+static uint16_t vhost_shadow_vring_used_flags(VhostShadowVirtqueue *svq)
+{
+ const struct vring_used *used = svq->hvq->used;
+ return virtio_tswap16(svq->vdev, used->flags);
+}
+
+static bool vhost_shadow_vring_should_kick(VhostShadowVirtqueue *vq)
+{
+ return !(vhost_shadow_vring_used_flags(vq) & VRING_USED_F_NO_NOTIFY);
+}
+
+static void vhost_shadow_vring_kick(VhostShadowVirtqueue *vq)
+{
+ if (vhost_shadow_vring_should_kick(vq)) {
+ event_notifier_set(&vq->kick_notifier);
+ }
+}
+
+static void handle_shadow_vq(VirtIODevice *vdev, VirtQueue *vq)
+{
+ struct vhost_dev *hdev = vhost_dev_from_virtio(vdev);
+ uint16_t idx = virtio_get_queue_index(vq);
+
+ VhostShadowVirtqueue *svq = hdev->shadow_vqs[idx];
+
+ vhost_shadow_vring_kick(svq);
+}
+
+/*
+ * Start shadow virtqueue operation.
+ * @dev vhost device
+ * @svq Shadow Virtqueue
+ *
+ * Run in RCU context
+ */
+bool vhost_shadow_vq_start_rcu(struct vhost_dev *dev,
+ VhostShadowVirtqueue *svq)
+{
+ const VirtioDeviceClass *k = VIRTIO_DEVICE_GET_CLASS(dev->vdev);
+ EventNotifier *vq_host_notifier = virtio_queue_get_host_notifier(svq->vq);
+ unsigned idx = virtio_queue_get_idx(svq->vdev, svq->vq);
+ struct vhost_vring_file kick_file = {
+ .index = idx,
+ .fd = event_notifier_get_fd(&svq->kick_notifier),
+ };
+ int r;
+ bool ok;
+
+ /* Check that notifications are still going directly to vhost dev */
+ assert(virtio_queue_host_notifier_status(svq->vq));
+
+ ok = k->set_vq_handler(dev->vdev, idx, handle_shadow_vq);
+ if (!ok) {
+ error_report("Couldn't set the vq handler");
+ goto err_set_kick_handler;
+ }
+
+ r = dev->vhost_ops->vhost_set_vring_kick(dev, &kick_file);
+ if (r != 0) {
+ error_report("Couldn't set kick fd: %s", strerror(errno));
+ goto err_set_vring_kick;
+ }
+
+ event_notifier_set_handler(vq_host_notifier,
+ virtio_queue_host_notifier_read);
+ virtio_queue_set_host_notifier_enabled(svq->vq, false);
+ virtio_queue_host_notifier_read(vq_host_notifier);
+
+ return true;
+
+err_set_vring_kick:
+ k->set_vq_handler(dev->vdev, idx, NULL);
+
+err_set_kick_handler:
+ return false;
+}
+
+/*
+ * Stop shadow virtqueue operation.
+ * @dev vhost device
+ * @svq Shadow Virtqueue
+ *
+ * Run in RCU context
+ */
+void vhost_shadow_vq_stop_rcu(struct vhost_dev *dev,
+ VhostShadowVirtqueue *svq)
+{
+ const VirtioDeviceClass *k = VIRTIO_DEVICE_GET_CLASS(svq->vdev);
+ unsigned idx = virtio_queue_get_idx(svq->vdev, svq->vq);
+ EventNotifier *vq_host_notifier = virtio_queue_get_host_notifier(svq->vq);
+ struct vhost_vring_file kick_file = {
+ .index = idx,
+ .fd = event_notifier_get_fd(vq_host_notifier),
+ };
+ int r;
+
+ /* Restore vhost kick */
+ r = dev->vhost_ops->vhost_set_vring_kick(dev, &kick_file);
+ /* Cannot do a lot of things */
+ assert(r == 0);
+
+ event_notifier_set_handler(vq_host_notifier, NULL);
+ virtio_queue_set_host_notifier_enabled(svq->vq, true);
+ k->set_vq_handler(svq->vdev, idx, NULL);
+}
+
/*
* Creates vhost shadow virtqueue, and instruct vhost device to use the
shadow
* methods and file descriptors.
