[PATCH net-next v11 00/11] Implement devlink-rate API and extend it

[PATCH net-next v11 00/11] Implement devlink-rate API and extend it

[Michal Wilczynski]

This patch series implements devlink-rate for ice driver. Unfortunately
current API isn't flexible enough for our use case, so there is a need to
extend it. Some functions have been introduced to enable the driver to
export current Tx scheduling configuration.

Pasting justification for this series from commit implementing devlink-rate
in ice driver(that is a part of this series):

There is a need to support modification of Tx scheduler tree, in the
ice driver. This will allow user to control Tx settings of each node in
the internal hierarchy of nodes. As a result user will be able to use
Hierarchy QoS implemented entirely in the hardware.

This patch implemenents devlink-rate API. It also exports initial
default hierarchy. It's mostly dictated by the fact that the tree
can't be removed entirely, all we can do is enable the user to modify
it. For example root node shouldn't ever be removed, also nodes that
have children are off-limits.

Example initial tree with 2 VF's:

[root@fedora ~]# devlink port function rate show
pci/0000:4b:00.0/node_27: type node parent node_26
pci/0000:4b:00.0/node_26: type node parent node_0
pci/0000:4b:00.0/node_34: type node parent node_33
pci/0000:4b:00.0/node_33: type node parent node_32
pci/0000:4b:00.0/node_32: type node parent node_16
pci/0000:4b:00.0/node_19: type node parent node_18
pci/0000:4b:00.0/node_18: type node parent node_17
pci/0000:4b:00.0/node_17: type node parent node_16
pci/0000:4b:00.0/node_21: type node parent node_20
pci/0000:4b:00.0/node_20: type node parent node_3
pci/0000:4b:00.0/node_14: type node parent node_5
pci/0000:4b:00.0/node_5: type node parent node_3
pci/0000:4b:00.0/node_13: type node parent node_4
pci/0000:4b:00.0/node_12: type node parent node_4
pci/0000:4b:00.0/node_11: type node parent node_4
pci/0000:4b:00.0/node_10: type node parent node_4
pci/0000:4b:00.0/node_9: type node parent node_4
pci/0000:4b:00.0/node_8: type node parent node_4
pci/0000:4b:00.0/node_7: type node parent node_4
pci/0000:4b:00.0/node_6: type node parent node_4
pci/0000:4b:00.0/node_4: type node parent node_3
pci/0000:4b:00.0/node_3: type node parent node_16
pci/0000:4b:00.0/node_16: type node parent node_15
pci/0000:4b:00.0/node_15: type node parent node_0
pci/0000:4b:00.0/node_2: type node parent node_1
pci/0000:4b:00.0/node_1: type node parent node_0
pci/0000:4b:00.0/node_0: type node
pci/0000:4b:00.0/1: type leaf parent node_27
pci/0000:4b:00.0/2: type leaf parent node_27

Let me visualize part of the tree:

                        +---------+
                        |  node_0 |
                        +---------+
                             |
                        +----v----+
                        | node_26 |
                        +----+----+
                             |
                        +----v----+
                        | node_27 |
                        +----+----+
                             |
                    |-----------------|
               +----v----+       +----v----+
               |   VF 1  |       |   VF 2  |
               +----+----+       +----+----+

So at this point there is a couple things that can be done.
For example we could only assign parameters to VF's.

[root@fedora ~]# devlink port function rate set pci/0000:4b:00.0/1 \
                 tx_max 5Gbps

This would cap the VF 1 BW to 5Gbps.

But let's say you would like to create a completely new branch.
This can be done like this:

[root@fedora ~]# devlink port function rate add \
                 pci/0000:4b:00.0/node_custom parent node_0
[root@fedora ~]# devlink port function rate add \
                 pci/0000:4b:00.0/node_custom_1 parent node_custom
[root@fedora ~]# devlink port function rate set \
                 pci/0000:4b:00.0/1 parent node_custom_1

This creates a completely new branch and reassigns VF 1 to it.

A number of parameters is supported per each node: tx_max, tx_share,
tx_priority and tx_weight.

V11:
 - updated documentation in ice.rst to clarify how new parameters
   are used, and describe in more detail how and when the hierarchy
   is exported/deleted, and how mutual exclusivity with DCB/ADQ
   works
 - added documentation of new parameters in devlink-port.rst
 - refactored ice_set_object_tx_priority, ice_set_object_tx_max,
   ice_set_object_tx_weight, ice_set_object_tx_share to avoid
   code duplication

V10:
 - changed attributes type from u16 to u32 as they are padded to u32
   anyway
 - changed NL_SET_ERR_MSG_MOD to NL_SET_ERR_MSG_ATTR as it points to
   a problem with specific attribute
 - fixed function parameter not described
 - added documentation in ice.rst

V9:
 - changed misleading name from 'parameter' to 'attribute'
 - removed mechanism referring for kernel object by string,
   now it's referring to them as pointers
 - removed limiting name size in devl_rate_node_create()
 - removed commit that allowed for change of priv in set_parent
   callback
 - added commit that allows for pre-allocation of ice_sched
   elements

V8:
- address minor formatting issues
- fix memory leak
- address warnings

V7:
- split into smaller commits
- paste justification for this series to cover letter

V6:
- replaced strncpy with strscpy
- renamed rate_vport -> rate_leaf

V5:
- removed queue support per community request
- fix division of 64bit variable with 32bit divisor by using div_u64()
- remove RDMA, ADQ exlusion as it's not necessary anymore
- changed how driver exports configuration, as queues are not supported
  anymore
- changed IDA to Xarray for unique node identification

V4:
- changed static variable counter to per port IDA to
  uniquely identify nodes

V3:
- removed shift macros, since FIELD_PREP is used
- added static_assert for struct
- removed unnecessary functions
- used tab instead of space in define

V2:
- fixed Alexandr comments
- refactored code to fix checkpatch issues
- added mutual exclusion for RDMA, DCB

Michal Wilczynski (11):
  devlink: Introduce new attribute 'tx_priority' to devlink-rate
  devlink: Introduce new attribute 'tx_weight' to devlink-rate
  devlink: Enable creation of the devlink-rate nodes from the driver
  devlink: Allow for devlink-rate nodes parent reassignment
  devlink: Allow to set up parent in devl_rate_leaf_create()
  ice: Introduce new parameters in ice_sched_node
  ice: Add an option to pre-allocate memory for ice_sched_node
  ice: Implement devlink-rate API
  ice: Prevent ADQ, DCB coexistence with Custom Tx scheduler
  ice: Add documentation for devlink-rate implementation
  Documentation: Add documentation for new devlink-rate attributes

 .../networking/devlink/devlink-port.rst       |  33 +-
 Documentation/networking/devlink/ice.rst      | 115 ++++
 .../net/ethernet/intel/ice/ice_adminq_cmd.h   |   4 +-
 drivers/net/ethernet/intel/ice/ice_common.c   |   7 +-
 drivers/net/ethernet/intel/ice/ice_dcb.c      |   2 +-
 drivers/net/ethernet/intel/ice/ice_dcb_lib.c  |   7 +
 drivers/net/ethernet/intel/ice/ice_devlink.c  | 498 ++++++++++++++++++
 drivers/net/ethernet/intel/ice/ice_devlink.h  |   3 +
 drivers/net/ethernet/intel/ice/ice_main.c     |   6 +
 drivers/net/ethernet/intel/ice/ice_repr.c     |  18 +
 drivers/net/ethernet/intel/ice/ice_sched.c    | 104 +++-
 drivers/net/ethernet/intel/ice/ice_sched.h    |  31 +-
 drivers/net/ethernet/intel/ice/ice_type.h     |   9 +
 .../mellanox/mlx5/core/esw/devlink_port.c     |   4 +-
 drivers/net/netdevsim/dev.c                   |   2 +-
 include/net/devlink.h                         |  18 +-
 include/uapi/linux/devlink.h                  |   3 +
 net/core/devlink.c                            | 130 ++++-
 18 files changed, 966 insertions(+), 28 deletions(-)

-- 
2.37.2

[PATCH net-next v11 01/11] devlink: Introduce new attribute 'tx_priority' to devlink-rate

[Michal Wilczynski]

To fully utilize offload capabilities of Intel 100G card QoS capabilities
new attribute 'tx_priority' needs to be introduced. This attribute allows
for usage of strict priority arbiter among siblings. This arbitration
scheme attempts to schedule nodes based on their priority as long as the
nodes remain within their bandwidth limit.

Introduce new attribute in devlink-rate that will allow for configuration
of strict priority. New attribute is optional.

Signed-off-by: Michal Wilczynski <michal.wilczynski@intel.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
---
 include/net/devlink.h        |  6 ++++++
 include/uapi/linux/devlink.h |  1 +
 net/core/devlink.c           | 31 +++++++++++++++++++++++++++++++
 3 files changed, 38 insertions(+)

diff --git a/include/net/devlink.h b/include/net/devlink.h
index 611a23a3deb2..90d59d673cb1 100644
--- a/include/net/devlink.h
+++ b/include/net/devlink.h
@@ -114,6 +114,8 @@ struct devlink_rate {
 			refcount_t refcnt;
 		};
 	};
+
+	u32 tx_priority;
 };
 
 struct devlink_port {
@@ -1511,10 +1513,14 @@ struct devlink_ops {
 				      u64 tx_share, struct netlink_ext_ack *extack);
 	int (*rate_leaf_tx_max_set)(struct devlink_rate *devlink_rate, void *priv,
 				    u64 tx_max, struct netlink_ext_ack *extack);
+	int (*rate_leaf_tx_priority_set)(struct devlink_rate *devlink_rate, void *priv,
+					 u32 tx_priority, struct netlink_ext_ack *extack);
 	int (*rate_node_tx_share_set)(struct devlink_rate *devlink_rate, void *priv,
 				      u64 tx_share, struct netlink_ext_ack *extack);
 	int (*rate_node_tx_max_set)(struct devlink_rate *devlink_rate, void *priv,
 				    u64 tx_max, struct netlink_ext_ack *extack);
+	int (*rate_node_tx_priority_set)(struct devlink_rate *devlink_rate, void *priv,
+					 u32 tx_priority, struct netlink_ext_ack *extack);
 	int (*rate_node_new)(struct devlink_rate *rate_node, void **priv,
 			     struct netlink_ext_ack *extack);
 	int (*rate_node_del)(struct devlink_rate *rate_node, void *priv,
diff --git a/include/uapi/linux/devlink.h b/include/uapi/linux/devlink.h
index 2f24b53a87a5..1a9214d35ef5 100644
--- a/include/uapi/linux/devlink.h
+++ b/include/uapi/linux/devlink.h
@@ -607,6 +607,7 @@ enum devlink_attr {
 
 	DEVLINK_ATTR_SELFTESTS,			/* nested */
 
+	DEVLINK_ATTR_RATE_TX_PRIORITY,		/* u32 */
 	/* add new attributes above here, update the policy in devlink.c */
 
 	__DEVLINK_ATTR_MAX,
diff --git a/net/core/devlink.c b/net/core/devlink.c
index 7f789bbcbbd7..bf6d3a3c28bb 100644
--- a/net/core/devlink.c
+++ b/net/core/devlink.c
@@ -1203,6 +1203,9 @@ static int devlink_nl_rate_fill(struct sk_buff *msg,
 			      devlink_rate->tx_max, DEVLINK_ATTR_PAD))
 		goto nla_put_failure;
 
