Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.os.linux.development.system > #237

Re: how to debug a kernel crash, using only the syslog dump?

Show all headers | View raw

On Jul 25, 11:35 am, Gorlash <dan.miller....@gmail.com> wrote:
> I've got a crash occurring in a driver here (Ubuntu server 10.10,
> kernel 2.6.35-22-generic-pae)
> I haven't been able to reproduce it on my development machine yet, but
> two machines in the lab are sporadically producing it, so I have the
> kernel dump from the system log, but nothing else.
> The dump tells me which function contained the failing code, but
> nothing more detailed (such as a line number).
>
> Is there a utility which will read the kernel dump and give me a
> pointer back to my code?
> I tried the 'crash' utility, but it appears to be looking for some
> file format other than a syslog dump.  I've found several other
> utilities mentioned on the LKCD site, but none of them say that they
> will work with this input format.  There was some utility back in the
> k2.2 days which could do that, but I don't recall what it was...
>
> I'm guessing that the fault is occurring in copy_to_user(), but I'd
> prefer to confirm that explicitly, if possible, before I delve into
> more specific debugging techniques!
>
> Dan Miller
>
> Here is my kernel dump:
> (Legend: tilapia is the daemon which is calling my driver (ip3eth),
> and netnot_read is the read function, in ip3eth, which is failing.)
>
> *pdpt = 000000000df40001 *pde = 000000000df07067 *pte =
> 0000000000000000
> Modules linked in: ip3eth snd_intel8x0 snd_ac97_codec ac97_bus snd_pcm
> snd_timer intel_agp snd lp soundcore shpchp snd_page_alloc agpgart
> parport e100 mii
>
> Pid: 604, comm: tilapia Not tainted 2.6.35-22-generic-pae #33-Ubuntu /
> EIP: 0060:[<d09a190d>] EFLAGS: 00010286 CPU: 0
> EIP is at netnot_read+0x6d/0x120 [ip3eth]
> EAX: 00000282 EBX: cdbf54e0 ECX: d09e23d0 EDX: 00000282
> ESI: 00000000 EDI: 00000000 EBP: ce09ff64 ESP: ce09ff48
>  DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
>  00000000 ce0a3b18 00000800 b770e9e0 cde35180 00000800 b770e9e0
> ce09ff8c
> <0> c0222d1f ce09ff98 616c6974 00616970 d09a18a0 00000000 cde35180
> fffffff7
> <0> b770e9e0 ce09ffac c0222e52 ce09ff98 00000000 00000000 00000001
> 00000005
>  [<c0222d1f>] ? vfs_read+0x9f/0x190
>  [<d09a18a0>] ? netnot_read+0x0/0x120 [ip3eth]
>  [<c0222e52>] ? sys_read+0x42/0x70
>  [<c05f0074>] ? syscall_call+0x7/0xb
> ---[ end trace 231bdd2736c78c90 ]---

hmmm... I tried using "objdump -d -S ip3eth.ko > dumpfile.txt" to get
source files interlaced with with the disassembly (which theoretically
would accomplish what I want here), but the result doesn't have any
source code in it:
00020b30 <netnot_read>:
   20b30:       55                      push   %ebp
   20b31:       89 e5                   mov    %esp,%ebp
   20b33:       83 ec 1c                sub    $0x1c,%esp
   20b36:       89 5d f4                mov    %ebx,-0xc(%ebp)
   20b39:       89 75 f8                mov    %esi,-0x8(%ebp)
   20b3c:       89 7d fc                mov    %edi,-0x4(%ebp)
   20b3f:       e8 fc ff ff ff          call   20b40 <netnot_read
+0x10>
   20b44:       8b 3d 00 00 00 00       mov    0x0,%edi
   20b4a:       85 ff                   test   %edi,%edi
   20b4c:       89 55 f0                mov    %edx,-0x10(%ebp)
   20b4f:       0f 84 be 00 00 00       je     20c13 <netnot_read
+0xe3>
   20b55:       8d b7 c4 00 00 00       lea    0xc4(%edi),%esi

The source C files for this driver are all in the current directory.
What do I need to do to actually get my sources included??

Back to comp.os.linux.development.system | Previous | Next — Previous in thread | Next in thread | Find similar

Thread

how to debug a kernel crash, using only the syslog dump? Gorlash <dan.miller.x23@gmail.com> - 2011-07-25 11:35 -0700
  Re: how to debug a kernel crash, using only the syslog dump? Gorlash <dan.miller.x23@gmail.com> - 2011-07-25 13:12 -0700
    Re: how to debug a kernel crash, using only the syslog dump? Gorlash <dan.miller.x23@gmail.com> - 2011-07-25 13:44 -0700
    Re: how to debug a kernel crash, using only the syslog dump? Jorgen Grahn <grahn+nntp@snipabacken.se> - 2011-07-25 20:45 +0000
      Re: how to debug a kernel crash, using only the syslog dump? Gorlash <dan.miller.x23@gmail.com> - 2011-07-25 15:55 -0700
        Re: how to debug a kernel crash, using only the syslog dump? Jorgen Grahn <grahn+nntp@snipabacken.se> - 2011-07-26 08:04 +0000
        Re: how to debug a kernel crash, using only the syslog dump? Rainer Weikusat <rweikusat@mssgmbh.com> - 2011-07-26 11:37 +0100
  Re: how to debug a kernel crash, using only the syslog dump? "Ersek, Laszlo" <lacos@caesar.elte.hu> - 2011-07-26 18:46 +0200

csiph-web