I'm experimenting with Kernel Preemption

[Dave Jones]

On Wed, May 11, 2005 at 06:48:54PM -0500, john bray wrote:

 > dave, can you tell us a little more about the "...opportunity for
 > drivers to break...." comment?  i'd just like to understand a bit more
 > about the implications of it.

We have a lot of drivers in the tree that were written before
preempt was merged. Whilst some of them got fixed up, theres no
telling how many didn't because no-one tested/reported problems.

The number of times I've heard "We've auditted the drivers, we
don't think there are any more preempt related bugs" isn't funny.

Given what it brings to the table, its not worth the potential pain.
A better use of time is to investigate where latencies are occuring
and fixing the problems, instead of pretending to fix them with
artificial means (remember, all preempt can do is lower the _average_
latency). This work has been going on upstream for quite a
while (see http://people.redhat.com/mingo/realtime-preempt/), and
bits of this patchset have been slowly making their way into
-mm and Linus' trees.

		Dave