On Fri, Sep 6, 2013 at 4:19 PM, Will Deacon will.deacon@arm.com wrote:
On Fri, Sep 06, 2013 at 11:09:09AM +0100, Anup Patel wrote:
On Fri, Sep 6, 2013 at 2:36 PM, Will Deacon will.deacon@arm.com wrote:
On Fri, Sep 06, 2013 at 09:54:23AM +0100, Alexander Graf wrote:
On 06.09.2013, at 09:44, Anup Patel wrote:
it hard to come up with a use case where having a separate ioctl for "get best CPU for this host" would be better for user space.
It can store it and migrate it as part of its migration protocol (probably hard for QEMU's current work flow, but that's QEMU's issue).
It's also better from the point of view of devicetree generation. For example, kvmtool needs to generate the correct architected timer interrupt numbers for the target CPU, so munging this into KVM_ARM_VCPU_INIT means we have to go off and figure out the host CPU separately anyway.
Please look at the patch carefully we are returning VCPU target type back to user space so, you can generate correct architected timer interrupt numbers for the target CPU.
I did look at the patch, but I also looked at the overriding consensus in the feedback saying that you can't change the direction of the ioctl, so that would require what I said above.
Will
Instead of arguing more, I'll save everyone's time and revise this patch as needed by everyone.
--Anup