On 5 Aug 2003, Benjamin Herrenschmidt wrote:
> On Tue, 2003-08-05 at 10:53, Javier Achirica wrote:
> > I've integrated this patch in my code. I've done a major change: Instead
> > of using schedule_delayed_work(), I create a new workqueue and use
> > queue_work() on that queue. As all tasks sleep in the same lock, I can
> > queue them there and make them sleep instead of requeueing them.
> >
> > I haven't sent them to Jeff yet, as I want to do more testing. If you want
> > to help testing them, please tell me.
>
> Well... creating a work queue means you create one thread per CPU, that
> sucks a bit don't think ? Maybe using a single thread for the driver
> with your own queuing primitives...
I've been studying the problem for a while and I've implemented a solution
using a single kernel thread and a wait queue for synchronization. I've
tested it and looks like it works fine. It can be used both in 2.4
and 2.6 kernels. Before submitting a patch with it I'd like someone with
experience in this kind of code to take a look at it just in case I'm
doing something dumb. Jeff? :-)
Javier Achirica
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Aug 07 2003 - 22:00:36 EST