Date: Mon, 1 Feb 93 04:30:09 PST
From: Advanced Amateur Radio Networking Group <tcp-group@ucsd.edu>
Errors-To: TCP-Group-Errors@UCSD.Edu
Reply-To: TCP-Group@UCSD.Edu
Precedence: Bulk
Subject: TCP-Group Digest V93 #32
To: tcp-group-digest
TCP-Group Digest Mon, 1 Feb 93 Volume 93 : Issue 32
Today's Topics:
386 compile
rcs files updated on ucsd.edu
RDATE questions and comments (2 msgs)
Send Replies or notes for publication to: <TCP-Group@UCSD.Edu>.
Subscription requests to <TCP-Group-REQUEST@UCSD.Edu>.
Problems you can't solve otherwise to brian@ucsd.edu.
Archives of past issues of the TCP-Group Digest are available
(by FTP only) from UCSD.Edu in directory "mailarchives".
We trust that readers are intelligent enough to realize that all text
herein consists of personal comments and does not represent the official
policies or positions of any party. Your mileage may vary. So there.
----------------------------------------------------------------------
Date: Sun, 31 Jan 93 23:51:19 EST
From: crompton@NADC.NADC.NAVY.MIL (D. Crompton)
Subject: 386 compile
To: tcp-group@ucsd.edu
I made an attempt to compile the WG7J code using the 3.1 -3 compile
this weekend put I ran into all kinds of problems. I am sure it can
be done and I was almost there but I ran out of patience and time!!
Anyhow it seems the problem I was having was with the device drivers.
Phil's code is now SO different from the old base that it is difficult
to draw a line of where to stop making mods!! For instance the I8250.C
file has been completely overhauled. Of course using it means changing
something else and something else.... etc.
It seems to lower the compile size on a 217K compressed -1 compile to
about 201K -3. This is the compressed size and it looks like the running
size difference may approach 50K!! So it is definitly worthwhile. My
time is precious at the moment or I would have continued on, but for now
it will have to go on the back burner for abit. Hopefully someone else
will get to it.
Some notes -
pcgen.asm and command.asm need mods to pushall 32 bit registers (PUSH/POPAD)
ALL device drivers.asm need a similiar mod and the makefile needs to have
a 386 compile option added.
I basically took ideas from Phil's code. For instance he now uses asmglobal.h
in all .asm files to define global stuff. His code has no command.asm
though and pcgen.asm is very different in many parts. Also the device
drivers are changed slighly in both .c and .asm - the calling and return
parms are different. The control signal stuff in I8250 has been changed -
probably for the better as it seems simplified.
I did get it running but I could not get any ethernet (packet driver)
data to flow and it seemed that the AX25 (asy) driver would crash.
Everything else worked OK though. It seemed faster and as I said took
LOTS less memory.
Note for Johan -
I think the stopwatch (watch) code should be a config.h define option.
It is ridiculous to compile this in as the only way you can use it is if
you put breakpoints into the source. Users would never have a need for it.
I defined it here in config.h. It saves about 600bytes.
Doug
------------------------------
Date: Sun, 31 Jan 93 02:50:18 -0800
From: Phil Karn <karn@unix.ka9q.ampr.org>
Subject: rcs files updated on ucsd.edu
To: tcp-group@ucsd.edu
I've updated the RCS source files on ucsd.edu. Mostly minor changes
this time, a few minor bug fixes, etc. The files were uploaded in
ascii mode this time.
Phil
------------------------------
Date: Sun, 31 Jan 93 23:29:00 EST
From: crompton@NADC.NADC.NAVY.MIL (D. Crompton)
Subject: RDATE questions and comments
To: miltonm@inetnode.austin.ibm.com
Milton,
Thanks for the info. I did in fact implement the disable interrupts
around the stime function call and it cleared up all of my problems.
I have not observed any unusual timer behavior since the change.
Johan (and others)
In RDATE.C do the following around the stime function call -
char i_state;
.....
.....
i_state=dirps();
stime(&rdate);
restore(i_state);
I use the at command to update the clock a few times daily from a
known good source.
By the way the problem with not parsing the text -
at 0800 "writeall shutting down now"
Maybe the at parser should be able to detect say a single quote -
at 0800 "writeall 'shutting down now'"
or something like that and replace it with a double quote when passing it
the the command parser.
Doug
------------------------------
Date: Mon, 1 Feb 93 09:41:06 GMT
From: agodwin@acorn.co.uk (Adrian Godwin)
Subject: RDATE questions and comments
To: tcp-group@ucsd.edu
crompton@NADC.NAVY.MIL (D. Crompton) writes :
> For some reason PC clocks are lousey! My inexpensive wristwatch is
> far better. It is actually amazing! It stays within a few seconds over
> many months. Most PC's seem to drift in minutes/week or even minutes/day.
The timekeeping you see on a PC that's left on permanently is governed
by the crystal that feeds the programmable timer. This is usually a
standard (50ppm) microprocessor crstal with no special care taken to
get it to run at speed.
The calendar/clock chip uses a watch crystal - probably 25ppm and often
calibrated to br on-frequency. In some cases it's a Dallas clock device.
This is much more accurate (though perhaps only as good as a cheap watch),
but the time from it is only loaded into the DOS clocks at startup - due
to early PCs having no calendar, there's no tradition of reading that
clock for the real time.
Ideally, a background process should read the calendar chip regularly and
carefully modify the DOS timer, either by injecting or removing interrupts
from the timer chain, or by tweaking the time constant register in the
counter/timer chip.
-adrian
------------------------------
End of TCP-Group Digest V93 #32
******************************
******************************