Re: [RFC v7 1/7] KVM: api: pass the devid in the msi routing entry

["Radim Krčmář" <rkrcmar@redhat.com>]

2016-07-21 17:43+0100, Andre Przywara:
> Hi Radim,
> 
> On 21/07/16 17:01, Radim Krčmář wrote:
>> 2016-07-18 13:25+0000, Eric Auger:
>>> On ARM, the MSI msg (address and data) comes along with
>>> out-of-band device ID information. The device ID encodes the
>>> device that writes the MSI msg. Let's convey the device id in
>>> kvm_irq_routing_msi and use KVM_MSI_VALID_DEVID flag value in
>>> kvm_irq_routing_entry to indicate the msi devid is populated.
>>>
>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>> Reviewed-by: Andre Przywara <andre.przywara@arm.com>
>>>
>>> ---
>>>  
>>> +devid: If KVM_MSI_VALID_DEVID is set, contains a unique device identifier
>>> +       for the device that wrote the MSI message.
>>> +       For PCI, this is usually a BFD identifier in the lower 16 bits.
>>> +
>>> +The per-VM KVM_CAP_MSI_DEVID capability advertises the requirement to
>>> +provide the device ID. If this capability is not set, userland cannot
>>> +rely on the kernel to allow the KVM_MSI_VALID_DEVID flag being set.
>> 
>> It would be better to enforce this mentioned dependency on set
>> KVM_CAP_MSI_DEVID, but is the dependency even required?
>> It seems we were checking flags for zero, so KVM_MSI_VALID_DEVID
>> couldn't have been set by old userspaces, therefor it is ok to only make
>> it depend only on the presence of KVM_CAP_MSI_DEVID, like the patch does
>> now.  (I assume KVM_MSI_VALID_DEVID and KVM_CAP_MSI_DEVID are being
>> merged at the same time.)
>> 
>> Then there would be little point in having KVM_CAP_MSI_DEVID enableable,
>> so does enabling KVM_CAP_MSI_DEVID mean that every MSI must have a valid
>> devid?
> 
> KVM_CAP_MSI_DEVID tells userland that it's fine to set the
> KVM_MSI_VALID_DEVID flag (because the kernel would bark otherwise).
> 
> KVM_MSI_VALID_DEVID tells the kernel that there is some meaningful
> device ID data in the field formerly known as "pad".
> 
> IIRC we started with the VALID_DEVID flag, then found that we need the
> CAP because we repurposed the pad field.
> 
> Does that make sense? Admittedly this _is_ confusing ;-)

It does, thanks.
Some capability is need and I thought that KVM_CAP_MSI_DEVID has to be
enabled by userspace before KVM_MSI_VALID_DEVID can be used, which isn't
the case.  It is enabled conditionally based on vgic ITS ... my bad.