[Intel-gfx] [PATCH 1/4] drm: Protect master->unique with dev->master_mutex

[Emil Velikov]

On 13 December 2016 at 09:46, Daniel Vetter <daniel at ffwll.ch> wrote:
> On Tue, Dec 13, 2016 at 10:42 AM, Michel Dänzer <michel at daenzer.net> wrote:
>>> Hm, I thought the grand plan is to use -modesetting almost everywhere and
>>> forget about all the others?
>>
>> Maybe if you mean s/grand plan/pipe dream/ ...
>
> I said "almost everywhere", not "everywhere". I'm fully aware that
> there's tons of old chips that need dedicated drivers, and that amd
> wants to do some fancy stuff in their -pro version. But besides those
> I do see a fairly clear push towards having standardized/generic kms
> userspace, so don't really see that much value in updating all the
> vendor specific drivers. And outside of X the push is even stronger.
>
> So no pipe dream at all, it's happening.
Precisely.

As Michel pointed out:
There will [always] be compelling reasons why users/distros prefer
xf86-video-foo over the modesetting one.
So, barring time concerns, updating those is good idea, imho.

Pretty we don't want another DRIVER_KMS_LEGACY_CONTEXT, because user X
decided to stay with the final version of xf86-video-foo which still
uses nasty libdrm/ioctl Y ;-)

Thanks
Emil