[PATCH V2 0/4]

[PATCH V2 0/4]

[Jason Wang]

*** BLURB HERE ***

Eli Cohen (1):
  virtio/vdpa: clear the virtqueue state during probe

Jason Wang (3):
  vdpa: support packed virtqueue for set/get_vq_state()
  virtio-pci library: introduce vp_modern_get_driver_features()
  vp_vdpa: allow set vq state to initial state after reset

 drivers/vdpa/ifcvf/ifcvf_main.c        |  4 +--
 drivers/vdpa/mlx5/net/mlx5_vnet.c      |  8 ++---
 drivers/vdpa/vdpa_sim/vdpa_sim.c       |  4 +--
 drivers/vdpa/virtio_pci/vp_vdpa.c      | 42 ++++++++++++++++++++++++--
 drivers/vhost/vdpa.c                   |  4 +--
 drivers/virtio/virtio_pci_modern_dev.c | 21 +++++++++++++
 drivers/virtio/virtio_vdpa.c           | 15 +++++++++
 include/linux/vdpa.h                   | 25 +++++++++++++--
 include/linux/virtio_pci_modern.h      |  1 +
 9 files changed, 109 insertions(+), 15 deletions(-)

-- 
2.25.1

[PATCH V2 0/4]

[Jason Wang]

*** BLURB HERE ***

Eli Cohen (1):
  virtio/vdpa: clear the virtqueue state during probe

Jason Wang (3):
  vdpa: support packed virtqueue for set/get_vq_state()
  virtio-pci library: introduce vp_modern_get_driver_features()
  vp_vdpa: allow set vq state to initial state after reset

 drivers/vdpa/ifcvf/ifcvf_main.c        |  4 +--
 drivers/vdpa/mlx5/net/mlx5_vnet.c      |  8 ++---
 drivers/vdpa/vdpa_sim/vdpa_sim.c       |  4 +--
 drivers/vdpa/virtio_pci/vp_vdpa.c      | 42 ++++++++++++++++++++++++--
 drivers/vhost/vdpa.c                   |  4 +--
 drivers/virtio/virtio_pci_modern_dev.c | 21 +++++++++++++
 drivers/virtio/virtio_vdpa.c           | 15 +++++++++
 include/linux/vdpa.h                   | 25 +++++++++++++--
 include/linux/virtio_pci_modern.h      |  1 +
 9 files changed, 109 insertions(+), 15 deletions(-)

-- 
2.25.1

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH V2 1/4] vdpa: support packed virtqueue for set/get_vq_state()

[Jason Wang]

This patch extends the vdpa_vq_state to support packed virtqueue
state which is basically the device/driver ring wrap counters and the
avail and used index. This will be used for the virito-vdpa support
for the packed virtqueue and the future vhost/vhost-vdpa support for
the packed virtqueue.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/vdpa/ifcvf/ifcvf_main.c   |  4 ++--
 drivers/vdpa/mlx5/net/mlx5_vnet.c |  8 ++++----
 drivers/vdpa/vdpa_sim/vdpa_sim.c  |  4 ++--
 drivers/vhost/vdpa.c              |  4 ++--
 include/linux/vdpa.h              | 25 +++++++++++++++++++++++--
 5 files changed, 33 insertions(+), 12 deletions(-)

diff --git a/drivers/vdpa/ifcvf/ifcvf_main.c b/drivers/vdpa/ifcvf/ifcvf_main.c
index ab0ab5cf0f6e..5d3891b1ca28 100644
--- a/drivers/vdpa/ifcvf/ifcvf_main.c
+++ b/drivers/vdpa/ifcvf/ifcvf_main.c
@@ -264,7 +264,7 @@ static int ifcvf_vdpa_get_vq_state(struct vdpa_device *vdpa_dev, u16 qid,
 {
 	struct ifcvf_hw *vf = vdpa_to_vf(vdpa_dev);
 
-	state->avail_index = ifcvf_get_vq_state(vf, qid);
+	state->split.avail_index = ifcvf_get_vq_state(vf, qid);
 	return 0;
 }
 
@@ -273,7 +273,7 @@ static int ifcvf_vdpa_set_vq_state(struct vdpa_device *vdpa_dev, u16 qid,
 {
 	struct ifcvf_hw *vf = vdpa_to_vf(vdpa_dev);
 
-	return ifcvf_set_vq_state(vf, qid, state->avail_index);
+	return ifcvf_set_vq_state(vf, qid, state->split.avail_index);
 }
 
 static void ifcvf_vdpa_set_vq_cb(struct vdpa_device *vdpa_dev, u16 qid,
diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
index 189e4385df40..e5505d760bca 100644
--- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
+++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
@@ -1427,8 +1427,8 @@ static int mlx5_vdpa_set_vq_state(struct vdpa_device *vdev, u16 idx,
 		return -EINVAL;
 	}
 
-	mvq->used_idx = state->avail_index;
-	mvq->avail_idx = state->avail_index;
+	mvq->used_idx = state->split.avail_index;
+	mvq->avail_idx = state->split.avail_index;
 	return 0;
 }
 
@@ -1449,7 +1449,7 @@ static int mlx5_vdpa_get_vq_state(struct vdpa_device *vdev, u16 idx, struct vdpa
 		 * Since both values should be identical, we take the value of
 		 * used_idx which is reported correctly.
 		 */
-		state->avail_index = mvq->used_idx;
+		state->split.avail_index = mvq->used_idx;
 		return 0;
 	}
 
@@ -1458,7 +1458,7 @@ static int mlx5_vdpa_get_vq_state(struct vdpa_device *vdev, u16 idx, struct vdpa
 		mlx5_vdpa_warn(mvdev, "failed to query virtqueue\n");
 		return err;
 	}
-	state->avail_index = attr.used_index;
+	state->split.avail_index = attr.used_index;
 	return 0;
 }
 
diff --git a/drivers/vdpa/vdpa_sim/vdpa_sim.c b/drivers/vdpa/vdpa_sim/vdpa_sim.c
index 98f793bc9376..14e024de5cbf 100644
--- a/drivers/vdpa/vdpa_sim/vdpa_sim.c
+++ b/drivers/vdpa/vdpa_sim/vdpa_sim.c
@@ -374,7 +374,7 @@ static int vdpasim_set_vq_state(struct vdpa_device *vdpa, u16 idx,
 	struct vringh *vrh = &vq->vring;
 
 	spin_lock(&vdpasim->lock);
-	vrh->last_avail_idx = state->avail_index;
+	vrh->last_avail_idx = state->split.avail_index;
 	spin_unlock(&vdpasim->lock);
 
 	return 0;
@@ -387,7 +387,7 @@ static int vdpasim_get_vq_state(struct vdpa_device *vdpa, u16 idx,
 	struct vdpasim_virtqueue *vq = &vdpasim->vqs[idx];
 	struct vringh *vrh = &vq->vring;
 
-	state->avail_index = vrh->last_avail_idx;
+	state->split.avail_index = vrh->last_avail_idx;
 	return 0;
 }
 
diff --git a/drivers/vhost/vdpa.c b/drivers/vhost/vdpa.c
index fb41db3da611..210ab35a7ebf 100644
--- a/drivers/vhost/vdpa.c
+++ b/drivers/vhost/vdpa.c
@@ -383,7 +383,7 @@ static long vhost_vdpa_vring_ioctl(struct vhost_vdpa *v, unsigned int cmd,
 		if (r)
 			return r;
 
-		vq->last_avail_idx = vq_state.avail_index;
+		vq->last_avail_idx = vq_state.split.avail_index;
 		break;
 	}
 
@@ -401,7 +401,7 @@ static long vhost_vdpa_vring_ioctl(struct vhost_vdpa *v, unsigned int cmd,
 		break;
 
 	case VHOST_SET_VRING_BASE:
-		vq_state.avail_index = vq->last_avail_idx;
+		vq_state.split.avail_index = vq->last_avail_idx;
 		if (ops->set_vq_state(vdpa, idx, &vq_state))
 			r = -EINVAL;
 		break;
diff --git a/include/linux/vdpa.h b/include/linux/vdpa.h
index f311d227aa1b..3357ac98878d 100644
--- a/include/linux/vdpa.h
+++ b/include/linux/vdpa.h
@@ -28,13 +28,34 @@ struct vdpa_notification_area {
 };
 
 /**
- * struct vdpa_vq_state - vDPA vq_state definition
+ * struct vdpa_vq_state_split - vDPA split virtqueue state
  * @avail_index: available index
  */
-struct vdpa_vq_state {
+struct vdpa_vq_state_split {
 	u16	avail_index;
 };
 
