Re: POSIX

[Jonathan S. Shapiro]

On Wed, 2005-10-26 at 15:35 +0200, Alfred M. Szmidt wrote:
>       If the sub-hurd is going to be the basic mechanism of security,
>       then EVERY new execution of every application should be
>       performed in a freshly instantiated sub-hurd.
> 
> You are assuming that each and every application is hostile, that
> isn't the case.  If you have something that can be considered hostile
> (say, something that needs root privs), you can run it in a seperate
> enviroment.  Enclosing each and every process into its own jail-like
> enviroment is beyond absurd.

Yes, I am definitely assuming this, because in my experience this is
actually true. Let's look at the three most common applications that
real users use:

  Web browsers
  Email readers
  Word processors
  Document browsers (e.g. acrobat, xpdf, ghostview)

Each of these runs code written by a very large number of untrusted
developers, and each downloads "plugins" (or equivalently: can spawn
local commands at the direction of documents) that I know nothing about.
The plugin code very often *is* hostile, and the programs that run them
very often contain security bugs.

So I would say that for the vast majority of program executions that I
do in a given day, yes, I would need a subhurd for every single one.

On the server side, things are even worse -- for those I need a new
sub-hurd for every page request that involves any sort of active
content.


shap