On 2016-11-23 17:51, Mark Rutland wrote: > On Wed, Nov 23, 2016 at 05:24:55PM +0100, Jan Kiszka wrote: >>> On 11/23/2016 03:46 PM, Mark Rutland wrote: >>>> I think jailhouse needs to gain some PSCI_CPU_SUSPEND support. It should >>>> be able to at least use WFI safely, and ideally it would call into the >>>> real PSCI implementation for deeper states. >> >> Good point! wfi would be simple and safe, but is going into deeper >> states side-effect free for other cores? That's always the major concern >> about power saving, not only on ARM... > > In platform coordinated mode, it *should* be side-effect free for otehr > cores per the spec. > > The idle states requested by CPUs are treated as a vote; the > implementation must respect the shallowest idle state requested and go > no deeper, so resources shouldn't be reset unexpectedly. > > In OS initiated mode, that's not the case, and the last core standing > can override previous votes. It should be possible to inhibit that, > however. We can transition the root cell back to platform coordinated > mode and never expose OS initiated mode to other inmates, or jailhouse > could reject/restrict calls that would violate cross-inmate > expectations.
Thanks for the explanation. Looks like someone needs to dig deeper into the details and how to implement them. For now, I will write a little patch that does the wfi - if that is simple: we are in EL2 then, not getting the usual vmexit on an interrupts... Jan -- Siemens AG, Corporate Technology, CT RDA ITP SES-DE Corporate Competence Center Embedded Linux -- You received this message because you are subscribed to the Google Groups "Jailhouse" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
