Re: [Qemu-devel] q35 and sysbus devices

[Markus Armbruster <armbru@redhat.com>]

Peter Maydell <peter.maydell@linaro.org> writes:

> On 24 March 2017 at 16:58, Markus Armbruster <armbru@redhat.com> wrote:
>> "Sysbus" isn't a bus.  In qdev's original design, every device had to
>> plug into a bus, period.  The ones that really didn't were made to plug
>> into "sysbus".
>>
>> Pretty much the only thing "sysbus" devices had in common was that they
>> couldn't be used with device_add and device_del.
>
> This isn't really true. Sysbus devices support having MMIO regions
> and IRQ lines and GPIO lines. If you need those you're a
> sysbus device; otherwise you can probably just be a plain old Device.

Well, "device has MMIO regions, IRQ lines and GPIO lines" is about as
"device contains virtual silicon".  What would a device without any of
these *do*?

Devices plugging into a bus have to expose their MMIO regions, IRQ
lines, etc. in a certain way dictated by the bus.  In return, you don't
have to wire up each resource manually, you simply plug into the bus and
are done.  That's what makes a bus a bus for me.

"Sysbus" does nothing of the sort.

>> We fixed the design to permit bus-less devices, but we didn't get rid of
>> "sysbus".
>
> Call it what you want, but we should have some common code support
> for "I want to have MMIOs and IRQs and GPIO lines".

Of course.

>                                                     You could
> argue for moving all that into Device I suppose.

Yup.

>> We got a "platform bus", which is really not the same as "sysbus", but
>> we shoehorned it into "sysbus" anyway.
>
> I agree 'platform bus' is a mess, and I'd rather it didn't exist.
> Unfortunately people really really want to be able to do device
> pass through of random memory-mapped devices :-(

The "platform" bus adds certain constraints over "sysbus", precisely to
make these uses possible.  And that's precisely why it should be its own
thing instead of complicating "sysbus".