+/**
+ * struct vdpa_vq_state_packed - vDPA packed virtqueue state
+ * @last_avail_counter: last driver ring wrap counter observed by device
+ * @last_avail_idx: device available index
+ * @last_used_counter: device ring wrap counter
+ * @last_used_idx: used index
+ */
+struct vdpa_vq_state_packed {
+        u16	last_avail_counter:1;
+        u16	last_avail_idx:15;
+        u16	last_used_counter:1;
+        u16	last_used_idx:15;
+};
+
+struct vdpa_vq_state {
+     union {
+          struct vdpa_vq_state_split split;
+          struct vdpa_vq_state_packed packed;
+     };
+};
+
 struct vdpa_mgmt_dev;
 
 /**
-- 
2.25.1

[PATCH V2 1/4] vdpa: support packed virtqueue for set/get_vq_state()

[Jason Wang]

This patch extends the vdpa_vq_state to support packed virtqueue
state which is basically the device/driver ring wrap counters and the
avail and used index. This will be used for the virito-vdpa support
for the packed virtqueue and the future vhost/vhost-vdpa support for
the packed virtqueue.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/vdpa/ifcvf/ifcvf_main.c   |  4 ++--
 drivers/vdpa/mlx5/net/mlx5_vnet.c |  8 ++++----
 drivers/vdpa/vdpa_sim/vdpa_sim.c  |  4 ++--
 drivers/vhost/vdpa.c              |  4 ++--
 include/linux/vdpa.h              | 25 +++++++++++++++++++++++--
 5 files changed, 33 insertions(+), 12 deletions(-)

diff --git a/drivers/vdpa/ifcvf/ifcvf_main.c b/drivers/vdpa/ifcvf/ifcvf_main.c
index ab0ab5cf0f6e..5d3891b1ca28 100644
--- a/drivers/vdpa/ifcvf/ifcvf_main.c
+++ b/drivers/vdpa/ifcvf/ifcvf_main.c
@@ -264,7 +264,7 @@ static int ifcvf_vdpa_get_vq_state(struct vdpa_device *vdpa_dev, u16 qid,
 {
 	struct ifcvf_hw *vf = vdpa_to_vf(vdpa_dev);
 
-	state->avail_index = ifcvf_get_vq_state(vf, qid);
+	state->split.avail_index = ifcvf_get_vq_state(vf, qid);
 	return 0;
 }
 
@@ -273,7 +273,7 @@ static int ifcvf_vdpa_set_vq_state(struct vdpa_device *vdpa_dev, u16 qid,
 {
 	struct ifcvf_hw *vf = vdpa_to_vf(vdpa_dev);
 
-	return ifcvf_set_vq_state(vf, qid, state->avail_index);
+	return ifcvf_set_vq_state(vf, qid, state->split.avail_index);
 }
 
 static void ifcvf_vdpa_set_vq_cb(struct vdpa_device *vdpa_dev, u16 qid,
diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
index 189e4385df40..e5505d760bca 100644
--- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
+++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
@@ -1427,8 +1427,8 @@ static int mlx5_vdpa_set_vq_state(struct vdpa_device *vdev, u16 idx,
 		return -EINVAL;
 	}
 
-	mvq->used_idx = state->avail_index;
-	mvq->avail_idx = state->avail_index;
+	mvq->used_idx = state->split.avail_index;
+	mvq->avail_idx = state->split.avail_index;
 	return 0;
 }
 
@@ -1449,7 +1449,7 @@ static int mlx5_vdpa_get_vq_state(struct vdpa_device *vdev, u16 idx, struct vdpa
 		 * Since both values should be identical, we take the value of
 		 * used_idx which is reported correctly.
 		 */
-		state->avail_index = mvq->used_idx;
+		state->split.avail_index = mvq->used_idx;
 		return 0;
 	}
 
@@ -1458,7 +1458,7 @@ static int mlx5_vdpa_get_vq_state(struct vdpa_device *vdev, u16 idx, struct vdpa
 		mlx5_vdpa_warn(mvdev, "failed to query virtqueue\n");
 		return err;
 	}
-	state->avail_index = attr.used_index;
+	state->split.avail_index = attr.used_index;
 	return 0;
 }
 
diff --git a/drivers/vdpa/vdpa_sim/vdpa_sim.c b/drivers/vdpa/vdpa_sim/vdpa_sim.c
index 98f793bc9376..14e024de5cbf 100644
--- a/drivers/vdpa/vdpa_sim/vdpa_sim.c
+++ b/drivers/vdpa/vdpa_sim/vdpa_sim.c
@@ -374,7 +374,7 @@ static int vdpasim_set_vq_state(struct vdpa_device *vdpa, u16 idx,
 	struct vringh *vrh = &vq->vring;
 
 	spin_lock(&vdpasim->lock);
-	vrh->last_avail_idx = state->avail_index;
+	vrh->last_avail_idx = state->split.avail_index;
 	spin_unlock(&vdpasim->lock);
 
 	return 0;
@@ -387,7 +387,7 @@ static int vdpasim_get_vq_state(struct vdpa_device *vdpa, u16 idx,
 	struct vdpasim_virtqueue *vq = &vdpasim->vqs[idx];
 	struct vringh *vrh = &vq->vring;
 
-	state->avail_index = vrh->last_avail_idx;
+	state->split.avail_index = vrh->last_avail_idx;
 	return 0;
 }
 
diff --git a/drivers/vhost/vdpa.c b/drivers/vhost/vdpa.c
index fb41db3da611..210ab35a7ebf 100644
--- a/drivers/vhost/vdpa.c
+++ b/drivers/vhost/vdpa.c
@@ -383,7 +383,7 @@ static long vhost_vdpa_vring_ioctl(struct vhost_vdpa *v, unsigned int cmd,
 		if (r)
 			return r;
 
-		vq->last_avail_idx = vq_state.avail_index;
+		vq->last_avail_idx = vq_state.split.avail_index;
 		break;
 	}
 
@@ -401,7 +401,7 @@ static long vhost_vdpa_vring_ioctl(struct vhost_vdpa *v, unsigned int cmd,
 		break;
 
 	case VHOST_SET_VRING_BASE:
-		vq_state.avail_index = vq->last_avail_idx;
+		vq_state.split.avail_index = vq->last_avail_idx;
 		if (ops->set_vq_state(vdpa, idx, &vq_state))
 			r = -EINVAL;
 		break;
diff --git a/include/linux/vdpa.h b/include/linux/vdpa.h
index f311d227aa1b..3357ac98878d 100644
--- a/include/linux/vdpa.h
+++ b/include/linux/vdpa.h
@@ -28,13 +28,34 @@ struct vdpa_notification_area {
 };
 
 /**
- * struct vdpa_vq_state - vDPA vq_state definition
+ * struct vdpa_vq_state_split - vDPA split virtqueue state
  * @avail_index: available index
  */
-struct vdpa_vq_state {
+struct vdpa_vq_state_split {
 	u16	avail_index;
 };
 
+/**
+ * struct vdpa_vq_state_packed - vDPA packed virtqueue state
+ * @last_avail_counter: last driver ring wrap counter observed by device
+ * @last_avail_idx: device available index
+ * @last_used_counter: device ring wrap counter
+ * @last_used_idx: used index
+ */
+struct vdpa_vq_state_packed {
+        u16	last_avail_counter:1;
+        u16	last_avail_idx:15;
+        u16	last_used_counter:1;
+        u16	last_used_idx:15;
+};
+
+struct vdpa_vq_state {
+     union {
+          struct vdpa_vq_state_split split;
+          struct vdpa_vq_state_packed packed;
+     };
+};
+
 struct vdpa_mgmt_dev;
 
 /**
-- 
2.25.1

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH V2 2/4] virtio-pci library: introduce vp_modern_get_driver_features()

[Jason Wang]

This patch introduce a helper to get driver/guest features from the
device.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/virtio/virtio_pci_modern_dev.c | 21 +++++++++++++++++++++
 include/linux/virtio_pci_modern.h      |  1 +
 2 files changed, 22 insertions(+)

diff --git a/drivers/virtio/virtio_pci_modern_dev.c b/drivers/virtio/virtio_pci_modern_dev.c
index 54f297028586..e11ed748e661 100644
--- a/drivers/virtio/virtio_pci_modern_dev.c
+++ b/drivers/virtio/virtio_pci_modern_dev.c
@@ -383,6 +383,27 @@ u64 vp_modern_get_features(struct virtio_pci_modern_device *mdev)
 }
 EXPORT_SYMBOL_GPL(vp_modern_get_features);
 
