[GSOC] Paravirtualization Layer in RTEMS.
claudiodcsilva at gmail.com
Thu May 1 12:08:09 UTC 2014
Pending interrupts are managed by Hardware (CPU or Interrupt Manager).
Lets keep in mind that the hypervisor (POK in this case) is supplying
a virtual CPU interface to the operating system executing inside each
partition (RTEMS). POK should mimic the behaviour of the real hardware
when supplying this interface.
If an hardware interrupt arrives, POK should vector it to the
virtualized RTEMS running in the partition. The virtualized interrupt
handler from RTEMS will only be called if the interrupt is unmasked
and virtual interrupts are enabled. If this is not the case, POK
should keep the interrupt as pending and vector it to RTEMS once it
enables or unmasks the interrupt.
On Wed, Apr 30, 2014 at 12:03 PM, Youren Shen <shenyouren at gmail.com> wrote:
> Hi, Cláudio:
> Thank you for your summary, if we want to mask or unmask an interrupt, we
> can just add a hypercall in hypercall system corresponding to the function,
> ( This is based on the premise that the interrupt structure has been adapt
> to support this function.) However, could you direct me about how to track
> pending interrupt in a normal OS(then I can consider how to deal with this
> part in paravirtualization)? I'm not clear about this point.
> Thank you very much.
> Best Regards.
> Youren Shen.
More information about the devel