This makes sense. If one node is used, the program listening on that end would
have to differentiate between different types of data (Ethernet, Sound, etc.). This
would have to be a fairly complicated program and it wouldn't be such an easy thing to
add functionality for a new device. If plex86 has it's own major, though, it can make
many devices so that each 'virtual device' can have a device to communicate to the
host OS with. Then separate daemons/programs can be written to take the data from the
real devices and move them where they need to go.
Justin Patrin
On Thu Mar 30, 2000 at 10:29:12AM -0800, Drew Northup wrote:
>
>
> --- Ramon van Handel <[EMAIL PROTECTED]> wrote:
> > >For reference, VMWare has a minor for /dev/vmmon
> > ("VMWare virtual
> > >machine monitor") and a major for /dev/vmnet*
> > ("VMWare virtual network
> > >control").
> >
> > I've never quite understood what vmnet was for, but
> > I guess we'll figure
> > out what the problems are with other approaches as
> > we try to implement them
> > ;)
> >
> > -- Ramon
>
> If I am interpreting their strategy correctly, they
> seem to have set up different nodes for each network
> port that the VMWare's client wishes to access. I am
> not sure if this is really necessary.... especially
> since we are not using the same internals as VMWare in
> our code. I would guess that since one node has been
> enough thus far, we may be able to continue the
> project using only one node. The only thing that
> comes to mind (to me, and I am no OS genuis) is that
> their may be a possible performance benefit to having
> multiple nodes to pass information to the host OS.
> For example:
>
> LINUX) the host OS passes things like video and
> keyboard/mouse through normal channels and the device
> node we currently use is explioted by the monitor;
> there may be a benefit to having other nodes through
> which the plugins may communicate with code _outside_
> of the monitor directly as a standard device--like a
> node to dump sound to a plex86 sound device client
> program the runs as a separate entity, and handles the
> inevitability that the guest and something else
> running on the host may want to grab the sound device
> at close enough intervals to cause a problem
>
> Windows) the host looks for specific types of data
> from specific places and thus the major/minor thing is
> not directly a question..... it is more a matter of
> how the plex86 VM emulated devices may benefit from
> having their own real "devices" to pass specific types
> of data through--especially important considering the
> device models & drivers needed by the
> dos/windows-shell environment.
>
> As I said, I am no OS genuis, but these are the issues
> at hand (that I know of to any, even small, extent)
> when one talks about how many device nodes we may
> really need in the end. Also, I don't know if it
> would be possible to use a different number of devices
> for different configurations (it would certainly make
> windows happier, when we get to that juncture).
>
> Drew Northup, N1XIM
>
> __________________________________________________
> Do You Yahoo!?
> Talk to your friends online with Yahoo! Messenger.
> http://im.yahoo.com
>
