Re: [PATCH 1/5] pxa: poodle: remove unnecessary parent assignment to the fb device

[Richard Purdie <rpurdie@xxxxxxxxx>]

On Fri, 2008-07-18 at 16:04 +0800, eric miao wrote:
> On Fri, Jul 18, 2008 at 3:45 PM, Richard Purdie <rpurdie@xxxxxxxxx> wrote:
> >
> > On Fri, 2008-07-18 at 09:12 +0800, eric miao wrote:
> >> This caused kref_get() to complain about incorrect kobject reference
> >> in set_pxa_fb_info(), due to the parent being registered later.
> >>
> >> And is this necessary anyway?
> >
> > This was present to ensure that the framebuffer was a child of the
> > locomo device. This ensured at suspend/resume that the framebuffer was
> > suspended first, then the locomo device. Have you tested suspend/resume?
> 
> Not really, so it should really be RFC. Otherwise, the parent device
> should be registered earlier.

Registering the parent device earlier would be better otherwise the
framebuffer can potentially make calls into locomo after its suspended
and I seem to remember that causing some issues. There is a similar
setup with scoop on corgi/spitz if I remember rightly.

Cheers,

Richard


-------------------------------------------------------------------
List admin: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel
FAQ:        http://www.arm.linux.org.uk/mailinglists/faq.php
Etiquette:  http://www.arm.linux.org.uk/mailinglists/etiquette.php