[Coco] Bad Driver Combination?

[Robert Gault]

K. Pruitt wrote:
> When I make a NitrOS9 boot disk using the no-halt driver for the Disto Super
> Controller II (rb1773_scii_ff74.dr) and the smartwatch driver (clock2_smart)
> together, the Coco 3 will crash instantly when the system tries to access the
> clock chip.
>
> Anyone else had this experience and is there a solution (other than not using
> the smart watch driver, which is the solution I am currently using)?
>
> Also, when I use the getclk utility to get the date/time from the chip, date -t
> reports the date as 1912. This is confusing to me because as far as I know the
> watch chip doesn't hold the century but rather only holds the year, and getclk
> doesn't seem to be passing a "19" to F$Time - at least not as far as I can tell
> from a disassembly of getclk. Where is this 19 coming from? Does NitrOS9 default
> to 19 in the absence of a century digit? Does the NitrOS9 F$Time system call use
> the same 6 digit scheme as the OS-9 version or does it actually have space for a
> century digit? I'm using Y2K updated modules in OS-9 so there it is not an
> issue, but under NitrOS9 that 19 always shows up.
>
> Thanks for any info you might pass on.
>
>

I'll start with the date issue as that ought to work correctly with the current 
NitrOS-9 version.

Clock2_smart has some code that makes an assumption about the century. The year 
is read from the clock chip which comes out as 0-99. The clock module assumes 
that if the year is equal or greater than 50, then the actual year is 1950-1999. 
If the year is less than 50, then the actual date is 2000-2049. As Gene 
mentioned in another message, I don't expect any of us need to worry about dates 
beyond 2049. :)

L006F    dec   ,s
          bne   L0050       get the next byte from clock
          lda   1,x         get year
          cmpa  #50         half-assed test for century
          bhs   c19
          adda  #100        make it 20th
c19      sta   1,x

Now the date command looks at the year value and tries to make sense of the 
century. It checks to see if the year is 100 or greater and if so, sets the 
century to 20xx else 19xx.

          lda   <sysyear		get year
          ldb   #19-1		century in B
CntyLp   incb			add a century
          suba  #100		subtract 100 yrs
          bhs   CntyLp		until yr<0
          adda  #100		restore year to 00-99 range

Remember that sysyear will be >100 if clock2_smart thinks the century is 20xx. 
So if the year is 2012, then sysyear should be 112 and you will get:
regA=112
regB=18
incb   regB=19
suba #100  regA=12
branch as regA>=0
incb   regB=20
suba #100  regA=-88
regA<0
adda #100  regA=12

Now date prints out regB and then regA so you should see 2012.

================
I can work with you by private email to solve the SC-II + Smartwatch problem. I 
believe the current code should work but I can easily generate custom modules 
for you.
Is the Smartwatch installed in the Disto SC-II controller or another pak in your 
MPI? Are you using any other modules that might reset the active MPI slot?

If we can't get the SC-II driver and clock2_smart modules to work, I can supply 
the program swread which can be set to read the Smartwatch on call or say every 
5 minutes. Then you just use clock2_soft.

Robert