+/*
+ * vp_modern_get_driver_features - get driver features from device
+ * @mdev: the modern virtio-pci device
+ *
+ * Returns the driver features read from the device
+ */
+u64 vp_modern_get_driver_features(struct virtio_pci_modern_device *mdev)
+{
+	struct virtio_pci_common_cfg __iomem *cfg = mdev->common;
+
+	u64 features;
+
+	vp_iowrite32(0, &cfg->guest_feature_select);
+	features = vp_ioread32(&cfg->guest_feature);
+	vp_iowrite32(1, &cfg->guest_feature_select);
+	features |= ((u64)vp_ioread32(&cfg->guest_feature) << 32);
+
+	return features;
+}
+EXPORT_SYMBOL_GPL(vp_modern_get_driver_features);
+
 /*
  * vp_modern_set_features - set features to device
  * @mdev: the modern virtio-pci device
diff --git a/include/linux/virtio_pci_modern.h b/include/linux/virtio_pci_modern.h
index 6a95b58fd0f4..eb2bd9b4077d 100644
--- a/include/linux/virtio_pci_modern.h
+++ b/include/linux/virtio_pci_modern.h
@@ -79,6 +79,7 @@ static inline void vp_iowrite64_twopart(u64 val,
 }
 
 u64 vp_modern_get_features(struct virtio_pci_modern_device *mdev);
+u64 vp_modern_get_driver_features(struct virtio_pci_modern_device *mdev);
 void vp_modern_set_features(struct virtio_pci_modern_device *mdev,
 		     u64 features);
 u32 vp_modern_generation(struct virtio_pci_modern_device *mdev);
-- 
2.25.1

[PATCH V2 2/4] virtio-pci library: introduce vp_modern_get_driver_features()

[Jason Wang]

This patch introduce a helper to get driver/guest features from the
device.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/virtio/virtio_pci_modern_dev.c | 21 +++++++++++++++++++++
 include/linux/virtio_pci_modern.h      |  1 +
 2 files changed, 22 insertions(+)

diff --git a/drivers/virtio/virtio_pci_modern_dev.c b/drivers/virtio/virtio_pci_modern_dev.c
index 54f297028586..e11ed748e661 100644
--- a/drivers/virtio/virtio_pci_modern_dev.c
+++ b/drivers/virtio/virtio_pci_modern_dev.c
@@ -383,6 +383,27 @@ u64 vp_modern_get_features(struct virtio_pci_modern_device *mdev)
 }
 EXPORT_SYMBOL_GPL(vp_modern_get_features);
 
+/*
+ * vp_modern_get_driver_features - get driver features from device
+ * @mdev: the modern virtio-pci device
+ *
+ * Returns the driver features read from the device
+ */
+u64 vp_modern_get_driver_features(struct virtio_pci_modern_device *mdev)
+{
+	struct virtio_pci_common_cfg __iomem *cfg = mdev->common;
+
+	u64 features;
+
+	vp_iowrite32(0, &cfg->guest_feature_select);
+	features = vp_ioread32(&cfg->guest_feature);
+	vp_iowrite32(1, &cfg->guest_feature_select);
+	features |= ((u64)vp_ioread32(&cfg->guest_feature) << 32);
+
+	return features;
+}
+EXPORT_SYMBOL_GPL(vp_modern_get_driver_features);
+
 /*
  * vp_modern_set_features - set features to device
  * @mdev: the modern virtio-pci device
diff --git a/include/linux/virtio_pci_modern.h b/include/linux/virtio_pci_modern.h
index 6a95b58fd0f4..eb2bd9b4077d 100644
--- a/include/linux/virtio_pci_modern.h
+++ b/include/linux/virtio_pci_modern.h
@@ -79,6 +79,7 @@ static inline void vp_iowrite64_twopart(u64 val,
 }
 
 u64 vp_modern_get_features(struct virtio_pci_modern_device *mdev);
+u64 vp_modern_get_driver_features(struct virtio_pci_modern_device *mdev);
 void vp_modern_set_features(struct virtio_pci_modern_device *mdev,
 		     u64 features);
 u32 vp_modern_generation(struct virtio_pci_modern_device *mdev);
-- 
2.25.1

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH V2 3/4] vp_vdpa: allow set vq state to initial state after reset

[Jason Wang]

We used to fail the set_vq_state() since it was not supported yet by
the virtio spec. But if the bus tries to set the state which is equal
to the device initial state after reset, we can let it go.

This is a must for virtio_vdpa() to set vq state during probe which is
required for some vDPA parents.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/vdpa/virtio_pci/vp_vdpa.c | 42 ++++++++++++++++++++++++++++---
 1 file changed, 39 insertions(+), 3 deletions(-)

diff --git a/drivers/vdpa/virtio_pci/vp_vdpa.c b/drivers/vdpa/virtio_pci/vp_vdpa.c
index c76ebb531212..18bf4a422772 100644
--- a/drivers/vdpa/virtio_pci/vp_vdpa.c
+++ b/drivers/vdpa/virtio_pci/vp_vdpa.c
@@ -210,13 +210,49 @@ static int vp_vdpa_get_vq_state(struct vdpa_device *vdpa, u16 qid,
 	return -EOPNOTSUPP;
 }
 
+static int vp_vdpa_set_vq_state_split(struct vdpa_device *vdpa,
+				      const struct vdpa_vq_state *state)
+{
+	const struct vdpa_vq_state_split *split = &state->split;
+
+	if (split->avail_index == 0)
+		return 0;
+
+	return -EOPNOTSUPP;
+}
+
+static int vp_vdpa_set_vq_state_packed(struct vdpa_device *vdpa,
+				       const struct vdpa_vq_state *state)
+{
+	const struct vdpa_vq_state_packed *packed = &state->packed;
+
+	if (packed->last_avail_counter == 1 &&
+	    packed->last_avail_idx == 0 &&
+	    packed->last_used_counter == 1 &&
+	    packed->last_used_idx == 0)
+		return 0;
+
+	return -EOPNOTSUPP;
+}
+
 static int vp_vdpa_set_vq_state(struct vdpa_device *vdpa, u16 qid,
 				const struct vdpa_vq_state *state)
 {
-	/* Note that this is not supported by virtio specification, so
-	 * we return -ENOPOTSUPP here. This means we can't support live
-	 * migration, vhost device start/stop.
+	struct virtio_pci_modern_device *mdev = vdpa_to_mdev(vdpa);
+
+	/* Note that this is not supported by virtio specification.
+	 * But if the state is by chance equal to the device initial
+	 * state, we can let it go.
 	 */
+	if ((vp_modern_get_status(mdev) & VIRTIO_CONFIG_S_FEATURES_OK) &&
+	    !vp_modern_get_queue_enable(mdev, qid)) {
+		if (vp_modern_get_driver_features(mdev) &
+		    BIT_ULL(VIRTIO_F_RING_PACKED))
+			return vp_vdpa_set_vq_state_packed(vdpa, state);
+		else
+			return vp_vdpa_set_vq_state_split(vdpa,	state);
+	}
+
 	return -EOPNOTSUPP;
 }
 
-- 
2.25.1

[PATCH V2 3/4] vp_vdpa: allow set vq state to initial state after reset

[Jason Wang]

We used to fail the set_vq_state() since it was not supported yet by
the virtio spec. But if the bus tries to set the state which is equal
to the device initial state after reset, we can let it go.

This is a must for virtio_vdpa() to set vq state during probe which is
required for some vDPA parents.

Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/vdpa/virtio_pci/vp_vdpa.c | 42 ++++++++++++++++++++++++++++---
 1 file changed, 39 insertions(+), 3 deletions(-)

diff --git a/drivers/vdpa/virtio_pci/vp_vdpa.c b/drivers/vdpa/virtio_pci/vp_vdpa.c
index c76ebb531212..18bf4a422772 100644
--- a/drivers/vdpa/virtio_pci/vp_vdpa.c
+++ b/drivers/vdpa/virtio_pci/vp_vdpa.c
@@ -210,13 +210,49 @@ static int vp_vdpa_get_vq_state(struct vdpa_device *vdpa, u16 qid,
 	return -EOPNOTSUPP;
 }
 
+static int vp_vdpa_set_vq_state_split(struct vdpa_device *vdpa,
+				      const struct vdpa_vq_state *state)
+{
+	const struct vdpa_vq_state_split *split = &state->split;
+
+	if (split->avail_index == 0)
+		return 0;
+
+	return -EOPNOTSUPP;
+}
+
+static int vp_vdpa_set_vq_state_packed(struct vdpa_device *vdpa,
+				       const struct vdpa_vq_state *state)
+{
+	const struct vdpa_vq_state_packed *packed = &state->packed;
+
+	if (packed->last_avail_counter == 1 &&
+	    packed->last_avail_idx == 0 &&
+	    packed->last_used_counter == 1 &&
+	    packed->last_used_idx == 0)
+		return 0;
+
+	return -EOPNOTSUPP;
+}
+
 static int vp_vdpa_set_vq_state(struct vdpa_device *vdpa, u16 qid,
 				const struct vdpa_vq_state *state)
 {
-	/* Note that this is not supported by virtio specification, so
-	 * we return -ENOPOTSUPP here. This means we can't support live
-	 * migration, vhost device start/stop.
+	struct virtio_pci_modern_device *mdev = vdpa_to_mdev(vdpa);
+
+	/* Note that this is not supported by virtio specification.
+	 * But if the state is by chance equal to the device initial
+	 * state, we can let it go.
 	 */
+	if ((vp_modern_get_status(mdev) & VIRTIO_CONFIG_S_FEATURES_OK) &&
+	    !vp_modern_get_queue_enable(mdev, qid)) {
+		if (vp_modern_get_driver_features(mdev) &
+		    BIT_ULL(VIRTIO_F_RING_PACKED))
+			return vp_vdpa_set_vq_state_packed(vdpa, state);
+		else
+			return vp_vdpa_set_vq_state_split(vdpa,	state);
+	}
+
 	return -EOPNOTSUPP;
 }
 
