Messages in this thread | | | Date | Wed, 15 Apr 1998 11:50:41 +0000 | From | Richard Jones <> | Subject | Re: Wlinux vs. LWin95, looking at the alternative |
| |
nuke@bayside.net wrote: > > On Tue, 14 Apr 1998, Rik van Riel wrote: > > > On Thu, 9 Apr 1998, Richard Jones wrote: > > > As I understand it, there are only a few commands that > > > can't be properly virtualized on the i386 processors. Is > > > > They will trap with 'invalid instruction', and can (in > > theory) be deassembled and emulated JIT... > > [well, _can_ they??? I don't know enough about x86 to really know] > > anything can be emulated. don't ever think otherwise. there is currently a > working nintendo 64 emulator for win95. *anything* can be emulated on > *any* hardware (provided the display is suitable). > > > > it possible that there are only a few tiny patches to the > > > Win95 `binary' that need to be made to make it behave itself > > > in a virtualized 386 environment? Perhaps someone who knows > > > more about this than me can comment ...
Sure, anything can be *emulated*, but the real problem is running the emulated binary at near to full speed as possible. This is why virtualizing the 386 is a better approach (given that writing a really good JIT compiler for 386 opcodes is difficult).
I don't know the full intricacies of this, so this should be taken with a pinch of salt, but as I understood it, various instructions which should cause traps don't. An example is popping a stack word into the flags register which could alter certain system-specific flags in kernel mode. The hypothesis (not proven) is that Win95 probably doesn't do such things in many places, and so it might be feasible to patch the binary where it does this to make it behave in a virtualized 386 environment.
Of course, you are then open to MS changing the way future releases of Windows work to make these patches more and more difficult to achieve, particularly if this approach were to be successful and everyone started running their Win programs under Linux :-)
Rich.
[Paul K. - I'm cc-ing this discussion on Linux-kernel to you, since it might be in your field]
-- Richard Jones rjones@orchestream.com Tel: +44 171 460 6141 Fax: .. 4461 Orchestream Ltd. 262a Fulham Rd. London SW10 9EL. "you'll write in PGP: www.four11.com telegraphic, or you won't write at all" [Céline] Copyright © 1998 Richard W.M. Jones
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |