Re: Format of an 'oops' call trace (in show_trace)

["Randy.Dunlap" <rddunlap@osdl.org>]

Install ksymoops.  Its man page has info on module debugging
and ooops reporting.

~Randy


On Thu, 9 Oct 2003 19:20:28 +0200 "Amir Hermelin" <amir@montilio.com> wrote:

| Thanks Randy.
| 
| Since the information doesn't tell me much other than which function
| dropped, is there any way I could get more than a hunch of where the fault
| occurred, assuming this is a loadable module?
| 
| Thanks,
| Amir.
| 
| 
| 
| -----Original Message-----
| From: Randy.Dunlap [mailto:rddunlap@osdl.org] 
| Sent: Wednesday, October 08, 2003 7:06 PM
| To: Amir Hermelin
| Cc: linux-kernel@vger.kernel.org
| Subject: Re: Format of an 'oops' call trace (in show_trace)
| 
| 
| On Wed, 8 Oct 2003 13:25:44 +0200 "Amir Hermelin" <amir@montilio.com> wrote:
| 
| | Oops,
| | I forgot that part :)  It's RH 2.4.20-8
| | 
| | Thanks,
| | Amir.
| | 
| | 
| | On Tue, 7 Oct 2003 17:56:38 +0200 "Amir Hermelin" <amir@montilio.com> 
| | wrote:
| | 
| | | Hi,
| | | Can someone please point me to a description of what I see in the 
| | | Call
| | | of the oops dump?  I tried looking into show_trace and lookup_symbol 
| | | functions, but I couldn't understand some things.  For example, in 
| | | this following trace:
| 
| The basic format (in RH 2.4.20-8) is:
| 
| [<address>] symbol_name [module_name] 0xoffset_from_symbol (where address is
| on stack)
| 
| offset_from_symbol is hex bytes from symbol to <address>, so 0x0 is an exact
| match.
| 
| | | joji kernel: [<e01bae00>] reqrdata [mymod] 0x0 (0xd5543fb4)) joji 
| | | kernel: [<e01a5220>] mymod [mymod] 0x0 (0xd5543fe0))
| 
| | | I don't understand the relevance to reqrdata (since it's not a 
| | | function,
| | but
| | | a data structure, and isn't the parameter to the mymod function).
| 
| It looks for any addresses in the kernel text (code) space and tries to find
| symbol names for them.
| 
| | | And could
| | | someone please explain what the 0x0 in the lines mean? From the code 
| | | I understood it to be the offset of the symbol within the module, 
| | | but that can't be right if both symbols translate to the same offset 
| | | - so I must've understood it wrong.
| 
| See above.
| 
| | | joji kernel:  printing eip:
| | | joji kernel: e01b090b
| | | joji kernel: *pde = 00000000
| | | joji kernel: Oops: 0002
| | | joji kernel: CPU:    0
| | | joji kernel: EIP:    0060:[<e01b090b>]    Not tainted
| | | joji kernel: EFLAGS: 00010282
| | | joji kernel:
| | | joji kernel: EIP is at rtp_recv [mymod] 0x5b (2.4.20-8custom)
| | | joji kernel: eax: 00000000   ebx: d5542000   ecx: 00000001
| | | edx: c0374c88
| | | joji kernel: esi: e01bae00   edi: d76aa400   ebp: d5543fcc
| | | esp: d5543f98
| | | joji kernel: ds: 0068   es: 0068   ss: 0068
| | | joji kernel: Process mymod (pid: 6978, stackpage=d5543000)
| | 
| | | joji kernel: Stack: e01bae00 d76aa400 d5542000 00000000 d76aa400 
| | | ffffffff e01a5308 e01bae00
| | | joji kernel:        d76aa400 d5543fcc d5542000 d5542000
| | | dbd15900 00000000 d54f3fd0 d5533fd0 
| | | joji kernel:        d5542000 00000000 e01a5220 00000000
| | | 00000000 00000000 c010742d d76aa400
| | | joji kernel: Call Trace:
| | |   [<e01bae00>] reqrdata [mymod] 0x0 (0xd5543f98))
| | | joji kernel: [<e01a5308>] mymod [mymod] 0xe8 (0xd5543fb0))
| | 
| | | joji kernel: [<e01bae00>] reqrdata [mymod] 0x0 (0xd5543fb4))
| | | 
| | | joji kernel: [<e01a5220>] mymod [mymod] 0x0 (0xd5543fe0)) joji 
| | | kernel: [<c010742d>] kernel_thread_helper [kernel] 0x5 (0xd5543ff0))
| 
| HTH.
| 
| --
| ~Randy