Problems with the blackfin port

[André Keller Abadie]

Now I see what you meant, Allan! Your idea is great, actually I didn't
know of this self-nesting interruption capability. The manual doesn't
say much about this either. We should test how many times the same
interruption can be called though, to check if this is suitable for
RTEMS. I'd like to have your patches, could you please send it to my
email?

Joel's idea of leaving a fake interrupt stack frame seems feasible
too, but I have to study the m68k port to be sure of this. The nice
part is that this approach would not limit the number of threads in
the system.


Lately i've been a little busy with academic stuff, but as soon as
possible I'll evaluate these ideas. I'll keep you informed.

Regards

André--



On 4/1/08, Allan Hessenflow <allanh-rtems2 at kallisti.com> wrote:
> Joel Sherrill wrote:
>  > self-nesting?  Is this "restoring the interrupt disable
>  > level to that of the task"?
>
>
> In this context self-nesting is a Blackfin specific term.  It means
>  that an interrupt of a given priority can interrupt another of the same
>  priority.
>
>
>  > The context switch code is supposed to include interrupt
>  > disable level per task.
>
>
> Yes, the Blackfin code does that.  However, the whole approach of using
>  a software interrupt to do the thread dispatch relies on the lowest
>  priority interrupt being enabled at any time that any other interrupt
>  is also enabled.  That requirement is currently met by the port only
>  supporting two levels, all enabled and all disabled.  If it were to be
>  expanded to support multiple levels, it could still be met by just
>  insuring the lowest priority level is enabled for all but the case
>  where all other levels are disabled.
>
>
>  allan
>
>  --
>  Allan N. Hessenflow      allanh at kallisti.com
>  _______________________________________________
>  rtems-users mailing list
>  rtems-users at rtems.com
>  http://rtems.rtems.org/mailman/listinfo/rtems-users
>