Avi Kivity wrote: > Anthony Liguori wrote: >> Avi Kivity wrote: >>> Anthony Liguori wrote: >>> >>>> + case VIRTIO_PCI_QUEUE_NOTIFY: >>>> + if (val < VIRTIO_PCI_QUEUE_MAX) >>>> + virtio_ring_kick(vdev, &vdev->vq[val]); >>>> + break; >>>> >>> >>> I see you're not using hypercalls for this, presumably for >>> compatibility >>> with -no-kvm. >> >> More than just that. By stick to PIO, we are compatible with just >> about any VMM. For instance, we get Xen support for free. If we >> used hypercalls, even if we agreed on a way to determine which number >> to use and how to make those calls, it would still be difficult to >> implement in something like Xen. >> > > But pio through the config space basically means you're committed to > handling it in qemu. We want a more flexible mechanism.
There's no reason that the PIO operations couldn't be handled in the kernel. You'll already need some level of cooperation in userspace unless you plan on implementing the PCI bus in kernel space too. It's easy enough in the pci_map function in QEMU to just notify the kernel that it should listen on a particular PIO range. > Detecting how to make hypercalls can be left to paravirt_ops. > > (for Xen you'd use an event channel; and for kvm the virtio kick > hypercall). > >>> Well I think I have a solution: advertise vmcall, >>> vmmcall, pio to some port, and int $some_vector as hypercall feature >>> bits in cpuid (for kvm, kvm, qemu, and kvm-lite respectively). Early >>> setup code could patch the instruction as appropriate (I hear code >>> patching is now taught in second grade). >>> >> >> That ties our device to our particular hypercall implementation. If >> we were going to do this, I'd prefer to advertise it in the device I >> think. I really would need to look at the performance though of >> vmcall and an edge triggered interrupt. It would have to be pretty >> compelling to warrant the additional complexity I think. > > vmcall costs will go down, and we don't want to use different > mechanisms for high bandwidth and low bandwidth devices. vmcalls will certainly get faster but I doubt that the cost difference between vmcall and pio will ever be greater than a few hundred cycles. The only performance sensitive operation here would be the kick and I don't think a few hundred cycles in the kick path is ever going to be that significant for overall performance. So why introduce the extra complexity? Regards, Anthony Liguori > > ------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/ _______________________________________________ kvm-devel mailing list kvm-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/kvm-devel
