What I am actually worried about is Tejun's rework, I am not sure
cmwq has the "this thread services that wq" property...
On 05/27, Sridhar Samudrala wrote:
>
> On Thu, 2010-05-27 at 14:44 +0200, Oleg Nesterov wrote:
> >
> > Instead, cgroup.c (or whoever needs this) can do
> >
> > struct move_struct {
> > struct work_struct work;
> > int ret;
> > };
> >
> > static void move_func(struct work_struct *work)
> > {
> > struct move_struct *move = container_of(...);
> >
> > if (cgroup_attach_task_current_cg(current))
>
> We are trying to attach the task associated with workqueue to the
> current task's cgroup. So what we need is
> cgroup_attach_task_current_cg(wq->task);
I do not see cgroup_attach_task_current_cg() in Linus's tree and thus I do
not now what exactly it does, and of course the code above is only template.
But I think this is easy. Just add "struct cgroup *cgrp" into move_struct
and then move_func() can do cgroup_attach_task(move->cgrp, current) ?
> > Or. Just export wq_per_cpu() from workqueue.c (probably with a better name)
> > and
> > use it like the patch does.
> This requires that struct cpu_workqueue_struct and struct
> workqueue_struct are made externally visible by moving them to
> kernel/workqueue.h.
no, no,
> Instead what about adding the simple helper get_singlethread_wq_task()
> in workqueue.c and exporting it.
Indeed, this is what I meant.
But. I disagree with get_singlethread_wq_task(). If we add this helper,
it should work with the multi-threaded wq's too, and needs the "int cpu"
parameter, ignored when is_wq_single_threaded().
So. Either rename wq_per_cpu() and export it (once again, I do not
mean we should move the body to workqueue.h!), or create the new
helper which just calls wq_per_cpu().
> I can add create_singlethread_workqueue_in_current_cg() to cgroup.c
> using this helper routine.
Imho, this is better.
But please note that it is possible to do without any changes in
workqueue.[ch] afaics, see above.
Oleg.
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
