Re: åå: [PATCH v5] workqueue: Remove unnecessary kfree() call in rcu_free_wq()

From: Lai Jiangshan
Date: Thu May 28 2020 - 09:28:04 EST


On Thu, May 28, 2020 at 8:27 PM Dan Carpenter <dan.carpenter@xxxxxxxxxx> wrote:
>
> On Thu, May 28, 2020 at 08:08:06PM +0800, Lai Jiangshan wrote:
> > On Thu, May 28, 2020 at 5:57 PM Dan Carpenter <dan.carpenter@xxxxxxxxxx> wrote:
> > >
> > > Guys, the patch is wrong. The kfree is harmless when this is called
> > > from destroy_workqueue() and required when it's called from
> > > pwq_unbound_release_workfn(). Lai Jiangshan already explained this
> > > already. Why are we still discussing this?
> > >
> >
> > I'm also confused why they have been debating about the changelog
> > after the patch was queued. My statement was about "the patch is
> > a correct cleanup, but the changelog is totally misleading".
> >
> > destroy_workqueue(percpu_wq) -> rcu_free_wq()
> > or
> > destroy_workqueue(unbound_wq) -> put_pwq() ->
> > pwq_unbound_release_workfn() -> rcu_free_wq()
> >
> > So the patch is correct to me. Only can destroy_workqueue()
> > lead to rcu_free_wq().
>
> It looks like there are lots of paths which call put_pwq() and
> put_pwq_unlocked().
>
> 1168 static void pwq_dec_nr_in_flight(struct pool_workqueue *pwq, int color)
> 1169 {
> 1170 /* uncolored work items don't participate in flushing or nr_active */
> 1171 if (color == WORK_NO_COLOR)
> 1172 goto out_put;
> 1173
>
> We don't take an extra reference in this function.
>
> 1200 out_put:
> 1201 put_pwq(pwq);
> 1202 }
>
> I don't know this code well, so I will defer to your expertise if you
> say it is correct.

wq owns the ultimate or permanent references to itself by
owning references to wq->numa_pwq_tbl[node], wq->dfl_pwq.
The pwq's references keep the pwq in wq->pwqs.

Only destroy_workqueue() can release these ultimate references
and then (after maybe a period of time) deplete the wq->pwqs
finally and then free itself in rcu callback.

Actually, in short, we don't need the care about the above
detail. The responsibility to free rescuer is on
destroy_workqueue(), and since it does the free early,
it doesn't need to do it again later.

e2dca7adff8f moved the free of rescuer into rcu callback,
and I didn't check all the changes between then and now.
But at least now, the rescuer is not accessed in rcu mananer,
so we don't need to free it in rcu mananer.

>
> >
> > Still, the kfree(NULL) is harmless. But it is cleaner
> > to have the patch. But the changelog is wrong, even after
> > the lengthened debating, and English is not my mother tongue,
> > so I just looked on.
>
> We have tried to tell Markus not to advise people about commit messages
> but he doesn't listen. He has discouraged some contributors. :/
>
> regards,
> dan carpenter