Re: [PATCH v2 2/3] qom: move user_creatable_add_opts logic to vl.c and QAPIfy it

[Markus Armbruster]

David Hildenbrand <david@redhat.com> writes:

> On 13.04.21 06:41, Markus Armbruster wrote:
>> David Hildenbrand <david@redhat.com> writes:
>> 
>>> On 12.03.21 18:35, Paolo Bonzini wrote:
>>>> Emulators are currently using OptsVisitor (via user_creatable_add_opts)
>>>> to parse the -object command line option.  This has one extra feature,
>>>> compared to keyval, which is automatic conversion of integers to lists
>>>> as well as support for lists as repeated options:
>>>>     -object
>>>> memory-backend-ram,id=pc.ram,size=1048576000,host-nodes=0,policy=bind
>>>> So we cannot replace OptsVisitor with keyval right now.  Still, this
>>>> patch moves the user_creatable_add_opts logic to vl.c since it is
>>>> not needed anywhere else, and makes it go through user_creatable_add_qapi.
>>>> In order to minimize code changes, the predicate still takes a
>>>> string.
>>>> This can be changed later to use the ObjectType QAPI enum directly.
>>>>
>>>
>>> Rebasing my "noreserve"[1] series on this, I get weird errors from
>>> QEMU when specifying the new "reserve=off" option for a
>>> memory-backend-ram:
>>>
>>> "Invalid parameter 'reserve'"
>>>
>>> And it looks like this is the case for any new properties. Poking
>>> around, I fail to find what's causing this -- or how to unlock new
>>> properties. What is the magic toggle to make it work?
>>>
>>> Thanks!
>>>
>>> [1] https://lkml.kernel.org/r/20210319101230.21531-1-david@redhat.com
>> Wild guess: you didn't add your new properties in the QAPI schema.
>> For a not-so-wild-guess, send us a git-fetch argument for your
>> rebased
>> series.
>> 
>
> Oh, there is qapi/qom.json -- maybe that does the trick.
>
> (I have mixed feelings about having to specify the same thing twice at
> different locations)

With reason.

We've talked about generating QOM boilerplate from the QAPI schema, but
haven't progressed to actual patches.

> I'll have a look if that makes it fly.