[Dnsmasq-discuss] dnsmasq crashing when serving multiple prefixes

[Simon Kelley]

On 04/09/13 18:16, Kaspar Schleiser wrote:
> Hi,
>
> we're experiencing a crash in dnsmasq 2.66 (also newest git) using it as
> DHCPv6 server serving multiple prefixes.
>
> This is on a slightly changed OpenWrt box with ARM/kirkwood architecture
> and uClibc.
>
> Linux router.ip 3.8.13 #42 Tue Sep 3 13:53:54 CEST 2013 armv5tel GNU/Linux
>
> Here's the dnsmasq output:
> root at router:/tmp# gdbserver :9000 /usr/sbin/dnsmasq
> --conf-dir=/var/etc/dnsmasq.d -d
> Process /usr/sbin/dnsmasq created; pid = 10423
> Listening on port 9000
> Remote debugging from host 192.168.2.5
> dnsmasq: started, version 2.66 cachesize 150
> dnsmasq: compile time options: IPv6 GNU-getopt no-DBus no-i18n no-IDN
> DHCP DHCPv6 no-Lua TFTP no-conntrack no-ipset no-auth
> dnsmasq-dhcp: DHCP, IP range 192.168.2.100 -- 192.168.2.249, lease time 12h
> dnsmasq-dhcp: DHCPv6 stateless on fd46:b82e:3f6d::1
> dnsmasq-dhcp: router advertisement on fd46:b82e:3f6d::
> dnsmasq-dhcp: DHCPv6, IP range 2a00:c31:8533:900::2 --
> 2a00:c31:8533:900:ffff:ffff:ffff:ffff, lease time 1h
> dnsmasq-dhcp: DHCPv6 stateless on 2a00:c31:8533:900::2
> dnsmasq-dhcp: router advertisement on 2a00:c31:8533:900::
> dnsmasq-dhcp: DHCPv6, IP range 2a00:c31:1133:900::2 --
> 2a00:c31:1133:900:ffff:ffff:ffff:ffff, lease time 1h
> dnsmasq-dhcp: DHCPv6, IP range 2a00:c31:1933:900::2 --
> 2a00:c31:1933:900:ffff:ffff:ffff:ffff, lease time 1h
> dnsmasq: using local addresses only for domain lan
> dnsmasq: reading /tmp/resolv.conf.auto
> dnsmasq: using nameserver 2a00:c31:e404:0:ffff:ffff:ffff:fc72#53
> dnsmasq: using nameserver 2a00:c31:e404:0:ffff:ffff:ffff:fc71#53
> dnsmasq: using local addresses only for domain lan
> dnsmasq: read /etc/hosts - 1 addresses
> dnsmasq-dhcp: read /etc/ethers - 0 addresses
> dnsmasq-dhcp: RTR-ADVERT(br-lan) 2a00:c31:8533:900::
> dnsmasq-dhcp: RTR-ADVERT(br-lan) fd46:b82e:3f6d::
> dnsmasq-dhcp: RTR-ADVERT(br-lan) 2a00:c31:8533:900::
> dnsmasq-dhcp: RTR-ADVERT(br-lan) fd46:b82e:3f6d::
> dnsmasq-dhcp: RTR-ADVERT(br-lan) 2a00:c31:8533:900::
> dnsmasq-dhcp: RTR-ADVERT(br-lan) fd46:b82e:3f6d::
> dnsmasq-dhcp: DHCPSOLICIT(br-lan) 00:01:00:01:19:ba:81:2c:70:54:d2:7b:32:45
>
> And here's the exact spot & backtrace:
> 0xb6ff0e38 in _start ()
>     from
> /home/kaspar/tlabs/buildbot/openrouter3/ea4500_noclean/build/scripts/../staging_dir/target-arm_v5te_uClibc-0.9.33.2_eabi/root-kirkwood/lib/ld-uClibc.so.0
>
> (gdb) c
> Continuing.
>
> Program received signal SIGSEGV, Segmentation fault.
> 0x0002c8e0 in match_netid (pool=<optimized out>, check=0x3fb40,
> tagnotneeded=<optimized out>) at dhcp-common.c:170
> 170          if (check->net[0] != '!' && check->net[0] != '#')
> (gdb) bt
