Re: [dm-devel] [PATCH -next v2 2/6] md: refactor action_store() for 'idle' and 'frozen'

[Xiao Ni]

在 2023/6/13 下午8:44, Yu Kuai 写道:
> Hi,
>
> 在 2023/06/13 20:25, Xiao Ni 写道:
> > On Tue, Jun 13, 2023 at 8:00 PM Yu Kuai <yukuai1@xxxxxxxxxxxxxxx> wrote:
> > > Hi,
> > >
> > > 在 2023/06/13 16:02, Xiao Ni 写道:
> > > > 在 2023/5/29 下午9:20, Yu Kuai 写道:
> > > > > From: Yu Kuai <yukuai3@xxxxxxxxxx>
> > > > >
> > > > > Prepare to handle 'idle' and 'frozen' differently to fix a deadlock,
> > > > > there
> > > > > are no functional changes except that MD_RECOVERY_RUNNING is checked
> > > > > again after 'reconfig_mutex' is held.
> > > >
> > > >
> > > > Can you explain more about why it needs to check MD_RECOVERY_RUNNING
> > > > again here?
> > >
> > > As I explain in the following comment:
> >
> > Hi
> >
> > Who can clear the flag before the lock is held?
>
> Basically every where that can clear the flag...
>
> // This context     // Other context
>             mutex_lock
>             ...
> test_bit -> pass
>             clear_bit
>             mutex_unlock
> mutex_lock
> test_bit -> check again
>
> Thanks,
> Kuai

At first, I wanted to figure out a specific case. Now I have the answer. 
Maybe there are two people that want to stop

the sync action at the same time. So this is the case that can be 
checked by the codes.

Regards

Xiao

> > Regards
> > Xiao
> > > > > +    /*
> > > > > +     * Check again in case MD_RECOVERY_RUNNING is cleared before 
> > > > > lock is
> > > > > +     * held.
> > > > > +     */
> > > > > +    if (!test_bit(MD_RECOVERY_RUNNING, &mddev->recovery)) {
> > > > > +        mddev_unlock(mddev);
> > > > > +        return;
> > > > > +    }
> > >
> > > Thanks,
> > > Kuai
> >
> > .