On 03/11/2015 04:23 PM, Ahmed S. Darwish wrote: > From: Ahmed S. Darwish > > 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. > > start_xmit() tx_acknowledge() > ............ ................ > atomic_inc(&tx_urbs); > if (atomic_read(&tx_urbs) >= 16) { > URB completion IRQ! > --> > atomic_dec(&tx_urbs); > netif_wake_queue(); > return; > <-- > end of IRQ! > 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 protection. I'm missing a spin_lock_init(), right? Please compile and test your code with everything switch on in Kernel hacking -> Lock Debugging. Marc -- Pengutronix e.K. | Marc Kleine-Budde | Industrial Linux Solutions | Phone: +49-231-2826-924 | Vertretung West/Dortmund | Fax: +49-5121-206917-5555 | Amtsgericht Hildesheim, HRA 2686 | http://www.pengutronix.de |