On Tue, 5 Dec 2006, Andrew Morton wrote: > But running flush_scheduled_work() from within dev_close() is a very > sensible thing to do, and dev_close is called under rtnl_lock(). > davem is -> thattaway ;)
And when within dev_close() there is quite a chance there is linkwatch_event() somewhere in the event queue already. ;-) > Ah. The point is that the phy code doesn't want to flush _all_ pending > callbacks. It only wants to flush its own one. And its own one doesn't > take rtnl_lock(). > > IOW, the phy code has no interest in running some random other subsystem's > callback - it just wants to run its own. Hence no deadlock. Both are true. It's linkwatch_event() that's somewhere in the queue already that makes the trouble here. > Maybe the lesson here is that flush_scheduled_work() is a bad function. > It should really be flush_this_work(struct work_struct *w). That is in > fact what approximately 100% of the flush_scheduled_work() callers actually > want to do. There may be cases where flush_scheduled_work() is indeed needed, but likely outside drivers, and I agree such a specific call would be useful and work here. Maciej - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
