Additionally, remove it from .ndo_stop().
This ensures that the worker is not called after being freed, and that
the UART TX queue remains active to send final commands when the netdev
is stopped.
Thanks to Jiri Slaby for finding this in slcan:
https://lore.kernel.org/linux-can/[email protected]/
A variant of this patch for slcan, with the flush in .ndo_stop() still
present, has been tested successfully on physical hardware:
https://bugzilla.suse.com/show_bug.cgi?id=1205597
Fixes: 43da2f07622f ("can: can327: CAN/ldisc driver for ELM327 based OBD-II adapters")
Cc: "Jiri Slaby (SUSE)" <[email protected]>
Cc: Max Staudt <[email protected]>
Cc: Wolfgang Grandegger <[email protected]>
Cc: Marc Kleine-Budde <[email protected]>
Cc: "David S. Miller" <[email protected]>
Cc: Eric Dumazet <[email protected]>
Cc: Jakub Kicinski <[email protected]>
Cc: Paolo Abeni <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Signed-off-by: Max Staudt <[email protected]>
---
drivers/net/can/can327.c | 17 ++++++++++-------
1 file changed, 10 insertions(+), 7 deletions(-)
diff --git a/drivers/net/can/can327.c b/drivers/net/can/can327.c
index ed3d0b8989a0..dc7192ecb001 100644
--- a/drivers/net/can/can327.c
+++ b/drivers/net/can/can327.c
@@ -796,9 +796,9 @@ static int can327_netdev_close(struct net_device *dev)
netif_stop_queue(dev);
- /* Give UART one final chance to flush. */
- clear_bit(TTY_DO_WRITE_WAKEUP, &elm->tty->flags);
- flush_work(&elm->tx_work);
+ /* We don't flush the UART TX queue here, as we want final stop
+ * commands (like the above dummy char) to be flushed out.
+ */
can_rx_offload_disable(&elm->offload);
elm->can.state = CAN_STATE_STOPPED;
@@ -1069,12 +1069,15 @@ static void can327_ldisc_close(struct tty_struct *tty)
{
struct can327 *elm = (struct can327 *)tty->disc_data;
- /* unregister_netdev() calls .ndo_stop() so we don't have to.
- * Our .ndo_stop() also flushes the TTY write wakeup handler,
- * so we can safely set elm->tty = NULL after this.
- */
+ /* unregister_netdev() calls .ndo_stop() so we don't have to. */
unregister_candev(elm->dev);
+ /* Give UART one final chance to flush.
+ * No need to clear TTY_DO_WRITE_WAKEUP since .write_wakeup() is
+ * serialised against .close() and will not be called once we return.
+ */
+ flush_work(&elm->tx_work);
+
/* Mark channel as dead */
spin_lock_bh(&elm->lock);
tty->disc_data = NULL;
--
2.30.2
On 03.12.2022 01:01:48, Max Staudt wrote:
> Additionally, remove it from .ndo_stop().
>
> This ensures that the worker is not called after being freed, and that
> the UART TX queue remains active to send final commands when the netdev
> is stopped.
>
> Thanks to Jiri Slaby for finding this in slcan:
>
> https://lore.kernel.org/linux-can/[email protected]/
>
> A variant of this patch for slcan, with the flush in .ndo_stop() still
> present, has been tested successfully on physical hardware:
>
> https://bugzilla.suse.com/show_bug.cgi?id=1205597
>
> Fixes: 43da2f07622f ("can: can327: CAN/ldisc driver for ELM327 based OBD-II adapters")
> Cc: "Jiri Slaby (SUSE)" <[email protected]>
> Cc: Max Staudt <[email protected]>
> Cc: Wolfgang Grandegger <[email protected]>
> Cc: Marc Kleine-Budde <[email protected]>
> Cc: "David S. Miller" <[email protected]>
> Cc: Eric Dumazet <[email protected]>
> Cc: Jakub Kicinski <[email protected]>
> Cc: Paolo Abeni <[email protected]>
> Cc: [email protected]
> Cc: [email protected]
> Cc: [email protected]
> Signed-off-by: Max Staudt <[email protected]>
Applied to linux-can.
regards,
Marc
--
Pengutronix e.K. | Marc Kleine-Budde |
Embedded Linux | https://www.pengutronix.de |
Vertretung West/Dortmund | Phone: +49-231-2826-924 |
Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |