RE: [PATCH 1/1] rcu/rcuscale: Stop kfree_scale_thread thread(s) after unloading rcuscale

[Zhuo, Qiuxu]

> From: Joel Fernandes <joel@xxxxxxxxxxxxxxxxx>
> [...]
> > >  kernel/rcu/rcuscale.c | 7 +++++++
> > >  1 file changed, 7 insertions(+)
> > >
> > > diff --git a/kernel/rcu/rcuscale.c b/kernel/rcu/rcuscale.c index
> > > 91fb5905a008..5e580cd08c58 100644
> > > --- a/kernel/rcu/rcuscale.c
> > > +++ b/kernel/rcu/rcuscale.c
> > > @@ -522,6 +522,8 @@ rcu_scale_print_module_parms(struct
> rcu_scale_ops *cur_ops, const char *tag)
> > >                scale_type, tag, nrealreaders, nrealwriters, verbose,
> > > shutdown);  }
> > >
> > > +static void kfree_scale_cleanup(void);
> > > +
> >
> > I do applaud minmimizing the size of the patch, but in this case could
> > you please pull the kfree_scale_cleanup() function ahead of its first use?
> 
> The only trouble with moving the function like that is, the file is mostly split
> across kfree and non-kfree functions. So moving a kfree function to be
> among the non-kfree ones would look a bit weird.

Yes, this would look a bit weird ...

Please see the reply to Paul in another e-mail: 
"Pull the rcu_scale_cleanup() function after kfree_scale_cleanup().
This groups kfree_* functions and groups rcu_scale_* functions. 
Then the code would look cleaner."

> Perhaps a better place for the function declaration could be a new
> "rcuscale.h". But I am really Ok with Paul's suggestion as well.
> 
> Reviewed-by: Joel Fernandes (Google) <joel@xxxxxxxxxxxxxxxxx>

Thanks for the review. :-)

> thanks,
> 
>  - Joel