+	if (nla_put_u32(msg, DEVLINK_ATTR_RATE_TX_PRIORITY,
+			devlink_rate->tx_priority))
+		goto nla_put_failure;
 	if (devlink_rate->parent)
 		if (nla_put_string(msg, DEVLINK_ATTR_RATE_PARENT_NODE_NAME,
 				   devlink_rate->parent->name))
@@ -1936,6 +1939,7 @@ static int devlink_nl_rate_set(struct devlink_rate *devlink_rate,
 {
 	struct nlattr *nla_parent, **attrs = info->attrs;
 	int err = -EOPNOTSUPP;
+	u32 priority;
 	u64 rate;
 
 	if (attrs[DEVLINK_ATTR_RATE_TX_SHARE]) {
@@ -1964,6 +1968,20 @@ static int devlink_nl_rate_set(struct devlink_rate *devlink_rate,
 		devlink_rate->tx_max = rate;
 	}
 
+	if (attrs[DEVLINK_ATTR_RATE_TX_PRIORITY]) {
+		priority = nla_get_u32(attrs[DEVLINK_ATTR_RATE_TX_PRIORITY]);
+		if (devlink_rate_is_leaf(devlink_rate))
+			err = ops->rate_leaf_tx_priority_set(devlink_rate, devlink_rate->priv,
+							     priority, info->extack);
+		else if (devlink_rate_is_node(devlink_rate))
+			err = ops->rate_node_tx_priority_set(devlink_rate, devlink_rate->priv,
+							     priority, info->extack);
+
+		if (err)
+			return err;
+		devlink_rate->tx_priority = priority;
+	}
+
 	nla_parent = attrs[DEVLINK_ATTR_RATE_PARENT_NODE_NAME];
 	if (nla_parent) {
 		err = devlink_nl_rate_parent_node_set(devlink_rate, info,
@@ -1995,6 +2013,12 @@ static bool devlink_rate_set_ops_supported(const struct devlink_ops *ops,
 			NL_SET_ERR_MSG_MOD(info->extack, "Parent set isn't supported for the leafs");
 			return false;
 		}
+		if (attrs[DEVLINK_ATTR_RATE_TX_PRIORITY] && !ops->rate_leaf_tx_priority_set) {
+			NL_SET_ERR_MSG_ATTR(info->extack,
+					    attrs[DEVLINK_ATTR_RATE_TX_PRIORITY],
+					    "TX priority set isn't supported for the leafs");
+			return false;
+		}
 	} else if (type == DEVLINK_RATE_TYPE_NODE) {
 		if (attrs[DEVLINK_ATTR_RATE_TX_SHARE] && !ops->rate_node_tx_share_set) {
 			NL_SET_ERR_MSG_MOD(info->extack, "TX share set isn't supported for the nodes");
@@ -2009,6 +2033,12 @@ static bool devlink_rate_set_ops_supported(const struct devlink_ops *ops,
 			NL_SET_ERR_MSG_MOD(info->extack, "Parent set isn't supported for the nodes");
 			return false;
 		}
+		if (attrs[DEVLINK_ATTR_RATE_TX_PRIORITY] && !ops->rate_node_tx_priority_set) {
+			NL_SET_ERR_MSG_ATTR(info->extack,
+					    attrs[DEVLINK_ATTR_RATE_TX_PRIORITY],
+					    "TX priority set isn't supported for the nodes");
+			return false;
+		}
 	} else {
 		WARN(1, "Unknown type of rate object");
 		return false;
@@ -9187,6 +9217,7 @@ static const struct nla_policy devlink_nl_policy[DEVLINK_ATTR_MAX + 1] = {
 	[DEVLINK_ATTR_LINECARD_INDEX] = { .type = NLA_U32 },
 	[DEVLINK_ATTR_LINECARD_TYPE] = { .type = NLA_NUL_STRING },
 	[DEVLINK_ATTR_SELFTESTS] = { .type = NLA_NESTED },
+	[DEVLINK_ATTR_RATE_TX_PRIORITY] = { .type = NLA_U32 },
 };
 
 static const struct genl_small_ops devlink_nl_ops[] = {
-- 
2.37.2

[PATCH net-next v11 02/11] devlink: Introduce new attribute 'tx_weight' to devlink-rate

[Michal Wilczynski]

To fully utilize offload capabilities of Intel 100G card QoS capabilities
new attribute 'tx_weight' needs to be introduced. This attribute allows
for usage of Weighted Fair Queuing arbitration scheme among siblings.
This arbitration scheme can be used simultaneously with the strict
priority.

Introduce new attribute in devlink-rate that will allow for configuration
of Weighted Fair Queueing. New attribute is optional.

Signed-off-by: Michal Wilczynski <michal.wilczynski@intel.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
---
 include/net/devlink.h        |  5 +++++
 include/uapi/linux/devlink.h |  2 ++
 net/core/devlink.c           | 33 +++++++++++++++++++++++++++++++++
 3 files changed, 40 insertions(+)

diff --git a/include/net/devlink.h b/include/net/devlink.h
index 90d59d673cb1..366b23d3f973 100644
--- a/include/net/devlink.h
+++ b/include/net/devlink.h
@@ -116,6 +116,7 @@ struct devlink_rate {
 	};
 
 	u32 tx_priority;
+	u32 tx_weight;
 };
 
 struct devlink_port {
@@ -1515,12 +1516,16 @@ struct devlink_ops {
 				    u64 tx_max, struct netlink_ext_ack *extack);
 	int (*rate_leaf_tx_priority_set)(struct devlink_rate *devlink_rate, void *priv,
 					 u32 tx_priority, struct netlink_ext_ack *extack);
+	int (*rate_leaf_tx_weight_set)(struct devlink_rate *devlink_rate, void *priv,
+				       u32 tx_weight, struct netlink_ext_ack *extack);
 	int (*rate_node_tx_share_set)(struct devlink_rate *devlink_rate, void *priv,
 				      u64 tx_share, struct netlink_ext_ack *extack);
 	int (*rate_node_tx_max_set)(struct devlink_rate *devlink_rate, void *priv,
 				    u64 tx_max, struct netlink_ext_ack *extack);
 	int (*rate_node_tx_priority_set)(struct devlink_rate *devlink_rate, void *priv,
 					 u32 tx_priority, struct netlink_ext_ack *extack);
+	int (*rate_node_tx_weight_set)(struct devlink_rate *devlink_rate, void *priv,
+				       u32 tx_weight, struct netlink_ext_ack *extack);
 	int (*rate_node_new)(struct devlink_rate *rate_node, void **priv,
 			     struct netlink_ext_ack *extack);
 	int (*rate_node_del)(struct devlink_rate *rate_node, void *priv,
diff --git a/include/uapi/linux/devlink.h b/include/uapi/linux/devlink.h
index 1a9214d35ef5..498d0d5d0957 100644
--- a/include/uapi/linux/devlink.h
+++ b/include/uapi/linux/devlink.h
@@ -608,6 +608,8 @@ enum devlink_attr {
 	DEVLINK_ATTR_SELFTESTS,			/* nested */
 
 	DEVLINK_ATTR_RATE_TX_PRIORITY,		/* u32 */
+	DEVLINK_ATTR_RATE_TX_WEIGHT,		/* u32 */
+
 	/* add new attributes above here, update the policy in devlink.c */
 
 	__DEVLINK_ATTR_MAX,
diff --git a/net/core/devlink.c b/net/core/devlink.c
index bf6d3a3c28bb..525bdf426163 100644
--- a/net/core/devlink.c
+++ b/net/core/devlink.c
@@ -1206,6 +1206,11 @@ static int devlink_nl_rate_fill(struct sk_buff *msg,
 	if (nla_put_u32(msg, DEVLINK_ATTR_RATE_TX_PRIORITY,
 			devlink_rate->tx_priority))
 		goto nla_put_failure;
+
+	if (nla_put_u32(msg, DEVLINK_ATTR_RATE_TX_WEIGHT,
+			devlink_rate->tx_weight))
+		goto nla_put_failure;
+
 	if (devlink_rate->parent)
 		if (nla_put_string(msg, DEVLINK_ATTR_RATE_PARENT_NODE_NAME,
 				   devlink_rate->parent->name))
@@ -1940,6 +1945,7 @@ static int devlink_nl_rate_set(struct devlink_rate *devlink_rate,
 	struct nlattr *nla_parent, **attrs = info->attrs;
 	int err = -EOPNOTSUPP;
 	u32 priority;
+	u32 weight;
 	u64 rate;
 
 	if (attrs[DEVLINK_ATTR_RATE_TX_SHARE]) {
@@ -1982,6 +1988,20 @@ static int devlink_nl_rate_set(struct devlink_rate *devlink_rate,
 		devlink_rate->tx_priority = priority;
 	}
 
+	if (attrs[DEVLINK_ATTR_RATE_TX_WEIGHT]) {
+		weight = nla_get_u32(attrs[DEVLINK_ATTR_RATE_TX_WEIGHT]);
+		if (devlink_rate_is_leaf(devlink_rate))
+			err = ops->rate_leaf_tx_weight_set(devlink_rate, devlink_rate->priv,
+							   weight, info->extack);
+		else if (devlink_rate_is_node(devlink_rate))
+			err = ops->rate_node_tx_weight_set(devlink_rate, devlink_rate->priv,
+							   weight, info->extack);
+
+		if (err)
+			return err;
+		devlink_rate->tx_weight = weight;
+	}
+
 	nla_parent = attrs[DEVLINK_ATTR_RATE_PARENT_NODE_NAME];
 	if (nla_parent) {
 		err = devlink_nl_rate_parent_node_set(devlink_rate, info,
@@ -2019,6 +2039,12 @@ static bool devlink_rate_set_ops_supported(const struct devlink_ops *ops,
 					    "TX priority set isn't supported for the leafs");
 			return false;
 		}
+		if (attrs[DEVLINK_ATTR_RATE_TX_WEIGHT] && !ops->rate_leaf_tx_weight_set) {
+			NL_SET_ERR_MSG_ATTR(info->extack,
+					    attrs[DEVLINK_ATTR_RATE_TX_WEIGHT],
+					    "TX weight set isn't supported for the leafs");
+			return false;
+		}
 	} else if (type == DEVLINK_RATE_TYPE_NODE) {
 		if (attrs[DEVLINK_ATTR_RATE_TX_SHARE] && !ops->rate_node_tx_share_set) {
 			NL_SET_ERR_MSG_MOD(info->extack, "TX share set isn't supported for the nodes");
@@ -2039,6 +2065,12 @@ static bool devlink_rate_set_ops_supported(const struct devlink_ops *ops,
 					    "TX priority set isn't supported for the nodes");
 			return false;
 		}
+		if (attrs[DEVLINK_ATTR_RATE_TX_WEIGHT] && !ops->rate_node_tx_weight_set) {
+			NL_SET_ERR_MSG_ATTR(info->extack,
+					    attrs[DEVLINK_ATTR_RATE_TX_WEIGHT],
+					    "TX weight set isn't supported for the nodes");
+			return false;
+		}
 	} else {
 		WARN(1, "Unknown type of rate object");
 		return false;
@@ -9218,6 +9250,7 @@ static const struct nla_policy devlink_nl_policy[DEVLINK_ATTR_MAX + 1] = {
 	[DEVLINK_ATTR_LINECARD_TYPE] = { .type = NLA_NUL_STRING },
 	[DEVLINK_ATTR_SELFTESTS] = { .type = NLA_NESTED },
 	[DEVLINK_ATTR_RATE_TX_PRIORITY] = { .type = NLA_U32 },
+	[DEVLINK_ATTR_RATE_TX_WEIGHT] = { .type = NLA_U32 },
 };
 
 static const struct genl_small_ops devlink_nl_ops[] = {
-- 
2.37.2

[PATCH net-next v11 03/11] devlink: Enable creation of the devlink-rate nodes from the driver

[Michal Wilczynski]

Intel 100G card internal firmware hierarchy for Hierarchicial QoS is very
rigid and can't be easily removed. This requires an ability to export
default hierarchy to allow user to modify it. Currently the driver is
only able to create the 'leaf' nodes, which usually represent the vport.
This is not enough for HQoS implemented in Intel hardware.

