Am 07.01.2014 11:35, schrieb Markus Armbruster:
> Stefan Hajnoczi <stefa...@redhat.com> writes:
>> On Tue, Dec 31, 2013 at 01:19:02AM +0100, Andreas Färber wrote:
>>> Am 30.12.2013 09:46, schrieb Kewei Yu:
>>>> diff --git a/tests/fdc-test.c b/tests/fdc-test.c
>>>> index 38b5b17..80dbdfb 100644
>>>> --- a/tests/fdc-test.c
>>>> +++ b/tests/fdc-test.c
>>>> @@ -538,7 +538,11 @@ int main(int argc, char **argv)
>>>> /* Run the tests */
>>>> g_test_init(&argc, &argv, NULL);
>>>>
>>>> +#ifdef CONFIG_VNC
>>>> cmdline = g_strdup_printf("-vnc none ");
>>>> +#else
>>>> + cmdline = NULL;
>>>> +#endif
>>>>
>>>> qtest_start(cmdline);
>>>> qtest_irq_intercept_in(global_qtest, "ioapic");
>>>
>>> I have already made -display none a default used by all qtests.
>>> http://git.qemu.org/?p=qemu.git;a=commit;h=2ad645d2854746b55ddfd1d8e951f689cca5d78f
>>>
>>> Is there a reason not to do the same for -vnc none?
>>> That would consolidate the #ifdef'ery to one central point.
>>
>> Agreed, let's set -vnc none by default in one place (if enabled by
>> ./configure).
>>
>> I wonder if we should use -nodefaults too? That may require adding some
>> new options to explicitly set up devices used by existing tests.
>
> Try and see what explodes?
I fear that will render my qom-test useless, which is supposed to test
our defaults. If -nodefaults drops serial ports, floppy drives, etc.
that will reduce its usefulness. Unless of course we find a way to
override -nodefaults with some generic -defaults options. :)
I haven't noticed anyone actually adding qtests along with new machines,
adding them to qom-test already required some poking of mine.
End of last year I was preparing some qtests for optional PCI devices,
but even such stubs are hard to share across multiple shoulders due to
resulting Makefile conflicts...
Andreas
--
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg
