On Tue, Jul 27, 2004 at 10:47:13 +0200, Tomasz Paszkowski wrote:
> Hello,
>
> I'am running hfsc qdisc with huge set of rules loaded. Always when I delete
> the root qdisc the machine stop responding for about 5-6 seconds. As I
> think it's due the hfsc_destory_qdisc is executed in main kernel thread.
>
> Is there any solutions which can after completing each run in loop, put
> itself at the end of run queue ?
The code would have to be rewritten.
If it's 2.6 kernel, recompile it with preemption. It should help
Actualy, I am not sure whether it runs in an event thread (ie. from work
queue, as the subject suggests), or from the syscall (as I would
consider more likely).
If it's from syscall, fixing it would be easy -- just add something
like:
if(current->need_resched) schedule()
into the loop.
If it's run from a work queue, it would have to be completely rewritten.
Anyway, please post it on the linux-kernel@xxxxxxxxxxxxxxx list as
a bug.
> hfsc_destroy_qdisc(struct Qdisc *sch)
> {
> struct hfsc_sched *q = (struct hfsc_sched *)sch->data;
> struct hfsc_class *cl, *next;
> unsigned int i;
>
> for (i = 0; i < HFSC_HSIZE; i++) {
> list_for_each_entry_safe(cl, next, &q->clhash[i], hlist)
> hfsc_destroy_class(sch, cl);
> }
> __skb_queue_purge(&q->requeue);
> del_timer(&q->wd_timer);
> MOD_DEC_USE_COUNT;
> }
>
>
>
> --
> Tomasz Paszkowski
-------------------------------------------------------------------------------
Jan 'Bulb' Hudec <bulb@xxxxxx>
Attachment:
signature.asc
Description: Digital signature