Introduce new function devl_rate_node_create() that allows for creation
of the devlink-rate nodes from the driver.

Signed-off-by: Michal Wilczynski <michal.wilczynski@intel.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
---
 include/net/devlink.h |  3 +++
 net/core/devlink.c    | 45 +++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 48 insertions(+)

diff --git a/include/net/devlink.h b/include/net/devlink.h
index 366b23d3f973..339a2ed02d36 100644
--- a/include/net/devlink.h
+++ b/include/net/devlink.h
@@ -1618,6 +1618,9 @@ void devlink_port_attrs_pci_sf_set(struct devlink_port *devlink_port,
 				   u32 controller, u16 pf, u32 sf,
 				   bool external);
 int devl_rate_leaf_create(struct devlink_port *port, void *priv);
+struct devlink_rate *
+devl_rate_node_create(struct devlink *devlink, void *priv, char *node_name,
+		      struct devlink_rate *parent);
 void devl_rate_leaf_destroy(struct devlink_port *devlink_port);
 void devl_rate_nodes_destroy(struct devlink *devlink);
 void devlink_port_linecard_set(struct devlink_port *devlink_port,
diff --git a/net/core/devlink.c b/net/core/devlink.c
index 525bdf426163..3dfee7cd9929 100644
--- a/net/core/devlink.c
+++ b/net/core/devlink.c
@@ -10384,6 +10384,51 @@ void devlink_port_attrs_pci_sf_set(struct devlink_port *devlink_port, u32 contro
 }
 EXPORT_SYMBOL_GPL(devlink_port_attrs_pci_sf_set);
 
+/**
+ * devl_rate_node_create - create devlink rate node
+ * @devlink: devlink instance
+ * @priv: driver private data
+ * @node_name: name of the resulting node
+ * @parent: parent devlink_rate struct
+ *
+ * Create devlink rate object of type node
+ */
+struct devlink_rate *
+devl_rate_node_create(struct devlink *devlink, void *priv, char *node_name,
+		      struct devlink_rate *parent)
+{
+	struct devlink_rate *rate_node;
+
+	rate_node = devlink_rate_node_get_by_name(devlink, node_name);
+	if (!IS_ERR(rate_node))
+		return ERR_PTR(-EEXIST);
+
+	rate_node = kzalloc(sizeof(*rate_node), GFP_KERNEL);
+	if (!rate_node)
+		return ERR_PTR(-ENOMEM);
+
+	if (parent) {
+		rate_node->parent = parent;
+		refcount_inc(&rate_node->parent->refcnt);
+	}
+
+	rate_node->type = DEVLINK_RATE_TYPE_NODE;
+	rate_node->devlink = devlink;
+	rate_node->priv = priv;
+
+	rate_node->name = kstrdup(node_name, GFP_KERNEL);
+	if (!rate_node->name) {
+		kfree(rate_node);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	refcount_set(&rate_node->refcnt, 1);
+	list_add(&rate_node->list, &devlink->rate_list);
+	devlink_rate_notify(rate_node, DEVLINK_CMD_RATE_NEW);
+	return rate_node;
+}
+EXPORT_SYMBOL_GPL(devl_rate_node_create);
+
 /**
  * devl_rate_leaf_create - create devlink rate leaf
  * @devlink_port: devlink port object to create rate object on
-- 
2.37.2

[PATCH net-next v11 04/11] devlink: Allow for devlink-rate nodes parent reassignment

[Michal Wilczynski]

Currently it's not possible to reassign the parent of the node using one
command. As the previous commit introduced a way to export entire
hierarchy from the driver, being able to modify and reassign parents
become important. This way user might easily change QoS settings without
interrupting traffic.

Example command:
devlink port function rate set pci/0000:4b:00.0/1 parent node_custom_1

This reassigns leaf node parent to node_custom_1.

Signed-off-by: Michal Wilczynski <michal.wilczynski@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
---
 net/core/devlink.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/net/core/devlink.c b/net/core/devlink.c
index 3dfee7cd9929..61d431578f5f 100644
--- a/net/core/devlink.c
+++ b/net/core/devlink.c
@@ -1887,10 +1887,8 @@ devlink_nl_rate_parent_node_set(struct devlink_rate *devlink_rate,
 	int err = -EOPNOTSUPP;
 
 	parent = devlink_rate->parent;
-	if (parent && len) {
-		NL_SET_ERR_MSG_MOD(info->extack, "Rate object already has parent.");
-		return -EBUSY;
-	} else if (parent && !len) {
+
+	if (parent && !len) {
 		if (devlink_rate_is_leaf(devlink_rate))
 			err = ops->rate_leaf_parent_set(devlink_rate, NULL,
 							devlink_rate->priv, NULL,
@@ -1904,7 +1902,7 @@ devlink_nl_rate_parent_node_set(struct devlink_rate *devlink_rate,
 
 		refcount_dec(&parent->refcnt);
 		devlink_rate->parent = NULL;
-	} else if (!parent && len) {
+	} else if (len) {
 		parent = devlink_rate_node_get_by_name(devlink, parent_name);
 		if (IS_ERR(parent))
 			return -ENODEV;
@@ -1931,6 +1929,10 @@ devlink_nl_rate_parent_node_set(struct devlink_rate *devlink_rate,
 		if (err)
 			return err;
 
+		if (devlink_rate->parent)
+			/* we're reassigning to other parent in this case */
+			refcount_dec(&devlink_rate->parent->refcnt);
+
 		refcount_inc(&parent->refcnt);
 		devlink_rate->parent = parent;
 	}
-- 
2.37.2

[PATCH net-next v11 05/11] devlink: Allow to set up parent in devl_rate_leaf_create()

[Michal Wilczynski]

Currently the driver is able to create leaf nodes for the devlink-rate,
but is unable to set parent for them. This wasn't as issue before the
possibility to export hierarchy from the driver. After adding the export
feature, in order for the driver to supply correct hierarchy, it's
necessary for it to be able to supply a parent name to
devl_rate_leaf_create().

Introduce a new parameter 'parent_name' in devl_rate_leaf_create().

Signed-off-by: Michal Wilczynski <michal.wilczynski@intel.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
---
 .../net/ethernet/mellanox/mlx5/core/esw/devlink_port.c   | 4 ++--
 drivers/net/netdevsim/dev.c                              | 2 +-
 include/net/devlink.h                                    | 4 +++-
 net/core/devlink.c                                       | 9 ++++++++-
 4 files changed, 14 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c
index 9bc7be95db54..084a910bb4e7 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c
@@ -91,7 +91,7 @@ int mlx5_esw_offloads_devlink_port_register(struct mlx5_eswitch *esw, u16 vport_
 	if (err)
 		goto reg_err;
 
-	err = devl_rate_leaf_create(dl_port, vport);
+	err = devl_rate_leaf_create(dl_port, vport, NULL);
 	if (err)
 		goto rate_err;
 
@@ -160,7 +160,7 @@ int mlx5_esw_devlink_sf_port_register(struct mlx5_eswitch *esw, struct devlink_p
 	if (err)
 		return err;
 
-	err = devl_rate_leaf_create(dl_port, vport);
+	err = devl_rate_leaf_create(dl_port, vport, NULL);
 	if (err)
 		goto rate_err;
 
diff --git a/drivers/net/netdevsim/dev.c b/drivers/net/netdevsim/dev.c
index 387c05953a8b..a5bd6dcca980 100644
--- a/drivers/net/netdevsim/dev.c
+++ b/drivers/net/netdevsim/dev.c
@@ -1401,7 +1401,7 @@ static int __nsim_dev_port_add(struct nsim_dev *nsim_dev, enum nsim_dev_port_typ
 
 	if (nsim_dev_port_is_vf(nsim_dev_port)) {
 		err = devl_rate_leaf_create(&nsim_dev_port->devlink_port,
-					    nsim_dev_port);
+					    nsim_dev_port, NULL);
 		if (err)
 			goto err_nsim_destroy;
 	}
diff --git a/include/net/devlink.h b/include/net/devlink.h
index 339a2ed02d36..074a79b8933f 100644
--- a/include/net/devlink.h
+++ b/include/net/devlink.h
@@ -1617,10 +1617,12 @@ void devlink_port_attrs_pci_vf_set(struct devlink_port *devlink_port, u32 contro
 void devlink_port_attrs_pci_sf_set(struct devlink_port *devlink_port,
 				   u32 controller, u16 pf, u32 sf,
 				   bool external);
-int devl_rate_leaf_create(struct devlink_port *port, void *priv);
 struct devlink_rate *
 devl_rate_node_create(struct devlink *devlink, void *priv, char *node_name,
 		      struct devlink_rate *parent);
+int
+devl_rate_leaf_create(struct devlink_port *devlink_port, void *priv,
+		      struct devlink_rate *parent);
 void devl_rate_leaf_destroy(struct devlink_port *devlink_port);
 void devl_rate_nodes_destroy(struct devlink *devlink);
 void devlink_port_linecard_set(struct devlink_port *devlink_port,
diff --git a/net/core/devlink.c b/net/core/devlink.c
index 61d431578f5f..d93bc95cd7cb 100644
--- a/net/core/devlink.c
+++ b/net/core/devlink.c
@@ -10435,10 +10435,12 @@ EXPORT_SYMBOL_GPL(devl_rate_node_create);
  * devl_rate_leaf_create - create devlink rate leaf
  * @devlink_port: devlink port object to create rate object on
  * @priv: driver private data
+ * @parent: parent devlink_rate struct
  *
  * Create devlink rate object of type leaf on provided @devlink_port.
  */
-int devl_rate_leaf_create(struct devlink_port *devlink_port, void *priv)
+int devl_rate_leaf_create(struct devlink_port *devlink_port, void *priv,
+			  struct devlink_rate *parent)
 {
 	struct devlink *devlink = devlink_port->devlink;
 	struct devlink_rate *devlink_rate;
@@ -10452,6 +10454,11 @@ int devl_rate_leaf_create(struct devlink_port *devlink_port, void *priv)
 	if (!devlink_rate)
 		return -ENOMEM;
 
+	if (parent) {
+		devlink_rate->parent = parent;
+		refcount_inc(&devlink_rate->parent->refcnt);
+	}
+
 	devlink_rate->type = DEVLINK_RATE_TYPE_LEAF;
 	devlink_rate->devlink = devlink;
 	devlink_rate->devlink_port = devlink_port;
-- 
2.37.2

[PATCH net-next v11 06/11] ice: Introduce new parameters in ice_sched_node

[Michal Wilczynski]

To support new devlink-rate API ice_sched_node struct needs to store
a number of additional parameters. This includes tx_max, tx_share,
tx_weight, and tx_priority.

Add new fields to ice_sched_node struct. Add new functions to configure
the hardware with new parameters. Introduce new xarray to identify
nodes uniquely.

