Avi Kivity wrote:
> Han, Weidong wrote:
>>> This is how KVM_CAP_ works:
>>> 
>>> - kvm.h defines KVM_CAP_DEVICE_ASSIGNMENT
>>> - userspace compiles device assignment code only if it sees
>>> KVM_CAP_DEVICE_ASSIGNMENT in kvm.h
>>> - kvm.ko returns nonzero to ioctl(KVM_CHECK_EXTENSION,
>>> KVM_CAP_DEVICE_ASSIGNMENT) if it supports device assignment (recent
>>> enough module, iommu found) - userspace runs device assignment code
>>> only if above ioctl passes 
>>> 
>>> We need something similar to deassignment, since it didn't arrive
>>> at the same time as assignment. 
>>> 
>> 
>> In my patches, deassignment already does the similar thing as
>> assignment. Can you point out where it is missed? 
>> 
> 
> It needs a new KVM_CAP_ symbol.  2.6.28 has KVM_CAP_DEVICE_ASSIGNMENT
> defined, but will not have the deassignment code.

In fact, without multiple device assignment, VT-d hotplug cannot be supported 
cleanly. Given hot add a device under a different iommu, it will be mapped in a 
wrong iommu, thus it won't work. I will start to implement multiple device 
assignment first. 

Regards,
Weidong--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to