- From: <rhk_at_newimage.com>
- Date: Fri, 11 Oct 2002 06:19:25 -0700
Message-Id: <20021011131924.01286102A1_at_newimage.com> Date: Fri, 11 Oct 2002 09:19:24 -0400 (EDT) From: rhk_at_newimage.com (Apache User) > The main class of races you have to worry about in drivers are between > the interrupt and the ioctls. For example, an analog input interrupt may > occur while the analog output command is being set up. And that's what is happening. I increased the rate that I was triggering the analog output and it was stopping every 10 seconds. I tried the same program with one of the CPUs disabled, and it ran fine overnight. I think the windowed register access needs a spinlock. I'm going to try that this morning to see if it fixes the problem, and if the performance is still acceptable. -Ray
Received on 2002-10-11Z12:19:25