-- 
2.25.1

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH V2 3/4] vp_vdpa: allow set vq state to initial state after reset

[Eli Cohen]

On Wed, Jun 02, 2021 at 10:10:42AM +0800, Jason Wang wrote:
> We used to fail the set_vq_state() since it was not supported yet by
> the virtio spec. But if the bus tries to set the state which is equal
> to the device initial state after reset, we can let it go.
> 
> This is a must for virtio_vdpa() to set vq state during probe which is
> required for some vDPA parents.
> 
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>  drivers/vdpa/virtio_pci/vp_vdpa.c | 42 ++++++++++++++++++++++++++++---
>  1 file changed, 39 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/vdpa/virtio_pci/vp_vdpa.c b/drivers/vdpa/virtio_pci/vp_vdpa.c
> index c76ebb531212..18bf4a422772 100644
> --- a/drivers/vdpa/virtio_pci/vp_vdpa.c
> +++ b/drivers/vdpa/virtio_pci/vp_vdpa.c
> @@ -210,13 +210,49 @@ static int vp_vdpa_get_vq_state(struct vdpa_device *vdpa, u16 qid,
>  	return -EOPNOTSUPP;
>  }
>  
> +static int vp_vdpa_set_vq_state_split(struct vdpa_device *vdpa,
> +				      const struct vdpa_vq_state *state)
> +{
> +	const struct vdpa_vq_state_split *split = &state->split;
> +
> +	if (split->avail_index == 0)
> +		return 0;
> +
> +	return -EOPNOTSUPP;
> +}
> +
> +static int vp_vdpa_set_vq_state_packed(struct vdpa_device *vdpa,
> +				       const struct vdpa_vq_state *state)
> +{
> +	const struct vdpa_vq_state_packed *packed = &state->packed;
> +
> +	if (packed->last_avail_counter == 1 &&
Can you elaborate on the requirement on last_avail_counter and
last_used_counter?

> +	    packed->last_avail_idx == 0 &&
> +	    packed->last_used_counter == 1 &&
> +	    packed->last_used_idx == 0)
> +		return 0;
> +
> +	return -EOPNOTSUPP;
> +}
> +
>  static int vp_vdpa_set_vq_state(struct vdpa_device *vdpa, u16 qid,
>  				const struct vdpa_vq_state *state)
>  {
> -	/* Note that this is not supported by virtio specification, so
> -	 * we return -ENOPOTSUPP here. This means we can't support live
> -	 * migration, vhost device start/stop.
> +	struct virtio_pci_modern_device *mdev = vdpa_to_mdev(vdpa);
> +
> +	/* Note that this is not supported by virtio specification.
> +	 * But if the state is by chance equal to the device initial
> +	 * state, we can let it go.
>  	 */
> +	if ((vp_modern_get_status(mdev) & VIRTIO_CONFIG_S_FEATURES_OK) &&
> +	    !vp_modern_get_queue_enable(mdev, qid)) {
> +		if (vp_modern_get_driver_features(mdev) &
> +		    BIT_ULL(VIRTIO_F_RING_PACKED))
> +			return vp_vdpa_set_vq_state_packed(vdpa, state);
> +		else
> +			return vp_vdpa_set_vq_state_split(vdpa,	state);
> +	}
> +
>  	return -EOPNOTSUPP;
>  }
>  
> -- 
> 2.25.1
> 

Re: [PATCH V2 3/4] vp_vdpa: allow set vq state to initial state after reset

[Jason Wang]

在 2021/6/2 下午2:13, Eli Cohen 写道:
> On Wed, Jun 02, 2021 at 10:10:42AM +0800, Jason Wang wrote:
>> We used to fail the set_vq_state() since it was not supported yet by
>> the virtio spec. But if the bus tries to set the state which is equal
>> to the device initial state after reset, we can let it go.
>>
>> This is a must for virtio_vdpa() to set vq state during probe which is
>> required for some vDPA parents.
>>
>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>> ---
>>   drivers/vdpa/virtio_pci/vp_vdpa.c | 42 ++++++++++++++++++++++++++++---
>>   1 file changed, 39 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/vdpa/virtio_pci/vp_vdpa.c b/drivers/vdpa/virtio_pci/vp_vdpa.c
>> index c76ebb531212..18bf4a422772 100644
>> --- a/drivers/vdpa/virtio_pci/vp_vdpa.c
>> +++ b/drivers/vdpa/virtio_pci/vp_vdpa.c
>> @@ -210,13 +210,49 @@ static int vp_vdpa_get_vq_state(struct vdpa_device *vdpa, u16 qid,
>>   	return -EOPNOTSUPP;
>>   }
>>   
>> +static int vp_vdpa_set_vq_state_split(struct vdpa_device *vdpa,
>> +				      const struct vdpa_vq_state *state)
>> +{
>> +	const struct vdpa_vq_state_split *split = &state->split;
>> +
>> +	if (split->avail_index == 0)
>> +		return 0;
>> +
>> +	return -EOPNOTSUPP;
>> +}
>> +
>> +static int vp_vdpa_set_vq_state_packed(struct vdpa_device *vdpa,
>> +				       const struct vdpa_vq_state *state)
>> +{
>> +	const struct vdpa_vq_state_packed *packed = &state->packed;
>> +
>> +	if (packed->last_avail_counter == 1 &&
> Can you elaborate on the requirement on last_avail_counter and
> last_used_counter?


This is required by the virtio spec:

"
2.7.1 Driver and Device Ring Wrap Counters
Each of the driver and the device are expected to maintain, internally, 
a single-bit ring wrap counter initialized to 1.
"

For virtio-pci device, since there's no way to assign the value of those 
counters, the counters will be reset to 1 after reset, otherwise the 
driver can't work.

Thanks


>
>> +	    packed->last_avail_idx == 0 &&
>> +	    packed->last_used_counter == 1 &&
>> +	    packed->last_used_idx == 0)
>> +		return 0;
>> +
>> +	return -EOPNOTSUPP;
>> +}
>> +
>>   static int vp_vdpa_set_vq_state(struct vdpa_device *vdpa, u16 qid,
>>   				const struct vdpa_vq_state *state)
>>   {
>> -	/* Note that this is not supported by virtio specification, so
>> -	 * we return -ENOPOTSUPP here. This means we can't support live
>> -	 * migration, vhost device start/stop.
>> +	struct virtio_pci_modern_device *mdev = vdpa_to_mdev(vdpa);
>> +
>> +	/* Note that this is not supported by virtio specification.
>> +	 * But if the state is by chance equal to the device initial
>> +	 * state, we can let it go.
>>   	 */
>> +	if ((vp_modern_get_status(mdev) & VIRTIO_CONFIG_S_FEATURES_OK) &&
>> +	    !vp_modern_get_queue_enable(mdev, qid)) {
>> +		if (vp_modern_get_driver_features(mdev) &
>> +		    BIT_ULL(VIRTIO_F_RING_PACKED))
>> +			return vp_vdpa_set_vq_state_packed(vdpa, state);
>> +		else
>> +			return vp_vdpa_set_vq_state_split(vdpa,	state);
>> +	}
>> +
>>   	return -EOPNOTSUPP;
>>   }
>>   
>> -- 
>> 2.25.1
>>

Re: [PATCH V2 3/4] vp_vdpa: allow set vq state to initial state after reset

[Jason Wang]

在 2021/6/2 下午2:13, Eli Cohen 写道:
> On Wed, Jun 02, 2021 at 10:10:42AM +0800, Jason Wang wrote:
>> We used to fail the set_vq_state() since it was not supported yet by
>> the virtio spec. But if the bus tries to set the state which is equal
>> to the device initial state after reset, we can let it go.
>>
>> This is a must for virtio_vdpa() to set vq state during probe which is
>> required for some vDPA parents.
>>
>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>> ---
>>   drivers/vdpa/virtio_pci/vp_vdpa.c | 42 ++++++++++++++++++++++++++++---
>>   1 file changed, 39 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/vdpa/virtio_pci/vp_vdpa.c b/drivers/vdpa/virtio_pci/vp_vdpa.c
>> index c76ebb531212..18bf4a422772 100644
>> --- a/drivers/vdpa/virtio_pci/vp_vdpa.c
>> +++ b/drivers/vdpa/virtio_pci/vp_vdpa.c
>> @@ -210,13 +210,49 @@ static int vp_vdpa_get_vq_state(struct vdpa_device *vdpa, u16 qid,
>>   	return -EOPNOTSUPP;
>>   }
>>   
>> +static int vp_vdpa_set_vq_state_split(struct vdpa_device *vdpa,
>> +				      const struct vdpa_vq_state *state)
>> +{
>> +	const struct vdpa_vq_state_split *split = &state->split;
>> +
>> +	if (split->avail_index == 0)
>> +		return 0;
>> +
>> +	return -EOPNOTSUPP;
>> +}
>> +
>> +static int vp_vdpa_set_vq_state_packed(struct vdpa_device *vdpa,
>> +				       const struct vdpa_vq_state *state)
>> +{
>> +	const struct vdpa_vq_state_packed *packed = &state->packed;
>> +
>> +	if (packed->last_avail_counter == 1 &&
> Can you elaborate on the requirement on last_avail_counter and
> last_used_counter?


