vdpa/mlx5: Implement susupend virtqueue callback

Implement the suspend callback allowing to suspend the virtqueues so
they stop processing descriptors. This is required to allow to query a
consistent state of the virtqueue while live migration is taking place.

Signed-off-by: Eli Cohen <elic@nvidia.com>
Message-Id: <20220714113927.85729-2-elic@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This commit is contained in:
Eli Cohen 2022-07-14 14:39:26 +03:00 committed by Michael S. Tsirkin
parent ad146355bf
commit cae15c2ed8
2 changed files with 89 additions and 4 deletions

View File

@ -164,6 +164,7 @@ struct mlx5_vdpa_net {
bool setup;
u32 cur_num_vqs;
u32 rqt_size;
bool nb_registered;
struct notifier_block nb;
struct vdpa_callback config_cb;
struct mlx5_vdpa_wq_ent cvq_ent;
@ -895,6 +896,7 @@ static int create_virtqueue(struct mlx5_vdpa_net *ndev, struct mlx5_vdpa_virtque
if (err)
goto err_cmd;
mvq->fw_state = MLX5_VIRTIO_NET_Q_OBJECT_STATE_INIT;
kfree(in);
mvq->virtq_id = MLX5_GET(general_obj_out_cmd_hdr, out, obj_id);
@ -922,6 +924,7 @@ static void destroy_virtqueue(struct mlx5_vdpa_net *ndev, struct mlx5_vdpa_virtq
mlx5_vdpa_warn(&ndev->mvdev, "destroy virtqueue 0x%x\n", mvq->virtq_id);
return;
}
mvq->fw_state = MLX5_VIRTIO_NET_Q_OBJECT_NONE;
umems_destroy(ndev, mvq);
}
@ -1121,6 +1124,20 @@ err_cmd:
return err;
}
static bool is_valid_state_change(int oldstate, int newstate)
{
switch (oldstate) {
case MLX5_VIRTIO_NET_Q_OBJECT_STATE_INIT:
return newstate == MLX5_VIRTIO_NET_Q_OBJECT_STATE_RDY;
case MLX5_VIRTIO_NET_Q_OBJECT_STATE_RDY:
return newstate == MLX5_VIRTIO_NET_Q_OBJECT_STATE_SUSPEND;
case MLX5_VIRTIO_NET_Q_OBJECT_STATE_SUSPEND:
case MLX5_VIRTIO_NET_Q_OBJECT_STATE_ERR:
default:
return false;
}
}
static int modify_virtqueue(struct mlx5_vdpa_net *ndev, struct mlx5_vdpa_virtqueue *mvq, int state)
{
int inlen = MLX5_ST_SZ_BYTES(modify_virtio_net_q_in);
@ -1130,6 +1147,12 @@ static int modify_virtqueue(struct mlx5_vdpa_net *ndev, struct mlx5_vdpa_virtque
void *in;
int err;
if (mvq->fw_state == MLX5_VIRTIO_NET_Q_OBJECT_NONE)
return 0;
if (!is_valid_state_change(mvq->fw_state, state))
return -EINVAL;
in = kzalloc(inlen, GFP_KERNEL);
if (!in)
return -ENOMEM;
@ -1992,6 +2015,7 @@ static void mlx5_vdpa_set_vq_ready(struct vdpa_device *vdev, u16 idx, bool ready
struct mlx5_vdpa_dev *mvdev = to_mvdev(vdev);
struct mlx5_vdpa_net *ndev = to_mlx5_vdpa_ndev(mvdev);
struct mlx5_vdpa_virtqueue *mvq;
int err;
if (!mvdev->actual_features)
return;
@ -2005,8 +2029,16 @@ static void mlx5_vdpa_set_vq_ready(struct vdpa_device *vdev, u16 idx, bool ready
}
mvq = &ndev->vqs[idx];
if (!ready)
if (!ready) {
suspend_vq(ndev, mvq);
} else {
err = modify_virtqueue(ndev, mvq, MLX5_VIRTIO_NET_Q_OBJECT_STATE_RDY);
if (err) {
mlx5_vdpa_warn(mvdev, "modify VQ %d to ready failed (%d)\n", idx, err);
ready = false;
}
}
mvq->ready = ready;
}
@ -2733,6 +2765,37 @@ out_err:
return err;
}
static void mlx5_vdpa_cvq_suspend(struct mlx5_vdpa_dev *mvdev)
