🚀 go-pugleaf

> I wrote a small program to read the BIOS clock ($0040:$006C) but this
clock
> is the wrong one. Can anybody help me to get the right CMOS-time?! Is
there
> a way to find it??

What you're getting is the BIOS tick count.  If this count is correct, it
can be used to reset the DOS clock.  It's fairly easy to figure the time
from the tick count, there are 65536 (2^16) ticks per hour.

Int 1A, Function 02 reads the time directly from the real-time hardware
clock.  The returned data is in BCD format which you will have to decode.
This should be the date/time that is displayed in your CMOS setup. It is
possible to go directly to the hardware itself but it's rather tricky
programming and there's not much point in it if the BIOS will do the job
for you.

The DOS clock is set on startup and maintained thereafter by the OS.  This
DOS clock is Y2K compliant but the timer hardware in many (if not most) PCs
has some problems.  A cold shutdown and re-start is often required to
re-set the hardware timer after the turn of the century.  Until that
happens, your software will likely be showing something like Jan. 1, 1900
if you get date/time from hardware.  All part of the Y2K mania.

-- 
Ernie Deel, EFD Systems
------------------------------------------

On 9 Dec 1997 10:07:16 GMT, "PB-Systems" <pb-systems@wxs.nl> wrote:

>Hi All,
>
>I have a problem with my BIOS-clock. The clock runs too fast. The clock of
>the Operating System runs too fast. If I reset the computer the time is Ok,
>but if the PC keeps runnings the clock will move 10 seconds (a day) into
>the future. A strange but irritation problem. 
>I wrote a small program to read the BIOS clock ($0040:$006C) but this clock
>is the wrong one. Can anybody help me to get the right CMOS-time?! Is there
>a way to find it??

The CMOS clock runs independently of the BIOS clock.  It is actually
located on the CMOS chip and is read by the BIOS at bootup.  To access
the CMOS memory locations (some of which are updated continuously by
the CMOS clock) you need to do it through the CPU I/O lines 0x70 and
0x71 using the in/out instructions.  The desired address (one byte) of
the CMOS memory is accessed by writing the address to port 0x70 then
either reading from or writing to port 0x71.

The following table maps the CMOS addresses asscociated with the
realtime clock:

CMOS address    Contents
~~~~~~~~~~~~    ~~~~~~~~
00H             Current second
01H             Second alarm
02H             Current minute
03H             Minute alarm
04H             Current hour
05H             Hour alarm
06H             Day of the week
07H             Day of the month
08H             Month
09H             Year (00-99)
0AH             Clock status register A
0BH             Clock status register B
0CH             Clock status register C
0DH             Clock status register D
32H             For NON-PS/2's: Century in BCD (19 or 20)
                for PS/2's, the Century byte is at 37H, and 
                32H and 33H are used for a CRC value for bytes
                10H through 31H.




-- 
Wayne D. Hoxsie Jr.  KB9NWK
wayne@hoxnet.com
http://www.hoxnet.com

Ernie Deel wrote:
<snip>
> The DOS clock is set on startup and maintained thereafter by the OS.  This
> DOS clock is Y2K compliant but the timer hardware in many (if not most) PCs
> has some problems.  A cold shutdown and re-start is often required to
> re-set the hardware timer after the turn of the century.  Until that
> happens, your software will likely be showing something like Jan. 1, 1900
> if you get date/time from hardware.  All part of the Y2K mania.
 
Actually, I think it is more like Jan. 1, 1980.  At least with the
early PCs, the 100 year calendar started with 1980, and not 1900.
Therefore, PCs should not have a problem until 2079.   :)

Surely the long int problem is encountered before this?

2037 and a few months..

:-)

In article <348E9CF9.5DF0@Early.com>, Walt <Walt@Early.com> writes
>Ernie Deel wrote:
><snip>
>> The DOS clock is set on startup and maintained thereafter by the OS.  This
>> DOS clock is Y2K compliant but the timer hardware in many (if not most) PCs
>> has some problems.  A cold shutdown and re-start is often required to
>> re-set the hardware timer after the turn of the century.  Until that
>> happens, your software will likely be showing something like Jan. 1, 1900
>> if you get date/time from hardware.  All part of the Y2K mania.
> 
>Actually, I think it is more like Jan. 1, 1980.  At least with the
>early PCs, the 100 year calendar started with 1980, and not 1900.
>Therefore, PCs should not have a problem until 2079.   :)

-- 
James Usmar

James Usmar wrote:
> 
> Surely the long int problem is encountered before this?
> 
> 2037 and a few months..
> 
> :-)
 
Damn it!  I might still be alive.   I thought with 2079, I
would not have to ever worry about it.

:-)