This is required by the virtio spec:

"
2.7.1 Driver and Device Ring Wrap Counters
Each of the driver and the device are expected to maintain, internally, 
a single-bit ring wrap counter initialized to 1.
"

For virtio-pci device, since there's no way to assign the value of those 
counters, the counters will be reset to 1 after reset, otherwise the 
driver can't work.

Thanks


>
>> +	    packed->last_avail_idx == 0 &&
>> +	    packed->last_used_counter == 1 &&
>> +	    packed->last_used_idx == 0)
>> +		return 0;
>> +
>> +	return -EOPNOTSUPP;
>> +}
>> +
>>   static int vp_vdpa_set_vq_state(struct vdpa_device *vdpa, u16 qid,
>>   				const struct vdpa_vq_state *state)
>>   {
>> -	/* Note that this is not supported by virtio specification, so
>> -	 * we return -ENOPOTSUPP here. This means we can't support live
>> -	 * migration, vhost device start/stop.
>> +	struct virtio_pci_modern_device *mdev = vdpa_to_mdev(vdpa);
>> +
>> +	/* Note that this is not supported by virtio specification.
>> +	 * But if the state is by chance equal to the device initial
>> +	 * state, we can let it go.
>>   	 */
>> +	if ((vp_modern_get_status(mdev) & VIRTIO_CONFIG_S_FEATURES_OK) &&
>> +	    !vp_modern_get_queue_enable(mdev, qid)) {
>> +		if (vp_modern_get_driver_features(mdev) &
>> +		    BIT_ULL(VIRTIO_F_RING_PACKED))
>> +			return vp_vdpa_set_vq_state_packed(vdpa, state);
>> +		else
>> +			return vp_vdpa_set_vq_state_split(vdpa,	state);
>> +	}
>> +
>>   	return -EOPNOTSUPP;
>>   }
>>   
>> -- 
>> 2.25.1
>>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH V2 3/4] vp_vdpa: allow set vq state to initial state after reset

[Eli Cohen]

On Wed, Jun 02, 2021 at 03:05:35PM +0800, Jason Wang wrote:
> 
> 在 2021/6/2 下午2:13, Eli Cohen 写道:
> > On Wed, Jun 02, 2021 at 10:10:42AM +0800, Jason Wang wrote:
> > > We used to fail the set_vq_state() since it was not supported yet by
> > > the virtio spec. But if the bus tries to set the state which is equal
> > > to the device initial state after reset, we can let it go.
> > > 
> > > This is a must for virtio_vdpa() to set vq state during probe which is
> > > required for some vDPA parents.
> > > 
> > > Signed-off-by: Jason Wang <jasowang@redhat.com>
> > > ---
> > >   drivers/vdpa/virtio_pci/vp_vdpa.c | 42 ++++++++++++++++++++++++++++---
> > >   1 file changed, 39 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/drivers/vdpa/virtio_pci/vp_vdpa.c b/drivers/vdpa/virtio_pci/vp_vdpa.c
> > > index c76ebb531212..18bf4a422772 100644
> > > --- a/drivers/vdpa/virtio_pci/vp_vdpa.c
> > > +++ b/drivers/vdpa/virtio_pci/vp_vdpa.c
> > > @@ -210,13 +210,49 @@ static int vp_vdpa_get_vq_state(struct vdpa_device *vdpa, u16 qid,
> > >   	return -EOPNOTSUPP;
> > >   }
> > > +static int vp_vdpa_set_vq_state_split(struct vdpa_device *vdpa,
> > > +				      const struct vdpa_vq_state *state)
> > > +{
> > > +	const struct vdpa_vq_state_split *split = &state->split;
> > > +
> > > +	if (split->avail_index == 0)
> > > +		return 0;
> > > +
> > > +	return -EOPNOTSUPP;
> > > +}
> > > +
> > > +static int vp_vdpa_set_vq_state_packed(struct vdpa_device *vdpa,
> > > +				       const struct vdpa_vq_state *state)
> > > +{
> > > +	const struct vdpa_vq_state_packed *packed = &state->packed;
> > > +
> > > +	if (packed->last_avail_counter == 1 &&
> > Can you elaborate on the requirement on last_avail_counter and
> > last_used_counter?
> 
> 
> This is required by the virtio spec:
> 
> "
> 2.7.1 Driver and Device Ring Wrap Counters
> Each of the driver and the device are expected to maintain, internally, a
> single-bit ring wrap counter initialized to 1.
> "
> 
> For virtio-pci device, since there's no way to assign the value of those
> counters, the counters will be reset to 1 after reset, otherwise the driver
> can't work.

I see, thanks for the explanation.

> 
> Thanks
> 
> 
> > 
> > > +	    packed->last_avail_idx == 0 &&
> > > +	    packed->last_used_counter == 1 &&
> > > +	    packed->last_used_idx == 0)
> > > +		return 0;
> > > +
> > > +	return -EOPNOTSUPP;
> > > +}
> > > +
> > >   static int vp_vdpa_set_vq_state(struct vdpa_device *vdpa, u16 qid,
> > >   				const struct vdpa_vq_state *state)
> > >   {
> > > -	/* Note that this is not supported by virtio specification, so
> > > -	 * we return -ENOPOTSUPP here. This means we can't support live
> > > -	 * migration, vhost device start/stop.
> > > +	struct virtio_pci_modern_device *mdev = vdpa_to_mdev(vdpa);
> > > +
> > > +	/* Note that this is not supported by virtio specification.
> > > +	 * But if the state is by chance equal to the device initial
> > > +	 * state, we can let it go.
> > >   	 */
> > > +	if ((vp_modern_get_status(mdev) & VIRTIO_CONFIG_S_FEATURES_OK) &&
> > > +	    !vp_modern_get_queue_enable(mdev, qid)) {
> > > +		if (vp_modern_get_driver_features(mdev) &
> > > +		    BIT_ULL(VIRTIO_F_RING_PACKED))
> > > +			return vp_vdpa_set_vq_state_packed(vdpa, state);
> > > +		else
> > > +			return vp_vdpa_set_vq_state_split(vdpa,	state);
> > > +	}
> > > +
> > >   	return -EOPNOTSUPP;
> > >   }
> > > -- 
> > > 2.25.1
> > > 
> 

[PATCH V2 4/4] virtio/vdpa: clear the virtqueue state during probe

[Jason Wang]

From: Eli Cohen <elic@nvidia.com>

Clear the available index as part of the initialization process to
clear and values that might be left from previous usage of the device.
For example, if the device was previously used by vhost_vdpa and now
probed by vhost_vdpa, you want to start with indices.

Fixes: c043b4a8cf3b ("virtio: introduce a vDPA based transport")
Signed-off-by: Eli Cohen <elic@nvidia.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/virtio/virtio_vdpa.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/drivers/virtio/virtio_vdpa.c b/drivers/virtio/virtio_vdpa.c
index e28acf482e0c..e1a141135992 100644
--- a/drivers/virtio/virtio_vdpa.c
+++ b/drivers/virtio/virtio_vdpa.c
@@ -142,6 +142,8 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
 	struct vdpa_callback cb;
 	struct virtqueue *vq;
 	u64 desc_addr, driver_addr, device_addr;
+	/* Assume split virtqueue, switch to packed if necessary */
+	struct vdpa_vq_state state = {0};
 	unsigned long flags;
 	u32 align, num;
 	int err;
@@ -191,6 +193,19 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
 		goto err_vq;
 	}
 
+	/* reset virtqueue state index */
+	if (virtio_has_feature(vdev, VIRTIO_F_RING_PACKED)) {
+		struct vdpa_vq_state_packed *s = &state.packed;
+
+		s->last_avail_counter = 1;
+		s->last_avail_idx = 0;
+		s->last_used_counter = 1;
+		s->last_used_idx = 0;
+	}
+	err = ops->set_vq_state(vdpa, index, &state);
+	if (err)
+		goto err_vq;
+
 	ops->set_vq_ready(vdpa, index, 1);
 
 	vq->priv = info;
-- 
2.25.1

[PATCH V2 4/4] virtio/vdpa: clear the virtqueue state during probe

[Jason Wang]

From: Eli Cohen <elic@nvidia.com>

Clear the available index as part of the initialization process to
clear and values that might be left from previous usage of the device.
For example, if the device was previously used by vhost_vdpa and now
probed by vhost_vdpa, you want to start with indices.

