> -----Original Message-----
> From: Jan Kiszka [mailto:[email protected]]
> Sent: 2019年1月11日 9:30
> To: Lokesh Vutla <[email protected]>; Peng Fan <[email protected]>; Peng
> Fan <[email protected]>
> Cc: Jailhouse <[email protected]>
> Subject: Re: Jailhouse on iMX8M
>
> On 10.01.19 15:58, 'Lokesh Vutla' via Jailhouse wrote:
> > On 10/01/19 11:08 AM, Peng Fan wrote:
> >> Hi Lokesh
> >>
> >>> -----Original Message-----
> >>> From: 'Lokesh Vutla' via Jailhouse
> >>> [mailto:[email protected]]
> >>> Sent: 2019年1月8日 21:24
> >>> To: Peng Fan <[email protected]>
> >>> Cc: [email protected]
> >>> Subject: Re: Jailhouse on iMX8M
> >>>
> >>> Hi Peng,
> >>> Sorry to bring up an old thread.
> >>>
> >>> On 13/03/18 9:43 AM, Peng Fan wrote:
> >>>> Hi Anders,
> >>>> On Mon, Mar 12, 2018 at 04:52:35PM +0100, Anders T??rnqvist wrote:
> >>>>> I noticed that there are development ongoing for a iMX8M based
> board.
> >>> Great!
> >>>>>
> >>>>> When reading in the reference manual for iMX8M it does not have
> >>> information about that the chip has virtualization extensions.
> >>>>> The web information about another chip - iMX8 - clearly talks
> >>>>> about
> >>> virtualization extentions.
> >>>>>
> >>>>> The existing Jailhouse development for the iMX8M must be a proof
> >>>>> of
> >>> some support for virtualization.
> >>>>> Right?
> >>>>
> >>>> The i.MX8M supports ARM virtualization extension.
> >>>
> >>> I am bit exited to see a doc[1] saying that iMX8Qm doesn't need a SW
> >>> hypervisor for running two OSs on the same SoC. Out of curiosity,
> >>> was wondering how does the IRQ routing happens without hypervisor?
> >>> Also I am interested to see more details on this problem. Are there
> >>> any public docs that I take a look?
> >>
> >> It is i.MX8QM, not i.MX8M :)
> >>
> >> It was developed by other team, I do not have much info.
> >> There is only one GIC controller, so need to config GIC distribute to
> >> make sure SPI are routed to cores correctly, and linux gic code also needs
> to be modified.
> >
> > If linux driver is used for configuring gic, then one VM can
> > potentially screw other VM interrupts no? But anyways it is
> > interesting to see how all the these problems are solved without
> > hypervisor :)
>
> I suppose it boils down to the questions who is in charge of configuring the
> shared resources and who can control the access to it - or it is just standard
> collaborative AMP, something everyone did before the arrival of
> virtualization.
To run without hypervisor, first need to disable coherency between the OSes,
on i.MX8QM, it is cluster coherency disabled.
The major issue is GIC. About gic, OS only need to touch gic cpu interface which
is per cpu, no need touch gic distributor. Anyway this has no chance to be
accepted by Linux Kernel.
Regards,
Peng.
>
> Jan
>
> --
> --
> Siemens AG, Corporate Technology, CT RDA IOT 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.
