DragonFly users List (threaded) for 2007-01
Re: write news article about virtual kernel
:One question I am surprised not to find asked is: how do virtual
:kernels work on SMP systems? A layered question I suppose; how would
:the virtual kernel itself scale, and also how would its processes
:scale, and how would the threads of those processes scale? The first
:thing I think when I hear of a new (to me) virtualization system is
:how well it would work with SMP.
At the moment the virtual kernel builds as a UP kernel. A
SMP virtual kernel isn't too much of a leap... all it really
has to do is rfork(). Certain mechanisms (IPI messaging mainly)
would have to be implemented.
:Yet another question is, is there any long-term plan to allow virtual
:kernels to control specific items of host hardware? This could be
:extremely useful for fast development of drivers, if the host memory
:protection is done properly. I think Xen has some facilities for this.
:It's part of the way towards DragonFly having safe userland drivers:
:using userland kernels with (what appears to them as) kernel drivers.
:This seems a much simpler initial step than a whole new interaction
:layer and architecture for userland drivers, even if it's not nearly
:as simple or efficient as what the final result should be.
No. Frankly I do not think it is a good idea to allow any
production virtualization mechanism to ever have direct access
to hardware. It destroys the layering that gives virtualization
stability and security... no matter how good the implement is.
As a debugging tool it might be useful, but that is about as
far as I would ever consider taking it.
:Thanks in advance for any responses.
:Centre for Synchrotron Science
:Victoria 3800, Australia