Fixes: c043b4a8cf3b ("virtio: introduce a vDPA based transport")
Signed-off-by: Eli Cohen <elic@nvidia.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
 drivers/virtio/virtio_vdpa.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/drivers/virtio/virtio_vdpa.c b/drivers/virtio/virtio_vdpa.c
index e28acf482e0c..e1a141135992 100644
--- a/drivers/virtio/virtio_vdpa.c
+++ b/drivers/virtio/virtio_vdpa.c
@@ -142,6 +142,8 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
 	struct vdpa_callback cb;
 	struct virtqueue *vq;
 	u64 desc_addr, driver_addr, device_addr;
+	/* Assume split virtqueue, switch to packed if necessary */
+	struct vdpa_vq_state state = {0};
 	unsigned long flags;
 	u32 align, num;
 	int err;
@@ -191,6 +193,19 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
 		goto err_vq;
 	}
 
+	/* reset virtqueue state index */
+	if (virtio_has_feature(vdev, VIRTIO_F_RING_PACKED)) {
+		struct vdpa_vq_state_packed *s = &state.packed;
+
+		s->last_avail_counter = 1;
+		s->last_avail_idx = 0;
+		s->last_used_counter = 1;
+		s->last_used_idx = 0;
+	}
+	err = ops->set_vq_state(vdpa, index, &state);
+	if (err)
+		goto err_vq;
+
 	ops->set_vq_ready(vdpa, index, 1);
 
 	vq->priv = info;
-- 
2.25.1

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH V2 4/4] virtio/vdpa: clear the virtqueue state during probe

[Eli Cohen]

On Wed, Jun 02, 2021 at 10:10:43AM +0800, Jason Wang wrote:
> From: Eli Cohen <elic@nvidia.com>
> 
> Clear the available index as part of the initialization process to
> clear and values that might be left from previous usage of the device.
> For example, if the device was previously used by vhost_vdpa and now
> probed by vhost_vdpa, you want to start with indices.
> 
> Fixes: c043b4a8cf3b ("virtio: introduce a vDPA based transport")
> Signed-off-by: Eli Cohen <elic@nvidia.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
>  drivers/virtio/virtio_vdpa.c | 15 +++++++++++++++
>  1 file changed, 15 insertions(+)
> 
> diff --git a/drivers/virtio/virtio_vdpa.c b/drivers/virtio/virtio_vdpa.c
> index e28acf482e0c..e1a141135992 100644
> --- a/drivers/virtio/virtio_vdpa.c
> +++ b/drivers/virtio/virtio_vdpa.c
> @@ -142,6 +142,8 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
>  	struct vdpa_callback cb;
>  	struct virtqueue *vq;
>  	u64 desc_addr, driver_addr, device_addr;
> +	/* Assume split virtqueue, switch to packed if necessary */
> +	struct vdpa_vq_state state = {0};
>  	unsigned long flags;
>  	u32 align, num;
>  	int err;
> @@ -191,6 +193,19 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
>  		goto err_vq;
>  	}
>  
> +	/* reset virtqueue state index */
> +	if (virtio_has_feature(vdev, VIRTIO_F_RING_PACKED)) {
> +		struct vdpa_vq_state_packed *s = &state.packed;
> +
> +		s->last_avail_counter = 1;
> +		s->last_avail_idx = 0;

It's already 0

> +		s->last_used_counter = 1;
> +		s->last_used_idx = 0;

already 0

> +	}
> +	err = ops->set_vq_state(vdpa, index, &state);
> +	if (err)
> +		goto err_vq;
> +
>  	ops->set_vq_ready(vdpa, index, 1);
>  
>  	vq->priv = info;
> -- 
> 2.25.1
> 

Re: [PATCH V2 4/4] virtio/vdpa: clear the virtqueue state during probe

[Jason Wang]

在 2021/6/2 下午2:17, Eli Cohen 写道:
> On Wed, Jun 02, 2021 at 10:10:43AM +0800, Jason Wang wrote:
>> From: Eli Cohen <elic@nvidia.com>
>>
>> Clear the available index as part of the initialization process to
>> clear and values that might be left from previous usage of the device.
>> For example, if the device was previously used by vhost_vdpa and now
>> probed by vhost_vdpa, you want to start with indices.
>>
>> Fixes: c043b4a8cf3b ("virtio: introduce a vDPA based transport")
>> Signed-off-by: Eli Cohen <elic@nvidia.com>
>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>> ---
>>   drivers/virtio/virtio_vdpa.c | 15 +++++++++++++++
>>   1 file changed, 15 insertions(+)
>>
>> diff --git a/drivers/virtio/virtio_vdpa.c b/drivers/virtio/virtio_vdpa.c
>> index e28acf482e0c..e1a141135992 100644
>> --- a/drivers/virtio/virtio_vdpa.c
>> +++ b/drivers/virtio/virtio_vdpa.c
>> @@ -142,6 +142,8 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
>>   	struct vdpa_callback cb;
>>   	struct virtqueue *vq;
>>   	u64 desc_addr, driver_addr, device_addr;
>> +	/* Assume split virtqueue, switch to packed if necessary */
>> +	struct vdpa_vq_state state = {0};
>>   	unsigned long flags;
>>   	u32 align, num;
>>   	int err;
>> @@ -191,6 +193,19 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
>>   		goto err_vq;
>>   	}
>>   
>> +	/* reset virtqueue state index */
>> +	if (virtio_has_feature(vdev, VIRTIO_F_RING_PACKED)) {
>> +		struct vdpa_vq_state_packed *s = &state.packed;
>> +
>> +		s->last_avail_counter = 1;
>> +		s->last_avail_idx = 0;
> It's already 0
>
>> +		s->last_used_counter = 1;
>> +		s->last_used_idx = 0;
> already 0


Yes, but for completeness and make code easy to read, it's no harm to 
keep them I think.

Thanks


>
>> +	}
>> +	err = ops->set_vq_state(vdpa, index, &state);
>> +	if (err)
>> +		goto err_vq;
>> +
>>   	ops->set_vq_ready(vdpa, index, 1);
>>   
>>   	vq->priv = info;
>> -- 
>> 2.25.1
>>

Re: [PATCH V2 4/4] virtio/vdpa: clear the virtqueue state during probe

[Jason Wang]

在 2021/6/2 下午2:17, Eli Cohen 写道:
> On Wed, Jun 02, 2021 at 10:10:43AM +0800, Jason Wang wrote:
>> From: Eli Cohen <elic@nvidia.com>
>>
>> Clear the available index as part of the initialization process to
>> clear and values that might be left from previous usage of the device.
>> For example, if the device was previously used by vhost_vdpa and now
>> probed by vhost_vdpa, you want to start with indices.
>>
>> Fixes: c043b4a8cf3b ("virtio: introduce a vDPA based transport")
>> Signed-off-by: Eli Cohen <elic@nvidia.com>
>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>> ---
>>   drivers/virtio/virtio_vdpa.c | 15 +++++++++++++++
>>   1 file changed, 15 insertions(+)
>>
>> diff --git a/drivers/virtio/virtio_vdpa.c b/drivers/virtio/virtio_vdpa.c
>> index e28acf482e0c..e1a141135992 100644
>> --- a/drivers/virtio/virtio_vdpa.c
>> +++ b/drivers/virtio/virtio_vdpa.c
>> @@ -142,6 +142,8 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
>>   	struct vdpa_callback cb;
>>   	struct virtqueue *vq;
>>   	u64 desc_addr, driver_addr, device_addr;
>> +	/* Assume split virtqueue, switch to packed if necessary */
>> +	struct vdpa_vq_state state = {0};
>>   	unsigned long flags;
>>   	u32 align, num;
>>   	int err;
>> @@ -191,6 +193,19 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
>>   		goto err_vq;
>>   	}
>>   
>> +	/* reset virtqueue state index */
>> +	if (virtio_has_feature(vdev, VIRTIO_F_RING_PACKED)) {
>> +		struct vdpa_vq_state_packed *s = &state.packed;
>> +
>> +		s->last_avail_counter = 1;
>> +		s->last_avail_idx = 0;
> It's already 0
>
>> +		s->last_used_counter = 1;
>> +		s->last_used_idx = 0;
> already 0


Yes, but for completeness and make code easy to read, it's no harm to 
keep them I think.

