Article View: comp.security.misc
Article #117Re: Advice on (potential?) security problem
From: Valdis Kletnieks
Date: Sat, 08 Aug 1992 08:37
Date: Sat, 08 Aug 1992 08:37
27 lines
1518 bytes
1518 bytes
In article <1992Aug7.022422.9878@afterlife.ncsc.mil> sjr@afterlife.ncsc.mil (Stephen J. Roznowski) writes:
>Currently, the program does the following:
>
> system("/bin/stty -echo ...");
>
>Now suppose that I do "all the correct things" before issuing the
>"system" call (set IFS, etc.) is this worse/same as the following:
>
> tcgetattr(fd, &t);
> t.c_lflag &= ~ECHO;
> tcsetattr(fd, TCSANOW, &t);
Using 'system()' is a big loss security-wise. You can easily
verify the security impact of an ioctl to turn on raw mode. However,
system() is open to SO many abuses it isn't funny. For instance,
if PATH has '.' in it, you better hope your /bin/sh is one that
automatically resets IFS - otherwise, the guy can put a program
called 'bin' in the current directory, and nuke you - this was how
the infamous 'expreserve' hole worked - it did a 'system("/bin/mail"...)"
I'm sure the security mavens in this group can add another 5-6 different
ways to break a "system()" call.. ;)
Valdis Kletnieks
Computer Systems Engineer
Virginia Tech
Technical Details
Message-ID:
<5465@vtserf.cc.vt.edu>Path:
arc1-ca.pugleaf.net!sparky!uunet!vtserf!vttcf.cc.vt.edu!valdisReferences:
<1992Aug7.022422.9878@afterlife.ncsc.mil>Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!vtserf!vttcf.cc.vt.edu!valdis From: valdis@vttcf.cc.vt.edu (Valdis Kletnieks) Newsgroups: comp.security.misc,alt.security Subject: Re: Advice on (potential?) security problem Message-ID: <5465@vtserf.cc.vt.edu> Date: 8 Aug 92 08:37:42 GMT References: <1992Aug7.022422.9878@afterlife.ncsc.mil> Sender: usenet@vtserf.cc.vt.edu Followup-To: comp.security.misc Organization: Virginia Tech, Blacksburg, VA Lines: 27