NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/alt/sys/sun/3587

Internet Message Format | 1993-01-24 | 1.6 KB

Xref: sparky alt.sys.sun:3587 comp.sys.sun.misc:6465
Path: sparky!uunet!auspex-gw!guy
From: [e-mail] (Guy Harris)
Newsgroups: alt.sys.sun,comp.sys.sun.misc
Subject: Re: Can "normal" users mount under Solaris 2.X
Message-ID: <[e-mail]>
Date: 24 Jan 93 02:03:38 GMT
References: <[e-mail]> <[e-mail]>
Sender: [e-mail]
Followup-To: alt.sys.sun
Organization: Auspex Systems, Santa Clara
Lines: 21
Nntp-Posting-Host: auspex.auspex.com

>It is generally 'unsecure' to have shellscripts setuid root. So it might be
>possible that setuid for scripts was disabled in Solaris.

No, it's not disabled; what may be the most infamous security hole is
closed in SVR4, although there may still be holes left (do *you* trust
your favorite shell not to leave any holes open? Do *you* trust there
not to be any *additional* kernel holes, even though the most infamous
one has been closed?).

You *do* have to put a "-p" flag in the "#! /bin/sh" line; that flag
tells the shell not to surrender set-UID privileges, and also, at least
in part, prevents subsequent arguments that begin with "-" from spoofing
the shell (similar to the "-b" flag in the 4.3BSD C shell and
derivatives thereof, including the SunOS 4.x C shell and the SVR4 C
shell).

Also, be aware that even if the kernel and shell don't have any such
holes, the script *itself* may have holes.  Does your script set the
path before running *any* commands (other than builtin commands)?  Does
it set IFS if it's a Bourne/Korn/Bourne-again shell script and the shell
in question imports IFS from the environment?