Bug#274756: libc6: abort() breaks gdb backtraces with 2.6 kernel

[Daniel Jacobowitz <dan@debian.org>]

On Mon, Oct 04, 2004 at 05:12:11AM +0300, Timo Sirainen wrote:
> On 4.10.2004, at 05:04, Daniel Jacobowitz wrote:
> 
> >I took the test case from your first URL.  I tried to reproduce the
> >problem.
> ..
> >0xffffe410 in __kernel_vsyscall ()
> >(gdb) bt
> >#0  0xffffe410 in __kernel_vsyscall ()
> >#1  0x4005ce33 in raise () from /lib/tls/i686/cmov/libc.so.6
> >#2  0x4005e72c in abort () from /lib/tls/i686/cmov/libc.so.6
> >#3  0x08048399 in main () at q.c:1
> 
> Wish that would happen with me too :)
> 
> >You need to work out how this does not happen for you.  What version of
> >GDB you're using, which you've never said.  What other confounding
> >factors, like kernel patches, may be involved.
> 
> I've tried with nonpatched 2.6.x kernels, although I can't remember the 
> exact versions. Currently I'm running 2.6.9-rc2-mm1-VP-S1 (VP-S1 = 
> voluntary re-emption). I've tried with Debian's gdb v6.1-3 as well as 
> sourced compiled gdb v6.2.1
> 
> I can give access to the computer if it helps (sending in separate 
> mail).

Sorry, but I don't really have the time to do this myself... does GDB
open the libraries in /usr/lib/debug?  Try stracing GDB and looking at
what files it opens.

-- 
Daniel Jacobowitz