Re: [patch] gpiolib: fix oops in gpio_get_value_cansleep()

[Andrew Morton <akpm@linux-foundation.org>]

On Thu, 16 Oct 2008 17:44:33 -0700 David Brownell <david-b@pacbell.net> wrote:

> On Thursday 16 October 2008, Andrew Morton wrote:
> > > From: David Brownell <dbrownell@users.sourceforge.net>
> > > 
> > > We can get the following oops from gpio_get_value_cansleep()
> > > when a GPIO controller doesn't provide a get() callback:
> > 
> > We can, but do we? ;)
> 
> I think it's unlikely without the sysfs interface.
> 
> 
> > iow: is this needed in any -stable release?
> 
> The bug has been there since 2.6.25 but nobody else seems
> to have reported it.  Is the general policy to fix all
> oopses that *could* appear?  I'd send it for 2.6.27-stable,
> since that's got the sysfs hooks.  And older kernels if
> bug likelihood isn't a major concern.

OK.  2.6.27 definitely (major distros are basing on that).

As for earlier kernels: I'd say so.  An oops is farily serious. 
Although an oops in a sysfs handler tends to be fairly tame, as the
code usually doesn't hold locks or many allocated resources.

Anyway - making decisions like this is why we pay stable@kernel.org the
big bucks :)