[PATCH 11/13] drm: remove redundant minor->device field

[David Herrmann]

Hi

On Fri, Feb 21, 2014 at 8:30 AM, Thierry Reding
<thierry.reding at gmail.com> wrote:
> On Wed, Feb 12, 2014 at 02:36:24PM +0100, Daniel Vetter wrote:
>> On Wed, Jan 29, 2014 at 03:01:58PM +0100, David Herrmann wrote:
>> > Whenever we access minor->device, we are in a minor->kdev->...->fops
>> > callback so the minor->kdev pointer *must* be valid. Thus, simply use
>> > minor->kdev->devt instead of minor->device and remove the redundant field.
>> >
>> > Signed-off-by: David Herrmann <dh.herrmann at gmail.com>
>>
>> I think this is simply compat cruft from the days when the drm core was
>> still shared with the *bsds. With the one patch I've commented on all
>> patches up to this one are
>>
>> Reviewed-by: Daniel Vetter <daniel.vetter at ffwll.ch>
>>
>> As discussed on irc I think we don't want to have stable minor ids really,
>> userspace simply needs to inquire udev to get at the right
>> render/control/legacy node it wants.
>
> Does that mean we should go all the way and don't keep the +64 (for
> control) and +128 (for render nodes) offsets either? Should it be
> possible to have a /dev/dri directory that looks somewhat like this:
>
>         /dev/dri/card0    (GPU#0, legacy)
>         /dev/dri/card1    (GPU#1, legacy)
>         /dev/dri/render0  (GPU#1, render)

That might break backwards compat, but may be worth it. However, we
*have* to keep the +64 / +128 offsets for minor numbers. There's
already user-space using that for dev-type testing (which is fine!).

Thanks
David