Re: Extensions Infrastructure Work

["Jasper St. Pierre" <jstpierre mecheye net>]

On Wed, Jun 22, 2011 at 3:46 PM, Olav Vitters <olav vitters nl> wrote:
> On Wed, Jun 22, 2011 at 03:39:19PM -0400, Jasper St. Pierre wrote:
>> On Wed, Jun 22, 2011 at 3:27 PM, Olav Vitters <olav vitters nl> wrote:
>> > On Tue, Jun 21, 2011 at 07:12:53PM -0400, Jasper St. Pierre wrote:
>> >> As I played around with it, I found the HTTP approach more feasible
>> >> and less ugly than the mimetype handler approach. At first I figured
>> >> the idea of running a local HTTP server would be a bit ugly, and Owen
>> >> thought of some security concerns, but there's nothing too critical
>> >> (or unsolvable) that I know of. The only "ugly" thing from a code
>> >> perspective is that there's a magic port number: 16269. It's not on
>> >> the IANA Registered Ports list, so I doubt there's going to be a
>> >> collision.
>> >
>> > Won't that break down in two cases:
>> > 1. Proxy set in the browser
>> >   User/sysadmin has to explicitly exclude localhost from being proxied
>>
>> I'm unsure how or why localhost would be proxied. If it's some DNS
>> quirk would 127.0.0.1 get around it? If not, is this something we can
>> put in the sysadmin documentation?
>
> Why not? If you put in a proxy setting, everything is proxied, including
> localhost, 127.0.0.1, etc. The browser will just connect to the proxy
> machine (which is pretty handy btw).

Good to know.

> I don't know what the default for 'do not proxy for' is in the various
> browsers, but I know I make use of the fact that localhost is proxied.
>
>> > 2. Multiple users or sessions on the same machine
>> >   Only the first session can use it.
>>
>> My idea was that log-out would stop the HTTP daemon for that session
>> and open one for the current user. Unless there's a special case (I
>> didn't think of virt) where two users can be securely both actively
>> having GNOME sessions at the same time, I don't think this is a
>> problem. The only security issue I can think of that arises out of
>> this compromise is that a user could ssh in to the same machine and
>> frob the HTTP server to... install, enable/disable and list extensions
>> from the official GNOME3 site.
>
> That does not seem ideal. If I give someone access to my machine, I
> don't want them being able to change anything belonging to my account. I
> don't care if it is only official extensions. I just don't think it
> should be possible.

Er, sorry, the tone came off wrong there -- I agree that it's a serious issue.

Owen and I thought of another hack to deal with security, though: We
can inspect the port that it was connected from and scrape
/proc/net/tcp for the UID, and make sure it matches the one the
current section.

>> I assume there's no magic way to tie a TCP socket to a user's session
>> (paging Dr. Lennart Poettering)
>
> --
> Regards,
> Olav
>



-- 
 Jasper