Signed-off-by: Michal Wilczynski <michal.wilczynski@intel.com>
---
 .../net/ethernet/intel/ice/ice_adminq_cmd.h   |  4 +-
 drivers/net/ethernet/intel/ice/ice_common.c   |  3 +
 drivers/net/ethernet/intel/ice/ice_sched.c    | 81 +++++++++++++++++--
 drivers/net/ethernet/intel/ice/ice_sched.h    | 27 +++++++
 drivers/net/ethernet/intel/ice/ice_type.h     |  8 ++
 5 files changed, 116 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_adminq_cmd.h b/drivers/net/ethernet/intel/ice/ice_adminq_cmd.h
index 1bdc70aa979d..958c1e435232 100644
--- a/drivers/net/ethernet/intel/ice/ice_adminq_cmd.h
+++ b/drivers/net/ethernet/intel/ice/ice_adminq_cmd.h
@@ -848,9 +848,9 @@ struct ice_aqc_txsched_elem {
 	u8 generic;
 #define ICE_AQC_ELEM_GENERIC_MODE_M		0x1
 #define ICE_AQC_ELEM_GENERIC_PRIO_S		0x1
-#define ICE_AQC_ELEM_GENERIC_PRIO_M	(0x7 << ICE_AQC_ELEM_GENERIC_PRIO_S)
+#define ICE_AQC_ELEM_GENERIC_PRIO_M	        GENMASK(3, 1)
 #define ICE_AQC_ELEM_GENERIC_SP_S		0x4
-#define ICE_AQC_ELEM_GENERIC_SP_M	(0x1 << ICE_AQC_ELEM_GENERIC_SP_S)
+#define ICE_AQC_ELEM_GENERIC_SP_M	        GENMASK(4, 4)
 #define ICE_AQC_ELEM_GENERIC_ADJUST_VAL_S	0x5
 #define ICE_AQC_ELEM_GENERIC_ADJUST_VAL_M	\
 	(0x3 << ICE_AQC_ELEM_GENERIC_ADJUST_VAL_S)
diff --git a/drivers/net/ethernet/intel/ice/ice_common.c b/drivers/net/ethernet/intel/ice/ice_common.c
index 039342a0ed15..e2e661010176 100644
--- a/drivers/net/ethernet/intel/ice/ice_common.c
+++ b/drivers/net/ethernet/intel/ice/ice_common.c
@@ -1105,6 +1105,9 @@ int ice_init_hw(struct ice_hw *hw)
 
 	hw->evb_veb = true;
 
+	/* init xarray for identifying scheduling nodes uniquely */
+	xa_init_flags(&hw->port_info->sched_node_ids, XA_FLAGS_ALLOC);
+
 	/* Query the allocated resources for Tx scheduler */
 	status = ice_sched_query_res_alloc(hw);
 	if (status) {
diff --git a/drivers/net/ethernet/intel/ice/ice_sched.c b/drivers/net/ethernet/intel/ice/ice_sched.c
index 118595763bba..980543074ddb 100644
--- a/drivers/net/ethernet/intel/ice/ice_sched.c
+++ b/drivers/net/ethernet/intel/ice/ice_sched.c
@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 /* Copyright (c) 2018, Intel Corporation. */
 
+#include <net/devlink.h>
 #include "ice_sched.h"
 
 /**
@@ -355,6 +356,9 @@ void ice_free_sched_node(struct ice_port_info *pi, struct ice_sched_node *node)
 	/* leaf nodes have no children */
 	if (node->children)
 		devm_kfree(ice_hw_to_dev(hw), node->children);
+
+	kfree(node->name);
+	xa_erase(&pi->sched_node_ids, node->id);
 	devm_kfree(ice_hw_to_dev(hw), node);
 }
 
@@ -875,7 +879,7 @@ void ice_sched_cleanup_all(struct ice_hw *hw)
  *
  * This function add nodes to HW as well as to SW DB for a given layer
  */
-static int
+int
 ice_sched_add_elems(struct ice_port_info *pi, struct ice_sched_node *tc_node,
 		    struct ice_sched_node *parent, u8 layer, u16 num_nodes,
 		    u16 *num_nodes_added, u32 *first_node_teid)
@@ -940,6 +944,22 @@ ice_sched_add_elems(struct ice_port_info *pi, struct ice_sched_node *tc_node,
 
 		new_node->sibling = NULL;
 		new_node->tc_num = tc_node->tc_num;
+		new_node->tx_weight = ICE_SCHED_DFLT_BW_WT;
+		new_node->tx_share = ICE_SCHED_DFLT_BW;
+		new_node->tx_max = ICE_SCHED_DFLT_BW;
+		new_node->name = kzalloc(SCHED_NODE_NAME_MAX_LEN, GFP_KERNEL);
+		if (!new_node->name)
+			return -ENOMEM;
+
+		status = xa_alloc(&pi->sched_node_ids, &new_node->id, NULL, XA_LIMIT(0, UINT_MAX),
+				  GFP_KERNEL);
+		if (status) {
+			ice_debug(hw, ICE_DBG_SCHED, "xa_alloc failed for sched node status =%d\n",
+				  status);
+			break;
+		}
+
+		snprintf(new_node->name, SCHED_NODE_NAME_MAX_LEN, "node_%u", new_node->id);
 
 		/* add it to previous node sibling pointer */
 		/* Note: siblings are not linked across branches */
@@ -2154,7 +2174,7 @@ ice_sched_get_free_vsi_parent(struct ice_hw *hw, struct ice_sched_node *node,
  * This function removes the child from the old parent and adds it to a new
  * parent
  */
-static void
+void
 ice_sched_update_parent(struct ice_sched_node *new_parent,
 			struct ice_sched_node *node)
 {
@@ -2188,7 +2208,7 @@ ice_sched_update_parent(struct ice_sched_node *new_parent,
  *
  * This function move the child nodes to a given parent.
  */
-static int
+int
 ice_sched_move_nodes(struct ice_port_info *pi, struct ice_sched_node *parent,
 		     u16 num_items, u32 *list)
 {
@@ -3560,7 +3580,7 @@ ice_sched_set_eir_srl_excl(struct ice_port_info *pi,
  * node's RL profile ID of type CIR, EIR, or SRL, and removes old profile
  * ID from local database. The caller needs to hold scheduler lock.
  */
-static int
+int
 ice_sched_set_node_bw(struct ice_port_info *pi, struct ice_sched_node *node,
 		      enum ice_rl_type rl_type, u32 bw, u8 layer_num)
 {
@@ -3596,6 +3616,57 @@ ice_sched_set_node_bw(struct ice_port_info *pi, struct ice_sched_node *node,
 				       ICE_AQC_RL_PROFILE_TYPE_M, old_id);
 }
 
+/**
+ * ice_sched_set_node_priority - set node's priority
+ * @pi: port information structure
+ * @node: tree node
+ * @priority: number 0-7 representing priority among siblings
+ *
+ * This function sets priority of a node among it's siblings.
+ */
+int
+ice_sched_set_node_priority(struct ice_port_info *pi, struct ice_sched_node *node,
+			    u16 priority)
+{
+	struct ice_aqc_txsched_elem_data buf;
+	struct ice_aqc_txsched_elem *data;
+
+	buf = node->info;
+	data = &buf.data;
+
+	data->valid_sections |= ICE_AQC_ELEM_VALID_GENERIC;
+	data->generic |= FIELD_PREP(ICE_AQC_ELEM_GENERIC_PRIO_M, priority);
+
+	return ice_sched_update_elem(pi->hw, node, &buf);
+}
+
+/**
+ * ice_sched_set_node_weight - set node's weight
+ * @pi: port information structure
+ * @node: tree node
+ * @weight: number 1-200 representing weight for WFQ
+ *
+ * This function sets weight of the node for WFQ algorithm.
+ */
+int
+ice_sched_set_node_weight(struct ice_port_info *pi, struct ice_sched_node *node, u16 weight)
+{
+	struct ice_aqc_txsched_elem_data buf;
+	struct ice_aqc_txsched_elem *data;
+
+	buf = node->info;
+	data = &buf.data;
+
+	data->valid_sections = ICE_AQC_ELEM_VALID_CIR | ICE_AQC_ELEM_VALID_EIR |
+			       ICE_AQC_ELEM_VALID_GENERIC;
+	data->cir_bw.bw_alloc = cpu_to_le16(weight);
+	data->eir_bw.bw_alloc = cpu_to_le16(weight);
+
+	data->generic |= FIELD_PREP(ICE_AQC_ELEM_GENERIC_SP_M, 0x0);
+
+	return ice_sched_update_elem(pi->hw, node, &buf);
+}
+
 /**
  * ice_sched_set_node_bw_lmt - set node's BW limit
  * @pi: port information structure
@@ -3606,7 +3677,7 @@ ice_sched_set_node_bw(struct ice_port_info *pi, struct ice_sched_node *node,
  * It updates node's BW limit parameters like BW RL profile ID of type CIR,
  * EIR, or SRL. The caller needs to hold scheduler lock.
  */
-static int
+int
 ice_sched_set_node_bw_lmt(struct ice_port_info *pi, struct ice_sched_node *node,
 			  enum ice_rl_type rl_type, u32 bw)
 {
diff --git a/drivers/net/ethernet/intel/ice/ice_sched.h b/drivers/net/ethernet/intel/ice/ice_sched.h
index 4f91577fed56..920db43ed4fa 100644
--- a/drivers/net/ethernet/intel/ice/ice_sched.h
+++ b/drivers/net/ethernet/intel/ice/ice_sched.h
@@ -6,6 +6,8 @@
 
 #include "ice_common.h"
 
+#define SCHED_NODE_NAME_MAX_LEN 32
+
 #define ICE_QGRP_LAYER_OFFSET	2
 #define ICE_VSI_LAYER_OFFSET	4
 #define ICE_AGG_LAYER_OFFSET	6
@@ -69,6 +71,28 @@ int
 ice_aq_query_sched_elems(struct ice_hw *hw, u16 elems_req,
 			 struct ice_aqc_txsched_elem_data *buf, u16 buf_size,
 			 u16 *elems_ret, struct ice_sq_cd *cd);
+
+int
+ice_sched_set_node_bw_lmt(struct ice_port_info *pi, struct ice_sched_node *node,
+			  enum ice_rl_type rl_type, u32 bw);
+
+int
+ice_sched_set_node_bw(struct ice_port_info *pi, struct ice_sched_node *node,
+		      enum ice_rl_type rl_type, u32 bw, u8 layer_num);
+
+int
+ice_sched_add_elems(struct ice_port_info *pi, struct ice_sched_node *tc_node,
+		    struct ice_sched_node *parent, u8 layer, u16 num_nodes,
+		    u16 *num_nodes_added, u32 *first_node_teid);
+
+int
+ice_sched_move_nodes(struct ice_port_info *pi, struct ice_sched_node *parent,
+		     u16 num_items, u32 *list);
+
+int ice_sched_set_node_priority(struct ice_port_info *pi, struct ice_sched_node *node,
+				u16 priority);
+int ice_sched_set_node_weight(struct ice_port_info *pi, struct ice_sched_node *node, u16 weight);
+
 int ice_sched_init_port(struct ice_port_info *pi);
 int ice_sched_query_res_alloc(struct ice_hw *hw);
 void ice_sched_get_psm_clk_freq(struct ice_hw *hw);
@@ -82,6 +106,9 @@ ice_sched_find_node_by_teid(struct ice_sched_node *start_node, u32 teid);
 int
 ice_sched_add_node(struct ice_port_info *pi, u8 layer,
 		   struct ice_aqc_txsched_elem_data *info);
+void
+ice_sched_update_parent(struct ice_sched_node *new_parent,
+			struct ice_sched_node *node);
 void ice_free_sched_node(struct ice_port_info *pi, struct ice_sched_node *node);
 struct ice_sched_node *ice_sched_get_tc_node(struct ice_port_info *pi, u8 tc);
 struct ice_sched_node *
diff --git a/drivers/net/ethernet/intel/ice/ice_type.h b/drivers/net/ethernet/intel/ice/ice_type.h
index e1abfcee96dc..daf86cf561bc 100644
--- a/drivers/net/ethernet/intel/ice/ice_type.h
+++ b/drivers/net/ethernet/intel/ice/ice_type.h
@@ -524,7 +524,14 @@ struct ice_sched_node {
 	struct ice_sched_node *sibling; /* next sibling in the same layer */
 	struct ice_sched_node **children;
 	struct ice_aqc_txsched_elem_data info;
+	char *name;
+	struct devlink_rate *rate_node;
+	u64 tx_max;
+	u64 tx_share;
 	u32 agg_id;			/* aggregator group ID */
+	u32 id;
+	u32 tx_priority;
+	u32 tx_weight;
 	u16 vsi_handle;
 	u8 in_use;			/* suspended or in use */
 	u8 tx_sched_layer;		/* Logical Layer (1-9) */
@@ -706,6 +713,7 @@ struct ice_port_info {
 	/* List contain profile ID(s) and other params per layer */
 	struct list_head rl_prof_list[ICE_AQC_TOPO_MAX_LEVEL_NUM];
 	struct ice_qos_cfg qos_cfg;
+	struct xarray sched_node_ids;
 	u8 is_vf:1;
 };
 
-- 
2.37.2

[PATCH net-next v11 07/11] ice: Add an option to pre-allocate memory for ice_sched_node

[Michal Wilczynski]

devlink-rate API requires a priv object to be allocated when node still
doesn't have a parent. This is problematic, because ice_sched_node can't
be currently created without a parent.

