Re: [Qemu-devel] [Qemu-block] [PATCH] iotests: drop unnecessary accel=kvm

[Thomas Huth]

On 19/02/2019 14.36, Vladimir Sementsov-Ogievskiy wrote:
> 19.02.2019 16:20, Thomas Huth wrote:
>> On 19/02/2019 14.07, Stefan Hajnoczi wrote:
>>> On Tue, Feb 19, 2019 at 12:12 PM Vladimir Sementsov-Ogievskiy
>>> <address@hidden> wrote:
>>>> 19.02.2019 15:02, Thomas Huth wrote:
>>>>> On 19/02/2019 12.59, Stefan Hajnoczi wrote:
>>>>>> Tests 235 and 238 do not require the kvm accelerator.  TCG works fine.
>>>>>>
>>>>>> Use the default accelerator instead of requiring kvm.
>>>>>>
>>>>>> Suggested-by: Thomas Huth <address@hidden>
>>>>>> Signed-off-by: Stefan Hajnoczi <address@hidden>
>>>>>> ---
>>>>>>    tests/qemu-iotests/235 | 1 -
>>>>>>    tests/qemu-iotests/238 | 1 -
>>>>>>    2 files changed, 2 deletions(-)
>>>>>>
>>>>>> diff --git a/tests/qemu-iotests/235 b/tests/qemu-iotests/235
>>>>>> index d6edd97ab4..329da8f0c2 100755
>>>>>> --- a/tests/qemu-iotests/235
>>>>>> +++ b/tests/qemu-iotests/235
>>>>>> @@ -49,7 +49,6 @@ qemu_img_create('-f', iotests.imgfmt, '-o', 
>>>>>> 'preallocation=metadata', disk,
>>>>>>                    str(size))
>>>>>>
>>>>>>    vm = QEMUMachine(iotests.qemu_prog)
>>>>>> -vm.add_args('-machine', 'accel=kvm')
>>>>>
>>>>> According to the initial commit log of 235:
>>>>>
>>>>>    "iotests: simple mirror test with kvm on 1G image"
>>>>>
>>>>> ... so I assume KVM was used on purpose here?
>>>>
>>>> As I remember kvm is not really necessary, and test have a comment about 
>>>> it:
>>>> # And it didn't reproduce if at least one of the following:
>>>> ...
>>>> # 3. drop kvm and use iotests.VM() (maybe, because of qtest) (however, it 
>>>> still
>>>> #    reproduces, if just drop kvm, but gdb failed to produce full 
>>>> backtraces
>>>> #    for me)
>>>>
>>>> But the comment should be updated with this patch.
>>>
>>> Will fix in v2.
>>
>> Any chance that you could even make it somehow work with -M accel=qtest
>> instead? ... in case someone compiled their QEMU with --disable-tcg, too
>> ...?
> 
> I didn't investigate why bug not reproduced with qtest.. I think, the simplest
> option should be a helper like iotests.needs_tcg(), to skip the test if
> it is unsupported.
> 
> Or you case is that kvm is OK for you but tcg is not? Stefan, do you have 
> strong reason
> for removing kvm from iotests? Could it be something like
> 
>    if not iotests.supports_tcg():
>       vm.add_args('-machine', 'accel=kvm')
> 
> ?

You can also use "accel=kvm:tcg" which tries to use kvm first and then
falls back to tcg if KVM is not available.

I just have this weird condition where I try to compile qemu with
--disable-tcg, and then run the iotests on a CI system where KVM is also
not available at runtime. So qtest is the only available accelerator
there...

 Thomas