[Dnsmasq-discuss] dnsmasq rc5 segfault

[Carlos Carvalho]

Carlos Carvalho wrote:
> Today dnsmasq died. It was the first time since I use it :-(
> 
> In its log there's no sign of problems. The last entry is for a
> DHCPNAK; it's the only such record with 2.43. The kernel log says
> 
> Jul 11 14:27:38 hoggar dnsmasq[11797]: segfault at 10 ip 08061e91 sp bfffe770 error 4 in dnsmasq[8048000+22000]

It died again today. Again, the last line in the log was a DHCPNAK:

Jul 14 13:19:18 dnsmasq[6873]: DHCP packet: transaction-id is 2892506353
Jul 14 13:19:18 dnsmasq[6873]: Available DHCP subnet: 192.168.25.1/255.255.255.0
Jul 14 13:19:18 dnsmasq[6873]: Vendor class: MSFT 5.0
Jul 14 13:19:18 dnsmasq[6873]: DHCPREQUEST(eth0.25) 10.1.1.6 00:50:c2:50:81:15 
Jul 14 13:19:18 dnsmasq[6873]: DHCPNAK(eth0.25) 10.1.1.6 00:50:c2:50:81:15 wrong network

It's the second time it segfaults after this. There are no other NAKs
in the logs, so maybe it's reproducible. gdb says it stopped at

Program terminated with signal 11, Segmentation fault.
[New process 6873]
#0  0x080640df in server_id (context=0x0, override={s_addr = 0})
    at rfc2131.c:968
968                     return context->local;

This is the stack:

#0  0x080640df in server_id (context=0x0, override={s_addr = 0}) at rfc2131.c:968
#1  0x08063463 in dhcp_reply (context=0x0, iface_name=0xbffff234 "eth0.25", int_index=25, sz=300, now=1216052358, unicast_dest=0, is_inform=0xbffff1f8) at rfc2131.c:779
#2  0x0805d933 in dhcp_packet (now=1216052358) at dhcp.c:231
#3  0x0805bdd4 in main (argc=Cannot access memory at address 0xb
) at dnsmasq.c:617

In 2.42 option_put(mess, end, OPTION_SERVER_IDENTIFIER,...) tests for
context being null, 2.43 doesn't. Maybe this is why 2.42 survives,
which doesn't mean it's correct however.