Add an option to pre-allocate memory for ice_sched_node struct. Add
new arguments to ice_sched_add() and ice_sched_add_elems() that allow
for pre-allocation of memory for ice_sched_node struct.

Signed-off-by: Michal Wilczynski <michal.wilczynski@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_common.c |  4 ++--
 drivers/net/ethernet/intel/ice/ice_dcb.c    |  2 +-
 drivers/net/ethernet/intel/ice/ice_sched.c  | 23 +++++++++++++++------
 drivers/net/ethernet/intel/ice/ice_sched.h  |  6 ++++--
 4 files changed, 24 insertions(+), 11 deletions(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_common.c b/drivers/net/ethernet/intel/ice/ice_common.c
index e2e661010176..216370ec60d4 100644
--- a/drivers/net/ethernet/intel/ice/ice_common.c
+++ b/drivers/net/ethernet/intel/ice/ice_common.c
@@ -4603,7 +4603,7 @@ ice_ena_vsi_txq(struct ice_port_info *pi, u16 vsi_handle, u8 tc, u16 q_handle,
 	q_ctx->q_teid = le32_to_cpu(node.node_teid);
 
 	/* add a leaf node into scheduler tree queue layer */
-	status = ice_sched_add_node(pi, hw->num_tx_sched_layers - 1, &node);
+	status = ice_sched_add_node(pi, hw->num_tx_sched_layers - 1, &node, NULL);
 	if (!status)
 		status = ice_sched_replay_q_bw(pi, q_ctx);
 
@@ -4838,7 +4838,7 @@ ice_ena_vsi_rdma_qset(struct ice_port_info *pi, u16 vsi_handle, u8 tc,
 	for (i = 0; i < num_qsets; i++) {
 		node.node_teid = buf->rdma_qsets[i].qset_teid;
 		ret = ice_sched_add_node(pi, hw->num_tx_sched_layers - 1,
-					 &node);
+					 &node, NULL);
 		if (ret)
 			break;
 		qset_teid[i] = le32_to_cpu(node.node_teid);
diff --git a/drivers/net/ethernet/intel/ice/ice_dcb.c b/drivers/net/ethernet/intel/ice/ice_dcb.c
index 0b146a0d4205..6be02f9b0b8c 100644
--- a/drivers/net/ethernet/intel/ice/ice_dcb.c
+++ b/drivers/net/ethernet/intel/ice/ice_dcb.c
@@ -1580,7 +1580,7 @@ ice_update_port_tc_tree_cfg(struct ice_port_info *pi,
 		/* new TC */
 		status = ice_sched_query_elem(pi->hw, teid2, &elem);
 		if (!status)
-			status = ice_sched_add_node(pi, 1, &elem);
+			status = ice_sched_add_node(pi, 1, &elem, NULL);
 		if (status)
 			break;
 		/* update the TC number */
diff --git a/drivers/net/ethernet/intel/ice/ice_sched.c b/drivers/net/ethernet/intel/ice/ice_sched.c
index 980543074ddb..6d08b397df2a 100644
--- a/drivers/net/ethernet/intel/ice/ice_sched.c
+++ b/drivers/net/ethernet/intel/ice/ice_sched.c
@@ -143,12 +143,14 @@ ice_aq_query_sched_elems(struct ice_hw *hw, u16 elems_req,
  * @pi: port information structure
  * @layer: Scheduler layer of the node
  * @info: Scheduler element information from firmware
+ * @prealloc_node: preallocated ice_sched_node struct for SW DB
  *
  * This function inserts a scheduler node to the SW DB.
  */
 int
 ice_sched_add_node(struct ice_port_info *pi, u8 layer,
-		   struct ice_aqc_txsched_elem_data *info)
+		   struct ice_aqc_txsched_elem_data *info,
+		   struct ice_sched_node *prealloc_node)
 {
 	struct ice_aqc_txsched_elem_data elem;
 	struct ice_sched_node *parent;
@@ -177,7 +179,10 @@ ice_sched_add_node(struct ice_port_info *pi, u8 layer,
 	if (status)
 		return status;
 
-	node = devm_kzalloc(ice_hw_to_dev(hw), sizeof(*node), GFP_KERNEL);
+	if (prealloc_node)
+		node = prealloc_node;
+	else
+		node = devm_kzalloc(ice_hw_to_dev(hw), sizeof(*node), GFP_KERNEL);
 	if (!node)
 		return -ENOMEM;
 	if (hw->max_children[layer]) {
@@ -876,13 +881,15 @@ void ice_sched_cleanup_all(struct ice_hw *hw)
  * @num_nodes: number of nodes
  * @num_nodes_added: pointer to num nodes added
  * @first_node_teid: if new nodes are added then return the TEID of first node
+ * @prealloc_nodes: preallocated nodes struct for software DB
  *
  * This function add nodes to HW as well as to SW DB for a given layer
  */
 int
 ice_sched_add_elems(struct ice_port_info *pi, struct ice_sched_node *tc_node,
 		    struct ice_sched_node *parent, u8 layer, u16 num_nodes,
-		    u16 *num_nodes_added, u32 *first_node_teid)
+		    u16 *num_nodes_added, u32 *first_node_teid,
+		    struct ice_sched_node **prealloc_nodes)
 {
 	struct ice_sched_node *prev, *new_node;
 	struct ice_aqc_add_elem *buf;
@@ -928,7 +935,11 @@ ice_sched_add_elems(struct ice_port_info *pi, struct ice_sched_node *tc_node,
 	*num_nodes_added = num_nodes;
 	/* add nodes to the SW DB */
 	for (i = 0; i < num_nodes; i++) {
-		status = ice_sched_add_node(pi, layer, &buf->generic[i]);
+		if (prealloc_nodes)
+			status = ice_sched_add_node(pi, layer, &buf->generic[i], prealloc_nodes[i]);
+		else
+			status = ice_sched_add_node(pi, layer, &buf->generic[i], NULL);
+
 		if (status) {
 			ice_debug(hw, ICE_DBG_SCHED, "add nodes in SW DB failed status =%d\n",
 				  status);
@@ -1023,7 +1034,7 @@ ice_sched_add_nodes_to_hw_layer(struct ice_port_info *pi,
 	}
 
 	return ice_sched_add_elems(pi, tc_node, parent, layer, num_nodes,
-				   num_nodes_added, first_node_teid);
+				   num_nodes_added, first_node_teid, NULL);
 }
 
 /**
@@ -1288,7 +1299,7 @@ int ice_sched_init_port(struct ice_port_info *pi)
 			    ICE_AQC_ELEM_TYPE_ENTRY_POINT)
 				hw->sw_entry_point_layer = j;
 
-			status = ice_sched_add_node(pi, j, &buf[i].generic[j]);
+			status = ice_sched_add_node(pi, j, &buf[i].generic[j], NULL);
 			if (status)
 				goto err_init_port;
 		}
diff --git a/drivers/net/ethernet/intel/ice/ice_sched.h b/drivers/net/ethernet/intel/ice/ice_sched.h
index 920db43ed4fa..9c100747445a 100644
--- a/drivers/net/ethernet/intel/ice/ice_sched.h
+++ b/drivers/net/ethernet/intel/ice/ice_sched.h
@@ -83,7 +83,8 @@ ice_sched_set_node_bw(struct ice_port_info *pi, struct ice_sched_node *node,
 int
 ice_sched_add_elems(struct ice_port_info *pi, struct ice_sched_node *tc_node,
 		    struct ice_sched_node *parent, u8 layer, u16 num_nodes,
-		    u16 *num_nodes_added, u32 *first_node_teid);
+		    u16 *num_nodes_added, u32 *first_node_teid,
+		    struct ice_sched_node **prealloc_node);
 
 int
 ice_sched_move_nodes(struct ice_port_info *pi, struct ice_sched_node *parent,
@@ -105,7 +106,8 @@ struct ice_sched_node *
 ice_sched_find_node_by_teid(struct ice_sched_node *start_node, u32 teid);
 int
 ice_sched_add_node(struct ice_port_info *pi, u8 layer,
-		   struct ice_aqc_txsched_elem_data *info);
+		   struct ice_aqc_txsched_elem_data *info,
+		   struct ice_sched_node *prealloc_node);
 void
 ice_sched_update_parent(struct ice_sched_node *new_parent,
 			struct ice_sched_node *node);
-- 
2.37.2

[PATCH net-next v11 08/11] ice: Implement devlink-rate API

[Michal Wilczynski]

There is a need to support modification of Tx scheduler tree, in the
ice driver. This will allow user to control Tx settings of each node in
the internal hierarchy of nodes. As a result user will be able to use
Hierarchy QoS implemented entirely in the hardware.

This patch implemenents devlink-rate API. It also exports initial
default hierarchy. It's mostly dictated by the fact that the tree
can't be removed entirely, all we can do is enable the user to modify
it. For example root node shouldn't ever be removed, also nodes that
have children are off-limits.

Example initial tree with 2 VF's:

[root@fedora ~]# devlink port function rate show

pci/0000:4b:00.0/node_27: type node parent node_26
pci/0000:4b:00.0/node_26: type node parent node_0
pci/0000:4b:00.0/node_34: type node parent node_33
pci/0000:4b:00.0/node_33: type node parent node_32
pci/0000:4b:00.0/node_32: type node parent node_16
pci/0000:4b:00.0/node_19: type node parent node_18
pci/0000:4b:00.0/node_18: type node parent node_17
pci/0000:4b:00.0/node_17: type node parent node_16
pci/0000:4b:00.0/node_21: type node parent node_20
pci/0000:4b:00.0/node_20: type node parent node_3
pci/0000:4b:00.0/node_14: type node parent node_5
pci/0000:4b:00.0/node_5: type node parent node_3
pci/0000:4b:00.0/node_13: type node parent node_4
pci/0000:4b:00.0/node_12: type node parent node_4
pci/0000:4b:00.0/node_11: type node parent node_4
pci/0000:4b:00.0/node_10: type node parent node_4
pci/0000:4b:00.0/node_9: type node parent node_4
pci/0000:4b:00.0/node_8: type node parent node_4
pci/0000:4b:00.0/node_7: type node parent node_4
pci/0000:4b:00.0/node_6: type node parent node_4
pci/0000:4b:00.0/node_4: type node parent node_3
pci/0000:4b:00.0/node_3: type node parent node_16
pci/0000:4b:00.0/node_16: type node parent node_15
pci/0000:4b:00.0/node_15: type node parent node_0
pci/0000:4b:00.0/node_2: type node parent node_1
pci/0000:4b:00.0/node_1: type node parent node_0
pci/0000:4b:00.0/node_0: type node
pci/0000:4b:00.0/1: type leaf parent node_27
pci/0000:4b:00.0/2: type leaf parent node_27

Let me visualize part of the tree:

                    +---------+
                    |  node_0 |
                    +---------+
                         |
                    +----v----+
                    | node_26 |
                    +----+----+
                         |
                    +----v----+
                    | node_27 |
                    +----+----+
                         |
                |-----------------|
           +----v----+       +----v----+
           |   VF 1  |       |   VF 2  |
           +----+----+       +----+----+

So at this point there is a couple things that can be done.
For example we could only assign parameters to VF's.

