Re: UMB and 386/486 --> status of FreeDOS emm386 etc.
Eric Auer <[email protected]>
| Newsgroups | gmane.os.freedos.devel |
|---|---|
| Message-ID | <[email protected]> |
Hi, you (Imre) over-criticize Tom. Here some facts: - no EMM386 can be loaded from the command line. Would be cool, but is not needed. When invoked from the command line, normal emm386 only allow to change settings of the copy loaded via config.sys ... - VCPI -DOES- run with EMS 3.2, but you do need 4k chunksize for VCPI, so implementing 4k chunksize for "alloc" is needed anyway and this in turn eases implementation of EMS 4.0 ... - No idea whether the UMB routines need optimization. Feel free to optimize. And finally: The code is hard to read because it is based on a monolithic ASM version without UMB that was published by a German PC magazine. Only very few people understand the workings of a v86 host and how to set up paging, multitasking, ... - basically if you have all that knowledge, you can write your own 32bit multitasking operating system -including- a DOS box for it. I think I have a rough understanding of most things, but as you can see, my emmsys.asm does not fully work. So I do not get it completely. By the way, emmsys.asm does not return to DOS yet. Making it work will be a bigger task! Tom has a rough understanding, too, but he cannot do much more than add stuff to the protected mode part -or- the v86 part. He is unable to do bigger changes to the interface between both, and just using the one used in the magazine version. Excuse me if I do false claims here. However, you can be sure that initializing protected multitasking paged mode and a v86 task inside it, plus managing its needs from protected mode is quite tricky. However, I agree that it should be possible to split the source as you describe, as long as we do not touch a thing about the order of code in memory (which would require us to understand it better). Eric