Hi Tirumalesh,
....
>> +/* The MOVI command moves an ITTE to a different collection. */
>> +static int vits_cmd_handle_movi(struct kvm *kvm, u64 *its_cmd)
>> +{
>> + struct vgic_its *its = &kvm->arch.vgic.its;
>> + u32 device_id = its_cmd_get_deviceid(its_cmd);
>> + u32 event_id = its_cmd_get_id(its_cmd);
>> + u32 coll_id = its_cmd_get_collection(its_cmd);
>> + struct its_itte *itte;
>> + struct its_collection *collection;
>> + int ret;
>> +
>> + spin_lock(&its->lock);
>> + itte = find_itte(kvm, device_id, event_id);
>> + if (!itte) {
>> + ret = E_ITS_MOVI_UNMAPPED_INTERRUPT;
>> + goto out_unlock;
>> + }
>> + if (!its_is_collection_mapped(itte->collection)) {
>> + ret = E_ITS_MOVI_UNMAPPED_COLLECTION;
>> + goto out_unlock;
>> + }
>> +
>> + collection = find_collection(kvm, coll_id);
>> + if (!its_is_collection_mapped(collection)) {
>> + ret = E_ITS_MOVI_UNMAPPED_COLLECTION;
>> + goto out_unlock;
>> + }
>> +
>> + itte->collection = collection;
>> + update_affinity(kvm, collection);
>> +
>
> Not planning to send this down to Physical ITS in case of VFIO for performance reasons?
Why should we?
So in general: this virtual ITS is a purely emulated one, it does not
depend on any hardware ITS to work. Also at this point we don't focus on
VFIO (in fact this doesn't even work without further patches).
So it would break abstraction to pass on commands. Besides I don't see
the benefit of passing a 32-byte structure around, which needs to be
validated (and thus touched) anyway.
Also in this particular case: Why would a change of the _virtual_
affinity affect a VFIO IRQ on the host at all?
Actually the collection IDs of the guests are _totally_ separate and
unrelated to a host collection ID.
Cheers,
Andre
_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
