Re: [PATCH] watchdog: core: factorize register error paths

[Damien Riegel <damien.riegel@savoirfairelinux.com>]

On Tue, Nov 24, 2015 at 12:45:48PM -0800, Guenter Roeck wrote:
> Hi Damien,
> 
> On Tue, Nov 24, 2015 at 03:15:35PM -0500, Damien Riegel wrote:
> > The commit adding the reboot notifier in the core introduced a new error
> > path in __watchdog_register_device, making error paths quite redondant.
> > 
> s/redondant/redundant/
> 
> > This commit factorizes all error paths that do some cleanup.
> > 
> > Signed-off-by: Damien Riegel <damien.riegel@savoirfairelinux.com>
> > ---
> >  drivers/watchdog/watchdog_core.c | 27 +++++++++++++++------------
> >  1 file changed, 15 insertions(+), 12 deletions(-)
> > 
> > diff --git a/drivers/watchdog/watchdog_core.c b/drivers/watchdog/watchdog_core.c
> > index 551af04..aff17b1 100644
> > --- a/drivers/watchdog/watchdog_core.c
> > +++ b/drivers/watchdog/watchdog_core.c
> > @@ -241,20 +241,16 @@ static int __watchdog_register_device(struct watchdog_device *wdd)
> >  		wdd->id = id;
> >  
> >  		ret = watchdog_dev_register(wdd);
> > -		if (ret) {
> > -			ida_simple_remove(&watchdog_ida, id);
> > -			return ret;
> > -		}
> > +		if (ret)
> > +			goto remove_ida;
> >  	}
> >  
> >  	devno = wdd->cdev.dev;
> >  	wdd->dev = device_create(watchdog_class, wdd->parent, devno,
> >  					wdd, "watchdog%d", wdd->id);
> >  	if (IS_ERR(wdd->dev)) {
> > -		watchdog_dev_unregister(wdd);
> > -		ida_simple_remove(&watchdog_ida, id);
> >  		ret = PTR_ERR(wdd->dev);
> > -		return ret;
> > +		goto unregister_device;
> >  	}
> >  
> >  	if (test_bit(WDOG_STOP_ON_REBOOT, &wdd->status)) {
> > @@ -264,11 +260,7 @@ static int __watchdog_register_device(struct watchdog_device *wdd)
> >  		if (ret) {
> >  			dev_err(wdd->dev, "Cannot register reboot notifier (%d)\n",
> >  				ret);
> > -			watchdog_dev_unregister(wdd);
> > -			device_destroy(watchdog_class, devno);
> > -			ida_simple_remove(&watchdog_ida, wdd->id);
> > -			wdd->dev = NULL;
> > -			return ret;
> > +			goto unregister_device;
> >  		}
> >  	}
> >  
> > @@ -282,6 +274,17 @@ static int __watchdog_register_device(struct watchdog_device *wdd)
> >  	}
> >  
> >  	return 0;
> > +
> > +unregister_device:
> > +	watchdog_dev_unregister(wdd);
> > +	if (!IS_ERR(wdd->dev)) {
> > +		device_destroy(watchdog_class, devno);
> > +		wdd->dev = NULL;
> 
> Oddly enough this leaves wdd->dev in place if it is an ERR_PTR.
> While that was the case before, it might be better and more
> consistent to always set it to NULL.
> 
> I also wonder if it is really necessary to call watchdog_dev_unregister()
> first. Something like
> 
> destroy_device:
> 	device_destroy(watchdog_class, devno);
> unregister_watchdog:
> 	wdd->dev = NULL;
> 	watchdog_dev_unregister(wdd);
> 
> would be cleaner. Wonder why it isn't done in that order to start with.
> It should be possible; after all, the device is created only after 
> watchdog_dev_register() succeeded, so one should think that it makes
> sense to remove it first. Any idea ?

No idea why the device_destroy was done after watchdog_dev_unregister in
__watchdog_unregister_device, I think there is no reasons for that, and
it seems safe to reorder them.