vc4: NULL pointer dereference in drm_client_dev_hotplug

[Stefan Wahren]

Hi,

> Andreas Färber <afaerber at suse.de> hat am 2. Oktober 2018 um 16:33 geschrieben:
> 
> 
> Hi Stefan and Daniel,
> 
> Am 02.10.18 um 11:48 schrieb Stefan Wahren:
> > Hi Daniel,
> SNIP
> > personally i didn't know this option before this issue, but i cannot
> > speak for all the distributions. I checked Raspbian and they don't use
> > this option. I had a better feeling to have at least the feedback from
> > Peter and Andreas this isn't used in their distributions.
> 
> For openSUSE kernel-source.git master branch I see:
> 
> $ grep -r CONFIG_DRM_FBDEV_OVERALLOC config/
> config/armv7hl/default:CONFIG_DRM_FBDEV_OVERALLOC=100
> config/armv7hl/lpae:CONFIG_DRM_FBDEV_OVERALLOC=100
> config/ppc64le/default:CONFIG_DRM_FBDEV_OVERALLOC=100
> config/i386/pae:CONFIG_DRM_FBDEV_OVERALLOC=100
> config/ppc64/default:CONFIG_DRM_FBDEV_OVERALLOC=100
> config/x86_64/default:CONFIG_DRM_FBDEV_OVERALLOC=100
> config/armv6hl/default:CONFIG_DRM_FBDEV_OVERALLOC=100
> config/arm64/default:CONFIG_DRM_FBDEV_OVERALLOC=100
> 
> Similar to what Peter said, on the Raspberry Pi we would usually use vc4
> drm; but some users have run into issues with vc4 not working with
> certain monitors, so they may blacklist vc4 and use efifb instead.

thanks for clarification.

> 
> Adding Alex and Petr in case more discussion is needed.
> 
> Regards,
> Andreas
> 
> >>  The new generic fbdev stuff has slightly more
> >> strict error checking, and the overalloc thing is somewhat of a hack to
> >> support mali blobs. If this goes boom now there's a good chance it didn't
> >> work beforehand either.
> >> -Daniel
> 
> -- 
> SUSE Linux GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
> GF: Felix Imendörffer, Jane Smithard, Graham Norton
> HRB 21284 (AG Nürnberg)