Subversion Repositories or1k

From: Linus Torvalds 

How to track down an Oops.. [originally a mail to linux-kernel]

The main trick is having 5 years of experience with those pesky oops

messages ;-)

Actually, there are things you can do that make this easier. I have two

separate approaches:

        gdb /usr/src/linux/vmlinux

        gdb> disassemble 

That's the easy way to find the problem, at least if the bug-report is

well made (like this one was - run through ksymoops to get the

information of which function and the offset in the function that it

happened in).

Oh, it helps if the report happens on a kernel that is compiled with the

same compiler and similar setups.

The other thing to do is disassemble the "Code:" part of the bug report:

ksymoops will do this too with the correct tools (and new version of

ksymoops), but if you don't have the tools you can just do a silly

program:

        char str[] = "\xXX\xXX\xXX...";

        main(){}

and compile it with gcc -g and then do "disassemble str" (where the "XX"

stuff are the values reported by the Oops - you can just cut-and-paste

and do a replace of spaces to "\x" - that's what I do, as I'm too lazy

to write a program to automate this all).

Finally, if you want to see where the code comes from, you can do

        cd /usr/src/linux

        make fs/buffer.s        # or whatever file the bug happened in

and then you get a better idea of what happens than with the gdb

disassembly.

Now, the trick is just then to combine all the data you have: the C

sources (and general knowledge of what it _should_ do, the assembly

listing and the code disassembly (and additionally the register dump you

also get from the "oops" message - that can be useful to see _what_ the

corrupted pointers were, and when you have the assembler listing you can

also match the other registers to whatever C expressions they were used

for).

Essentially, you just look at what doesn't match (in this case it was the

"Code" disassembly that didn't match with what the compiler generated).

Then you need to find out _why_ they don't match. Often it's simple - you

see that the code uses a NULL pointer and then you look at the code and

wonder how the NULL pointer got there, and if it's a valid thing to do

you just check against it..

Now, if somebody gets the idea that this is time-consuming and requires

some small amount of concentration, you're right. Which is why I will

mostly just ignore any panic reports that don't have the symbol table

info etc looked up: it simply gets too hard to look it up (I have some

programs to search for specific patterns in the kernel code segment, and

sometimes I have been able to look up those kinds of panics too, but

that really requires pretty good knowledge of the kernel just to be able

to pick out the right sequences etc..)

_Sometimes_ it happens that I just see the disassembled code sequence

from the panic, and I know immediately where it's coming from. That's when

I get worried that I've been doing this for too long ;-)

                Linus

---------------------------------------------------------------------------

Notes on Oops tracing with klogd:

In order to help Linus and the other kernel developers there has been

substantial support incorporated into klogd for processing protection

faults.  In order to have full support for address resolution at least

version 1.3-pl3 of the sysklogd package should be used.

When a protection fault occurs the klogd daemon automatically

translates important addresses in the kernel log messages to their

symbolic equivalents.  This translated kernel message is then

forwarded through whatever reporting mechanism klogd is using.  The

protection fault message can be simply cut out of the message files

and forwarded to the kernel developers.

Two types of address resolution are performed by klogd.  The first is

static translation and the second is dynamic translation.  Static

translation uses the System.map file in much the same manner that

ksymoops does.  In order to do static translation the klogd daemon

must be able to find a system map file at daemon initialization time.

See the klogd man page for information on how klogd searches for map

files.

Dynamic address translation is important when kernel loadable modules

are being used.  Since memory for kernel modules is allocated from the

kernel's dynamic memory pools there are no fixed locations for either

the start of the module or for functions and symbols in the module.

The kernel supports system calls which allow a program to determine

which modules are loaded and their location in memory.  Using these

system calls the klogd daemon builds a symbol table which can be used

to debug a protection fault which occurs in a loadable kernel module.

At the very minimum klogd will provide the name of the module which

generated the protection fault.  There may be additional symbolic

information available if the developer of the loadable module chose to

export symbol information from the module.

Since the kernel module environment can be dynamic there must be a

mechanism for notifying the klogd daemon when a change in module

environment occurs.  There are command line options available which

allow klogd to signal the currently executing daemon that symbol

information should be refreshed.  See the klogd manual page for more

information.

A patch is included with the sysklogd distribution which modifies the

modules-2.0.0 package to automatically signal klogd whenever a module

is loaded or unloaded.  Applying this patch provides essentially

seamless support for debugging protection faults which occur with

kernel loadable modules.

The following is an example of a protection fault in a loadable module

processed by klogd:

---------------------------------------------------------------------------

Aug 29 09:51:01 blizard kernel: Unable to handle kernel paging request at virtual address f15e97cc

Aug 29 09:51:01 blizard kernel: current->tss.cr3 = 0062d000, %cr3 = 0062d000

Aug 29 09:51:01 blizard kernel: *pde = 00000000

Aug 29 09:51:01 blizard kernel: Oops: 0002

Aug 29 09:51:01 blizard kernel: CPU:    0

Aug 29 09:51:01 blizard kernel: EIP:    0010:[oops:_oops+16/3868]

Aug 29 09:51:01 blizard kernel: EFLAGS: 00010212

Aug 29 09:51:01 blizard kernel: eax: 315e97cc   ebx: 003a6f80   ecx: 001be77b   edx: 00237c0c

Aug 29 09:51:01 blizard kernel: esi: 00000000   edi: bffffdb3   ebp: 00589f90   esp: 00589f8c

Aug 29 09:51:01 blizard kernel: ds: 0018   es: 0018   fs: 002b   gs: 002b   ss: 0018

Aug 29 09:51:01 blizard kernel: Process oops_test (pid: 3374, process nr: 21, stackpage=00589000)

Aug 29 09:51:01 blizard kernel: Stack: 315e97cc 00589f98 0100b0b4 bffffed4 0012e38e 00240c64 003a6f80 00000001

Aug 29 09:51:01 blizard kernel:        00000000 00237810 bfffff00 0010a7fa 00000003 00000001 00000000 bfffff00

Aug 29 09:51:01 blizard kernel:        bffffdb3 bffffed4 ffffffda 0000002b 0007002b 0000002b 0000002b 00000036

Aug 29 09:51:01 blizard kernel: Call Trace: [oops:_oops_ioctl+48/80] [_sys_ioctl+254/272] [_system_call+82/128]

Aug 29 09:51:01 blizard kernel: Code: c7 00 05 00 00 00 eb 08 90 90 90 90 90 90 90 90 89 ec 5d c3

---------------------------------------------------------------------------

Dr. G.W. Wettstein           Oncology Research Div. Computing Facility

Roger Maris Cancer Center    INTERNET: greg@wind.rmcc.com

820 4th St. N.

Fargo, ND  58122

Phone: 701-234-7556