Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rutgers!sri-spam!mordor!styx!ptsfa!hoptoad!academ!killer!tsp
From: tsp@killer.UUCP (Tom Poindexter)
Newsgroups: comp.os.minix
Subject: Re: Clock setting program for AST Six Pack
Message-ID: <805@killer.UUCP>
Date: Mon, 27-Apr-87 14:32:49 EDT
Article-I.D.: killer.805
Posted: Mon Apr 27 14:32:49 1987
Date-Received: Wed, 6-May-87 05:25:27 EDT
References: <771@killer.UUCP> <9490005@hpfclp.HP.COM>
Organization: The Unix(tm) Connection, Dallas, Texas
Lines: 54
Summary: astclock.c rationale




[ In <9490005@hpfclp.HP.COM>, diamant@hpfclp.HP.COM (John Diamant) writes: ]

>I have also written an astclock replacement program (for a MegaPlus, but it
>probably works the same for the other AST cards).  I also took the port_in
>routine the same as Tom.  The compilation command is identical. Tom, mine
>is considerably simpler than yours.  There are two main reasons I see for this:
>
>1) You copied the code for date into astclock.  I just called date.
>2) You seem to have bullet proofed your code against conditions I don't
>   understand (such as why do you need to call the clock twice -- do you
>   get garbage back sometimes -- I never have).
>
>Tom, maybe you'd care to comment on why you do as much as you do in your
>version.

John, here are the reasons for writing my version of astclock.c as I did:

1.  The code from `date.c' to compute the number of seconds since
epoch and calling stime() was used because of a comment to an earlier posting
of an AT clock reading & setting program.  The commentator cautioned against
using pipes during init becase of possible (or prior) damage to the file 
system.  I don't have the posting any longer to pinpoint the possible hazards,
but I took the advice anyway.  I will also have to hack both date.c and my
astclock.c in Dec. 2099, assuming I still have my PC & AST board and I am 
not {a higher life form/a frog/dust}, since date.c doesn't figure leap
years quite correctly. :-)

2.  I read the clock chip twice based on the programming suggestions in my
AST Six Pack manual.  I never experienced a situation where two readings were
different in my testing, but just in case.....

3.  I didn't experiment with the clock chip much, so I didn't know one 
could read directly from the clock registers.  The valid clock i/o addressing
is listed as 0x02c0 through 0x02c7 in the manual; your program addresses
ports beyond 0x02c7 (CLOCK_PORT + Y3 to address year, where CLOCK_PORT is
0x02c0 and Y3 is 10).  Could be that my documenation doesn't tell all or 
your MegaPlus is wired up differently, or uses a different clock chip.
My Six Pack clock chip always returned 'F' as the high nibble, so I and'ed
the value returned with LOW_4 (0x0f) after reading it.

4.  As a matter of my programming style, I tend to bullet-proof, write 
verbosely, add comments, etc., especially on code that I release 
publicly, so that others can make changes, find bugs, etc.  My previous 
iterations of the program were rather terse and clunky.


In any case, thanks for your comments and code; I always look for easier
ways of doing things. Perhaps someone can verify if the Six Pack Plus
clock chip & i/o addressing are the same for the MegaPlus.  -Tom
  
  Tom Poindexter     ..!ihnp4!killer!tsp