[root@fedora ~]# devlink port function rate set pci/0000:4b:00.0/1 \
                 tx_max 5Gbps

This would cap the VF 1 BW to 5Gbps.

But let's say you would like to create a completely new branch.
This can be done like this:

[root@fedora ~]# devlink port function rate add \
                 pci/0000:4b:00.0/node_custom parent node_0
[root@fedora ~]# devlink port function rate add \
                 pci/0000:4b:00.0/node_custom_1 parent node_custom
[root@fedora ~]# devlink port function rate set \
                 pci/0000:4b:00.0/1 parent node_custom_1

This creates a completely new branch and reassigns VF 1 to it.

A number of parameters is supported per each node: tx_max, tx_share,
tx_priority and tx_weight.

Signed-off-by: Michal Wilczynski <michal.wilczynski@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_devlink.c | 407 +++++++++++++++++++
 drivers/net/ethernet/intel/ice/ice_devlink.h |   2 +
 drivers/net/ethernet/intel/ice/ice_repr.c    |  13 +
 3 files changed, 422 insertions(+)

diff --git a/drivers/net/ethernet/intel/ice/ice_devlink.c b/drivers/net/ethernet/intel/ice/ice_devlink.c
index 455489e9457d..3e93f20fec44 100644
--- a/drivers/net/ethernet/intel/ice/ice_devlink.c
+++ b/drivers/net/ethernet/intel/ice/ice_devlink.c
@@ -713,6 +713,396 @@ ice_devlink_port_unsplit(struct devlink *devlink, struct devlink_port *port,
 	return ice_devlink_port_split(devlink, port, 1, extack);
 }
 
+/**
+ * ice_traverse_tx_tree - traverse Tx scheduler tree
+ * @devlink: devlink struct
+ * @node: current node, used for recursion
+ * @tc_node: tc_node struct, that is treated as a root
+ * @pf: pf struct
+ *
+ * This function traverses Tx scheduler tree and exports
+ * entire structure to the devlink-rate.
+ */
+static void ice_traverse_tx_tree(struct devlink *devlink, struct ice_sched_node *node,
+				 struct ice_sched_node *tc_node, struct ice_pf *pf)
+{
+	struct devlink_rate *rate_node = NULL;
+	struct ice_vf *vf;
+	int i;
+
+	if (node->parent == tc_node) {
+		/* create root node */
+		rate_node = devl_rate_node_create(devlink, node, node->name, NULL);
+	} else if (node->vsi_handle &&
+		   pf->vsi[node->vsi_handle]->vf) {
+		vf = pf->vsi[node->vsi_handle]->vf;
+		if (!vf->devlink_port.devlink_rate)
+			/* leaf nodes doesn't have children
+			 * so we don't set rate_node
+			 */
+			devl_rate_leaf_create(&vf->devlink_port, node,
+					      node->parent->rate_node);
+	} else if (node->info.data.elem_type != ICE_AQC_ELEM_TYPE_LEAF &&
+		   node->parent->rate_node) {
+		rate_node = devl_rate_node_create(devlink, node, node->name,
+						  node->parent->rate_node);
+	}
+
+	if (rate_node && !IS_ERR(rate_node))
+		node->rate_node = rate_node;
+
+	for (i = 0; i < node->num_children; i++)
+		ice_traverse_tx_tree(devlink, node->children[i], tc_node, pf);
+}
+
+/**
+ * ice_devlink_rate_init_tx_topology - export Tx scheduler tree to devlink rate
+ * @devlink: devlink struct
+ * @vsi: main vsi struct
+ *
+ * This function finds a root node, then calls ice_traverse_tx tree, which
+ * traverses the tree and export it's contents to devlink rate.
+ */
+int ice_devlink_rate_init_tx_topology(struct devlink *devlink, struct ice_vsi *vsi)
+{
+	struct ice_port_info *pi = vsi->port_info;
+	struct ice_sched_node *tc_node;
+	struct ice_pf *pf = vsi->back;
+	int i;
+
+	tc_node = pi->root->children[0];
+	mutex_lock(&pi->sched_lock);
+	devl_lock(devlink);
+	for (i = 0; i < tc_node->num_children; i++)
+		ice_traverse_tx_tree(devlink, tc_node->children[i], tc_node, pf);
+	devl_unlock(devlink);
+	mutex_unlock(&pi->sched_lock);
+
+	return 0;
+}
+
+/**
+ * ice_set_object_tx_share - sets node scheduling parameter
+ * @pi: devlink struct instance
+ * @node: node struct instance
+ * @extack: extended netdev ack structure
+ *
+ * This function sets ICE_MIN_BW scheduling BW limit.
+ */
+static int ice_set_object_tx_share(struct ice_port_info *pi, struct ice_sched_node *node,
+				   u64 bw, struct netlink_ext_ack *extack)
+{
+	int status;
+
+	mutex_lock(&pi->sched_lock);
+	node->tx_share = div_u64(bw, 125);
+	status = ice_sched_set_node_bw_lmt(pi, node, ICE_MIN_BW, node->tx_share);
+	mutex_unlock(&pi->sched_lock);
+
+	if (status)
+		NL_SET_ERR_MSG_MOD(extack, "Can't set scheduling node tx_share");
+
+	return status;
+}
+
+/**
+ * ice_set_object_tx_max - sets node scheduling parameter
+ * @pi: devlink struct instance
+ * @node: node struct instance
+ * @extack: extended netdev ack structure
+ *
+ * This function sets ICE_MAX_BW scheduling BW limit.
+ */
+static int ice_set_object_tx_max(struct ice_port_info *pi, struct ice_sched_node *node,
+				 u64 bw, struct netlink_ext_ack *extack)
+{
+	int status;
+
+	mutex_lock(&pi->sched_lock);
+	node->tx_max = div_u64(bw, 125);
+	status = ice_sched_set_node_bw_lmt(pi, node, ICE_MAX_BW, node->tx_max);
+	mutex_unlock(&pi->sched_lock);
+
+	if (status)
+		NL_SET_ERR_MSG_MOD(extack, "Can't set scheduling node tx_max");
+
+	return status;
+}
+
+/**
+ * ice_set_object_tx_priority - sets node scheduling parameter
+ * @pi: devlink struct instance
+ * @node: node struct instance
+ * @extack: extended netdev ack structure
+ *
+ * This function sets priority of node among siblings.
+ */
+static int ice_set_object_tx_priority(struct ice_port_info *pi, struct ice_sched_node *node,
+				      u32 priority, struct netlink_ext_ack *extack)
+{
+	int status;
+
+	if (node->tx_priority >= 8) {
+		NL_SET_ERR_MSG_MOD(extack, "Priority should be less than 8");
+		return -EINVAL;
+	}
+
+	mutex_lock(&pi->sched_lock);
+	node->tx_priority = priority;
+	status = ice_sched_set_node_priority(pi, node, node->tx_priority);
+	mutex_unlock(&pi->sched_lock);
+
+	if (status)
+		NL_SET_ERR_MSG_MOD(extack, "Can't set scheduling node tx_priority");
+
+	return status;
+}
+
+/**
+ * ice_set_object_tx_weight - sets node scheduling parameter
+ * @pi: devlink struct instance
+ * @node: node struct instance
+ * @extack: extended netdev ack structure
+ *
+ * This function sets node weight for WFQ algorithm.
+ */
+static int ice_set_object_tx_weight(struct ice_port_info *pi, struct ice_sched_node *node,
+				    u32 weight, struct netlink_ext_ack *extack)
+{
+	int status;
+
+	if (node->tx_weight > 200 || node->tx_weight < 1) {
+		NL_SET_ERR_MSG_MOD(extack, "Weight must be between 1 and 200");
+		return -EINVAL;
+	}
+
+	mutex_lock(&pi->sched_lock);
+	node->tx_weight = weight;
+	status = ice_sched_set_node_weight(pi, node, node->tx_weight);
+	mutex_unlock(&pi->sched_lock);
+
+	if (status)
+		NL_SET_ERR_MSG_MOD(extack, "Can't set scheduling node tx_weight");
+
+	return status;
+}
+
+/**
+ * ice_get_pi_from_dev_rate - get port info from devlink_rate
+ * @rate_node: devlink struct instance
+ *
+ * This function returns corresponding port_info struct of devlink_rate
+ */
+static struct ice_port_info *ice_get_pi_from_dev_rate(struct devlink_rate *rate_node)
+{
+	struct ice_pf *pf = devlink_priv(rate_node->devlink);
+
+	return ice_get_main_vsi(pf)->port_info;
+}
+
+static int ice_devlink_rate_node_new(struct devlink_rate *rate_node, void **priv,
+				     struct netlink_ext_ack *extack)
+{
+	struct ice_sched_node *node;
+	struct ice_port_info *pi;
+
+	pi = ice_get_pi_from_dev_rate(rate_node);
+
+	/* preallocate memory for ice_sched_node */
+	node = devm_kzalloc(ice_hw_to_dev(pi->hw), sizeof(*node), GFP_KERNEL);
+	*priv = node;
+
+	return 0;
+}
+
+static int ice_devlink_rate_node_del(struct devlink_rate *rate_node, void *priv,
+				     struct netlink_ext_ack *extack)
+{
+	struct ice_sched_node *node, *tc_node;
+	struct ice_port_info *pi;
+
+	pi = ice_get_pi_from_dev_rate(rate_node);
+	tc_node = pi->root->children[0];
+	node = priv;
+
+	if (!rate_node->parent || !node || tc_node == node || !extack)
+		return 0;
+
+	/* can't allow to delete a node with children */
+	if (node->num_children)
+		return -EINVAL;
+
+	mutex_lock(&pi->sched_lock);
+	ice_free_sched_node(pi, node);
+	mutex_unlock(&pi->sched_lock);
+
+	return 0;
+}
+
+static int ice_devlink_rate_leaf_tx_max_set(struct devlink_rate *rate_leaf, void *priv,
+					    u64 tx_max, struct netlink_ext_ack *extack)
+{
+	struct ice_sched_node *node = priv;
+
+	if (!node)
+		return 0;
+
+	return ice_set_object_tx_max(ice_get_pi_from_dev_rate(rate_leaf),
+				     node, tx_max, extack);
+}
+
+static int ice_devlink_rate_leaf_tx_share_set(struct devlink_rate *rate_leaf, void *priv,
+					      u64 tx_share, struct netlink_ext_ack *extack)
+{
+	struct ice_sched_node *node = priv;
+
+	if (!node)
+		return 0;
+
+	return ice_set_object_tx_share(ice_get_pi_from_dev_rate(rate_leaf), node,
+				       tx_share, extack);
+}
+
+static int ice_devlink_rate_leaf_tx_priority_set(struct devlink_rate *rate_leaf, void *priv,
+						 u32 tx_priority, struct netlink_ext_ack *extack)
+{
+	struct ice_sched_node *node = priv;
+
+	if (!node)
+		return 0;
+
+	return ice_set_object_tx_priority(ice_get_pi_from_dev_rate(rate_leaf), node,
+					  tx_priority, extack);
+}
+
+static int ice_devlink_rate_leaf_tx_weight_set(struct devlink_rate *rate_leaf, void *priv,
+					       u32 tx_weight, struct netlink_ext_ack *extack)
+{
+	struct ice_sched_node *node = priv;
+
+	if (!node)
+		return 0;
+
+	return ice_set_object_tx_weight(ice_get_pi_from_dev_rate(rate_leaf), node,
+					tx_weight, extack);
+}
+
+static int ice_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node, void *priv,
+					    u64 tx_max, struct netlink_ext_ack *extack)
+{
+	struct ice_sched_node *node = priv;
+
+	if (!node)
+		return 0;
+
+	return ice_set_object_tx_max(ice_get_pi_from_dev_rate(rate_node),
+				     node, tx_max, extack);
+}
+
+static int ice_devlink_rate_node_tx_share_set(struct devlink_rate *rate_node, void *priv,
+					      u64 tx_share, struct netlink_ext_ack *extack)
+{
+	struct ice_sched_node *node = priv;
+
+	if (!node)
+		return 0;
+
+	return ice_set_object_tx_share(ice_get_pi_from_dev_rate(rate_node),
+				       node, tx_share, extack);
+}
+
+static int ice_devlink_rate_node_tx_priority_set(struct devlink_rate *rate_node, void *priv,
+						 u32 tx_priority, struct netlink_ext_ack *extack)
+{
+	struct ice_sched_node *node = priv;
+
+	if (!node)
+		return 0;
+
+	return ice_set_object_tx_priority(ice_get_pi_from_dev_rate(rate_node),
+					  node, tx_priority, extack);
+}
+
+static int ice_devlink_rate_node_tx_weight_set(struct devlink_rate *rate_node, void *priv,
+					       u32 tx_weight, struct netlink_ext_ack *extack)
+{
+	struct ice_sched_node *node = priv;
+
+	if (!node)
+		return 0;
+
+	return ice_set_object_tx_weight(ice_get_pi_from_dev_rate(rate_node),
+					node, tx_weight, extack);
+}
+
+static int ice_devlink_set_parent(struct devlink_rate *devlink_rate,
+				  struct devlink_rate *parent,
+				  void *priv, void *parent_priv,
+				  struct netlink_ext_ack *extack)
+{
+	struct ice_port_info *pi = ice_get_pi_from_dev_rate(devlink_rate);
+	struct ice_sched_node *tc_node, *node, *parent_node;
+	u16 num_nodes_added;
+	u32 first_node_teid;
+	u32 node_teid;
+	int status;
+
+	tc_node = pi->root->children[0];
+	node = priv;
+
+	if (!extack)
+		return 0;
+
+	if (!parent) {
+		if (!node || tc_node == node || node->num_children)
+			return -EINVAL;
+
+		mutex_lock(&pi->sched_lock);
+		ice_free_sched_node(pi, node);
+		mutex_unlock(&pi->sched_lock);
+
+		return 0;
+	}
+
+	parent_node = parent_priv;
+
+	/* if the node doesn't exist, create it */
+	if (!node->parent) {
+		mutex_lock(&pi->sched_lock);
+
+		status = ice_sched_add_elems(pi, tc_node, parent_node,
+					     parent_node->tx_sched_layer + 1,
+					     1, &num_nodes_added, &first_node_teid,
+					     &node);
+
+		mutex_unlock(&pi->sched_lock);
+
+		if (status) {
+			NL_SET_ERR_MSG_MOD(extack, "Can't add a new node");
+			return status;
+		}
+
+		if (devlink_rate->tx_share)
+			ice_set_object_tx_share(pi, node, devlink_rate->tx_share, extack);
+		if (devlink_rate->tx_max)
+			ice_set_object_tx_max(pi, node, devlink_rate->tx_max, extack);
+		if (devlink_rate->tx_priority)
+			ice_set_object_tx_priority(pi, node, devlink_rate->tx_priority, extack);
+		if (devlink_rate->tx_weight)
+			ice_set_object_tx_weight(pi, node, devlink_rate->tx_weight, extack);
+	} else {
+		node_teid = le32_to_cpu(node->info.node_teid);
+		mutex_lock(&pi->sched_lock);
+		status = ice_sched_move_nodes(pi, parent_node, 1, &node_teid);
+		mutex_unlock(&pi->sched_lock);
+
+		if (status)
+			NL_SET_ERR_MSG_MOD(extack, "Can't move existing node to a new parent");
+	}
+
+	return status;
+}
+
 static const struct devlink_ops ice_devlink_ops = {
 	.supported_flash_update_params = DEVLINK_SUPPORT_FLASH_UPDATE_OVERWRITE_MASK,
 	.reload_actions = BIT(DEVLINK_RELOAD_ACTION_FW_ACTIVATE),
@@ -725,6 +1115,22 @@ static const struct devlink_ops ice_devlink_ops = {
 	.eswitch_mode_set = ice_eswitch_mode_set,
 	.info_get = ice_devlink_info_get,
 	.flash_update = ice_devlink_flash_update,
+
+	.rate_node_new = ice_devlink_rate_node_new,
+	.rate_node_del = ice_devlink_rate_node_del,
+
+	.rate_leaf_tx_max_set = ice_devlink_rate_leaf_tx_max_set,
+	.rate_leaf_tx_share_set = ice_devlink_rate_leaf_tx_share_set,
+	.rate_leaf_tx_priority_set = ice_devlink_rate_leaf_tx_priority_set,
+	.rate_leaf_tx_weight_set = ice_devlink_rate_leaf_tx_weight_set,
+
+	.rate_node_tx_max_set = ice_devlink_rate_node_tx_max_set,
+	.rate_node_tx_share_set = ice_devlink_rate_node_tx_share_set,
+	.rate_node_tx_priority_set = ice_devlink_rate_node_tx_priority_set,
+	.rate_node_tx_weight_set = ice_devlink_rate_node_tx_weight_set,
+
+	.rate_leaf_parent_set = ice_devlink_set_parent,
+	.rate_node_parent_set = ice_devlink_set_parent,
 };
 