Thanks


>
>> +	}
>> +	err = ops->set_vq_state(vdpa, index, &state);
>> +	if (err)
>> +		goto err_vq;
>> +
>>   	ops->set_vq_ready(vdpa, index, 1);
>>   
>>   	vq->priv = info;
>> -- 
>> 2.25.1
>>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

Re: [PATCH V2 4/4] virtio/vdpa: clear the virtqueue state during probe

[Eli Cohen]

On Wed, Jun 02, 2021 at 03:07:00PM +0800, Jason Wang wrote:
> 
> 在 2021/6/2 下午2:17, Eli Cohen 写道:
> > On Wed, Jun 02, 2021 at 10:10:43AM +0800, Jason Wang wrote:
> > > From: Eli Cohen <elic@nvidia.com>
> > > 
> > > Clear the available index as part of the initialization process to
> > > clear and values that might be left from previous usage of the device.
> > > For example, if the device was previously used by vhost_vdpa and now
> > > probed by vhost_vdpa, you want to start with indices.
> > > 
> > > Fixes: c043b4a8cf3b ("virtio: introduce a vDPA based transport")
> > > Signed-off-by: Eli Cohen <elic@nvidia.com>
> > > Signed-off-by: Jason Wang <jasowang@redhat.com>
> > > ---
> > >   drivers/virtio/virtio_vdpa.c | 15 +++++++++++++++
> > >   1 file changed, 15 insertions(+)
> > > 
> > > diff --git a/drivers/virtio/virtio_vdpa.c b/drivers/virtio/virtio_vdpa.c
> > > index e28acf482e0c..e1a141135992 100644
> > > --- a/drivers/virtio/virtio_vdpa.c
> > > +++ b/drivers/virtio/virtio_vdpa.c
> > > @@ -142,6 +142,8 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
> > >   	struct vdpa_callback cb;
> > >   	struct virtqueue *vq;
> > >   	u64 desc_addr, driver_addr, device_addr;
> > > +	/* Assume split virtqueue, switch to packed if necessary */
> > > +	struct vdpa_vq_state state = {0};
> > >   	unsigned long flags;
> > >   	u32 align, num;
> > >   	int err;
> > > @@ -191,6 +193,19 @@ virtio_vdpa_setup_vq(struct virtio_device *vdev, unsigned int index,
> > >   		goto err_vq;
> > >   	}
> > > +	/* reset virtqueue state index */
> > > +	if (virtio_has_feature(vdev, VIRTIO_F_RING_PACKED)) {
> > > +		struct vdpa_vq_state_packed *s = &state.packed;
> > > +
> > > +		s->last_avail_counter = 1;
> > > +		s->last_avail_idx = 0;
> > It's already 0
> > 
> > > +		s->last_used_counter = 1;
> > > +		s->last_used_idx = 0;
> > already 0
> 
> 
> Yes, but for completeness and make code easy to read, it's no harm to keep
> them I think.
> 

OK.

> Thanks
> 
> 
> > 
> > > +	}
> > > +	err = ops->set_vq_state(vdpa, index, &state);
> > > +	if (err)
> > > +		goto err_vq;
> > > +
> > >   	ops->set_vq_ready(vdpa, index, 1);
> > >   	vq->priv = info;
> > > -- 
> > > 2.25.1
> > > 
> 

Re: [PATCH V2 0/4]

[Jason Wang]

在 2021/6/2 上午10:10, Jason Wang 写道:
> *** BLURB HERE ***


Missing blurb...

Will resend a new version.

Thanks


>
> Eli Cohen (1):
>    virtio/vdpa: clear the virtqueue state during probe
>
> Jason Wang (3):
>    vdpa: support packed virtqueue for set/get_vq_state()
>    virtio-pci library: introduce vp_modern_get_driver_features()
>    vp_vdpa: allow set vq state to initial state after reset
>
>   drivers/vdpa/ifcvf/ifcvf_main.c        |  4 +--
>   drivers/vdpa/mlx5/net/mlx5_vnet.c      |  8 ++---
>   drivers/vdpa/vdpa_sim/vdpa_sim.c       |  4 +--
>   drivers/vdpa/virtio_pci/vp_vdpa.c      | 42 ++++++++++++++++++++++++--
>   drivers/vhost/vdpa.c                   |  4 +--
>   drivers/virtio/virtio_pci_modern_dev.c | 21 +++++++++++++
>   drivers/virtio/virtio_vdpa.c           | 15 +++++++++
>   include/linux/vdpa.h                   | 25 +++++++++++++--
>   include/linux/virtio_pci_modern.h      |  1 +
>   9 files changed, 109 insertions(+), 15 deletions(-)
>

Re: [PATCH V2 0/4]

[Jason Wang]

在 2021/6/2 上午10:10, Jason Wang 写道:
> *** BLURB HERE ***


Missing blurb...

Will resend a new version.

Thanks


>
> Eli Cohen (1):
>    virtio/vdpa: clear the virtqueue state during probe
>
> Jason Wang (3):
>    vdpa: support packed virtqueue for set/get_vq_state()
>    virtio-pci library: introduce vp_modern_get_driver_features()
>    vp_vdpa: allow set vq state to initial state after reset
>
>   drivers/vdpa/ifcvf/ifcvf_main.c        |  4 +--
>   drivers/vdpa/mlx5/net/mlx5_vnet.c      |  8 ++---
>   drivers/vdpa/vdpa_sim/vdpa_sim.c       |  4 +--
>   drivers/vdpa/virtio_pci/vp_vdpa.c      | 42 ++++++++++++++++++++++++--
>   drivers/vhost/vdpa.c                   |  4 +--
>   drivers/virtio/virtio_pci_modern_dev.c | 21 +++++++++++++
>   drivers/virtio/virtio_vdpa.c           | 15 +++++++++
>   include/linux/vdpa.h                   | 25 +++++++++++++--
>   include/linux/virtio_pci_modern.h      |  1 +
>   9 files changed, 109 insertions(+), 15 deletions(-)
>

_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization

[PATCH v2 0/4]

[Max Filippov]

Hello David, Ben, Florian, Chris and everybody,

this series improves ethoc behavior in gigabit environment:
- first patch introduces two phylib setters for 'advertising' and 'supported'
  fields of struct phy_device;
- second patch disables gigabit advertisement in the attached PHY making
  possible to use gigabit link without any additional setup;
- third patch adds support to set up MII management bus frequency, adding
  new fields to platform data and to OF bindings;
- fourth patch adds basic ethtool support to ethoc driver.

These changes allow to use KC-705 board with 50MHz xtensa core and OpenCores
10/100 Mbps MAC connected to gigabit network without any additional setup.

Changes v1->v2:
- new patch "phy: provide accessors for 'advertising' and 'supported' fields";
- disable both gigabit advertisement and support;
- drop MDIO bus frequency configurability, always configure for standard
  2.5MHz;
- allow using common clock framework to provide ethoc clock;
- new patch: "net: ethoc: implement ethtool operations";
- drop device tree bindings documentation patch until common bindings format
  for network drivers is decided.

Max Filippov (4):
  phy: provide accessors for 'advertising' and 'supported' fields
  net: ethoc: don't advertise gigabit speed on attached PHY
  net: ethoc: set up MII management bus clock
  net: ethoc: implement ethtool operations

 drivers/net/ethernet/ethoc.c | 130 ++++++++++++++++++++++++++++++++++++++++++-
 include/linux/phy.h          |  12 ++++
 include/net/ethoc.h          |   1 +
 3 files changed, 141 insertions(+), 2 deletions(-)

-- 
1.8.1.4

[PATCH v2 0/4]

[Sebastian Schuberth]

This is the second iteration of the patches in

http://www.spinics.net/lists/git/msg222428.html
http://www.spinics.net/lists/git/msg222429.html

which

* adds a commit to use the term "builtin" instead of "internal command",
* also modifies the docs accordingly,
* moves the is_builtin() declaration to the existing builtin.h,
* finally moves all builtin-related definitions to a new builtin.c file.

Sebastian Schuberth (4):
  Consistently use the term "builtin" instead of "internal command"
  Call load_command_list() only when it is needed
  Speed up is_git_command() by checking early for internal commands
  Move builtin-related implementations to a new builtin.c file

 Documentation/technical/api-builtin.txt |   4 +-
 Makefile                                |   1 +
 builtin.c                               | 225 ++++++++++++++++++++++++++++++
 builtin.h                               |  23 +++
 builtin/help.c                          |   6 +-
 git.c                                   | 238 +-------------------------------
 6 files changed, 262 insertions(+), 235 deletions(-)
 create mode 100644 builtin.c

-- 
1.8.3-mingw-1

[PATCH v2 0/4]

[Murali Karicheri]

This patch series add a K2HK (K2 Hawking/Kepler) EVM
dts file and make some updates to the existing dts
files and Keystone PLL clk driver.

