Thanks a lot for the great input guys.
Bruce
From: Shidan <[EMAIL PROTECTED]>
To: "D. Hugh Redelmeier" <[EMAIL PROTECTED]>
CC: [email protected]
Subject: Re: [on-asterisk] x100p
Date: Mon, 17 Apr 2006 19:29:51 -0400
MIME-Version: 1.0
Received: from iran.unitycode.org ([205.207.27.18]) by bay0-mc7-f11.bay0.hotmail.com with Microsoft SMTPSVC(6.0.3790.1830); Mon, 17 Apr 2006 16:30:08 -0700
Received: (qmail 1864 invoked by alias); 17 Apr 2006 23:29:53 -0000
Received: (qmail 24549 invoked by uid 0); 17 Apr 2006 23:29:53 -0000
>VMWare does not give a virtual machine access to
>the raw PCI bus (nor the ensuing interrupts). That would be suicidal.
>It would damage security and sharing but would be great for efficiency
>and transparency. It would make it easy to support a lot of devices.
Thats exactly right, The problem is that to replicate the pci card,
you can't work with a generic abstracted memory model, you have to
keep a correspondance between the guests memory and the hosts (its
real, low level memory) memory, if they did have a "virtual pci bus"
then you could build it in a much more abstract way and you wouldn't
have to worry about things like busmastering. Sure its a hard job
............. but some company will do it eventually. Again the
important thing is now that the source is viewable, if companies like
digium or sangoma see a market in building drivers for vmware, then
they can do it.I don't think this will ever be one of those things
that someone would do as a hobby.
> On the other hand, Xen does in some cases give the client machines
> direct PCI bus access.
Which is the main reason why they will make it for xen before vmware.
The vmware kernel is very simple compared to xen, and there already
are t1 virtualizations for xen..........
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
Help protect your PC with Virus Guard and Firewall from MSN Premium. Join now and get the first two months FREE*