 static int
@@ -1089,6 +1495,7 @@ int ice_devlink_create_vf_port(struct ice_vf *vf)
  */
 void ice_devlink_destroy_vf_port(struct ice_vf *vf)
 {
+	devl_rate_leaf_destroy(&vf->devlink_port);
 	devlink_port_unregister(&vf->devlink_port);
 }
 
diff --git a/drivers/net/ethernet/intel/ice/ice_devlink.h b/drivers/net/ethernet/intel/ice/ice_devlink.h
index fe006d9946f8..8bfed9ee2c4c 100644
--- a/drivers/net/ethernet/intel/ice/ice_devlink.h
+++ b/drivers/net/ethernet/intel/ice/ice_devlink.h
@@ -18,4 +18,6 @@ void ice_devlink_destroy_vf_port(struct ice_vf *vf);
 void ice_devlink_init_regions(struct ice_pf *pf);
 void ice_devlink_destroy_regions(struct ice_pf *pf);
 
+int ice_devlink_rate_init_tx_topology(struct devlink *devlink, struct ice_vsi *vsi);
+
 #endif /* _ICE_DEVLINK_H_ */
diff --git a/drivers/net/ethernet/intel/ice/ice_repr.c b/drivers/net/ethernet/intel/ice/ice_repr.c
index 0483eb14c288..46f58d48318c 100644
--- a/drivers/net/ethernet/intel/ice/ice_repr.c
+++ b/drivers/net/ethernet/intel/ice/ice_repr.c
@@ -389,6 +389,7 @@ static void ice_repr_rem(struct ice_vf *vf)
  */
 void ice_repr_rem_from_all_vfs(struct ice_pf *pf)
 {
+	struct devlink *devlink;
 	struct ice_vf *vf;
 	unsigned int bkt;
 
@@ -396,6 +397,14 @@ void ice_repr_rem_from_all_vfs(struct ice_pf *pf)
 
 	ice_for_each_vf(pf, bkt, vf)
 		ice_repr_rem(vf);
+
+	/* since all port representors are destroyed, there is
+	 * no point in keeping the nodes
+	 */
+	devlink = priv_to_devlink(pf);
+	devl_lock(devlink);
+	devl_rate_nodes_destroy(devlink);
+	devl_unlock(devlink);
 }
 
 /**
@@ -404,6 +413,7 @@ void ice_repr_rem_from_all_vfs(struct ice_pf *pf)
  */
 int ice_repr_add_for_all_vfs(struct ice_pf *pf)
 {
+	struct devlink *devlink;
 	struct ice_vf *vf;
 	unsigned int bkt;
 	int err;
@@ -416,6 +426,9 @@ int ice_repr_add_for_all_vfs(struct ice_pf *pf)
 			goto err;
 	}
 
+	devlink = priv_to_devlink(pf);
+	ice_devlink_rate_init_tx_topology(devlink, ice_get_main_vsi(pf));
+
 	return 0;
 
 err:
-- 
2.37.2

[PATCH net-next v11 09/11] ice: Prevent ADQ, DCB coexistence with Custom Tx scheduler

[Michal Wilczynski]

ADQ, DCB might interfere with Custom Tx Scheduler changes that user
might introduce using devlink-rate API.

Check if ADQ, DCB is active, when user tries to change any setting
in exported Tx scheduler tree. If any of those are active block the user
from doing so, and log an appropriate message.

Remove the exported hierarchy if user enable ADQ or DCB.
Prevent ADQ or DCB from getting configured if user already made some
changes using devlink-rate API.

Signed-off-by: Michal Wilczynski <michal.wilczynski@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_dcb_lib.c |  7 ++
 drivers/net/ethernet/intel/ice/ice_devlink.c | 91 ++++++++++++++++++++
 drivers/net/ethernet/intel/ice/ice_devlink.h |  1 +
 drivers/net/ethernet/intel/ice/ice_main.c    |  6 ++
 drivers/net/ethernet/intel/ice/ice_repr.c    |  5 ++
 drivers/net/ethernet/intel/ice/ice_type.h    |  1 +
 6 files changed, 111 insertions(+)

diff --git a/drivers/net/ethernet/intel/ice/ice_dcb_lib.c b/drivers/net/ethernet/intel/ice/ice_dcb_lib.c
index add90e75f05c..9defb9d0fe88 100644
--- a/drivers/net/ethernet/intel/ice/ice_dcb_lib.c
+++ b/drivers/net/ethernet/intel/ice/ice_dcb_lib.c
@@ -3,6 +3,7 @@
 
 #include "ice_dcb_lib.h"
 #include "ice_dcb_nl.h"
+#include "ice_devlink.h"
 
 /**
  * ice_dcb_get_ena_tc - return bitmap of enabled TCs
@@ -364,6 +365,12 @@ int ice_pf_dcb_cfg(struct ice_pf *pf, struct ice_dcbx_cfg *new_cfg, bool locked)
 	/* Enable DCB tagging only when more than one TC */
 	if (ice_dcb_get_num_tc(new_cfg) > 1) {
 		dev_dbg(dev, "DCB tagging enabled (num TC > 1)\n");
+		if (pf->hw.port_info->is_custom_tx_enabled) {
+			dev_err(dev, "Custom Tx scheduler feature enabled, can't configure DCB\n");
+			return -EBUSY;
+		}
+		ice_tear_down_devlink_rate_tree(pf);
+
 		set_bit(ICE_FLAG_DCB_ENA, pf->flags);
 	} else {
 		dev_dbg(dev, "DCB tagging disabled (num TC = 1)\n");
diff --git a/drivers/net/ethernet/intel/ice/ice_devlink.c b/drivers/net/ethernet/intel/ice/ice_devlink.c
index 3e93f20fec44..13783b734d63 100644
--- a/drivers/net/ethernet/intel/ice/ice_devlink.c
+++ b/drivers/net/ethernet/intel/ice/ice_devlink.c
@@ -8,6 +8,7 @@
 #include "ice_devlink.h"
 #include "ice_eswitch.h"
 #include "ice_fw_update.h"
+#include "ice_dcb_lib.h"
 
 static int ice_active_port_option = -1;
 
