Wright,
Thanks for your comments. I have updated the qemu-options.hx and used
following new CAP. Updated version is also attached.
Index: qemu-kvm/hw/device-assignment.c
===================================================================
--- qemu-kvm.orig/hw/device-assignment.c
+++ qemu-kvm/hw/device-assignment.c
@@ -809,6 +809,16 @@ static int assign_device(AssignedDevice
struct kvm_assigned_pci_dev assigned_dev_data;
int r;
+#ifdef KVM_CAP_PCI_SEGMENT
+ /* Only pass non-zero PCI segment to capable module */
+ if (!kvm_check_extension(kvm_state, KVM_CAP_PCI_SEGMENT) &&
+ dev->h_segnr) {
+ fprintf(stderr, "Can't assign device inside non-zero PCI segment "
+ "as this KVM module doesn't support it.\n");
+ return -ENODEV;
+ }
+#endif
+
memset(&assigned_dev_data, 0, sizeof(assigned_dev_data));
assigned_dev_data.assigned_dev_id =
calc_assigned_dev_id(dev->h_segnr, dev->h_busnr, dev->h_devfn);
Index: qemu-kvm/kvm/include/linux/kvm.h
===================================================================
--- qemu-kvm.orig/kvm/include/linux/kvm.h
+++ qemu-kvm/kvm/include/linux/kvm.h
@@ -498,6 +498,8 @@ struct kvm_ioeventfd {
#define KVM_CAP_S390_PSW 42
#define KVM_CAP_PPC_SEGSTATE 43
+#define KVM_CAP_PCI_SEGMENT 47
+
#ifdef KVM_CAP_IRQ_ROUTING
struct kvm_irq_routing_irqchip {
Index: linux-2.6/include/linux/kvm.h
===================================================================
--- linux-2.6.orig/include/linux/kvm.h
+++ linux-2.6/include/linux/kvm.h
@@ -500,6 +500,7 @@ struct kvm_ioeventfd {
#define KVM_CAP_HYPERV 44
#define KVM_CAP_HYPERV_VAPIC 45
#define KVM_CAP_HYPERV_SPIN 46
+#define KVM_CAP_PCI_SEGMENT 47
#ifdef KVM_CAP_IRQ_ROUTING
Index: linux-2.6/arch/x86/kvm/x86.c
===================================================================
--- linux-2.6.orig/arch/x86/kvm/x86.c
+++ linux-2.6/arch/x86/kvm/x86.c
@@ -1569,6 +1569,7 @@ int kvm_dev_ioctl_check_extension(long e
case KVM_CAP_HYPERV:
case KVM_CAP_HYPERV_VAPIC:
case KVM_CAP_HYPERV_SPIN:
+ case KVM_CAP_PCI_SEGMENT:
r = 1;
break;
case KVM_CAP_COALESCED_MMIO:
Chris Wright wrote:
* Zhai, Edwin (edwin.zhai@xxxxxxxxx) wrote:> These 2 patches enable optional parameter(default 0) - PCI segment(or > domain) besides BDF, when assigning PCI device to guest.> > Signed-off-by: Zhai Edwin <edwin.zhai@xxxxxxxxx> Looks good. For compatibility, an old userspace will zero the padding and keep the default 0 segment when running w/ a new kernel, a new userspace won't be able to pass a non-zero segment value to an old kernel. If that latter bit matters, need a capability to express this change. Looks like qemu-options.hx needs an update too. Otherwise... Acked-by: Chris Wright <chrisw@xxxxxxxxxxxx>
Attachment:
kvm_vtd_multi_seg_qemu_v2.patch
Description: Binary data
Attachment:
kvm_vtd_multi_seg_kmod_v2.patch
Description: Binary data
