Re: [Freedos-32-dev] Removing the UseLfn parameter in the FAT driver
Status: Pre-Alpha
Brought to you by:
salvois
From: luca a. <l_...@ho...> - 2002-04-06 19:31:43
|
Hi all, today is a rainy saturday in oregon, hence I have a lot of time for sending mails :) [...] >What you have as the "FD32 device list" looks much like the int20h of >DOS386.EXE or KRNL386.EXE or whatever you call it, in Microsoft >terminology. The action of "registering a device" would be analogous to >"hook the appropriate subf. of int20h". Ok, I am beginning to get it... I (once again :) proposed something similar to this ``interrupt hooking mechanism'' at the beginning of the project, but it was refused because it was not efficient enough (ehhhh... At that time, in this mailing list there were a lot of very skilled programmers...). >You say that the DPMI redirects >the int 21h call of DOS calls to the appropriate fd-32 driver. >MS-DOSMGR >redirects (iirc) these calls to the appropriate VxD driver. Yes, I agree. Same concepts, different mechanisms. >Are these functions offered to applications as an API? If so, how? For the future, we plan to have a ``FD/32 native API'', but it is still not clear how it will look like... An FD/32 native program will be dynamically linked into the kernel (exaclty as a driver), and will be be able to invoke kernel functionalities just using simple function calls (as the DPMI driver does, for example). We still have to define the native API, and I am inclined to think that we could do it after that the DPMI support will be fully working: we could look at the functions used by the DPMI driver (and by other drivers, of course), and use them as a base for defining the FD/32 API. (for having a look at the functions that are currently esported by the kernel, see fd32/kernel/syscalls.c http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/freedos-32/fd32/kernel/syscalls.c?rev=1.5&content-type=text/vnd.viewcvs-markup >If it's via some interrupt, why not use int 20h? Currently, the INT-based ABI/API that we are supporting is DPMI. As already said, nothing prevent us from exporting another INT20 based ABI (through a different driver) in the future. >Just as an example, consider the VxD called VMM, the kernel of the >DOS/Windows of 32-bits, and see the API: >http://www.ctyme.com/intr/rb-2473.htm I am studying it in my spare time... Our problem, right now, is that we have a lot of things to do, but only few (2?) coders willing to work on them... Hence, I don't know when I'll have the time to start working more seriously on this VxD issue. >THe project in itself looks interesting. However, as I said in other >messages, I still need a lot to learn, specially about EMM386.EXE. As >far as I know, this driver, in the Microsoft version, offers EMS via >processor paging, so it's a potential focus of information about VMs, >and a potential focus of problems about installing the fd-32 once DOS is >booted. Yes, EMM386 has fairly different goals respect to FD/32 (for example, FD/32 is not interested in EMM & similar things)... Probably it could be a good source of information, but I never had the time to study it. >In DR-DOS, this seems to be an interesting and fundamental piece for >this all, please read this article that you can find easily searching >with Google: >http://www.drdos.net/quicknote/32mb_ems.htm I'll have a look at it, thanks. Thanks, Luca _________________________________________________________________ Chat with friends online, try MSN Messenger: http://messenger.msn.com |