NEW_XORG and vt(4) in stable branches

[John Baldwin]

On Friday, February 14, 2014 3:29:35 pm Kevin Oberman wrote:
> On Fri, Feb 14, 2014 at 12:24 AM, David Chisnall 
<theraven at freebsd.org>wrote:
> 
> > On 14 Feb 2014, at 02:12, Kevin Oberman <rkoberman at gmail.com> wrote:
> >
> > > I'm just slightly confused by this. I am unaware of any reason that the
> > use of NEW_XORG requires vt(4). KMS certainly does, but NEW_XORG should
> > not, as far as I can tell. At least it does not on my system. I do believe
> > that NEW_XORG will break some really old graphics cards, but I don't see
> > how vt(4) will help this.
> > >
> > > Am I missing something?
> > >
> > > And I am very anxious to see vt(4) merged into 9 and 10, but I don't see
> > how it impacts moving to NEW_XORG as default.
> >
> > KMS is required for several of the new drivers.  Without KMS, NEW_XORG
> > means Intel GPUs can only use VESA (the same is true for radeon, but the
> > old radeon driver doesn't work at all with newer cards so it's not as much
> > of a regression).  With KMS and without vt(4), starting X means losing
> > consoles.  The only way to introduce NEW_XORG without introducing feature
> > regressions is to have vt(4).  This is why we asked the Foundation to fund
> > the vt(4) work.
> >
> 
> Yes, I agre. I have an Intel GPU and it either must run KMS or VESA (which
> really sucks). But, AFAIK, NEW_XORG does not break VESA, so making NEW_XORG
> default regresses nothing. I am running with KMS and have no access to vtys
> once I start X. I can see no regression from "It's broken" to "It's
> broken". And VESA is pretty broken on it, though I lived with it for about
> 9 months until the first release of KMS for testing on 9.
> 
> Is it the case that NEW_XORG breaks syscons on some pre-Sandybridge systems
> that would work with UMS but won't work with NEW_XORG? If that is the case,
> I do understand. I was not aware of this issue.

I believe there are older Intel GPUs that work without VESA with the old
Xorg server (and can still switch vtys, etc.) that NEW_XORG would be a
regression for.  OTOH, I also want to push vt(4) to the stable branches in
general as an option.  I used NEW_XORG with KMS for a while without vt(4)
and compared to all my previous experiences with X and laptops on FreeBSD,
the lack of being able to exit X (or restart X to try to change settings,
etc.) was quite a PITA.

-- 
John Baldwin