> #0  0x0002c8e0 in match_netid (pool=<optimized out>, check=0x3fb40,
> tagnotneeded=<optimized out>) at dhcp-common.c:170
> #1  match_netid (check=<optimized out>, pool=0xbefff900,
> tagnotneeded=<optimized out>) at dhcp-common.c:160
> #2  0x0002b600 in dhcp6_no_relay (now=52, is_unicast=263256,
> sz=<optimized out>, inbuff=0xbefff9bc, fallback=0x2c758 <dhcp6_reply+156>,
>      iface_name=0x18 <Address 0x18 out of bounds>, interface=0,
> context=0x3fd90, tags=<optimized out>, link_address=<optimized out>,
> msg_type=<optimized out>)
>      at rfc3315.c:683
> #3  dhcp6_maybe_relay (link_address=<optimized out>,
> relay_tagsp=<optimized out>, context=0x3fd90, interface=0,
> iface_name=0xbefff9bc "br-lan",
>      fallback=0xbefff9e0, inbuff=0x40458, sz=52, is_unicast=0,
> now=1378314215) at rfc3315.c:149
> #4  0x0002c758 in dhcp6_reply (context=0x3fd90, interface=9,
> iface_name=0xbefff9bc "br-lan", fallback=0xbefff9e0, sz=52,
> is_unicast=0, now=1378314215)
>      at rfc3315.c:88
> #5  0x00028d14 in dhcp6_packet (now=1378314215) at dhcp6.c:173
> #6  0x0000b848 in main (argc=<optimized out>, argv=<optimized out>) at
> dnsmasq.c:884
> (gdb)
>
>
> Output from dhclient on wheezy:
> root at router666:~# dhclient -6 -sf /bin/echo --no-pid -d -v eth1
> Internet Systems Consortium DHCP Client 4.2.2
> Copyright 2004-2011 Internet Systems Consortium.
> All rights reserved.
> For info, please visit https://www.isc.org/software/dhcp/
>
>
> Bound to *:546
> Listening on Socket/eth1
> Sending on   Socket/eth1
> PRC: Soliciting for leases (INIT).
> XMT: Forming Solicit, 0 ms elapsed.
> XMT:  X-- IA_NA d2:7b:32:45
> XMT:  | X-- Request renew in  +3600
> XMT:  | X-- Request rebind in +5400
> XMT: Solicit on eth1, interval 1010ms.
> XMT: Forming Solicit, 1010 ms elapsed.
> XMT:  X-- IA_NA d2:7b:32:45
> XMT:  | X-- Request renew in  +3600
> XMT:  | X-- Request rebind in +5400
> XMT: Solicit on eth1, interval 1960ms.
> XMT: Forming Solicit, 2970 ms elapsed.
> XMT:  X-- IA_NA d2:7b:32:45
> XMT:  | X-- Request renew in  +3600
> XMT:  | X-- Request rebind in +5400
> XMT: Solicit on eth1, interval 3970ms.
> XMT: Forming Solicit, 6940 ms elapsed.
> XMT:  X-- IA_NA d2:7b:32:45
> XMT:  | X-- Request renew in  +3600
> XMT:  | X-- Request rebind in +5400
> XMT: Solicit on eth1, interval 8310ms.
> XMT: Forming Solicit, 15260 ms elapsed.
> XMT:  X-- IA_NA d2:7b:32:45
> XMT:  | X-- Request renew in  +3600
> XMT:  | X-- Request rebind in +5400
> XMT: Solicit on eth1, interval 17260ms.
>
>
> Ideas?
>

Fabulous bug report, thanks. I'll take 	a look at this tomorrow. It will 
probably make my life easier to have a copy of your dnsmasq config:
/var/etc/dnsmasq.d/* and /etc/dnsmasq.conf.

Would you mind sharing that? (off-list, if you prefer.)

Cheers,

Simon.