Polling and Sleep in a Driver

[Alfred Perlstein]

* Len Gross <sandiegobiker at gmail.com> [081019 12:28] wrote:
> Warner,
> 
> Thanks so much for your response.  This is really a case where I'm a
> bit over my head!
> 
> I thought on a a "busy" machine, there was a fundamental "no uppper
> bound" on the wait on a sleep; independent of the Hz?  Not real-time
> blah, blah, ......
> 
> I am trying to understand if a driver "sleep" gets special attention
> (i.e higher "priority") so that I see less slop in the timing.
> 
> Of course, I could hack up a driver and try some tests, but wanted
> some advice before diving into that pool.
> 
> Thanks again.

Yes, you should see less slop in a driver because the ithread
priority will be much higher than userland.  The reason your
userland is getting jitter is because other things may be
running before it is based on priority.

-Alfred