Murali Karicheri (4):
  clk: keystone: use clkod register bits for postdiv
  keystone: dts: add a k2hk-evm specific dts file
  keystone: dts: fix typo in the ddr3 pllclk node name
  keystone: dts: add paclk divider clock node

 .../devicetree/bindings/clock/keystone-pll.txt     |    8 +--
 arch/arm/boot/dts/k2hk-evm.dts                     |   55 ++++++++++++++++++++
 arch/arm/boot/dts/keystone-clocks.dtsi             |   36 ++++++-------
 arch/arm/boot/dts/{keystone.dts => keystone.dtsi}  |    2 -
 drivers/clk/keystone/pll.c                         |   24 +++++++--
 5 files changed, 96 insertions(+), 29 deletions(-)
 create mode 100644 arch/arm/boot/dts/k2hk-evm.dts
 rename arch/arm/boot/dts/{keystone.dts => keystone.dtsi} (98%)

-- 
1.7.9.5

[PATCH v2 0/4]

[Karicheri, Muralidharan]

>> -----Original Message-----
>> From: Karicheri, Muralidharan
>> Sent: Tuesday, November 19, 2013 7:43 PM
>> To: linux-arm-kernel at lists.infradead.org; mturquette at linaro.org; Shilimkar, Santosh; linux-
>> keystone at list.ti.com - Linux developers for Keystone family of devices (May contain non-
>> TIers)
>> Cc: Karicheri, Muralidharan
>> Subject: [PATCH v2 0/4]
>> 
>> This patch series add a K2HK (K2 Hawking/Kepler) EVM dts file and make some updates to
>> the existing dts files and Keystone PLL clk driver.
>> 
>> Murali Karicheri (4):
>>   clk: keystone: use clkod register bits for postdiv
>>   keystone: dts: add a k2hk-evm specific dts file
>>   keystone: dts: fix typo in the ddr3 pllclk node name
>>   keystone: dts: add paclk divider clock node
>> 
>>  .../devicetree/bindings/clock/keystone-pll.txt     |    8 +--
>>  arch/arm/boot/dts/k2hk-evm.dts                     |   55 ++++++++++++++++++++
>>  arch/arm/boot/dts/keystone-clocks.dtsi             |   36 ++++++-------
>>  arch/arm/boot/dts/{keystone.dts => keystone.dtsi}  |    2 -
>>  drivers/clk/keystone/pll.c                         |   24 +++++++--
>>  5 files changed, 96 insertions(+), 29 deletions(-)  create mode 100644
>> arch/arm/boot/dts/k2hk-evm.dts  rename arch/arm/boot/dts/{keystone.dts =>
>> keystone.dtsi} (98%)
>> 
>> --
>> 1.7.9.5

Will resend. Subject is missing.

[PATCH V2 0/4]

[Joseph Lo]

This series adds the LP1 support for Tegra124. (suspend to LP2 also
supported in this series)

Note:
This patch series depends on the patch series below.
* [PATCH 0/5] Tegra124 clock support
* [PATCH V2 0/6] ARM: tegra: basic support for Tegra124 SoC
* [PATCH] ARM: tegra: add clock properties for devices of Tegra124
* [PATCH] ARM: tegra: enable Tegra RTC as default for Tegra124
* [PATCH 0/4] ARM: tegra: add CPU hot-plug and idle support for Tegra124

Verified on Cardhu, Dalmore and Venice2 and with THUMB2_KERNEL as well.

V2:
* remove the patch of "enable Tegra RTC for Venice2"
* add new patch of "re-calculate the LP1 data for Tegra30/114"
* squash the 2 to 5 into patch 3/4 of this series

Joseph Lo (4):
  clk: tegra124: add suspend/resume function for tegra_cpu_car_ops
  ARM: tegra: re-calculate the LP1 data for Tegra30/114
  ARM: tegra: add LP1 support code for Tegra124
  ARM: tegra: enable LP1 suspend mode for Venice2

 arch/arm/boot/dts/tegra124-venice2.dts |  7 +++++
 arch/arm/mach-tegra/Makefile           |  1 +
 arch/arm/mach-tegra/flowctrl.c         |  2 ++
 arch/arm/mach-tegra/iomap.h            |  3 +++
 arch/arm/mach-tegra/pm.c               | 12 ++++++---
 arch/arm/mach-tegra/sleep-tegra30.S    | 49 +++++++++++++++++++++++++---------
 drivers/clk/tegra/clk-tegra124.c       | 27 +++++++++++++++++++
 7 files changed, 86 insertions(+), 15 deletions(-)

-- 
1.8.4

[PATCH V2 0/4]

[Joseph Lo]

This series adds the LP1 support for Tegra124. (suspend to LP2 also
supported in this series)

Note:
This patch series depends on the patch series below.
* [PATCH 0/5] Tegra124 clock support
* [PATCH V2 0/6] ARM: tegra: basic support for Tegra124 SoC
* [PATCH] ARM: tegra: add clock properties for devices of Tegra124
* [PATCH] ARM: tegra: enable Tegra RTC as default for Tegra124
* [PATCH 0/4] ARM: tegra: add CPU hot-plug and idle support for Tegra124

Verified on Cardhu, Dalmore and Venice2 and with THUMB2_KERNEL as well.

V2:
* remove the patch of "enable Tegra RTC for Venice2"
* add new patch of "re-calculate the LP1 data for Tegra30/114"
* squash the 2 to 5 into patch 3/4 of this series

Joseph Lo (4):
  clk: tegra124: add suspend/resume function for tegra_cpu_car_ops
  ARM: tegra: re-calculate the LP1 data for Tegra30/114
  ARM: tegra: add LP1 support code for Tegra124
  ARM: tegra: enable LP1 suspend mode for Venice2

 arch/arm/boot/dts/tegra124-venice2.dts |  7 +++++
 arch/arm/mach-tegra/Makefile           |  1 +
 arch/arm/mach-tegra/flowctrl.c         |  2 ++
 arch/arm/mach-tegra/iomap.h            |  3 +++
 arch/arm/mach-tegra/pm.c               | 12 ++++++---
 arch/arm/mach-tegra/sleep-tegra30.S    | 49 +++++++++++++++++++++++++---------
 drivers/clk/tegra/clk-tegra124.c       | 27 +++++++++++++++++++
 7 files changed, 86 insertions(+), 15 deletions(-)

-- 
1.8.4

Re: [PATCH V2 0/4]

[Stephen Warren]

On 10/11/2013 03:58 AM, Joseph Lo wrote:
> This series adds the LP1 support for Tegra124. (suspend to LP2 also
> supported in this series)
> 
> Note:
> This patch series depends on the patch series below.
> * [PATCH 0/5] Tegra124 clock support
> * [PATCH V2 0/6] ARM: tegra: basic support for Tegra124 SoC
> * [PATCH] ARM: tegra: add clock properties for devices of Tegra124
> * [PATCH] ARM: tegra: enable Tegra RTC as default for Tegra124
> * [PATCH 0/4] ARM: tegra: add CPU hot-plug and idle support for Tegra124
> 
> Verified on Cardhu, Dalmore and Venice2 and with THUMB2_KERNEL as well.

I have applied patches 2-3 to Tegra's for-3.13/soc branch, and patch 4
to Tegra's for-3.13/dt branch.

I have not applied patch 1, since that's a clock driver change. Please
work with Mike and/or Peter to send that patch through the clock tree.
In general, if there aren't any compile-time or run-time dependencies
(for existing features, not new ones) between patches, patches for
different subsystems should be sent separately, rather than as a
combined series.

Note that "[PATCH] ARM: tegra: add clock properties for devices of
Tegra124" is not yet applied; I'm waiting for Peter's Tegra124 clock
series to be finalized before applying that. There's no compile-time
dependency here, and the patch only enables new features without
breaking existing ones, so there's no bisection issue here.

[PATCH V2 0/4]

[Stephen Warren]

On 10/11/2013 03:58 AM, Joseph Lo wrote:
> This series adds the LP1 support for Tegra124. (suspend to LP2 also
> supported in this series)
> 
> Note:
> This patch series depends on the patch series below.
> * [PATCH 0/5] Tegra124 clock support
> * [PATCH V2 0/6] ARM: tegra: basic support for Tegra124 SoC
> * [PATCH] ARM: tegra: add clock properties for devices of Tegra124
> * [PATCH] ARM: tegra: enable Tegra RTC as default for Tegra124
> * [PATCH 0/4] ARM: tegra: add CPU hot-plug and idle support for Tegra124
> 
> Verified on Cardhu, Dalmore and Venice2 and with THUMB2_KERNEL as well.

I have applied patches 2-3 to Tegra's for-3.13/soc branch, and patch 4
to Tegra's for-3.13/dt branch.

I have not applied patch 1, since that's a clock driver change. Please
work with Mike and/or Peter to send that patch through the clock tree.
In general, if there aren't any compile-time or run-time dependencies
(for existing features, not new ones) between patches, patches for
different subsystems should be sent separately, rather than as a
combined series.

Note that "[PATCH] ARM: tegra: add clock properties for devices of
Tegra124" is not yet applied; I'm waiting for Peter's Tegra124 clock
series to be finalized before applying that. There's no compile-time
dependency here, and the patch only enables new features without
breaking existing ones, so there's no bisection issue here.