Re: [gentoo-desktop] Re: startx / X -config /root/xorg.conf.new crash on T410 intel Arrandale (custom kernel)

[Levon Ghazaryan]

Thanks Duncan for hints and the thorough reply, I am also pretty sure
it's a kernel crash.
No logs are leaved, and indeed as I read on wiki.x.org the only way
is to log remotely (there is also a guide how).

in meantime i tried to use a different kernel, though the same version but
vanilla source, the effect was the same.

But i could run ubuntu from the life-cd, it uses the 2.6.32-21, so
it seems to me you are right, and it is KMS "crashing" the kernel.

I'll post the results of my trials this weekend, and many thanks meanwhile,

~levon

On Thu, Aug 5, 2010 at 9:51 AM, Duncan <1i5t5.duncan@×××.net> wrote:
> Levon Ghazaryan posted on Wed, 04 Aug 2010 13:27:18 +0200 as excerpted:
>
>> on 2.6.34-r1, ThinkPad T410 with Intel Arrandale with a custom kernel
>> the laptop freezes on startx or X -retro -config /root/xorg.conf.new
>> (with a fresh created xorg.conf.new with Xorg -config).
>>
>> as this happens I'm not able to switch the tty, the laptop doesn't
>> respond to ping and caps lock is constantly blinking.
>
> Won't even respond to ping.  That's definitely a hard lockup. =:^(
>
> Unfortunately, such lockups often don't leave a lot in the way of logs,
> because whatever triggers them panics the kernel to the point it can't
> trust itself to write log entries (good thing too, at that point it's
> generally so confused who knows where it might scribble on the disk,
> precisely the reason it doesn't write anything if it doesn't trust itself
> to do so correctly) detailing what went wrong.
>
> But you can tell it doesn't respond to pings, which presumably means you
> have at least one other machine available.  It's often possible in such
> cases to take a log remotely, and sometimes get an entry with the problem
> via the remote connection as the kernel's going down, since the kernel
> knows that writing to a network connection can't scribble where it's not
> supposed to on the disk, like trying to right a log entry to disk might do
> at that point.
>
> Unfortunately I don't know those details, as it's only relatively recently
> that I have both a netbook and a desktop, to be able to do such things,
> and I've not actually done them yet.  But I do know it's possible, and you
> can look into it further if you find it necessary.
>
>> there is no error in /var/log/Xorg.0.log
>
> One wouldn't be expected in such a case, for the reasons I mention above.
>
>> and Xorg -config exits without complaining about anything.
>
> =:^(
>
>> I followed the guide at:
>> http://www.gentoo.org/doc/en/xorg-config.xml#using_startx
>>
>> so I completed all steps mentioned there.
>>
>> at this point I have no idea how to debug, get information what is going
>> an, or what causes the problem - there are no error messages and no log
>> left. the only thing I see is a black screen, a blinking cursor at the
>> top left corner for a moment and a completely black screen after wards
>> and laptop the responds to nothing as a result. the only way that i
>> could get out of this was to hard-reboot using the power button.
>>
>> please post here if you have any ideas what this could be or how can it
>> be figured out what causes the crash.
>>
>> at some point i tried:
>> emerge -e xorg-server
>>
>> but this changed nothing and startx crashes again in the same way.
>
> Altho my netbook has Intel graphics, I've not had it as long as my main
> machine (AMD, several generations of Radeon graphics), and know rather
> less about Intel graphics than I do Radeons.
>
> However, both the Radeon and Intel drivers now (with 2.6.34 on the Radeon
> side, I believe earlier for the Intel side) default to KMS, kernel mode
> setting, as opposed to the former UMS, user (xorg) mode setting.  KMS runs
> fine on both my main AMD/Opteron/Radeon machine and my Intel/Atom based
> Acer Aspire One netbook (ICH7 family chipset, 945GME graphics, rev 03),
> but I'm running ~arch on both (~amd64 on the workstation, ~x86 on the
> netbook), and in fact, running the x11 overlay as well, so getting X
> related packages before they're even in the main tree.  With a technology
> as new as is KMS, the newest versions are very likely more stable than
> earlier versions. as the technology itself is still developing and
> stabilizing.  FWIW, xorg-server-1.8.2 (from the x11 overlay), on both
> machines, here, and the xf86-video-*, drm, mesa, and other such packages,
> are equally current, some in-tree already, some from the overlay.
>
> I'd therefore suggest you look into the KMS/UMS thing.  That may well be
> your problem.    I believe there's a kernel parameter you can add in grub,
> to turn off KMS and see if that's it.  nokms or no-kms or some such, I've
> not had to use it so IDR for sure.
>
> Actually, now that I think of it, I believe I've read about a particular
> Intel chipset, the ICH-5 series, IIRC, that has had very serious problems
> with KMS, and either has /just/ fixed them (would likely be with the
> 2.6.35 kernel and/or xorg-1.8 or later and/or comparable xf86-video-intel
> driver, the fix would be that fresh, tho of course some of us have been
> running that stuff for months, now, so it's not necessarily /that/ new),
> or they're still affected and won't yet work with kms at all.  I've no
> clue where Arrandale is relative to ICH-5, but if that's it, you're very
> likely affected and need the nokms boot parameter, at least with xorg
> earlier than the very latest ~arch and x11 overlay stuff, and with kernels
> older than the just release 2.6.35, and might still need it with them.
>
> --
> Duncan - List replies preferred.   No HTML msgs.
> "Every nonfree program has a lord, a master --
> and if you use the program, he is your master."  Richard Stallman
>
>
>