[RFC] Deprecation and removal of the drm2 driver

[Matthew Macy]

On Mon, Jun 4, 2018 at 12:03 PM, Adrian Chadd <adrian.chadd at gmail.com> wrote:
> Hi,
>
> As a driver/framework developer - no, don't do that.
>
> It's worked mostly great for the video side of things because your
> touch points are "the VM system" and "linuxkpi". And they're all in
> one big driver pull from Linux.
>
> For wifi as an example - it has a bunch of userland components, a
> kernel framework component (net80211); it gets API churn from people
> who keep making networking API changes without making them opaque (i
> just got bitten by the STAILQ -> CK_STAILQ changes for multicast
> iteration, instead of us growing a multicast iterator function thing.)

We've had one for several years. You're just not using it.

 > Having it be multiple drivers/firmware means that anyone doing wifi
> development here would have to install /all/ of the relevant packages
> and the net80211 stuff and userland just to get any work done and hope
> it stays in sync.

This is the same old saw of people who can't be bothered to use ports.
It is more of a headache with ABI drift but it's certainly not a
fundamental impediment.

-M