Re: looking for the solution of rootless subhurd

[olafBuddenhagen]

Hi,

On Sat, Jan 17, 2009 at 01:38:11AM +0000, Zheng Da wrote:
> On Thu, Jan 15, 2009 at 11:03 AM, <olafBuddenhagen@gmx.net> wrote:

> > Indeed, while the primary goal is running subhurd as normal user, it
> > is evidently a desirable variation to have a completely isolated
> > subhurd.
[...]
> I don't really accept this option. I see subhurd as a substitute of
> the virtual machine where can only run Hurd, at least, it's what I
> hope subhurd should be.

This is not entirely true: At least in theory, it should be possible to
run *any* Mach-based system...

Admittedly, there are few others left ;-)

> So if I runs two subhurd, what I wants is an isolated environment to
> do some experiments, but somehow the program in one subhurd can crash
> the other subhurd. Isn't it very strange?

That really depends on the situation. Total isolation is not always
desirable. The flexibility is indeed the major advantage of Hurd-based
virtualization, as I pointed out in
http://tri-ceps.blogspot.com/2007/10/advanced-lightweight-virtualization.html

> Since I think subhurd as a virtual machine, I hope it can give more
> control of resource usage to the user. For example, limit its memory
> usage (It's the only example I can think of now:-).

Unfortunately this is not possible with the current Mach-based Hurd
implementation.

The impossibility of tracking memory usage in the existing
implementation was in fact the major motivation (or at least one of the
major motivations) for the Hurd/L4 port; and it's also the focus of
Neal's current research work.

I'm not saying that it's impossible to add such abilities -- but it
will require very fundamental changes both to Mach and to the Hurd...

-antrik-