From: Eugenio Pérez <eperezma@redhat.com>
Next patches will register the vhost_vdpa memory listener while the VM
is migrating at the destination, so we can map the memory to the device
before stopping the VM at the source. The main goal is to reduce the
downtime.
However, the destination QEMU is unaware of which vhost_vdpa device will
register its memory_listener. If the source guest has CVQ enabled, it
will be the CVQ device. Otherwise, it will be the first one.
Move the iova range to VhostVDPAShared so all vhost_vdpa can use it,
rather than always in the first or last vhost_vdpa.
Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20231221174322.3130442-4-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
---
include/hw/virtio/vhost-vdpa.h | 3 ++-
hw/virtio/vdpa-dev.c | 5 ++++-
hw/virtio/vhost-vdpa.c | 16 ++++++++++------
net/vhost-vdpa.c | 10 +++++-----
4 files changed, 21 insertions(+), 13 deletions(-)
diff --git a/include/hw/virtio/vhost-vdpa.h b/include/hw/virtio/vhost-vdpa.h
index ac036055d3..8d52a7e498 100644
--- a/include/hw/virtio/vhost-vdpa.h
+++ b/include/hw/virtio/vhost-vdpa.h
@@ -32,6 +32,8 @@ typedef struct VhostVDPAHostNotifier {
/* Info shared by all vhost_vdpa device models */
typedef struct vhost_vdpa_shared {
+ struct vhost_vdpa_iova_range iova_range;
+
/* IOVA mapping used by the Shadow Virtqueue */
VhostIOVATree *iova_tree;
} VhostVDPAShared;
@@ -43,7 +45,6 @@ typedef struct vhost_vdpa {
bool iotlb_batch_begin_sent;
uint32_t address_space_id;
MemoryListener listener;
- struct vhost_vdpa_iova_range iova_range;
uint64_t acked_features;
bool shadow_vqs_enabled;
/* Vdpa must send shadow addresses as IOTLB key for data queues, not GPA */
diff --git a/hw/virtio/vdpa-dev.c b/hw/virtio/vdpa-dev.c
index f22d5d5bc0..457960d28a 100644
--- a/hw/virtio/vdpa-dev.c
+++ b/hw/virtio/vdpa-dev.c
@@ -114,7 +114,8 @@ static void vhost_vdpa_device_realize(DeviceState *dev, Error **errp)
strerror(-ret));
goto free_vqs;
}
- v->vdpa.iova_range = iova_range;
+ v->vdpa.shared = g_new0(VhostVDPAShared, 1);
+ v->vdpa.shared->iova_range = iova_range;
ret = vhost_dev_init(&v->dev, &v->vdpa, VHOST_BACKEND_TYPE_VDPA, 0, NULL);
if (ret < 0) {
@@ -162,6 +163,7 @@ vhost_cleanup:
vhost_dev_cleanup(&v->dev);
free_vqs:
g_free(vqs);
+ g_free(v->vdpa.shared);
out:
qemu_close(v->vhostfd);
v->vhostfd = -1;
@@ -184,6 +186,7 @@ static void vhost_vdpa_device_unrealize(DeviceState *dev)
g_free(s->config);
g_free(s->dev.vqs);
vhost_dev_cleanup(&s->dev);
+ g_free(s->vdpa.shared);
qemu_close(s->vhostfd);
s->vhostfd = -1;
}
diff --git a/hw/virtio/vhost-vdpa.c b/hw/virtio/vhost-vdpa.c
index 95a179a082..5ff1d43ba9 100644
--- a/hw/virtio/vhost-vdpa.c
+++ b/hw/virtio/vhost-vdpa.c
@@ -213,10 +213,10 @@ static void vhost_vdpa_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
RCU_READ_LOCK_GUARD();
/* check if RAM section out of device range */
llend = int128_add(int128_makes64(iotlb->addr_mask), int128_makes64(iova));
- if (int128_gt(llend, int128_make64(v->iova_range.last))) {
+ if (int128_gt(llend, int128_make64(v->shared->iova_range.last))) {
error_report("RAM section out of device range (max=0x%" PRIx64
", end addr=0x%" PRIx64 ")",
- v->iova_range.last, int128_get64(llend));
+ v->shared->iova_range.last, int128_get64(llend));
return;
}
@@ -316,8 +316,10 @@ static void vhost_vdpa_listener_region_add(MemoryListener *listener,
int page_size = qemu_target_page_size();
int page_mask = -page_size;
- if (vhost_vdpa_listener_skipped_section(section, v->iova_range.first,
- v->iova_range.last, page_mask)) {
+ if (vhost_vdpa_listener_skipped_section(section,
+ v->shared->iova_range.first,
+ v->shared->iova_range.last,
+ page_mask)) {
return;
}
if (memory_region_is_iommu(section->mr)) {
@@ -403,8 +405,10 @@ static void vhost_vdpa_listener_region_del(MemoryListener *listener,
int page_size = qemu_target_page_size();
int page_mask = -page_size;
- if (vhost_vdpa_listener_skipped_section(section, v->iova_range.first,
- v->iova_range.last, page_mask)) {
+ if (vhost_vdpa_listener_skipped_section(section,
+ v->shared->iova_range.first,
+ v->shared->iova_range.last,
+ page_mask)) {
return;
}
if (memory_region_is_iommu(section->mr)) {
diff --git a/net/vhost-vdpa.c b/net/vhost-vdpa.c
index 10703e5833..7be2c30ad3 100644
--- a/net/vhost-vdpa.c
+++ b/net/vhost-vdpa.c
@@ -354,8 +354,8 @@ static void vhost_vdpa_net_data_start_first(VhostVDPAState *s)
migration_add_notifier(&s->migration_state,
vdpa_net_migration_state_notifier);
if (v->shadow_vqs_enabled) {
- v->shared->iova_tree = vhost_iova_tree_new(v->iova_range.first,
- v->iova_range.last);
+ v->shared->iova_tree = vhost_iova_tree_new(v->shared->iova_range.first,
+ v->shared->iova_range.last);
}
}
@@ -591,8 +591,8 @@ out:
* and it is not worth it for the moment.
*/
if (!v->shared->iova_tree) {
- v->shared->iova_tree = vhost_iova_tree_new(v->iova_range.first,
- v->iova_range.last);
+ v->shared->iova_tree = vhost_iova_tree_new(v->shared->iova_range.first,
+ v->shared->iova_range.last);
}
r = vhost_vdpa_cvq_map_buf(&s->vhost_vdpa, s->cvq_cmd_out_buffer,
@@ -1688,12 +1688,12 @@ static NetClientState *net_vhost_vdpa_init(NetClientState *peer,
s->always_svq = svq;
s->migration_state.notify = NULL;
s->vhost_vdpa.shadow_vqs_enabled = svq;
- s->vhost_vdpa.iova_range = iova_range;
s->vhost_vdpa.shadow_data = svq;
if (queue_pair_index == 0) {
vhost_vdpa_net_valid_svq_features(features,
&s->vhost_vdpa.migration_blocker);
s->vhost_vdpa.shared = g_new0(VhostVDPAShared, 1);
+ s->vhost_vdpa.shared->iova_range = iova_range;
} else if (!is_datapath) {
s->cvq_cmd_out_buffer = mmap(NULL, vhost_vdpa_net_cvq_cmd_page_len(),
PROT_READ | PROT_WRITE,
--
MST