From: bsa@kf8nh.wariat.org (Brandon S. Allbery) Subject: Re: DPMI under dosemu (was Re: DOSEMU Success list!!) Date: Sun, 28 Mar 1993 09:27:36 GMT
In article <90006@hydra.gatech.EDU> gt8134b@prism.gatech.EDU (Howlin' Bob) writes:
>Windows needs a lot more power than we can safely give it. I believe
>it may run under VCPI, which *from what I have heard* is not stingy
>enough in its granting of power. This means that it could seriously
>interfere with the inner workings of Linux.
win/s will use VCPI if available.
Keep in mind that VCPI's purpose is the opposite of "stingy enough in its
granting of power". Its intent is to allow multiple programs to share ring 0.
Which is why there's no consistent paging, etc. under VCPI: all it really
does is permit VCPI clients to *voluntarily* (ring 0, remember) not step on
each other. Linux with VCPI support would have to specifically *avoid*
touching any pages "owned" by other VCPI processes, or the system would crash
hard. DPMI, for all its difficulty of implementation, puts page allocation
under central control of a single ring-0 process (the DPMI server) and all
clients run in ring 3, so Linux would not have to give up control of the
processor to make dosemu programs happy.
What I've never heard is whether win/s will run under DPMI.... :-~ (win/3
probably will NOT, since it is a ring 0 system and DPMI server.)
Discussion of running Windows under dosemu is probably pointless anyway, since
making it understand graphics modes other than 640x480x16 would require having
it use the X drivers or something like that. Or the trick DESQview/X uses to
run Windows in an X window, which is miserably slow (Windows drivers that map
Windows I/O operations to X operations).
++Brandon
-- Brandon S. Allbery bsa@kf8nh.wariat.orgIt's not too late to turn back from the "Gates" of Hell... Linux. The FREE 32-bit operating system, available NOW.