mga with default depth 16 turns computer off???

Alex Deucher alexdeucher at gmail.com
Fri Dec 20 14:40:49 PST 2013


On Fri, Dec 20, 2013 at 5:03 PM, Felix Miata <mrmazda at earthlink.net> wrote:
> On 2013-12-20 08:53 (GMT-0500) Alex Deucher composed:
>
>> On Felix Miata wrote:
>
>
>>> This 1.13GHz i815 Piii with G400 AGP is multiboot which I've been
>>> updating
>>> over the past couple of days, Fedora 21, openSUSE 12.3, 13.1 & 13.2. It's
>>> been reliable for years, but tonight it has been acting like the power
>>> plug
>>> has been pulled shortly after KDE finishes initializing my desktop. This
>>> happened first in 13.2 a few times with 3.11.10 kernel, so I tried and
>>> got
>>> same result in 13.1 with 3.11.6 kernel. Previously xorg.conf was written
>>> with default depth 24, but tonight I changed it to 16, and only then did
>>> the
>>> crashing start. I've currently booted back to 13.2 but with xorg.conf
>>> changed back to default depth 24 and it hasn't crashed yet. So it seems
>>> like
>>> 16 bit X with MGA is broken, while 24 is not. Am I missing something? I
>>> don't see any open bugs that look anything like this, except for one
>>> ancient
>>> one: https://bugs.freedesktop.org/show_bug.cgi?id=8237
>
>
>> If you can narrow down which component caused the problem (xserver,
>> mesa, mga ddx) you may be able to bisect it and figure out what caused
>> it.  Unfortunately mga chips are pretty old and thus not well tested
>> on modern software stacks these days.
>
>
> Thanks! Further investigation revealed the problem is in openSUSE and/or
> KDE3, so I filed a Novell bug:
> https://bugzilla.novell.com/show_bug.cgi?id=856606
>
> Good thing there and not xserver, mesa, mga ddx, because I found no kind of
> recognizable clues in dmesg or any logs.
>
> FWIW, Dell T610 PowerEdge servers as recently as 5 years ago were shipped
> with G200eW chips on the motherboard. More importantly, MGA users do not
> suffer from https://bugs.freedesktop.org/show_bug.cgi?id=39949

There's actually a kms driver for the mga server chips that is
probably better tested at this point than the old ums ddx.

Alex


More information about the xorg-devel mailing list