Re: extract debuginfo for Xorg backtraces

[Elmar Stellnberger <estellnb at gmail dot com>]

 Well, I have already resolved the problem by either just taking the
offsets for the so-s or if available by using /proc/pid/maps to
calculate them (works better).
 Now the next riddle is on how to obtain the load base for a kernel
module in order to decode crashes related to kernel modules (lsmod
does not seem to be useful since it only shows the size but not the
load base).

  Why is it no more possible to save the kernel`s call interface per
dd from /proc/self/mem into a file? - dd if=/proc/self/mem
of=/dld/linux-gate.so.1 skip=$((0xffffe000)) count=1 bs=4096
How does ldd obtain the kernel address 0xffffe000 of the call gate
window - it apparently does not appear in /proc/self/maps (just the
window where it shows up.). Can I open this with libbfd?

2011/2/22, Nick Clifton <nickc@redhat.com>:
> Hi Elmar,
>
>> It works well for everything in /usr/bin/Xorg. However as the
>> backtrace meets a shared library (libc.so, radeon_drv.so,
>> evdev_drv.so) the symtab does not contain any section that would
>> include the given addresses/+offsets.
>
> How about the dynamic symbol table ?
>
> I suggest that you have a look at the GDB code base to see how symbols
> in shared libraries are handled there.
>
> Cheers
>    Nick
>
>