{
struct mlx5_control_vq *cvq;
if (!(mvdev->actual_features & BIT_ULL(VIRTIO_NET_F_CTRL_VQ)))
return;
cvq = &mvdev->cvq;
cvq->ready = false;
}
static int mlx5_vdpa_suspend(struct vdpa_device *vdev)
{
struct mlx5_vdpa_dev *mvdev = to_mvdev(vdev);
struct mlx5_vdpa_net *ndev = to_mlx5_vdpa_ndev(mvdev);
struct mlx5_vdpa_virtqueue *mvq;
int i;
down_write(&ndev->reslock);
mlx5_notifier_unregister(mvdev->mdev, &ndev->nb);
ndev->nb_registered = false;
flush_workqueue(ndev->mvdev.wq);
for (i = 0; i < ndev->cur_num_vqs; i++) {
mvq = &ndev->vqs[i];
suspend_vq(ndev, mvq);
}
mlx5_vdpa_cvq_suspend(mvdev);
up_write(&ndev->reslock);
return 0;
}
static const struct vdpa_config_ops mlx5_vdpa_ops = {
.set_vq_address = mlx5_vdpa_set_vq_address,
.set_vq_num = mlx5_vdpa_set_vq_num,
@ -2763,6 +2826,7 @@ static const struct vdpa_config_ops mlx5_vdpa_ops = {
.get_generation = mlx5_vdpa_get_generation,
.set_map = mlx5_vdpa_set_map,
.free = mlx5_vdpa_free,
.suspend = mlx5_vdpa_suspend,
};
static int query_mtu(struct mlx5_core_dev *mdev, u16 *mtu)
@ -2828,6 +2892,7 @@ static void init_mvqs(struct mlx5_vdpa_net *ndev)
mvq->index = i;
mvq->ndev = ndev;
mvq->fwqp.fw = true;
mvq->fw_state = MLX5_VIRTIO_NET_Q_OBJECT_NONE;
}
for (; i < ndev->mvdev.max_vqs; i++) {
mvq = &ndev->vqs[i];
@ -2902,13 +2967,21 @@ static int event_handler(struct notifier_block *nb, unsigned long event, void *p
switch (eqe->sub_type) {
case MLX5_PORT_CHANGE_SUBTYPE_DOWN:
case MLX5_PORT_CHANGE_SUBTYPE_ACTIVE:
wqent = kzalloc(sizeof(*wqent), GFP_ATOMIC);
if (!wqent)
down_read(&ndev->reslock);
if (!ndev->nb_registered) {
up_read(&ndev->reslock);
return NOTIFY_DONE;
}
wqent = kzalloc(sizeof(*wqent), GFP_ATOMIC);
if (!wqent) {
up_read(&ndev->reslock);
return NOTIFY_DONE;
}
wqent->mvdev = &ndev->mvdev;
INIT_WORK(&wqent->work, update_carrier);
queue_work(ndev->mvdev.wq, &wqent->work);
up_read(&ndev->reslock);
ret = NOTIFY_OK;
break;
default:
@ -3062,6 +3135,7 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name,
ndev->nb.notifier_call = event_handler;
mlx5_notifier_register(mdev, &ndev->nb);
ndev->nb_registered = true;
mvdev->vdev.mdev = &mgtdev->mgtdev;
err = _vdpa_register_device(&mvdev->vdev, max_vqs + 1);
if (err)
@ -3093,7 +3167,10 @@ static void mlx5_vdpa_dev_del(struct vdpa_mgmt_dev *v_mdev, struct vdpa_device *
struct mlx5_vdpa_net *ndev = to_mlx5_vdpa_ndev(mvdev);
struct workqueue_struct *wq;
mlx5_notifier_unregister(mvdev->mdev, &ndev->nb);
if (ndev->nb_registered) {
mlx5_notifier_unregister(mvdev->mdev, &ndev->nb);
ndev->nb_registered = false;
}
wq = mvdev->wq;
mvdev->wq = NULL;
destroy_workqueue(wq);

View File

@ -150,6 +150,14 @@ enum {
MLX5_VIRTIO_NET_Q_OBJECT_STATE_ERR = 0x3,
};
/* This indicates that the object was not created or has already
* been desroyed. It is very safe to assume that this object will never
* have so many states
*/
enum {
MLX5_VIRTIO_NET_Q_OBJECT_NONE = 0xffffffff
};
enum {
MLX5_RQTC_LIST_Q_TYPE_RQ = 0x0,
MLX5_RQTC_LIST_Q_TYPE_VIRTIO_NET_Q = 0x1,