Re: Looking for a boot disk for the Dectalk PC2

[Gary Lawrence Murphy <garym@xxxxxxxxxxx>]

>>>>> "J" == Jason White <jasonw@xxxxxxxxxxx> writes:

    J> On Tue, Jun 27, 2006 at 09:00:27PM -0700, T. V. Raman wrote:
    >>
    >> From memory, I believe the way that that card worked was to download
    >> some firware bits into the card from the PC, and I believe that
    >> was what the code Gary mentions emulated.

This is correct; the kit I received from Nicolas Pitre included a char
device driver for Linux-2.2, which appears to emulate the first stage
of the DOS boot, and a linux utility to load the data sections
extracted from these sys, exe and dic files.  This latter program
seems to behave, allocating and posting the data to reasonable
addresses without crashing, but that doesn't necessarily imply the
data being sent is correct or complete, or being dumped into empty
space.

the load script (not sure if I posted this before, but it's short)
seems to emulate the DOS autoexec pretty closely, outside of that
"dtpc_850.exe" file in Jason's set but missing in mine.

BASE=`dirname $0`
echo -n "Loading DECTALK ..."
sudo /sbin/insmod -f $BASE/dtpc.o 
$BASE/dtload -v -f /dev/dtpc0 \
    -K $BASE/kernel.sys -D $BASE/dtpc.dic \
    -xt $BASE/lts.exe $BASE/ph.exe $BASE/cmd.exe $BASE/usa.exe

dtload returns a trace that repeats load sequences which look like this:

dt_ioctl: DTIOSTARTEXE
dt_starttask: CS:IP = 33D2:0010
dt_ioctl: DTIOALLOCMEM
dt_allocmem: paras = 345 addr = 0x32780
dt_ioctl: DTIOLDMEM
dt_loadmem: addr = 00032780 size = 4956
dt_ioctl: DTIOSTARTEXE
dt_starttask: CS:IP = 3278:0010
dt_ioctl: DTIOTEST
dt_sendbuff: sent 34 bytes

ie, it allocates the memory, and transfers a seemingly reasonable amount,
and there's no real reason to think that a change in Linux kernel should
affect this particular part of the code (there were changes needed to
wait-queues and unloading the module)

turns out I don't have the DOS versions of these utilities, only the old
Linux versions; the manuals I have describe the DOS version as behaving
like a serial port /and/ parallel port you would basically just print
text to a virtual printer; the linux version creates a new device at
dev/dtpcN for each card discovered.

/proc/devices shows the card at 0x250, but I'm wondering if that might
be conflicting with some other device onboard -- it's a Dell, and an old
Dell, so just about anything could be happening in the port assignments;
I'll try moving the card up to 350 to see if maybe that might change
its tune.

Jason: would you be able to send me that dtpc_80.exe by email or
rapidshare.de or something?  It may be completely unimportant, but it
may also be a vital missing link, and it can't hurt to try it! :)

-- 
Gary Lawrence Murphy <garym at linux.ca>
"what I need is a job that doesn't interfere with my work"

-----------------------------------------------------------------------------
To unsubscribe from the emacspeak list or change your address on the
emacspeak list send mail to "emacspeak-request@xxxxxxxxxxx" with a
subject of "unsubscribe" or "help"