Re: [Qemu-devel] [PATCH for-2.3] util/qemu-config" fix regression of qmp_query_command_line_options

[Marcel Apfelbaum <marcel@redhat.com>]

On 04/01/2015 07:30 PM, Eric Blake wrote:
> On 04/01/2015 09:17 AM, Eric Blake wrote:
>> On 04/01/2015 08:08 AM, Marcel Apfelbaum wrote:
>>> Commit 49d2e64 (machine: remove qemu_machine_opts global list) made
>>> machine machine options specific to machine sub-type, leaving
>>
>> At the risk of sounding like a machine: s/machine machine/machine/
>>
>>> the qemu_machine_opts desc array empty. Sadly this is the place
>>> qmp_query_command_line_options is looking for supported options.
>>>
>>> As a fix for for 2.3 the machine_qemu_opts are restored only
>>> for the scope of qemu-config, bringing together all machines
>>
>> s/machines/machines'/
>>
>>> properties. We need to find a better fix for 2.4.
Hi Eric,
Thank you for your help.

>
> As discussed on the other thread, we may want a v2:
Submitted.

>
>
>> I also tested that query-command-line-options on qemu.git + this patch
>> provided a superset of the options exposed by
>> 2:qemu-system-x86-2.1.3-3.fc21.x86_64 as installed on Fedora 21, while
>> without this patch it had no options.
>
> The problem is that the superset includes per-machine options, even for
> machines where they don't apply.  I think it's better for 2.3 to be
> conservative (advertise no less than before, so we don't regress) than
> to lie (advertising something we don't support is risky).
I agree, this is the reason we wanted per-machine options, to not
lie to higher levels.

Thanks,
Marcel

>