[940] in WWW Security List Archive
re: Random seed
daemon@ATHENA.MIT.EDU (Stan Tazuma 206-865-4261)
Tue Sep 26 18:18:08 1995
Date: Tue, 26 Sep 95 12:14:26 PDT
From: Stan Tazuma 206-865-4261 <skt@espresso.rt.cs.boeing.com>
To: elgamal@netscape.com
Cc: stw@espresso.rt.cs.boeing.com, mignon@espresso.rt.cs.boeing.com,
karen@espresso.rt.cs.boeing.com, rpwhite@espresso.rt.cs.boeing.com,
www-security@ns2.rutgers.edu
Errors-To: owner-www-security@ns2.rutgers.edu
Here's another source of random information. The "vmstat -s"
command will usually generate some numbers that change
fairly rapidly. Running it several times in a row as fast as the shell
can run it shows that the following lines are good candidates for
random information:
(below is a portion of a "diff" on the output between 2 successive
runs of "vmstat -s")
25,29c25,29
< 98310893 cpu context switches
< 462167243 device interrupts
< 418211889 system calls
---
> 98311007 cpu context switches
> 462167742 device interrupts
> 418212244 system calls
"vmstat -s" works on Solaris 1.x (SunOS 4.x) and Solaris 2.x.
Should be available on most other Unix systems.
The numbers probably start at 0 when the system boots, and then
increment from there. The last 6-9 bits worth appear to be
fairly random, and more random than the output of "netstat -ni"
or "netstat -na" (at least in our environment).
Stan Tazuma
Boeing
skt@espresso.rt.cs.boeing.com
(I'm not on the www-security mailing list, so use direct e-mail
if you want to contact me.)
