kern/100080: tun(4) device close removes only IPv4 addresses and routes

Remi Denis-Courmont rdenis at simphalempin.com
Tue Jul 11 11:10:22 UTC 2006


>Number:         100080
>Category:       kern
>Synopsis:       tun(4) device close removes only IPv4 addresses and routes
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Tue Jul 11 11:10:14 GMT 2006
>Closed-Date:
>Last-Modified:
>Originator:     Remi Denis-Courmont
>Release:        6.1
>Organization:
>Environment:
FreeBSD fbsd 6.1-RELEASE FreeBSD 6.1-RELEASE #0: Sun May  7 04:32:43 UTC 2006
  root at opus.cse.buffalo.edu:/usr/obj/usr/src/sys/GENERIC  i386
>Description:
The userland layer-3 tunnel character device tun(4) "/dev/tun" should remove all addresses and routes through an allocated network interface whenever the file descriptor is closed (src/sys/net/if_tun.c:tunclose).

However, it currently seems to only remove IPv4 (AF_INET) addresses and routes, leaving dangling routes in the network stack when IPv6 is used.
>How-To-Repeat:
This problem is known to trigger when running Miredo 0.8.5 in Teredo relay mode. At first run, the program works fine, but if it is stopped/killed or crashes, we get a dangling route on the leftover tunnel device; restarting the program will then fail with EEXIST as it tries to add the same route to a new tunnel interface. Sample configuration (/usr/local/etc/miredo.conf) would be the following one-liner:

RelayType cone

Unfortunately, setting this up in non-trivial since said software is not in the ports tree. Sources are there: http://www.remlab.net/files/miredo/v0.8/

Other tunnel software using tun + IPv6 might also be affected.
>Fix:
Supposedly yemove the AF_INET check in tunclose() from src/sys/net/if_tun.c or add a check for all supported addresses family, including but maybe not limited to AF_INET6.
>Release-Note:
>Audit-Trail:
>Unformatted:


More information about the freebsd-bugs mailing list