On Tue, 25 Sep 2007 14:38:38 +1000,
Rusty Russell <rusty@rustcorp.com.au> wrote:

quoted text
> Have you tested that *this* path works?  Let's take your first change as
> an example:
> 
> +       mutex_lock(&gdev->reg_mutex);
> +       __ccwgroup_remove_symlinks(gdev);
> +       device_unregister(dev);
> +       mutex_unlock(&gdev->reg_mutex);
> 
> Now, are you sure that calling cleanup_ccwgroup just after
> device_unregister() works?
> 
> static void __exit
> cleanup_ccwgroup (void)
> {
> 	bus_unregister (&ccwgroup_bus_type);
> }
> 

ccwgroup is a bit special. The ccwgroup drivers (say, qeth) will
unregister their ccwgroup_driver in their exit function. ccwgroup will
then unregister all devices bound to this driver (a ccwgroup device
without a driver does not make sense, since they are artifically
created by writing to a 'group' attribute provided by the driver). This
makes sure that ccwgroup cannot be unloaded while there are still
devices on the bus, so your example won't hit.

quoted text
> > I think it's too much work for the
> > users of the API and it will be easy to pass the wrong @owner and go
> > unnoticed.
> 
> But your shortcut insists that all module authors be aware that
> functions can be running after exit() is called.  That's a recipe for
> instability and disaster.

There are already problems like this with ->release().

<And no, I still haven't gotten around to testing and reviewing all
those patchsets, sorry>
-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/