Subject: Re: systrace features?
To: Charles Blundell <cb@NetBSD.org>
From: Niels Provos <provos@citi.umich.edu>
List: tech-security
Date: 09/29/2003 12:35:58
On Wed, Sep 24, 2003 at 12:36:37PM +0100, Charles Blundell wrote:
> but by having the logic in systrace we can allow successive calls to
> work, then have one fail. (so we can probe deeper into the code.)

Actually, the connection to the frontend is persistent.  You can
keep whatever state you like in there.  So, allowing the first
ten invocations of a system call and then denying the 11th would
not be a problem at all.

I believe that keeping this out of Systrace is a good approach
as it allows you to support more complex and sophisticated fault
injection.

Niels.