@@ -24,8 +138,13 @@ typedef struct VhostShadowVirtqueue {
VhostShadowVirtqueue *vhost_shadow_vq_new(struct vhost_dev *dev, int idx)
{
g_autofree VhostShadowVirtqueue *svq = g_new0(VhostShadowVirtqueue, 1);
+ int vq_idx = dev->vhost_ops->vhost_get_vq_index(dev, dev->vq_index + idx);
int r;
+ svq->vq = virtio_get_queue(dev->vdev, vq_idx);
+ svq->hvq = &dev->vqs[idx];
+ svq->vdev = dev->vdev;
+
r = event_notifier_init(&svq->kick_notifier, 0);
if (r != 0) {
error_report("Couldn't create kick event notifier: %s",
@@ -40,7 +159,7 @@ VhostShadowVirtqueue *vhost_shadow_vq_new(struct vhost_dev
*dev, int idx)
goto err_init_call_notifier;
}
- return svq;
+ return g_steal_pointer(&svq);
err_init_call_notifier:
event_notifier_cleanup(&svq->kick_notifier);
diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
index 42836e45f3..bde688f278 100644
--- a/hw/virtio/vhost.c
+++ b/hw/virtio/vhost.c
@@ -25,6 +25,7 @@
#include "exec/address-spaces.h"
#include "hw/virtio/virtio-bus.h"
#include "hw/virtio/virtio-access.h"
+#include "hw/virtio/vhost-shadow-virtqueue.h"
#include "migration/blocker.h"
#include "migration/qemu-file-types.h"
#include "sysemu/dma.h"
@@ -945,6 +946,82 @@ static void vhost_log_global_stop(MemoryListener *listener)
}
}
+static int vhost_sw_live_migration_stop(struct vhost_dev *dev)
+{
+ int idx;
+
+ WITH_RCU_READ_LOCK_GUARD() {
+ dev->sw_lm_enabled = false;
+
+ for (idx = 0; idx < dev->nvqs; ++idx) {
+ vhost_shadow_vq_stop_rcu(dev, dev->shadow_vqs[idx]);
+ }
+ }
+
+ for (idx = 0; idx < dev->nvqs; ++idx) {
+ vhost_shadow_vq_free(dev->shadow_vqs[idx]);
+ }
+
+ g_free(dev->shadow_vqs);
+ dev->shadow_vqs = NULL;
+ return 0;
+}
+
+static int vhost_sw_live_migration_start(struct vhost_dev *dev)
+{
+ int idx;
+
+ dev->shadow_vqs = g_new0(VhostShadowVirtqueue *, dev->nvqs);
+ for (idx = 0; idx < dev->nvqs; ++idx) {
+ dev->shadow_vqs[idx] = vhost_shadow_vq_new(dev, idx);
+ if (unlikely(dev->shadow_vqs[idx] == NULL)) {
+ goto err;
+ }
+ }
+
+ WITH_RCU_READ_LOCK_GUARD() {
+ for (idx = 0; idx < dev->nvqs; ++idx) {
+ int stop_idx = idx;
+ bool ok = vhost_shadow_vq_start_rcu(dev,
+ dev->shadow_vqs[idx]);
+
+ if (!ok) {
+ while (--stop_idx >= 0) {
+ vhost_shadow_vq_stop_rcu(dev, dev->shadow_vqs[stop_idx]);
+ }
+
+ goto err;
+ }
+ }
+ }
+
+ dev->sw_lm_enabled = true;
+ return 0;
+
+err:
+ for (; idx >= 0; --idx) {
+ vhost_shadow_vq_free(dev->shadow_vqs[idx]);
+ }
+ g_free(dev->shadow_vqs[idx]);
+
+ return -1;
+}
+
+static int vhost_sw_live_migration_enable(struct vhost_dev *dev,
+ bool enable_lm)
