[comp.sys.pyramid] Should kmem be read to the world?

mb@rex.cs.tulane.edu (Mark Benard) (06/27/89)

We just noticed that w does not work.  It returns the message "no kmem".
Should /dev/kmem be made readable to the world or will that cause
security problems?

(We are running OSx4.4 on a 9815.)
-- 
Mark Benard
Department of Computer Science     INTERNET & BITNET: mb@rex.cs.tulane.edu
Tulane University                  USENET:   [{ames,bionet}!]rex!mb
New Orleans, LA 70118

karl@giza.cis.ohio-state.edu (Karl Kleinpaste) (06/27/89)

mb@rex.cs.tulane.edu writes:
   We just noticed that w does not work.  It returns the message "no kmem".
   Should /dev/kmem be made readable to the world or will that cause
   security problems?

That will cause security problems; the intelligent cracker will learn
all kinds of fascinating things by reading /dev/kmem.

We define a group "devkmem" with no members, and then chgrp all memory
devices to this group, and in turn chgrp and chmod g+s all the
memory-reading programs (ps, w, top, etc) so that they retain their
well-behaved access.

--Karl

csg@pyramid.pyramid.com (Carl S. Gutekunst) (06/27/89)

In article <856@rex.cs.tulane.edu> mb@rex.cs.tulane.edu (Mark Benard) writes:
>We just noticed that w does not work.  It returns the message "no kmem".
>Should /dev/kmem be made readable to the world or will that cause
>security problems?

Making /dev/kmem world readable is *definitely* a security problem. It looks
like someone has been careless when doing file copies or updates. As shipped,
the permissions on /dev/kmem are:

	cr--r-----  1 root     sys        3,   1 May  4  1988 /dev/kmem

and the permissions on the kmem-reader utilities are:

	-rwxr-sr-x  1 bin      sys         61440 Apr 29  1988 /bin/ps
	-rwxr-sr-x  1 bin      sys         45056 Apr 14  1988 /etc/pstat
	-rwxr-sr-x  1 bin      sys         32768 Apr 13  1988 /usr/ucb/vmstat

etc.

<csg>