> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf
> Of Jeroen Janssen
> Sent: Friday, August 04, 2000 10:32 AM
> To: [EMAIL PROTECTED]
> Subject: Re: video driver protocol
>
>
> > If you put all devices on the same port, that means that all
> device plugins
> > need to be invoked on every port access in order to check if
> the message was
> > meant for them. Slow!
>
> ah... put I meant something different:
>
> have a generic host <->guest plugin/plex86 part register a specific
> plex86 port.
>
> next plugins can request a host <-> guest callback by means of a
> specific ID.
>
> the packet which address was written to the plex86 port contains
> information about a specific ID & other data. now the plugin/plex86 part
> looks in a ID table for a callback and calls the plugins function.
>
> this way we have a one port entry point for the plex86 host <-> guest
> packet communication instead of having to register specific ports for
> every plugin.
> (but if you'd rather have the last option, I don't mind....)
> --
> Best regards,
>
> Jeroen Janssen
>
Might it not be a bad idea to create a "Plex86 Master device" to handle the
mass passing of data between the guest OS and the host code? I am just
throwing this idea out here because it [this thread] sounds a great deal
like what happened when manufacturers started to make multi-function
devices. I just think that it would seem logical to do this because then we
can have one sort of "master device handler driver" and then _inside_ of the
guest OS redirect the necessary data (accelerated video, sound, etc.)
through its port(s) instead of having huge chunks of space taken up inside
of something where "real" memory is a hard to come by commodity. Put
another way: create one "device" to ferry data around because then we need
to lock less memory away from the paging algorithm. Just another
hair-brained idea.........
Drew Northup, N1XIM
