Crash using tun device

[Peter Carah]

Victor Snezhko wrote:
> Pete Carah <pete at altadena.net> writes:
> 
> 
>>I now (cvsup as of a day ago, and also from a week ago) see a
>>spontaneous reboot when using the tun device from userland ppp.  This
>>occurs with either umodem or an ssh tunnel on the serial side, and
>>crashes immediately after the IP addresses are set.  I don't know if it
>>is in route setting or not; the ssh tunnel script doesn't set the
>>default route via the tunnel (shoot self in foot, anyone?).
> 
> 
> There is a known bug in netinet6 that might cause this behaviour
> (although, I didn't see reboots, only panics). The fix is known but
> has not been committed yet. Here it is:
> 
> --- netinet6/mld6.c	Wed Nov  9 08:27:14 2005
> ***************
> *** 640,645 ****
> --- 640,649 ----
>   		mld6_stop_listening(in6m);
>   		ifma->ifma_protospec = NULL;
>   		LIST_REMOVE(in6m, in6m_entry);
> + 		if (in6m->in6m_timer != IN6M_TIMER_UNDEF)
> + 			mld_stoptimer(in6m);
>   		free(in6m->in6m_timer_ch, M_IP6MADDR);
>   		free(in6m, M_IP6MADDR);
>   	}
> 
> 

....

It helps completely - I'm talking to my mail server using ppp over verizon
right now.  It may not have rebooted if you had invariant_support or invariants
enabled - those tend to panic in cases that would otherwise reboot or crash,
especially with free of a block with a pending callback :-); I normally don't
enable those (nor witness) on my laptop - too slow.  I run current on two other
systems, though, one of which sometimes gets invariant_support.  I don't care
about LOR or other such unless I see mysterious hangs - even then I normally
just mail the list...  Also my ppp never uses ip6 (but ethernet sometimes
does), so it wouldn't hurt most of the time to disable it.

> The bug I mentioned was committed on Oct 21.
> 
> If the fix don't help, try to disable INET6 and see if that helps.
Did help.  Thank you.

-- Pete