On Thu, 15 Jan 2009 16:15:46 -0500 (EST), Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote: > Then the submitting method must release its lock before doing the > submission. It has never been specified in the USB kernel > documentation that callbacks will occur in a different context from > submissions or unlinks. This is a thorny issue in some contexts. Sometimes some kind of a framework holds a lock for you. In ub I eventually ended bouncing USB callbacks off a tasklet. This is faster than workqueue, has no issues with cancelling. > But it does seem unlikely that any HCD will ever complete an URB > during the submission, because of the bus transmission time if nothing > else. I would rather not rely on this, with all the new fishy HCDs we have out there. I would not put it past them to busy-wait somewhere long enough to catch a quickly occuring error callback. I almost can swear it has actually happened to me in ub, too. -- Pete -- 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
