Hi,
I wrote a new blog post about the clock and interrupt issues
http://phipse.github.io/rtems/blog/2013/07/09/how-late-is-it/
Excerpt:
What we need to provide clock ticks to a partition ...
.. in POK
* Forward interrupts to registered partitions
* Register several partitions to the same interrupt, aka interrupt chaining
* Keep a count for every partition’s pending interrupt
* Checkpointing the time in the scheduler at partition entry and exit
.. in RTEMS
* A time warp function to cover the time, when the partition was not
scheduled
The post describes the current situation, proposes a meta-handler
structure for interrupts and how interrupts can be forwarded to the
partition.
# Questions to be discussed:
## Scheduler Time Checkpoints
The POK scheduler needs to keep track of each partition entry and exit,
so it can tell the partition how many clock ticks it missed. I haven’t
come around to take a look at the scheduler implementation and how this
can be inserted, but on the other hand I think, that pending clock
interrupts will pretty much have the same effect. They accumulate during
the time the partition is off-line and are delivered, when it comes
on-line again.
What do you think?
I would prefer to only mess around with one subsystem at a time.
Regards,
Philipp
_______________________________________________
rtems-devel mailing list
rtems-devel@rtems.org
http://www.rtems.org/mailman/listinfo/rtems-devel
