Menu

#407 MGA locks system on heavy AGP traffic

closed
MGA OpenGL (56)
5
2003-05-12
2002-02-09
Anonymous
No

I recently built a new system, and the graphics keep
locking up whenever I use hardware accelerated OpenGL.

The lockups appear to be related to the amount of data
passing over the AGP bus. Polygon or texture intensive
applications such as Return to Castle Wolfenstein crash
the system faster than simple applications (such
glxgears). Most of the time I see texture corruption
occuring before the crash.

On IRC yesterday evening (Sunday Feb 03) one other user
confirmed having the *exact* same problem.

My first thought was that it was the AMD/AGP bug, but
the workaround for that one does not bring any
noticeable relief.

Following are my system stats, and a copy of a kernel
trap from my logs.

Thanks in advance for *any* help.

Mart

System:

AMD Athlon XP 1600 CPU
MSI K7Pro2 motherboard with VIA KT266 chipset
Matrox G450DH 4x AGP graphics card
256M PC2100 DDR RAM
64M AGP aperture.
Kernel 2.4.17
XFree86 4.1
Debian unstable

Kernel log info attached.

Discussion

  • Nobody/Anonymous

    Logged In: NO

    Right.

    I got an error trying to attach my kernel log. I could try
    cutting and pasting it in here, but I don't know how the
    formatting turns up. If it's unreadable mail me for a copy.

    Mart (mvdwege@drebbelstraat20.dyndns.org)

    Kernel log:

    Feb 3 23:48:36 drebbelstraat20 kernel: invalid operand: 0000
    Feb 3 23:48:36 drebbelstraat20 kernel: CPU: 0
    Feb 3 23:48:36 drebbelstraat20 kernel: EIP:
    0010:[mga_g400_emit_state+826/1360] Not tainted
    Feb 3 23:48:36 drebbelstraat20 kernel: EFLAGS: 00210213
    Feb 3 23:48:36 drebbelstraat20 kernel: eax: ffff0207 ebx:
    d095d8cc ecx: d2974000 edx: 00000000
    Feb 3 23:48:36 drebbelstraat20 kernel: esi: d095d898 edi:
    cd53a681 ebp: d095d898 esp: c1ebff0c
    Feb 3 23:48:36 drebbelstraat20 kernel: ds: 0018 es: 0018
    ss: 0018
    Feb 3 23:48:36 drebbelstraat20 kernel: Process wolf.x86
    (pid: 27747, stackpage=c1ebf000)
    Feb 3 23:48:36 drebbelstraat20 kernel: Stack: cd53a680
    ca7cf7c0 000000c0 d095d898 00000000 ffffffff c01c11b4 cd53a680
    Feb 3 23:48:36 drebbelstraat20 kernel: cd53a680
    ca7cf7c0 ca80c1c8 00000000 00000000 e0168000 ca7cf7c0 c01c1bb5
    Feb 3 23:48:36 drebbelstraat20 kernel: cff44000
    ca80c1c8 00000045 cff44000 c5818180 400c6445 c4aaa800 cff44000
    Feb 3 23:48:36 drebbelstraat20 kernel: Call Trace:
    [mga_dma_dispatch_vertex+132/416] [mga_dma_vertex+389/416]
    [mga_ioctl+210/224] [sys_ioctl+359/384] [system_call+51/56]
    Feb 3 23:48:36 drebbelstraat20 kernel:
    Feb 3 23:48:36 drebbelstraat20 kernel: Code: 8f 8c 96 97 8b
    53 04 80 ce 80 89 54 08 04 8b 56 34 89 54 08
    Feb 3 23:48:36 drebbelstraat20 kernel: invalid operand: 0000
    Feb 3 23:48:36 drebbelstraat20 kernel: CPU: 0
    Feb 3 23:48:36 drebbelstraat20 kernel: EIP:
    0010:[mga_g400_emit_state+826/1360] Not tainted
    Feb 3 23:48:36 drebbelstraat20 kernel: EFLAGS: 00210213
    Feb 3 23:48:36 drebbelstraat20 kernel: eax: ffff0207 ebx:
    d095d8cc ecx: d2974000 edx: 00000000
    Feb 3 23:48:36 drebbelstraat20 kernel: esi: d095d898 edi:
    cd53a681 ebp: d095d898 esp: c1ebff0c
    Feb 3 23:48:36 drebbelstraat20 kernel: ds: 0018 es: 0018
    ss: 0018
    Feb 3 23:48:36 drebbelstraat20 kernel: Process wolf.x86
    (pid: 27747, stackpage=c1ebf000)
    Feb 3 23:48:36 drebbelstraat20 kernel: Stack: cd53a680
    ca7cf7c0 000000c0 d095d898 00000000 ffffffff c01c11b4 cd53a680
    Feb 3 23:48:36 drebbelstraat20 kernel: cd53a680
    ca7cf7c0 ca80c1c8 00000000 00000000 e0168000 ca7cf7c0 c01c1bb5
    Feb 3 23:48:36 drebbelstraat20 kernel: cff44000
    ca80c1c8 00000045 cff44000 c5818180 400c6445 c4aaa800 cff44000
    Feb 3 23:48:36 drebbelstraat20 kernel: Call Trace:
    [mga_dma_dispatch_vertex+132/416] [mga_dma_vertex+389/416]
    [mga_ioctl+210/224] [sys_ioctl+359/384] [system_call+51/56]
    Feb 3 23:48:36 drebbelstraat20 kernel:
    Feb 3 23:48:36 drebbelstraat20 kernel: Code: 8f 8c 96 97 8b
    53 04 80 ce 80 89 54 08 04 8b 56 34 89 54 08

     
  • Mart van de Wege

    Logged In: YES
    user_id=455122

    resubmitted as #522096, this time as a logged in user with
    some more debug info attached. This one can be closed now.

     
  • Ian Romanick

    Ian Romanick - 2003-05-12
    • assigned_to: nobody --> idr
    • status: open --> closed
     
  • Ian Romanick

    Ian Romanick - 2003-05-12

    Logged In: YES
    user_id=423974

    Duplicate of #546281.

     

Log in to post a comment.