cwr@pnet01.cts.com (Will Rose) (04/28/91)
With the improved responsiveness of KuB's scheduler it seemed worth trying
to edit/compile simultaneously under Minix. Running on an XT, I've never
bothered before; even if I could get the processes to fit in memory, long
response times did not make for happy editing. Rather than put the root
file system on the hard disk, I added a BocaRam EMS card, and moved the
root file system to expanded memory.
This message therefore is just to say that the recent reposting of the 1.5.10
ems driver works with no problems with the Boca Research board (at least, the
version currently on sale in the US). I put 256K of memory on the board, and
mapped it in at 0xE0000 (various device drivers took up 0xC0000-CFFFF and I
wasn't sure if they extended into the D bank). The initialisation bytes
(MAP1-4) for this were all 0, and I used the BocaRam's default port of 0x268.
The patches themselves (for a Zuckerboard) look as if they were made against
a slightly modified 1.5.10. I had to add the patch to config.h by hand, and
klib.x had a 19-line shift. The BocaRam doesn't need any of the special
code for the Zuckerboard; according to Boca Research documents, the LIM 3.2
standard defined actual hardware, but LIM 4.0 is restricted to software,
so different manufacturer's 4.0 boards may differ in their setup routines.
One odd thing is that I have several times seen it written that a ramdisk
larger than 256KB is automatically loaded into expanded/extended memory on
boot-up. In fact, I can't find the code for this anywhere; it looks as
if Minix loads the ramdisk into extended/expanded memory if this is available,
and otherwise into real memory. Does anyone know different? I've tried
ramdisks of 130KB and 256KB and both were loaded above 1MB without question.
I don't know how you initialise the BocaRam to appear at 0xD0000; someday
I'll try it, but meanwhile it looks as if the expanded memory driver is
pretty solid. Anyway, I now have space to run update, getty on tty2, a
console shell, a background compile and vi - ehh, when I were young, we used
to dream of running a compile in the background. But when you tell that to
the young folks of today, they just won't believe you.
-----------------------------------------------------------------------
"It is difficult to be a | Will Rose
ruler, and it is not easy | UUCP:{nosc ucsd hplabs!hp-sdd}!crash!pnet01!cwr
to be a subject, either." | ARPA:crash!pnet01!cwr@nosc.mil
Confucius. | INET:cwr@pnet01.cts.com
UUCP: {nosc ucsd hplabs!hp-sdd}!crash!pnet01!cwr
ARPA: crash!pnet01!cwr@nosc.mil
INET: cwr@pnet01.cts.com