FreeLists / openbeosnetteam / [openbeosnetteam] Re: Test!

["Bruno G. Albuquerque" <bga@xxxxxxxxxxxxx>]

> Can I ask all of you who are able to update your tree and see if 
> you're
> getting the same bug I am? This is probably the biggest bug we 
> currently
> have and it'd be really nice to get it squashed.
> 
> Basically, I'm able to reproduce it 100% of the time by simply 
> doing...
> 
> start new net stack
> run new ping using ping www.yahoo.com
> 
> That's all it takes for me. If anyone can debug it further that would 
> be
> great :)

I just did some testing, but just now saw your email so didn't include 
the test you mentioned... I was looking into the "route show" crash:

loading symbols
segment violation occurred
ea4f0b00 read_fault
route:sc
   frame         retaddr
fcffb554   ec72c6bd  __res_send + 00000945
fcffb7d4   ec72af04  res_query + 00000104
fcffbbfc   ec72524a  getnetbyaddr + 000001ce
fcffc448   8000328b  netname + 0000011f
fcffc478   80002159  p_sockaddr + 000001d9
fcffc520   80001eba  p_rtentry + 000000c2
fcffc544   80001d7d  show + 000001fd
fcffc588   80002a19  main + 000002ed
fcffc5a4   80001b71  _start + 00000061
route: 

The code that was being executed was the following one (from show.c):

if (buf) {
                for (next = buf; next < lim; next += rtm->rtm_msglen) {
                        rtm = (struct rt_msghdr *)next;
                        sa = (struct sockaddr *)(rtm + 1);
                        if (af && sa->sa_family != af)
                                continue;
                        p_rtentry(rtm); // <--- ** CRASH **
                }
                free(buf);
        }

And it crashed in the market line with a segment violation. I could 
debug it further and step into that function, but I will have lunch 
now. :P

On a related note, traceroute brought me to KDL (the cullprit was the 
network card tx thread).

Hope this helps.

-Bruno


--
Fortune Cookie Says:

"I'd love to go out with you, but there are important world issues that
need worrying about."