Bart Oldeman wrote:

> You mean the current CVS sources (which were indeed last updated three
> days ago)? Just so I know what you mean?

Exactly. Initially I downloaded them from the Web interface but then I
realised that I would have to do it many times from now on and did it
the *right* way (using CVS ;-)

> I'll have another look; for me a high-loaded tdsk.exe (2.30) works so you
> might want to keep it simple first by trying an empty config.sys which just
> has something like
>
> dos=high,umb
> device=fdxxms.sys
> device=umbpci.sys
> devicehigh=tdsk.exe
>
> and then try some permutations (himem instead of fdxxms, emm386 instead of
> umbpci) and so on.

You're right, but I usually play the Yes/No game with F8 instead. The
permutations are too many, but the essential result is that BITDISK
(TDSK Jr.) works for me too, when loaded low *and* high, unlike the
other drivers which appear to work when loaded low but not high. I
suspect this is because of a problem I found now, similar to the
previous one I found, about which in a while...

> Which compiler do you use to compile the kernel by the way?

Initially I played with Borland C++ 4/5 but when I figured out that its
code was buggy (or at least the kernel is not optimised for it) I moved
to Watcom. So I use Open Watcom 1.0 now. But I compile for 386.

And now to let you know what I've found. I've got a copy of a little but
excellent DOS utility called DUMPDOS (the author is unknown to me but he
had done a great job as you will see) which you can download from my
website (http://linux.tu-varna.acad.bg/~lig/dumpdos.exe) that dumps all
the DOS internal data structures including the LoL, MCBs, DPBs, buffers,
and many more. Using DUMPDOS with the original build 2030, I found that
all the installable drivers had dpb_unit of zero, but dpb_subunit was
correct. But now (with your patches applied) I've found exactly the
opposite - the dpb_unit is correct, but dpb_subunit is zero! And this is
persistent, no matter if the drivers are loaded high or low. I can't see
any updates of the global number of block devices blk_dev.hs_name[0]
after the initial call of dsk_init() in main.c (which is done before
installing the drivers from CONFIG.SYS). Maybe this is done indirectly
when the passed "dhp" to init_device is &blk_dev?! But then I don't see
such a call. I think that the problem may result from blk_dev.dh_name[0]
not being updated to reflect the real number of drives after config.sys
is processed.

Lucho

==^================================================================
This email was sent to: [EMAIL PROTECTED]

EASY UNSUBSCRIBE click here: http://topica.com/u/?b1ddyi.b3hwCs.YXJjaGl2
Or send an email to: [EMAIL PROTECTED]

TOPICA - Start your own email discussion group. FREE!
http://www.topica.com/partner/tag02/create/index2.html
==^================================================================


Reply via email to