Oliver Neukum <oliver@xxxxxxxxxx> writes: > Am Montag, 16. Januar 2012, 12:41:49 schrieb Bjørn Mork: >> Setting the WDM_READ bit in wdm_disconnect() is useless, as wdm_disconnect() >> won't be called at all as long as wdm_read() is blocking waiting for this bit. > > Yes, and this shows that wdm_read() must be woken at this point. >> Use wait_event_interruptible_timeout for blocking reads to allow the USB >> subsystem to call wdm_disconnect, and thereby allow a blocking read to >> complete on device removal. > > No, playing around with timeout is evil. > This shows that there's a wake_up_all() missing in disconnect() > between setting the flags and taking the mutex. right you are. There is a wake_up_all() after taking the mutex, but that seems pointless as noone should be waiting for anything if we can get the mutex with the WDM_DISCONNECTING set. I'll send a new version as soon as I've tested that it works. Bjørn -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
