With the caveat that I've never used of ppriv (and never heard of it 
till just now) things look fine to me. I agree that in read_only, it's 
the case that it shouldn't be able to fork and exec. That might change 
if/when we add a scratch area, or we'll need a different flag than 
read_only for that situation.  I've got no objections.

Brock

Dan Price wrote:
> I've prototyped an enhancement: the method script which launches the
> depot will now relinquish a substantial number of unneeded privileges.
> This is context-aware, dropping more privileges for depots running
> read-only.
>
> http://cr.opensolaris.org/~dp/pkg-priv/
>
> I'd like feedback on this idea.  I've implemented it, it seems to work.
> Whether it's the right approach for the long term or whether it fits
> the "correct SMF way to do things"-- since we could specify privs via
> method contexts-- I'm not sure.  I do think it will bolster our security
> in the short term.
>
> Stephen, Shawn & Brock I would like you to take a look.
>
>         -dp
>
>   

_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to