[Bug network/19791] res_query.c:262 asserts when dns contains unreachable entries

["ppadmavilasom at vmware dot com" <sourceware-bugzilla at sourceware dot org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=19791

--- Comment #10 from priyesh <ppadmavilasom at vmware dot com> ---
(In reply to Florian Weimer from comment #9)
> Thanks.  I can reproduce it now.  I was running the test against the name
> server from the reproducer, not a real-world nameserver.  I have not found
> the root cause yet, though.
> 
> From what I can tell, the resolver is pretty much unusable in this
> configuration (IPv6 addresses disabled, IPv6 name server configured). 
> Supposedly successful DNS lookups do not return any data, and
> NXDOMAIN/NODATA responses result in the reported assertion failure.  Does
> this match what you see?
I have an ipv4 nameserver as well and it's usable. However, when a name like
foo.bar.google.com is used is when i run into this issue. 
> 
> I assume you discovered this right after applying the patch because
> everything stopped working.  Or was the failure more subtle?
>
If it was right after the patch, the correlation to the patch would have been
stronger but unfortunately this got noticed after the patch was sort of
forgotten. Actually, i remembered to backtrace to 2.22 without the patch as
part of the dialogue with you here.
The reason why this was noticed is because tcserver will die and restart when
it hits this and people noticed when they were using web interfaces.

> (I'm asking these questions to determine if this constitutes a security bug
> in its own right, rather than a regression exposed by a security fix.)
I guess any bug is a security bug but this one definitely escapes the chances
someone would run into it right away. However, it is possible to have a system
otherwise working and have specific scenarios run into this issue.

-- 
You are receiving this mail because:
You are on the CC list for the bug.