On Sep 16, 2015, at 11:40 PM, Alex Wilson <[email protected]> wrote:

> Robert Seastrom <[email protected]> wrote:
> 
>> My $0.02 - I consider this a wobbler. If it weren't for the relatively
>> short notice that we're getting rid of it, I wouldn't particularly support
>> putting it back in, but the OpenSSH guys have good reasons for getting rid
>> of it.
>> 
>> Given that the consequences could be fairly severe if people are counting
>> on hosts.allow/deny to save them from "the Internet", if you put it back
>> as a migration strategy it might be worthwhile for sshd to complain on
>> interactive logins, system logs, and anywhere else that it might be
>> noticed if it finds non-default hosts.allow and hosts.deny files. You’ve
>> been able to implement the functionality via "Match" for literally
>> years…
> 
> Unfortunately the libwrap API doesn’t seem to provide a decent way to detect
> this. 

Fair enough, but it's simple enough to iterate over the appropriate files (if 
they exist) looking for lines that begin with [Ss][Ss][Hh][Dd]: and squawk if 
any are found.  I encourage this sort of behavior as an advance warning for a 
while if un-revert-patching it is in the offing.

> For now I’ve just put in a revert patch that adds the support for
> tcpwrappers back in as it was in older versions (no deprecation warning).
> It’s a pretty short patch anyway, so carrying it around for an extended
> period doesn’t seem like a huge technical risk. It should be in the release
> that gets cut tomorrow.

That sounds reasonable.

Thanks!

-r


-------------------------------------------
smartos-discuss
Archives: https://www.listbox.com/member/archive/184463/=now
RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=25769125&id_secret=25769125-7688e9fb
Powered by Listbox: http://www.listbox.com

Reply via email to