[ubuntu-x] [PPC] KMS is hurting. Can I help?

[Str8bs]

Mr. Harrington,

Thank you for the response, scripts, and advice. 
Sadly, my NV34 PPC issue does not appear to be 982889 as the workarounds do
not seem to help. I will check for it in other bug reports.
I will also look in to getting more involved with testing.

To everyone:
I will attempt to assist in triaging bugs related to graphics on PPC to the
best of my ability. Please don't hesitate to bluntly point out the many
mistakes I'm sure to make.

BTW, I've never been a "Mac Head" and have no explanation for my obsession
with PPC graphics. :) I think maybe I perceive a possible significant impact
there through FOSS driver improvement due to an understandable "lack of
understanding the purpose of a GPU" in the general PPC mac community. I
should probably stop messing with PPC to save myself frustration, but I want
to try and contribute knowing full well odds are against developers having
time to work on these. The iMacs are still great for kids.

I have a mishmash of cards around if anyone needs some testing. Admittedly,
I've only tried Linux on a few of these so far.
[AMD64]
Radeon Caicos/HD6450, RV710/HD4570, RV370/X300SE
Nouveau NVA5(GT216)/GT220, NV46(G72)/7300SE, and ... NV04/TNT2 :)
Intel GMA950, 965, GMAX4500HD
[PPC]
Nouveau NV34,FX5200
R128 Rage128Ultra
Mach64 RagePro

Thanks again,
Str8


-----Original Message-----
From: Bryce Harrington [mailto:bryce at canonical.com] 
Sent: Wednesday, April 17, 2013 5:20 PM
To: Str8bs
Cc: ubuntu-x at lists.ubuntu.com
Subject: Re: [ubuntu-x] [PPC] KMS is hurting. Can I help?

On Wed, Apr 17, 2013 at 03:15:11PM -0500, Str8bs wrote:
> Greetings all,
> 
> There are multiple bugs apparently related to KMS with both nouveau 
> and radeon. This is causing blank screen issues for many PPC users 
> from 12.10 through Raring daily tests getting failed. Those who are 
> able to fall back to FBDEV will experience "slow" performance due to
software rendering.

Black screen issues can have a wide variety of underlying causes, but be
aware of bug #982889, a race condition that exists between plymouth and
lightdm, which can cause certain kinds of boot failures quite similar to
what you're describing.  There is a known workaround (add sleep to
lightdm.conf) which can be used to rule in or out that bug as related to
what you're seeing.  There's a few other work arounds and checks - see the
bug report for details.

Assuming it to be the same issue as that, we're hoping to see a plymouth fix
soon.  If you're interested in helping test things, join #ubuntu-x on
freenode and talk to tjaalton.

> I am not qualified to become a part of your team and no use whatsoever 
> at creating patches. I am very new to Linux myself. However, I could 
> attempt to help triage bugs by cleaning / clarifying the titles, tags 
> and soliciting logs from other reporters. Maybe help them with SSH set 
> up to obtain more logs. Would that be of any help?
> 
> As an example, I attempted to follow the guidelines 
> https://wiki.ubuntu.com/X/Debugging in my own bug report -
> https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-nouveau/+
> bug/11
> 70054
> 
> If this would be helpful, could someone clarify the following:
> Should I -
> Just go ahead and retitle / retag bugs and mark bugs missing logs as 
> incomplete?
> Or, just suggest the reporter do so and link to debugging / triaging X
wiki?
> Link to other arches / upstream reports found?

Any/all of that would definitely be helpful.  Linking to upstream reports or
patches is the most helpful, since it makes it a lot easier for package
maintainers to chase down fixes.

Fwiw, I use a greasemonkey script that adds buttons to the bug report that
lets me easily paste in various standard debugging tips:

  https://code.launchpad.net/~bryce/launchpad-gm-scripts/main

Or if you just want to see the tips themselves:

 
http://bazaar.launchpad.net/~bryce/launchpad-gm-scripts/main/view/head:/ubun
tu-x-replies.xml

For many bugs, steps to reproduce are vital.  Often the logs and backtraces
don't paint a complete picture, so being able to reproduce the issue is key.


> Or, leave that to someone more experienced?

There's more than enough bug reports to go around!  :-)

Seriously though, there's scant few people experienced enough to do bug
work, and with Mir we're being stretched ever thinner.  We do try to stay
atop the bug reports against the current development release that were
reported through apport; anything else is probably not going to get
attention.

Bryce