Re: Timer issues?

["Michael L. Hitch" <mhitch%lightning.msu.montana.edu@localhost>]

On Sun, 16 Aug 2009, Frank Wille wrote:

> To verify your assumption I did the following hack to amiga/dev/clock.c:
>
> #endif
>        {
> +               static int old_hardclock_ticks = 0;
> +               static u_int old_interval = 0;
> +               int current_hardclock_ticks;
> +
>                hi  = clockcia->tahi;
>                lo  = clockcia->talo;
>                hi2 = clockcia->tahi;
> @@ -311,8 +315,14 @@
>                        hi = hi2;
>                }
>
> +               current_hardclock_ticks = hardclock_ticks;
>                interval = (amiga_clk_interval - 1) - ((hi<<8) | lo);
> +               if (interval < old_interval &&
> +                   old_hardclock_ticks == current_hardclock_ticks)
> +                       interval += amiga_clk_interval;
>
> +               old_hardclock_ticks = current_hardclock_ticks;
> +               old_interval = interval;
>                /*
>                 * should read ICR and if there's an int pending, adjust
>                 * interval. However, since reading ICR clears the interrupt,

  I did a similar thing on the vax, and stopped the running backwards 
warning.

> I would have hoped there is a nicer solution, but when nobody objects I
> could commit the change.

  I couldn't see any better solution either.  I did find out that the 
interval counter interrupt on the vax seems to use either IPL 22 or IPL 
24.  Some models use 24, but my architecture book mentioned that 22 was 
preferred.  The vax defined IPL_VM as 23 (which is where I saw the problem 
most of the time) and IPL_SCHED as 24 (which I saw the other times).

  I don't think your change is in the right place.  You put it in 
clk_gettick(), which I think is supposed to return the current interval 
from the CIA counter with no change.  It's also used in delay(), which 
specifically handles the case of the timer wrap.  I think the proper place 
would be in clk_getcounter() where the return value depends upon 
hardclock_ticks incrementing on the hardclock interrupt.

--
Michael L. Hitch                        mhitch%montana.edu@localhost
Computer Consultant
Information Technology Center
Montana State University        Bozeman, MT     USA