On 09/04/15 14:00, Joel Sherrill wrote:
On April 9, 2015 4:11:14 AM CDT, Alexander
Krutwig<[email protected]> wrote:
>Hello,
>
>the newest Timecounter patch was sent yesterday. As of now, the
>replacement of the nanoseconds extension with FreeBSD timecounters will
>
>also be started for the remaining BSPs. Exemplarily, the changes for
>the
>ERC32-BSP are given in the patch.
Nice to see this coming along so fast. Any thoughts on the impact of variable length
clock ticks? This is often referred to as tickless but I am not as concerned with the use
case of dynamically reprogramming the clock tick quantum as advancing time to account for
being in a virtualized environment and your "container" not running for a while.
In terms of support for tickless systems the situation is slightly
better. With the nanoseconds extensions this would simply not work. With
the timecounters it depends on the hardware capabilities. In a tickless
system large time steps can occur (e.g. you sleep one minute in
powerdown mode, then wake up, now you have a 1 minute time step). In
case this time step is larger than your hardware timer interval, then
you have a problem.
--
Sebastian Huber, embedded brains GmbH
Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone : +49 89 189 47 41-16
Fax : +49 89 189 47 41-09
E-Mail : [email protected]
PGP : Public key available on request.
Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.
_______________________________________________
devel mailing list
[email protected]
http://lists.rtems.org/mailman/listinfo/devel
