bug#4225 (was: unbound 1.9 respone wrong answer (from cache?) when dig +tcp used.)

[A. Schulze]

Am 08.02.19 um 16:58 schrieb Wouter Wijngaards via Unbound-users:
> This seems to be bug#4225, for which there is a fix in the code
> repository. It was just fixed an hour ago. The patch for it is this, I
> think:

Hello Wouter,

https://www.nlnetlabs.nl/bugs-script/show_bug.cgi?id=4225
also mention an other fix in qname minimisation.

do you suggest to apply both or will there a next version soon?

Andreas

> 
> 
> Index: services/listen_dnsport.c
> ===================================================================
> --- services/listen_dnsport.c    (revision 5105)
> +++ services/listen_dnsport.c    (working copy)
> @@ -1779,6 +1779,12 @@
>      /* If mesh failed(mallocfail) and called commpoint_send_reply with
>       * something like servfail then we pick up that reply below. */
>      if(req->is_reply) {
> +        /* reply from mesh is in the spool_buffer */
> +        sldns_buffer_clear(c->buffer);
> +        sldns_buffer_write(c->buffer,
> +            sldns_buffer_begin(req->spool_buffer),
> +            sldns_buffer_limit(req->spool_buffer));
> +        sldns_buffer_flip(c->buffer);
>          goto send_it;
>      }
>  
> 
> 
> Best regards, Wouter
> 
>>>
>>>  
>>>
>>> dig say “WARNING: recursion requested but not available”, 
>>>
>>> kdig say “WARNING: response QR bit not set”
>>>
>>>  
>>>
>>> Back to 1.8.3, everthing work fine.
>> can't reproduce:
>>
>> $ dig @::1 version.bind. txt ch +short
>> "unbound 1.9.0"
>>
>> $ for i in 1 2 3 ; do dig @::1 twitter.com +tcp +short; done
>> 104.244.42.193
>> 104.244.42.65
>> 104.244.42.193
>> 104.244.42.65
>> 104.244.42.65
>> 104.244.42.193
>>
>> any special configuration / compile options?
>>
>> Andreas
>>