Re: [PATCH] virtio_mmio: add ACPI probing

["Michael S. Tsirkin" <mst@redhat.com>]

On Wed, Jul 29, 2015 at 10:01:35AM +0800, Shannon Zhao wrote:
> 
> 
> On 2015/7/29 5:10, Peter Maydell wrote:
> > On 28 July 2015 at 21:28, G Gregory <graeme.gregory@linaro.org> wrote:
> >> > On 28 July 2015 at 21:12, Peter Maydell <peter.maydell@linaro.org> wrote:
> >>> >> Mmm. I'm not terribly happy about stuff being in QEMU before the
> >>> >> ACPI spec for it has been finalised. We should not be picking
> >>> >> stuff randomly on the fly...
> >>> >>
> >>> >> If we want to fix the ACPI IDs QEMU is using for 2.4 then we
> >>> >> really need to do that now (ie within the next day or two).
> >>> >>
> >> > It is upto the owner of the QEMU prefix to allocate numbers. This is
> >> > not an issue for ACPI spec at all.
> > I mean "the specification for how this device should be advertised
> > in an ACPI table". I don't care whether that's an official ACPI
> > consortium thing or something less official. The table is
> > constructed by QEMU and read by the kernel (and possibly
> > also by UEFI?), so everybody needs to agree on what the
> > string is...
> 
> I agree with Peter. Maybe we should record these IDs at some place.

Right. And it should apply even if you keep using LNRO.
Pls create docs/specs/acpi_ids.txt and list them there.

For simplicity, I'd just replace LNRO with QEMU.

> Since QEMU is owner of this device and we register QEMU in ASWG, the
> official ID can be assigned by QEMU and the ID could(or should) be
> "QEMUXXXX". But what's the exact ID for this virtio-mmio? That's what we
> need to agree on and record.

So far we only had PVPANIC with ACPI ID QEMU0001.

> P.S. I don't see "QEMU" in the list of approved Vendor IDs for ACPI.
> http://www.uefi.org/sites/default/files/resources/PNPID_List.pdf
> http://www.uefi.org/sites/default/files/resources/ACPIID_List.pdf

Should appear there any day now.

> -- 
> Shannon