can: kvaser_usb: Fix tx queue start/stop race conditions
authorAhmed S. Darwish <ahmed.darwish@valeo.com>
Sat, 14 Mar 2015 13:02:49 +0000 (09:02 -0400)
committerMarc Kleine-Budde <mkl@pengutronix.de>
Sat, 14 Mar 2015 08:20:07 +0000 (09:20 +0100)
commita9dc960c37b0d4eb192598dc4c94276270454514
tree769341072aeb277f71beed94b5b73f00b9dce46f
parent963a822b6d5fece27c88522ac34dd48928571c8b
can: kvaser_usb: Fix tx queue start/stop race conditions

A number of tx queue wake-up events went missing due to the
outlined scenario below. Start state is a pool of 16 tx URBs,
active tx_urbs count = 15, with the netdev tx queue open.

CPU #1 [softirq]                         CPU #2 [softirq]
start_xmit()                             tx_acknowledge()
................                         ................

atomic_inc(&tx_urbs);
if (atomic_read(&tx_urbs) >= 16) {
                        -->
                                         atomic_dec(&tx_urbs);
                                         netif_wake_queue();
                                         return;
                        <--
    netif_stop_queue();
}

At the end, the correct state expected is a 15 tx_urbs count
value with the tx queue state _open_. Due to the race, we get
the same tx_urbs value but with the tx queue state _stopped_.
The wake-up event is completely lost.

Thus avoid hand-rolled concurrency mechanisms and use a proper
lock for contexts and tx queue protection.

Signed-off-by: Ahmed S. Darwish <ahmed.darwish@valeo.com>
Cc: linux-stable <stable@vger.kernel.org>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
drivers/net/can/usb/kvaser_usb.c