[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#182773: xserver-xfree86: [mga] Any opengl app on 4.2.1-5 and 4.2.1-6 crashes the hardware hard.



On Thu, Feb 27, 2003 at 08:04:55PM +0000, Phil Armstrong wrote:
> Package: xserver-xfree86
> Version: 4.2.1-6
> Severity: normal
> 
> Just running "glxinfo" on 4.2.1-5 or 4.2.1-6 is enough to put my G400
> in an unusable state, with screen corruption all over the
> place. Switching to another console is possible, but the corruption
> remains; switching back to the X console hangs the system completely.
> 
> 4.2.1-4 appears to be entirely stable. I think I missed this bug on
> 4.2.1-5 due to running the experimental dri packages at the time.
> 
> Let me know if there's any diagnostics I can run.

Here's a summary of Debian patches to the mga driver:

debian/patches/000_stolen_from_HEAD.diff:

    + Fix pixmap cache corruption on VT switches in mga driver (#A.1279, Ross
      Mikosh).  (added in 4.2.1-4)

    + (xc/programs/Xserver/hw/xfree86/drivers/mga/mga_driver.c): Add reporting
      of G400/G450 in MGA driver (Andrew C. Aitchison).  (added in 4.2.1-6)

debian/patches/016_mga_manpage_updates.diff: manpage updates only,
  unlike to affect the MGA driver's operation (added in 4.2.1-0pre1v2)

debian/patches/023_mga_use_hal_modeinit_workarounds_for_g550.diff: use
  the buggy-HAL workarounds in mode initialization for G550 as well as
  previous Matrox cards (added in 4.2.1-1)

debian/patches/063_mga_texmode_restore_fix.diff:  fix for nonstandard
  textmode restoration on Matrox G450 (added in 4.1.0-13)

As you can see, not much has really changed in the MGA driver since
4.2.1-4.

If you're handy with Debian packaging, you could roll a custom XFree86
4.2.1-6 package with above patches removed, and see if that changes
anything.  If not, it's an upstream problem.

Note that the switch to the GCC 3.2-based toolchain has had somewhat
negative consequences on the Mesa 3D acceleration modules; for instance,
the Radeon accelerator is completely inoperative.  This, too, is an
upstream problem.  So another experiment might be to compile XFree86
4.2.1-6 with gcc-2.95 and see if that makes the problem go away.

I don't have a Matrox card, so I'm going to need help to track this down
(this message will be CCed to the debian-x mailing list.)

Alternatively, we can just hope that the problem is fixed in XFree86
4.3.0, which was just released, and for which there will be Debian
packages as soon as is feasible.

-- 
G. randen Robinson                |    If you make people think they're
Debian GNU/Linux                   |    thinking, they'll love you; but if
branden@debian.org                 |    you really make them think, they'll
http://people.debian.org/~branden/ |    hate you.

Attachment: pgpsefj0926sD.pgp
Description: PGP signature


Reply to: