2009-12-11 11:32:48

by Ivan Zahariev

[permalink] [raw]
Subject: EUID != root + EGID = root, and CAP_SETGID

Hi guys,

Currently, if a process is started with EUID which is non-root, and EGID
which IS root (for example by set-group-ID file permission + file group
owner "root", or an account in /etc/passwd with group=0), then the
processes is not granted CAP_SETGID.

As a result, such a process cannot change its EGID to an arbitrary one,
even though the current EGID is the super-user "root" one. Therefore,
such a process cannot easily drop its EGID "root" privileges to non-root
ones, for security reasons.

This is not the case if the process starts with EUID=0. Then the
processes is granted *both* CAP_SETUID and CAP_SETGID.

Is this an intended behavior? Shouldn't a process which is started with
EGID=0 get CAP_SETGID too?

Thank you.

Best regads,
Ivan Zahariev

P.S. For more detailed info:
http://blog.famzah.net/2009/12/11/linux-non-root-user-processes-which-run-with-group-root-cannot-change-their-process-group-to-an-arbitrary-one/


2009-12-11 12:04:52

by Andreas Schwab

[permalink] [raw]
Subject: Re: EUID != root + EGID = root, and CAP_SETGID

Ivan Zahariev <[email protected]> writes:

> As a result, such a process cannot change its EGID to an arbitrary one,
> even though the current EGID is the super-user "root" one.

There is no such thing as a "super-user group". No group has any
special privleges.

Andreas.

--
Andreas Schwab, [email protected]
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756 01D3 44D5 214B 8276 4ED5
"And now for something completely different."

2009-12-13 05:45:27

by David Wagner

[permalink] [raw]
Subject: Re: EUID != root + EGID = root, and CAP_SETGID

Ivan Zahariev wrote:
>Is this an intended behavior?

Yes. Setuid/setgid are a mess. For more details, you might
find the following research papers interesting:

http://www.cs.berkeley.edu/~daw/papers/setuid-usenix02.pdf
http://www.cs.berkeley.edu/~daw/papers/setuid-login08b.pdf

See, e.g., Section 5.2 of the former paper, which says:

"an effective group ID of zero does not accord any
special privileges to change groups. This is a potential
source of confusion: it is tempting to assume incorrectly
that since appropriate privileges are carried by the euid
in the setuid-like calls, they will be carried by the
egid in the setgid-like calls, but this is not how it
actually works. This misconception caused a mistake in
the manual page of setgid in Redhat Linux 7.2 (Section
6.4.1)."