Re: debugging fvwm/X11
Hello,
Simon Kainz, on Mon 13 Feb 2017 14:23:41 +0100, wrote:
> I added the repo (didn't know about that!)
I believe we have to do something in Debian about this. I found about
next to nobody that knows about this move to a separate repo.
> (EE) 0: /usr/lib/xorg/Xorg (xorg_backtrace+0x52) [0x81ceee2]
> (EE) 1: /usr/lib/xorg/Xorg (0x8048000+0x18ae12) [0x81d2e12]
> (EE) 2: /lib/i386-gnu/libc.so.0.3 (0x1276000+0x3fbd2) [0x12b5bd2]
>
> Using addr2line i got:
> $ addr2line -e /usr/lib/xorg/Xorg 81ceee2
> ./debian/build/main/os/../../../../os/backtrace.c:130
>
> $ addr2line -e /usr/lib/xorg/Xorg 81d2e12
> ./debian/build/main/os/../../../../os/osinit.c:139
>
> for libc.so.0.3 i only see
> ??:0
Note that you need to look up the offset (0x3fbd2), not the absolute
value (0x12b5bd2), since libc might be loaded anywhere in memory.
> Inspecting the code it seems to me that something inside lobc.so.0.3
> triggers SIgnal 10, but i have no ide waht, as i am unable to addr2line
> libc.
€ file /lib/i386-gnu/libc-2.24.so
/lib/i386-gnu/libc-2.24.so: ELF 32-bit LSB shared object, Intel 80386, version 1 (SYSV), dynamically linked, interpreter /lib/ld.so.1, BuildID[sha1]=7a7fafb866f1680656f7343e9d38fa76986bcfff, for GNU/Hurd 0.0.0, stripped
€ nm -n /usr/lib/debug/.build-id/7a/7fafb866f1680656f7343e9d38fa76986bcfff.debug
looking for ^0003fb shows:
0003fbc0 t rpc_wait_trampoline
0003fbd0 t trampoline
So the libc part is just the trampoline for the signal handler :/
That trampoline is unfortunately missing debugging information for
backtrace to step further. You could perhaps manage to dump a core so
that it can be inspected by gdb?
Samuel
Reply to: