Article View: comp.security.misc
Article #109Re: mounting security leak
From: Alan J Rosenthal
Date: Fri, 07 Aug 1992 03:10
Date: Fri, 07 Aug 1992 03:10
24 lines
1606 bytes
1606 bytes
holland@dutiws.tudelft.nl (Gerhard den Hollander) writes: >If machine B mounts machine A, >I can execute (on machine B) my root-script on machine A >meaning I am root on machine B. Generally speaking, if you import information at this kind of security level, the source had better be trustworthy. Even if you allow users to mount floppy disks, they can edit them with a sector editor on their own machine (esp. the same kind of machine running a different OS, like a mac running mac OS if we're talking about an a/ux disk) and create setuid shells. >How can I prevent this ???? >Mounting with the nosuid option does NOT work. Do you mean it's not implemented on your system (quite common), or that it fails to perform the documented function? What kind of system is it? Nosuid isn't good enough, though. For example, people can still create device files and mount them. Being able to write /dev/kmem should be good enough to get you in, and all you have to do (on "machine A") is mknod myfile c 3 1; chmod 666 myfile. Also, you may have root-run programs which "trust" files on some filesystems imported from machine A; or for that matter, if you're root on machine A you can edit any non-root file it imports (how to do this is left as an exercise for the reader).
Technical Details
Message-ID:
<1992Aug6.231006.21580@jarvis.csri.toronto.edu>Path:
arc1-ca.pugleaf.net!sparky!uunet!utcsri!dgp.toronto.edu!flapsReferences:
<1959@dutiws.tudelft.nl>Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!utcsri!dgp.toronto.edu!flaps Newsgroups: comp.security.misc From: flaps@dgp.toronto.edu (Alan J Rosenthal) Subject: Re: mounting security leak Message-ID: <1992Aug6.231006.21580@jarvis.csri.toronto.edu> References: <1959@dutiws.tudelft.nl> Date: 7 Aug 92 03:10:06 GMT Lines: 24