[Dnsmasq-discuss] rev-server=fe80::/10, 192.168.178.1 no longer accepted (version 2.86)

Simon Kelley simon at thekelleys.org.uk
Sat Sep 11 22:42:40 UTC 2021


On 11/09/2021 16:23, Dominik DL6ER wrote:
> Hey Simon,
> 
> On Fri, 2021-09-10 at 23:48 +0100, Simon Kelley wrote:
>> For IPv6, this seems like an obviously good thing to do, since
>> one
>> --rev-server can create as most eight .....ip6.arpa, but for
>> IPv4, that
>> goes up to 128 since for instance 192.168.1.0/25 expands to
>>
>> 0.1.168.192.in-addr.arpa
>> 1.1.168.192.in-addr.arpa
>> .
>> .
>> 127.1.168.192.in-addr.arpa
>>
>>
>> I guess that now we have efficient searches over domains, that's
>> more
>> likely to be OK, but it needs some consideration.
> 
> I see, however, this is surely the worst case and not something
> that'd happen very often. The typical use case I've seen being
> requested for Pi-hole is something like /22 which would result in
> 4 generated entries.
> 
> Patch attached, it allows arbitrary subnets and explains what it
> is doing to syslog (similar to how we do it for DHCP). It applies
> cleanly to the current master.
> 
> I guess it's pretty self-explanatory, more details can be found
> in my last submission:
> 
> https://lists.thekelleys.org.uk/pipermail/dnsmasq-discuss/2020q2/013985.html
> https://lists.thekelleys.org.uk/pipermail/dnsmasq-discuss/2020q2/013997.html
> 
> Everything is combined in one patch now.
> 
> Best,
> Dominik
> 


That looks good, but I hit a couple of problems.

1) gen_rev6 fails to allocate the *domains array, and crashes.

2) The explanation stuff falls foul of the problem that it's generally
running during whilst reading options, and before the logging system has
been initialised. To make that work sensibly, you need to store the
information and do the logging after around line 860 of dnsmasq.c
I'd be inclined just to omit this. The final results of the process get
logged, after all.

I've run out of time to look further now, and I'm away next week and
unlikely to get more time until the end of the week. Then, I'll do more
testing.


cheers,

Simon.




More information about the Dnsmasq-discuss mailing list