Re: [RFC 1/2] watchdog: update watchdog attributes atomically

[Don Zickus <dzickus@redhat.com>]

On Fri, Jul 19, 2013 at 06:33:23PM +0200, Michal Hocko wrote:
> On Fri 19-07-13 12:10:48, Don Zickus wrote:
> > On Fri, Jul 19, 2013 at 11:04:58AM +0200, Michal Hocko wrote:
> > > proc_dowatchdog doesn't synchronize multiple callers which
> > > might lead to confusion when two parallel callers might confuse
> > > watchdog_enable_all_cpus resp. watchdog_disable_all_cpus (e.g. watchdog
> > > gets enabled even if watchdog_thresh was set to 0 already).
> > > 
> > > This patch adds a local mutex which synchronizes callers to the sysctl
> > > handler.
> > 
> > Looks fine by me, except one little nitpick..
> > 
> > > 
> > > Signed-off-by: Michal Hocko <mhocko@suse.cz>
> > > ---
> > >  kernel/watchdog.c | 7 +++++--
> > >  1 file changed, 5 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/kernel/watchdog.c b/kernel/watchdog.c
> > > index 1241d8c..2d64c02 100644
> > > --- a/kernel/watchdog.c
> > > +++ b/kernel/watchdog.c
> > > @@ -520,13 +520,15 @@ int proc_dowatchdog(struct ctl_table *table, int write,
> > >  		    void __user *buffer, size_t *lenp, loff_t *ppos)
> > >  {
> > >  	int err, old_thresh, old_enabled;
> > > +	static DEFINE_MUTEX(watchdog_proc_mutex);
> > 
> > Should we just make this global instead of hiding it as a static inside a
> > function.  I don't know the kernel rules for deciding which approach makes
> > sense.  I know it is the same result in either case...
> 
> I've hidden it into the function to discourage from abusing it for
> something else and because the usage is nicely focused in this
> function. But I have no problem to pull it out.

I understand what you are saying.  I just remember in the past being asked
not to do that, which is why I had the question.  I guess we can leave it
as is.

Cheers,
Don