+{
+ int r;
+
+ if (enable_lm == dev->sw_lm_enabled) {
+ return 0;
+ }
+
+ r = enable_lm ? vhost_sw_live_migration_start(dev)
+ : vhost_sw_live_migration_stop(dev);
+
+ return r;
+}
+
static void vhost_log_start(MemoryListener *listener,
MemoryRegionSection *section,
int old, int new)
@@ -1389,6 +1466,7 @@ int vhost_dev_init(struct vhost_dev *hdev, void *opaque,
hdev->log = NULL;
hdev->log_size = 0;
hdev->log_enabled = false;
+ hdev->sw_lm_enabled = false;
hdev->started = false;
memory_listener_register(&hdev->memory_listener, &address_space_memory);
QLIST_INSERT_HEAD(&vhost_devices, hdev, entry);
@@ -1816,6 +1894,11 @@ void vhost_dev_stop(struct vhost_dev *hdev, VirtIODevice
*vdev)
hdev->vhost_ops->vhost_dev_start(hdev, false);
}
for (i = 0; i < hdev->nvqs; ++i) {
+ if (hdev->sw_lm_enabled) {
+ vhost_shadow_vq_stop_rcu(hdev, hdev->shadow_vqs[i]);
+ vhost_shadow_vq_free(hdev->shadow_vqs[i]);
+ }
+
vhost_virtqueue_stop(hdev,
vdev,
hdev->vqs + i,
@@ -1829,6 +1912,8 @@ void vhost_dev_stop(struct vhost_dev *hdev, VirtIODevice
*vdev)
memory_listener_unregister(&hdev->iommu_listener);
}
vhost_log_put(hdev, true);
+ g_free(hdev->shadow_vqs);
+ hdev->sw_lm_enabled = false;
hdev->started = false;
hdev->vdev = NULL;
}
@@ -1845,5 +1930,53 @@ int vhost_net_set_backend(struct vhost_dev *hdev,
void qmp_x_vhost_enable_shadow_vq(const char *name, bool enable, Error
**errp)
{
- error_setg(errp, "Shadow virtqueue still not implemented.");
+ struct vhost_dev *hdev;
+ const char *err_cause = NULL;
+ const VirtioDeviceClass *k;
+ int r;
+ ErrorClass err_class = ERROR_CLASS_GENERIC_ERROR;
+
+ QLIST_FOREACH(hdev, &vhost_devices, entry) {
+ if (hdev->vdev && 0 == strcmp(hdev->vdev->name, name)) {
+ break;
+ }
+ }
+
+ if (!hdev) {
+ err_class = ERROR_CLASS_DEVICE_NOT_FOUND;
+ err_cause = "Device not found";
+ goto err;
+ }
+
+ if (!hdev->started) {
+ err_cause = "Device is not started";
+ goto err;
+ }
+
+ if (hdev->acked_features & BIT_ULL(VIRTIO_F_RING_PACKED)) {
+ err_cause = "Use packed vq";
+ goto err;
+ }
+
+ if (vhost_dev_has_iommu(hdev)) {
+ err_cause = "Device use IOMMU";
+ goto err;
+ }
+
+ k = VIRTIO_DEVICE_GET_CLASS(hdev->vdev);
+ if (!k->set_vq_handler) {
+ err_cause = "Virtio device type does not support reset of vq handler";
+ goto err;
+ }
+
+ r = vhost_sw_live_migration_enable(hdev, enable);
+ if (unlikely(r)) {
+ err_cause = "Error enabling (see monitor)";
+ }
+
+err:
+ if (err_cause) {
+ error_set(errp, err_class,
+ "Can't enable shadow vq on %s: %s", name, err_cause);
+ }
}