Howto resolve crash address into function name / line number?

[Patrick O'Callaghan]

On Thu, 2012-11-01 at 18:08 +0100, Clemens Eisserer wrote:
> Hi,
> 
> Thanks for trying to help Patrick, but just installing
> debuginfo-packages won't tell me where a crash happend.
> 
> It would be very helpful if somebody could tell me, how to resolve a
> crash address to a function name - or even  better a line number, so
> that I can file a bug-report.
> 
> Thanks, Clemens
> 
> 2012/11/1, Patrick O'Callaghan <pocallaghan at gmail.com>:
> > On Wed, 2012-10-31 at 19:56 +0100, Clemens Eisserer wrote:
> >> Hi,
> >>
> >> I see usb_modeswitch frequently crashing with a segfault, and a message
> >> like:
> >> Oct 31 19:38:28 localhost kernel: [   33.953670] usb_modeswitch_[1373]:
> >> segfault at 7fffd64bfbe4 ip 0000003000874b06 sp 00007fffd64bfbd0 error 6
> >> in
> >> libc-2.15.so[3000800000+1ac000]
> >>
> >> Is there any way to trace back 7fffd64bfbe4 to the function name / line
> >> number?
> >>
> >> Thank you in advance, Clemens
> >
> > debuginfo-install glibc

[Please don't top-post on the Fedora lists. See the Guidelines]

Clearly debuginfo on its own isn't going to do anything. You need to run
a debugger (such as gdb) on the crash results (either generate a core
dump or use the debugger on the running process). And a line number on
its own may not be of much use (the segfault appears to happen in libc
after all). What you need is a stack trace.

Actually getting a core dump from usb_modeswitch may not be easy,
especially if it's being called from a udev event.

poc