On Tue, Sep 10, 2024 at 02:18:35PM +0200, Arthur Borsboom wrote:
> On Tue, 10 Sept 2024 at 10:33, Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> >
> > On Tue, Sep 10, 2024 at 10:13:01AM +0200, Roger Pau Monné wrote:
> > > On Tue, Sep 10, 2024 at 09:29:30AM +0200, Thomas Zimmermann wrote:
> > > > Hi
> > > >
> > > > Am 10.09.24 um 09:22 schrieb Roger Pau Monné:
> > > > > On Mon, Sep 09, 2024 at 10:09:16PM -0400, Jason Andryuk wrote:
> > > > > > From: Jason Andryuk <jason.andryuk@xxxxxxx>
> > > > > >
> > > > > > Probing xen-fbfront faults in video_is_primary_device(). The passed-in
> > > > > > struct device is NULL since xen-fbfront doesn't assign it and the
> > > > > > memory is kzalloc()-ed. Assign fb_info->device to avoid this.
> > > > > >
> > > > > > This was exposed by the conversion of fb_is_primary_device() to
> > > > > > video_is_primary_device() which dropped a NULL check for struct device.
> > > > > >
> > > > > > Fixes: f178e96de7f0 ("arch: Remove struct fb_info from video helpers")
> > > > > > Reported-by: Arthur Borsboom <arthurborsboom@xxxxxxxxx>
> > > > > > Closes: https://lore.kernel.org/xen-devel/CALUcmUncX=LkXWeiSiTKsDY-cOe8QksWhFvcCneOKfrKd0ZajA@xxxxxxxxxxxxxx/
> > > > > > Tested-by: Arthur Borsboom <arthurborsboom@xxxxxxxxx>
> > > > > > CC: stable@xxxxxxxxxxxxxxx
> > > > > > Signed-off-by: Jason Andryuk <jason.andryuk@xxxxxxx>
> > > > > Reviewed-by: Roger Pau Monné <roger.pau@xxxxxxxxxx>
> > > > >
> > > > > > ---
> > > > > > The other option would be to re-instate the NULL check in
> > > > > > video_is_primary_device()
> > > > > I do think this is needed, or at least an explanation. The commit
> > > > > message in f178e96de7f0 doesn't mention anything about
> > > > > video_is_primary_device() not allowing being passed a NULL device
> > > > > (like it was possible with fb_is_primary_device()).
> > > > >
> > > > > Otherwise callers of video_is_primary_device() would need to be
> > > > > adjusted to check for device != NULL.
> > > >
> > > > The helper expects a non-NULL pointer. We might want to document this.
> > >
> > > A BUG_ON(!dev); might be enough documentation that the function
> > > expected a non-NULL dev IMO.
> >
> > No need for that, don't check for things that will never happen.
>
> And yet, here we are, me reporting a kernel/VM crash due to a thing
> that will never happen, see 'Closes' above.
>
> I don't want to suggest BUG_ON is the right approach; I have no idea.
> I just want to mention that (!dev) did happen. :-)
A BUG_ON() will cause the same crash, so I don't see your point, sorry.
greg k-h
