Re: [kvm-unit-tests PATCH 3/7] s390x: virtio: CCW transport implementation

[Pierre Morel <pmorel@linux.ibm.com>]

On 11/9/21 08:01, Thomas Huth wrote:
> On 08/11/2021 13.34, Pierre Morel wrote:
>>
>>
>> On 11/3/21 08:49, Thomas Huth wrote:
>>> On 27/08/2021 12.17, Pierre Morel wrote:
>>>> This is the implementation of the virtio-ccw transport level.
>>>>
>>>> We only support VIRTIO revision 0.
>>>
>>> That means only legacy virtio? Wouldn't it be better to shoot for 
>>> modern virtio instead?
>>
>> Yes but can we do it in a second series?
> 
> Sure.
> 
>>>> +int virtio_ccw_read_features(struct virtio_ccw_device *vcdev, 
>>>> uint64_t *features)
>>>> +{
>>>> +    struct virtio_feature_desc *f_desc = &vcdev->f_desc;
>>>> +
>>>> +    f_desc->index = 0;
>>>> +    if (ccw_send(vcdev, CCW_CMD_READ_FEAT, f_desc, sizeof(*f_desc), 
>>>> 0))
>>>> +        return -1;
>>>> +    *features = swap32(f_desc->features);
>>>> +
>>>> +    f_desc->index = 1;
>>>> +    if (ccw_send(vcdev, CCW_CMD_READ_FEAT, f_desc, sizeof(*f_desc), 
>>>> 0))
>>>> +        return -1;
>>>> +    *features |= (uint64_t)swap32(f_desc->features) << 32;
>>>
>>> Weren't the upper feature bits only available for modern virtio anyway?
>>
>> Yes.
>> I have the intention to upgrade to Rev. 1 when I get enough time for it.
>> Should I remove this? It does not induce problem does it?
> 
> No problem - maybe simply add a comment that the upper bits are for 
> virtio 1.0 and later.

OK,
Thanks,

Pierre

> 
>   Thomas
> 

-- 
Pierre Morel
IBM Lab Boeblingen