Re: drm2 in base

On Fri, Sep 28, 2018 at 02:21:24PM -0600, Warner Losh wrote:
> On Fri, Sep 28, 2018 at 2:02 PM Steve Kargl <
> sgk_at_troutmask.apl.washington.edu> wrote:
> 
> > On Fri, Sep 28, 2018 at 02:47:39PM -0500, Matthew D. Fuller wrote:
> > > On Fri, Sep 28, 2018 at 11:00:02AM -0700 I heard the voice of
> > > Steve Kargl, and lo! it spake thus:
> > > >
> > > > I have a radeon HD 6450 video card (CAICOS firmware), and have been
> > > > informed that the hardware is too old for graphics/drm-stable-kmod
> > > > and that I should use graphics/drm-legacy-kmod.
> > >
> > > Well, please don't tell that to the 6450 I've got running on
> > > drm-next-kmod (which also spent some time on drm-stable)...
> > >
> >
> > See the URL I posted.
> >
> > How did you get around the "fence_wait returned with error -512"
> > messages filling /var/log/messages.
> >
> 
> Ah, that thread...  OK. I'm back with you...
> 
> > It also does not get around the problem that
> > drm-legacy-kmod has been advertised as the drop-in
> > replacement for those that use drm2 and have older
> > hardware.
> >
> 
> It should be. The fact that it's not is concerning and should be
> addressed...
> 

Yes, it is concerning, and the reason for my request to retain
drm2 in head.  I'm willing to work with Johannes and Niclas 
or whomever to get things working.  Unfortunately, kernel hacking
is a little beyond my skills. 

AFAICT, the drm-*-kmod ports all depend on the same gpu-firmware
port, which as the name indicates holds all the firmware.  It
seems that drm-legacy-kmod needs to learn how to parse the new
naming scheme (and may be a new scheme for packing the firmware
into a *.ko file).

-- 
Steve