[2040] in bugtraq
Re: Security Problem ftpd (includes wu.ftpd 2.4 and 2.4.2 beta 4)
daemon@ATHENA.MIT.EDU (Lyndon Nerenberg)
Thu Jul 13 14:00:19 1995
Date: Wed, 12 Jul 1995 23:35:25 -0700
Reply-To: Bugtraq List <BUGTRAQ@CRIMELAB.COM>
From: Lyndon Nerenberg <lyndon@multivac.orthanc.com>
X-To: Bugtraq List <BUGTRAQ@CRIMELAB.COM>
To: Multiple recipients of list BUGTRAQ <BUGTRAQ@CRIMELAB.COM>
In-Reply-To: Your message of "Wed, 12 Jul 1995 13:58:34 PDT."
<199507122058.NAA18469@hiroaki.nbs.nau.edu>
"James W. Abendschan" writes:
>Maybe I'm completely missing the point, but wouldn't this help?
>
> linux# chown root.kmem /proc
> linux# chmod 750 /proc
>
>And then sgid kmem all the binaries that need /proc access:
One of the big reasons for having /proc is to simplify the interface
for debuggers. Doing this means setgid dbx/gdb/adb/...
Wouldn't it be better to have files in /proc export the owner/group/perm
bits of the underlying file (where applicable), with a further restriction
that write access follow the semantics of ptrace(2)? Or perhaps have
/proc/pid/* use the underlying perms, but restrict /proc/pid/fd/* to
being writable only by the current process and the parent as per ptrace(2)?
Of course, write access to /proc/pid/mem effectively bypasses any
additional restrictions placed on /proc/pid/fd/*.
--lyndon
