gsmtty_flush_buffer() modifies the fifo but
do not guard this.
Add a guard here to prevent race conditions on parallel writes to the fifo.
Fixes: e1eaea46bb40 ("tty: n_gsm line discipline")
Cc: stable@...r.kernel.org
Signed-off-by: Daniel Starke <daniel.starke@...mens.com>
---
drivers/tty/n_gsm.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/drivers/tty/n_gsm.c b/drivers/tty/n_gsm.c
index f4ec48c0d6d7..1e135a71860f 100644
--- a/drivers/tty/n_gsm.c
+++ b/drivers/tty/n_gsm.c
@@ -1446,13 +1446,17 @@ static int gsm_control_wait(struct gsm_mux *gsm, struct gsm_control *control)
static void gsm_dlci_close(struct gsm_dlci *dlci)
{
+ unsigned long flags;
+
del_timer(&dlci->t1);
if (debug & 8)
pr_debug("DLCI %d goes closed.\n", dlci->addr);
dlci->state = DLCI_CLOSED;
if (dlci->addr != 0) {
tty_port_tty_hangup(&dlci->port, false);
+ spin_lock_irqsave(&dlci->lock, flags);
kfifo_reset(&dlci->fifo);
+ spin_unlock_irqrestore(&dlci->lock, flags);
/* Ensure that gsmtty_open() can return. */
tty_port_set_initialized(&dlci->port, 0);
wake_up_interruptible(&dlci->port.open_wait);
@@ -3150,13 +3154,17 @@ static unsigned int gsmtty_chars_in_buffer(struct tty_struct *tty)
static void gsmtty_flush_buffer(struct tty_struct *tty)
{
struct gsm_dlci *dlci = tty->driver_data;
+ unsigned long flags;
+
if (dlci->state == DLCI_CLOSED)
return;
/* Caution needed: If we implement reliable transport classes
then the data being transmitted can't simply be junked once
it has first hit the stack. Until then we can just blow it
away */
+ spin_lock_irqsave(&dlci->lock, flags);
kfifo_reset(&dlci->fifo);
+ spin_unlock_irqrestore(&dlci->lock, flags);
/* Need to unhook this DLCI from the transmit queue logic */
}
--
2.25.1
Powered by blists - more mailing lists