Re: [PATCH 0/9] namespaces: Introduction

From: Herbert Poetzl
Date: Sun May 21 2006 - 19:31:31 EST


On Sun, May 21, 2006 at 05:18:50PM -0600, Eric W. Biederman wrote:
> Pavel Machek <pavel@xxxxxx> writes:
>
> > Well, if pid #1 virtualization is only needed for pstree, we may want
> > to fix pstree instead :-).

yes, actually this and init itself (which uses the
pid to switch between init and telinit behaviour)
are the only two applications we found so far ...

and as far as I know, those work with non pid=1
values on other operating systems (inside containers)

a fix there would definitely be appreciated and
I think it would not hurt normal behaviour ...

> One thing that is not clear is if isolation by permission checks is
> any easier to implement than isolation with a namespace.

for the pid space, I'm not really sure if isolation
is really cheaper than virtualization, but for the
network space for example, a virtualization solution
which is as lightweigth as the isolation is probably
more challenging, although not impossible ...

> Isolation at permission checks may actually be more expensive in terms
> of execution time, and maintenance.

again, for the pid space, maintenance is quite low ..

best,
Herbert

> Eric
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/