Should standard binaries & directories revert from uid=root to bin ?

[Alfred Perlstein]

[[ Top post replying just for lols. ]]

Julian, please read up on how maproot works with NFS.

I think Adrian's paranoia trumps your convenience, but
I'm open to being convinced otherwise.

-Alfred

* Julian H. Stacey <jhs at berklix.com> [120330 13:17] wrote:
> Hi Adrian & arch@
> Please don't top post to arch at freebsd.org
> Please don't emit messy quoted-printable hex. '\xa0' for clean spaces.
> 
> Adrian Chadd wrote:
> > hi,
> > 
> > because id=0 defaults to being squashed via nfs.
> 
> Not a sentence.  Please clarify.
> 
> 
> > But if you have a
> > filesystem full of uid=bin/gid=bin binaries, a slightly insecure NFS
> > setup would allow NFS clients to simply set their uid=bin and change
> > these binaries. :-)
> 
> I don't understand your meaning. I do understand SUID though.
> Please clarify whay you mean.
> 
> Do you mean if something like /usr/sbin/lpd was uid=bin on one
> system, it might slip via a bad NFS to be seen as UID=0 on another ?
> & remotely excutable on 2nd system as a UID=0 ? 
> If that's what you mean, bear in mind /usr/sbin/lpd is currently already
> uid=0.  Also bear in mind NFS man exports -maproot
> 
> Are you stating? or just speculating ? if [flakey?] NFS was the
> reason FreeBSD changed from bin to root ?
> 
> I hadn't considered NFS lax security when I asked the question.
>   (I had merely mentioned NFS in context of explaining how I
>   (re-)noticed the wholesale conversion from bin to root.
> 
> It's possible NFS might have been a reason ?
> but I don't see you made an explanation [yet] as to how 
> a return from root to bin would be dangerous with a flakey NFS ?
> 
> Not that I'm saying it would/ wouldn't be an issue,
> I am just asking why we changed, & if a move back would be good ?
> As I see one loss from the change.
> There may have been other issues though ? Anyone know ?
> 
> 
> > On 30 March 2012 08:16, Julian H. Stacey <jhs at berklix.com> wrote:
> > > Hi arch@
> > > Time was, (& I can go back over 25 years here, but more recently too :-)
> > > When standard Unix non SUID executables such as wc would be UID=bin,
> > > GID=bin, & not root.  Ditto bin/ & lib/ etc directories.
> > >
> > > One advantage was:
> > >  Anything that showed up with ls -l as UID=0 was either a SUID
> > >  special, known to the admin's eye, or some administrative dropping,
> > >  mistakenly created by someone logged in as root, to be reviewed/
> > >  regenerated/ deleted.
> > >
> > > Now all is UID=0.  Why ? What advantage did it bring ?
> > >
> > > Obviously some SUID & SGID executables need 0 (some could need just bin!)
> > > but most files & directories do not need UID 0.
> > >
> > > BTW, How I noticed this :
> > >  I was tracing why
> > >        /usr/sbin/sshd -d -d -d -D
> > >  was erroring:
> > >        debug3: secure_filename: checking '/.amd_mnt/sshd_host/ad4s1/usr1/home'
> > >        Authentication refused: bad ownership or modes for directory
> > >                 /.amd_mnt/sshd_host/ad4s1/usr1/home
> > >  just because my ~/.ssh was symbolicaly linked via AMD+NFS mounted on another
> > >  host, & there an intermediate directory was owned by bin & not root,
> > >        ls -la /host/sshd_host/ad4s1/usr1/home
> > >                drwxr-xr-x  18 bin     bin       512 Mar  6 11:56 ./
> > >  so I had to
> > >        chown root:wheel /ad4s1/usr1/home
> > >  Just to satisfy sshd being pointlessly strict, as directory was 755.
> > >
> > > So we have sshd that's pointlessly strict, & ownerships that seem
> > > to have near all lost their precision. A funny combo ;-)
> > >
> > > Might others tackle the generic over use of root ?
> > > If so I could create a patch to send-pr ssh  ?
> > > (but as ssh is an import, maybe just report & not [yet?] patch ?)
> > >
> > > Cheers,
> > > Julian
> > > --
> > > Julian Stacey, BSD Unix Linux C Sys Eng Consultants Munich http://berklixcom
> > >  Reply below not above, cumulative like a play script, & indent with "> ".
> > >  Format: Plain text. Not HTML, multipart/alternative, base64, quoted-printable.
> > >        Mail from @yahoo dumped @berklix.  http://berklix.org/yahoo/
> > > _______________________________________________
> > > freebsd-arch at freebsd.org mailing list
> > > http://lists.freebsd.org/mailman/listinfo/freebsd-arch
> > > To unsubscribe, send any mail to "freebsd-arch-unsubscribe at freebsd.org"
> > _______________________________________________
> > freebsd-arch at freebsd.org mailing list
> > http://lists.freebsd.org/mailman/listinfo/freebsd-arch
> > To unsubscribe, send any mail to "freebsd-arch-unsubscribe at freebsd.org"
> > 
> > 
> 
> 
> Cheers,
> Julian
> -- 
> Julian Stacey, BSD Unix Linux C Sys Eng Consultants Munich http://berklix.com
>  Reply below not above, cumulative like a play script, & indent with "> ".
>  Format: Plain text. Not HTML, multipart/alternative, base64, quoted-printable.
> 	Mail from @yahoo dumped @berklix.  http://berklix.org/yahoo/
> _______________________________________________
> freebsd-arch at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-arch
> To unsubscribe, send any mail to "freebsd-arch-unsubscribe at freebsd.org"

-- 
- Alfred Perlstein
.- VMOA #5191, 03 vmax, 92 gs500, 85 ch250, 07 zx10
.- FreeBSD committer