[coyotos-dev] Usermode interrupt handling delayed until end of quanta?

[Jeroen C. Visser]

On Jan 23, 2008 4:18 PM, Jonathan S. Shapiro <shap at eros-os.com> wrote:
> Yes. It's most definitely wrong.

Because it does the exact opposite of the current hardcoded policy. Understood.

> The dispatched handler should preempt
> the currently running process exactly (and only) if it has higher
> priority. So your patch introduces exactly the wrong behavior. The
> correct approach is to run the serial driver at higher priority. You
> will then discover that priority-based dispatch isn't yet implemented,
> which is the real problem.
>
> In order to fix that:
>
>   (1) the ready queue needs a better design so that it will handle
>       priorities better, and
>   (2) insert onto the ready queue needs to do the preemption check.

Is that work scheduled already?

For now I will probably keep my patch so as to have a responsive driver :)

-JCV