Re: Panic in ip_input

Max Laier wrote:
> 
> Hello Andreas,
> 
> Monday, November 17, 2003, 8:11:47 AM, you wrote:
> AK> #7  0xc05ed8a9 in ip_input (m=0x0) at
> AK> /usr/src/sys/netinet/ip_input.c:364
> AK> #8  0xc05e6292 in div_output (so=0xc2f11d20, m=0xc16d6600,
> AK> sin=0xc2d834b0,
> AK>     control=0x0) at /usr/src/sys/netinet/ip_divert.c:364
> 
> AK> (kgdb) frame 7
> AK> #7  0xc05ed8a9 in ip_input (m=0x0) at
> AK> /usr/src/sys/netinet/ip_input.c:364
> AK> 364                             m_free(m0);
> AK> (kgdb) p m0
> AK> $1 = (struct mbuf *) 0x0
> AK> (kgdb) l
> AK> 359
> AK> 360                     m0 = m;
> AK> 361                     m = m->m_next;
> AK> 362                     /* XXX: This is set by ip_fastforward */
> AK> 363                     if (m0->m_nextpkt == (struct mbuf *)1)
> AK> 364                             m_free(m0);
> AK> 365             }
> AK> 366
> AK> 367             M_ASSERTPKTHDR(m);
> AK> 368
> 
> AK> This panic is relatively easy to recreate.
> 
> AK> Some data points:
> 
> AK> The machine is an Athlon Thunderbird 1200 (CPUTYPE=athlon-tbird in
> AK> make.conf), the NIC is a Realtek 8139.
> AK> net.inet.ip.fastforwarding is 0.
> 
> AK> I have a dump available (256M). What can I do to help fix this problem?
> 
> What rev. of ip_input.c is this? Looks like head. Rev. 1.253 mangled
> the for(;;) in a strange way and added that very strange check ... can
> somebody just kill these bastard MT_TAG thing in flavour for real
> mbuf_tags, now? Please!

Green fixed the problem a couple of hours ago in ip_divert.c.  The
m_nextpkt was uninitialized and happend to be 1 this time.  Please
re-cvsup.

-- 
Andre