Re: signed int for interrupts??

Marty Leisner (leisner@sdsp.mc.xerox.com)
Mon, 9 Sep 1996 09:29:05 PDT

> This was a disappointment. I was looking forward to seeing my Cyclades
> interrupt count wrap at 2^32 and start counting up again...but just
> noticed that procinfo was reporting it to be stuck at (2^31)-1. Looking
> into /proc revealed this:
>
> endor:~# cat /proc/interrupts
> 0

Now I'm confused...2**31 is a huge number.

If you have interrupts at 1 Megahertz, (10**6), you still have a big
enough number for 10*25 seconds of interrupts.

There are about 10*5 seconds in a day...so we can encode years and years
and years (how long has your linux system been running).

Methinks the problem ws the width of the field, not rollover...

Would it be useful to print the interrupt cound as a floating pointing format?
(do we he any floating point support in the kernel?)

Or probably printer it the other way

(level name count)

with infinite precision.

-- 
marty
leisner@sdsp.mc.xerox.com  
Member of the League for Programming Freedom