Bug in libnl-3.2.0?

[samriti katoch]

Think that might be the problem then.

Thanks!

On Mon, Aug 13, 2012 at 3:32 AM, Thomas Graf <tgraf at suug.ch> wrote:

> On Fri, Aug 10, 2012 at 02:34:04PM -0400, samriti katoch wrote:
> > Hi,
> >
> > I am using libnl-3-200_3.2.7-2_i386.deb and
> > libnl-route-3-200_3.2.7-2_i386.deb on Ubuntu to resolve IPv6 addresses
> into
> > corresponding MAC addresses, in a multi-threaded application. I see the
> > application crash with Segmentation fault after resolving close to 30,000
> > IPv6 addresses.
> > The application creates a cache manager for both route/link and
> route/neigh
> > and then uses nl_cache_mngr_poll function to receive neighbor cache
> > updates. Here is a backtrace from gdb after the segmentation fault:
> >
> > Program received signal SIGSEGV, Segmentation fault.
> > [Switching to Thread -378549344 (LWP 3159)]
> > 0xf7e081b3 in main_arena () from /lib/tls/i686/cmov/libc.so.6
> > (gdb) bt
> > #0  0xf7e081b3 in main_arena () from /lib/tls/i686/cmov/libc.so.6
> > #1  0xf7f968b6 in nl_recvmsgs () from /lib/libnl-3.so.200
> > #2  0xf7f96e52 in nl_recvmsgs_default () from /lib/libnl-3.so.200
> > #3  0xf7f929ee in nl_cache_mngr_data_ready () from /lib/libnl-3.so.200
> > #4  0xf7f92a8a in nl_cache_mngr_poll () from /lib/libnl-3.so.200
> >
> > I looked through logs of recent commits and did not find anything that
> > mentions this problem.
> > I would appreciate it if I could get your inputs on this, and also if
> libnl
> > can handle high load from a multi-threaded applications.
>
> Do you run updates to the cache manager from multiple threads? That is
> not safe. You can read from the cache manager from multiple threads if
> you ensure that the updating thread does not run while you are reading
> it. libnl only implements locking for static data.
>
> ~Thomas
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.infradead.org/pipermail/libnl/attachments/20120814/662769e6/attachment.html>