#72 emu10k1 Segfault Edit

open
Rui Sousa
install (14)
5
2012-07-08
2003-07-23
Anonymous
No

I've sucessfully compiled both emu10k1 .19a and .20a,
but when I try to modprobe emu10k1 (for either version)
I get a segfault. dmesg has a big spit about "Unable to
handle kernel NULL pointer deference at virtual address
00000000". lsmod shows me that the emu10k1 module is
stuck trying to initalize and if I try to play any
sounds the process locks up. Finaly when I try to
reboot I get stuck when it trys to unconfigure my
network card. I'm running Debian 3.0 and a 2.4.21
kernel. Any ideas?

Finn.
finn@finerdetails.com.au

dmesg result:

emu10k1: Audigy rev 3 model 0x0053 found, IO at
0xb800-0xb81f, IRQ 11
Unable to handle kernel NULL pointer dereference at
virtual address 00000000
printing eip:
00000000
*pde = 00000000
Oops: 0000
CPU: 0
EIP: 0010:[<00000000>] Tainted: PF
EFLAGS: 00010292
eax: 00000000 ebx: 00000000 ecx: c029a5ca edx:
c029a5c4
esi: d6f10000 edi: d6f140a0 ebp: e1a87ea4 esp:
d796de5c
ds: 0018 es: 0018 ss: 0018
Process insmod (pid: 1110, stackpage=d796d000)
Stack: c01e6bad d6f140a0 00000000 00000000 00000000
d6f10000 d6f140a0 e1a87ea4
e1a87fc0 ffffffff 00000000 00000080 c029a5ca
00000180 e1a7e7ac e1a7e7ff
d6f140a0 00000000 c159ec00 d6f10000 76697264
652f7265 3031756d 302f316b
Call Trace: [<c01e6bad>] [<e1a87ea4>] [<e1a87fc0>]
[<e1a7e7ac>] [<e1a7e7ff>]
[<e1a87e80>] [<e1a825bc>] [<e1a87ea4>] [<e1a87f00>]
[<c01f1d01>] [<e1a87ea4>]
[<e1a87f00>] [<c01f1d64>] [<e1a87f00>] [<e1a82768>]
[<e1a87f00>] [<c0116df1>]
[<e1a79060>] [<c0106bd3>]

Code: Bad EIP value.

Discussion

  • M.J. Blom
    M.J. Blom
    2003-08-22

    Logged In: YES
    user_id=829800

    Same thing happens over here with (Debian) kernel 2.4.21. I recently
    upgraded from 2.4.18 where I had no problems.

    Hope someone can help ...

    B10m
    menno@b10m.net

    My dmesg:
    Creative EMU10K1 PCI Audio Driver, version 0.20a, 19:47:32 Aug 22 2003
    emu10k1: Audigy rev 4 model 0x1005 found, IO at 0xdf00-0xdf3f, IRQ 20
    Unable to handle kernel NULL pointer dereference at virtual address 00000000
    printing eip:
    00000000
    *pde = 00000000
    Oops: 0000
    CPU: 0
    EIP: 0010:[<00000000>] Not tainted
    EFLAGS: 00010292
    eax: 00000000 ebx: 00000000 ecx: c02f69ea edx: c02f69e4
    esi: f4290000 edi: f429ds: 0018 es: 0018 ss: 0018
    Process modprobe (pid: 1231, stackpage=f5a83000)
    Stack: c02429bd f42940a0 00000000 00000000 00000000 f4290000
    f42940a0 f894d0e4
    f894d200 ffffffff 00000000 00000080 c02f69ea 00000180 f89432ec
    f894333f
    f42940a0 00000000 c1c31400 f4290000 76697264 652f7265 3031756d
    302f316b
    Call Trace: [<c02429bd>] [<f894d0e4>] [<f894d200>] [<f89432ec>]
    [<f894333f>]
    [<f894d0c0>] [<f894714d>] [<f894d0e4>] [<f894d140>] [<c0243ad1>]
    [<f894d0e4>]
    [<f894d140>] [<c0243b34>] [<f894d140>] [<f89472f8>] [<f894d140>]
    [<c0118f81>]
    [<f893d060>] [<c0106f17>]

    Code: Bad EIP value.
    40a0 ebp: f894d0e4 esp: f5a83e5c

     
  • Logged In: YES
    user_id=630341

    The same happens to me. I upgraded from 2.4.20 to 2.4.22. I can insert
    the module, but it make segmentation fault while the module is initializing

     
  • Logged In: NO

    Same thing happens to me on Debian unstable.

    Using both 0.20a and the CVS copy, both crash out with an
    Audigy Gamer.

    Worked with 2.4.19 and upgraded to 2.4.22 and now it dies.

     
  • C Fulmer
    C Fulmer
    2003-11-12

    Logged In: YES
    user_id=68384

    Check if you have soundcore built as a module or have it built into your kernel w/ sound support set to Y in menuconfig or xconfig. If you don't have soundcore.o in /lib/modules/uname -r/kernel/drivers/sound/ it is most likely built into your kernel. Also when modprobe emu10k1 seg faults, run lsmod if you can and it should show soundcore there if you have it, if you don't, then it was built in and modprobe didn't have to insmod it before trying to load emu10k1 and segfaulting.
    And yes the docs say its ok to set sound support (soundcore) to y/m, but w/ later kernels it appears to cause this problem if set to Y. So if at all possible set sound support to module, for "monolithic" (whatever definition you want to start a flamewar about) kernels you probably won't have a problem with both emu10k1 and soundcore built in since nothing gets loaded afterwards.
    Guess i should check my logs from #creative more often, original poster and poster immediately prior to this one came in asking about it.

    guess i should check my logs from #creative on freenode more often, original poster came in that same day.

     
  • Logged In: YES
    user_id=630341

    Hi!,

    Good news, selecting Sound core as a module in the kernel
    solves the problem. The only comment is that insmod emu10k1
    gives unresolved symbols, but if you make before insmod
    ac97_codec, then everything goes right.

    Than you very much!!!

     


Anonymous


Cancel   Add attachments