Re: [PATCH v1 02/12] vhost: Return number of free memslots

[David Hildenbrand]

On 27.10.21 18:11, Philippe Mathieu-Daudé wrote:
> On 10/27/21 17:45, David Hildenbrand wrote:
>> On 27.10.21 17:33, Michael S. Tsirkin wrote:
>>> On Wed, Oct 27, 2021 at 04:11:38PM +0200, Philippe Mathieu-Daudé wrote:
>>>> On 10/27/21 16:04, David Hildenbrand wrote:
>>>>> On 27.10.21 15:36, Philippe Mathieu-Daudé wrote:
>>>>>> On 10/27/21 14:45, David Hildenbrand wrote:
>>>>>>> Let's return the number of free slots instead of only checking if there
>>>>>>> is a free slot. Required to support memory devices that consume multiple
>>>>>>> memslots.
>>>>>>>
>>>>>>> Signed-off-by: David Hildenbrand <david@redhat.com>
>>>>>>> ---
>>>>>>>  hw/mem/memory-device.c    | 2 +-
>>>>>>>  hw/virtio/vhost-stub.c    | 2 +-
>>>>>>>  hw/virtio/vhost.c         | 4 ++--
>>>>>>>  include/hw/virtio/vhost.h | 2 +-
>>>>>>>  4 files changed, 5 insertions(+), 5 deletions(-)
>>>>
>>>>>>> -bool vhost_has_free_slot(void)
>>>>>>> +unsigned int vhost_get_free_memslots(void)
>>>>>>>  {
>>>>>>>      return true;
>>>>>>
>>>>>>        return 0;
>>>>>
>>>>> Oh wait, no. This actually has to be
>>>>>
>>>>> "return ~0U;" (see real vhost_get_free_memslots())
>>>>>
>>>>> ... because there is no vhost and consequently no limit applies.
>>>>
>>>> Indeed.
>>>>
>>>> Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
>>>
>>> confused. are you acking the theoretical patch with ~0 here?
>>>
>>
>> That's how I interpreted it.
> 
> ~0U doesn't seem harmful when comparing. However I haven't tested
> nor looked at the big picture. I wonder if vhost_has_free_slot()
> shouldn't take the Error* as argument and each implementation set
> the error message ("virtio/vhost support disabled" would be more
> explicit in the stub case). But I still don't understand why when
> built without virtio/vhost we return vhost_get_free_memslots() > 0.

For the same reason we faked infinite slots via
vhost_has_free_slot()->true for now. We call it unconditionally from
memory device code.

Sure, we could add a stub "vhost_available()-> false" (or
vhost_enabled() ?) instead and do

if (vhost_available())
        ... vhost_get_free_memslots()

similar to how we have

if (kvm_enabled())
        ... kvm_get_free_memslots()

Not sure if it's worth it, though.

-- 
Thanks,

David / dhildenb