Re: [PATCH v2] hw/arm: add versioning to sbsa-ref machine DT

[Rob Herring <robh@kernel.org>]

On Tue, May 17, 2022 at 11:50 AM Peter Maydell <peter.maydell@linaro.org> wrote:
>
> On Tue, 17 May 2022 at 14:27, Rob Herring <robh@kernel.org> wrote:
> >
> > On Thu, May 5, 2022 at 6:41 AM Leif Lindholm <quic_llindhol@quicinc.com> wrote:
> > >
> > > The sbsa-ref machine is continuously evolving. Some of the changes we
> > > want to make in the near future, to align with real components (e.g.
> > > the GIC-700), will break compatibility for existing firmware.
> > >
> > > Introduce two new properties to the DT generated on machine generation:
> > > - machine-version-major
> > >   To be incremented when a platform change makes the machine
> > >   incompatible with existing firmware.
> > > - machine-version-minor
> > >   To be incremented when functionality is added to the machine
> > >   without causing incompatibility with existing firmware.
> > >   to be reset to 0 when machine-version-major is incremented.
> >
> > Where's the binding documentation for this?
> >
> > We already have a way to version DTs and that's with compatible. I'm
> > not completely opposed to a version number though, but I am opposed to
> > it not being common. We've rejected vendors (QCom in fact) doing their
> > own thing here.
> >
> >
> > > This versioning scheme is *neither*:
> > > - A QEMU versioned machine type; a given version of QEMU will emulate
> > >   a given version of the platform.
> > > - A reflection of level of SBSA (now SystemReady SR) support provided.
> >
> > FYI, it's planned to certify the virt machine for SR-IR which will
> > include DT schema validation. Undocumented properties are a problem
> > for that.
>
> This isn't the 'virt' machine :-)

Ah, okay.

> This dtb fragment is a purely private communication between
> the QEMU model and the sbsa-ref EL3 firmware. We could in

And that interface is tightly coupled and always in sync?

> theory equally replace it with a set of hardwired
> "board revision" registers. There's a comment in the existing
> sources about this:
>
> /*
>  * Firmware on this machine only uses ACPI table to load OS, these limited
>  * device tree nodes are just to let firmware know the info which varies from
>  * command line parameters, so it is not necessary to be fully compatible
>  * with the kernel CPU and NUMA binding rules.
>  */
>
> Kernels running on sbsa-ref won't see a dtb (let alone one with this
> version information in it), because firmware will always boot them
> with ACPI.

Okay, if there's no chance this moves up the stack, NM.

Rob