@@ -713,6 +714,63 @@ ice_devlink_port_unsplit(struct devlink *devlink, struct devlink_port *port,
 	return ice_devlink_port_split(devlink, port, 1, extack);
 }
 
+/**
+ * ice_tear_down_devlink_rate_tree - removes devlink-rate exported tree
+ * @pf: pf struct
+ *
+ * This function tears down tree exported during VF's creation.
+ */
+void ice_tear_down_devlink_rate_tree(struct ice_pf *pf)
+{
+	struct devlink *devlink;
+	struct ice_vf *vf;
+	unsigned int bkt;
+
+	devlink = priv_to_devlink(pf);
+
+	devl_lock(devlink);
+	mutex_lock(&pf->vfs.table_lock);
+	ice_for_each_vf(pf, bkt, vf) {
+		if (vf->devlink_port.devlink_rate)
+			devl_rate_leaf_destroy(&vf->devlink_port);
+	}
+	mutex_unlock(&pf->vfs.table_lock);
+
+	devl_rate_nodes_destroy(devlink);
+	devl_unlock(devlink);
+}
+
+/**
+ * ice_enable_custom_tx - try to enable custom Tx feature
+ * @pf: pf struct
+ *
+ * This function tries to enable custom Tx feature,
+ * it's not possible to enable it, if DCB or ADQ is active.
+ */
+static bool ice_enable_custom_tx(struct ice_pf *pf)
+{
+	struct ice_port_info *pi = ice_get_main_vsi(pf)->port_info;
+	struct device *dev = ice_pf_to_dev(pf);
+
+	if (pi->is_custom_tx_enabled)
+		/* already enabled, return true */
+		return true;
+
+	if (ice_is_adq_active(pf)) {
+		dev_err(dev, "ADQ active, can't modify Tx scheduler tree\n");
+		return false;
+	}
+
+	if (ice_is_dcb_active(pf)) {
+		dev_err(dev, "DCB active, can't modify Tx scheduler tree\n");
+		return false;
+	}
+
+	pi->is_custom_tx_enabled = true;
+
+	return true;
+}
+
 /**
  * ice_traverse_tx_tree - traverse Tx scheduler tree
  * @devlink: devlink struct
@@ -908,6 +966,9 @@ static int ice_devlink_rate_node_new(struct devlink_rate *rate_node, void **priv
 
 	pi = ice_get_pi_from_dev_rate(rate_node);
 
+	if (!ice_enable_custom_tx(devlink_priv(rate_node->devlink)))
+		return -EBUSY;
+
 	/* preallocate memory for ice_sched_node */
 	node = devm_kzalloc(ice_hw_to_dev(pi->hw), sizeof(*node), GFP_KERNEL);
 	*priv = node;
@@ -928,6 +989,9 @@ static int ice_devlink_rate_node_del(struct devlink_rate *rate_node, void *priv,
 	if (!rate_node->parent || !node || tc_node == node || !extack)
 		return 0;
 
+	if (!ice_enable_custom_tx(devlink_priv(rate_node->devlink)))
+		return -EBUSY;
+
 	/* can't allow to delete a node with children */
 	if (node->num_children)
 		return -EINVAL;
@@ -944,6 +1008,9 @@ static int ice_devlink_rate_leaf_tx_max_set(struct devlink_rate *rate_leaf, void
 {
 	struct ice_sched_node *node = priv;
 
+	if (!ice_enable_custom_tx(devlink_priv(rate_leaf->devlink)))
+		return -EBUSY;
+
 	if (!node)
 		return 0;
 
@@ -956,6 +1023,9 @@ static int ice_devlink_rate_leaf_tx_share_set(struct devlink_rate *rate_leaf, vo
 {
 	struct ice_sched_node *node = priv;
 
+	if (!ice_enable_custom_tx(devlink_priv(rate_leaf->devlink)))
+		return -EBUSY;
+
 	if (!node)
 		return 0;
 
@@ -968,6 +1038,9 @@ static int ice_devlink_rate_leaf_tx_priority_set(struct devlink_rate *rate_leaf,
 {
 	struct ice_sched_node *node = priv;
 
+	if (!ice_enable_custom_tx(devlink_priv(rate_leaf->devlink)))
+		return -EBUSY;
+
 	if (!node)
 		return 0;
 
@@ -980,6 +1053,9 @@ static int ice_devlink_rate_leaf_tx_weight_set(struct devlink_rate *rate_leaf, v
 {
 	struct ice_sched_node *node = priv;
 
+	if (!ice_enable_custom_tx(devlink_priv(rate_leaf->devlink)))
+		return -EBUSY;
+
 	if (!node)
 		return 0;
 
@@ -992,6 +1068,9 @@ static int ice_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node, void
 {
 	struct ice_sched_node *node = priv;
 
+	if (!ice_enable_custom_tx(devlink_priv(rate_node->devlink)))
+		return -EBUSY;
+
 	if (!node)
 		return 0;
 
@@ -1004,6 +1083,9 @@ static int ice_devlink_rate_node_tx_share_set(struct devlink_rate *rate_node, vo
 {
 	struct ice_sched_node *node = priv;
 
+	if (!ice_enable_custom_tx(devlink_priv(rate_node->devlink)))
+		return -EBUSY;
+
 	if (!node)
 		return 0;
 
@@ -1016,6 +1098,9 @@ static int ice_devlink_rate_node_tx_priority_set(struct devlink_rate *rate_node,
 {
 	struct ice_sched_node *node = priv;
 
+	if (!ice_enable_custom_tx(devlink_priv(rate_node->devlink)))
+		return -EBUSY;
+
 	if (!node)
 		return 0;
 
@@ -1028,6 +1113,9 @@ static int ice_devlink_rate_node_tx_weight_set(struct devlink_rate *rate_node, v
 {
 	struct ice_sched_node *node = priv;
 
+	if (!ice_enable_custom_tx(devlink_priv(rate_node->devlink)))
+		return -EBUSY;
+
 	if (!node)
 		return 0;
 
@@ -1053,6 +1141,9 @@ static int ice_devlink_set_parent(struct devlink_rate *devlink_rate,
 	if (!extack)
 		return 0;
 
+	if (!ice_enable_custom_tx(devlink_priv(devlink_rate->devlink)))
+		return -EBUSY;
+
 	if (!parent) {
 		if (!node || tc_node == node || node->num_children)
 			return -EINVAL;
diff --git a/drivers/net/ethernet/intel/ice/ice_devlink.h b/drivers/net/ethernet/intel/ice/ice_devlink.h
index 8bfed9ee2c4c..6ec96779f52e 100644
--- a/drivers/net/ethernet/intel/ice/ice_devlink.h
+++ b/drivers/net/ethernet/intel/ice/ice_devlink.h
@@ -19,5 +19,6 @@ void ice_devlink_init_regions(struct ice_pf *pf);
 void ice_devlink_destroy_regions(struct ice_pf *pf);
 
 int ice_devlink_rate_init_tx_topology(struct devlink *devlink, struct ice_vsi *vsi);
+void ice_tear_down_devlink_rate_tree(struct ice_pf *pf);
 
 #endif /* _ICE_DEVLINK_H_ */
diff --git a/drivers/net/ethernet/intel/ice/ice_main.c b/drivers/net/ethernet/intel/ice/ice_main.c
index a9fc89aebebe..d6f460ff1b72 100644
--- a/drivers/net/ethernet/intel/ice/ice_main.c
+++ b/drivers/net/ethernet/intel/ice/ice_main.c
@@ -8580,6 +8580,12 @@ static int ice_setup_tc_mqprio_qdisc(struct net_device *netdev, void *type_data)
 	switch (mode) {
 	case TC_MQPRIO_MODE_CHANNEL:
 
+		if (pf->hw.port_info->is_custom_tx_enabled) {
+			dev_err(dev, "Custom Tx scheduler feature enabled, can't configure ADQ\n");
+			return -EBUSY;
+		}
+		ice_tear_down_devlink_rate_tree(pf);
+
 		ret = ice_validate_mqprio_qopt(vsi, mqprio_qopt);
 		if (ret) {
 			netdev_err(netdev, "failed to validate_mqprio_qopt(), ret %d\n",
diff --git a/drivers/net/ethernet/intel/ice/ice_repr.c b/drivers/net/ethernet/intel/ice/ice_repr.c
index 46f58d48318c..109761c8c858 100644
--- a/drivers/net/ethernet/intel/ice/ice_repr.c
+++ b/drivers/net/ethernet/intel/ice/ice_repr.c
@@ -6,6 +6,7 @@
 #include "ice_devlink.h"
 #include "ice_sriov.h"
 #include "ice_tc_lib.h"
+#include "ice_dcb_lib.h"
 
 /**
  * ice_repr_get_sw_port_id - get port ID associated with representor
@@ -426,6 +427,10 @@ int ice_repr_add_for_all_vfs(struct ice_pf *pf)
 			goto err;
 	}
 
+	/* only export if ADQ and DCB disabled */
+	if (ice_is_adq_active(pf) || ice_is_dcb_active(pf))
+		return 0;
+
 	devlink = priv_to_devlink(pf);
 	ice_devlink_rate_init_tx_topology(devlink, ice_get_main_vsi(pf));
 
diff --git a/drivers/net/ethernet/intel/ice/ice_type.h b/drivers/net/ethernet/intel/ice/ice_type.h
index daf86cf561bc..e3f622cad425 100644
--- a/drivers/net/ethernet/intel/ice/ice_type.h
+++ b/drivers/net/ethernet/intel/ice/ice_type.h
@@ -715,6 +715,7 @@ struct ice_port_info {
 	struct ice_qos_cfg qos_cfg;
 	struct xarray sched_node_ids;
 	u8 is_vf:1;
+	u8 is_custom_tx_enabled:1;
 };
 
 struct ice_switch_info {
-- 
2.37.2

[PATCH net-next v11 10/11] ice: Add documentation for devlink-rate implementation

[Michal Wilczynski]

Add documentation to a newly added devlink-rate feature. Provide some
examples on how to use the commands, which netlink attributes are
supported and descriptions of the attributes.

Signed-off-by: Michal Wilczynski <michal.wilczynski@intel.com>
---
 Documentation/networking/devlink/ice.rst | 115 +++++++++++++++++++++++
 1 file changed, 115 insertions(+)

diff --git a/Documentation/networking/devlink/ice.rst b/Documentation/networking/devlink/ice.rst
index 0c89ceb8986d..62bf353a4527 100644
--- a/Documentation/networking/devlink/ice.rst
+++ b/Documentation/networking/devlink/ice.rst
@@ -254,3 +254,118 @@ Users can request an immediate capture of a snapshot via the
     0000000000000210 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 
     $ devlink region delete pci/0000:01:00.0/device-caps snapshot 1
+
+Devlink Rate
+==========
+
+The ``ice`` driver implements devlink-rate API. It allows for offload of
+the Hierarchical QoS to the hardware. It enables user to group Virtual
+Functions in a tree structure and assign supported parameters: tx_share,
+tx_max, tx_priority and tx_weight to each node in a tree. So effectively
+user gains an ability to control how much bandwidth is allocated for each
+VF group. This is later enforced by the HW.
+
+It is assumed that this feature is mutually exclusive with DCB performed
+in FW and ADQ, or any driver feature that would trigger changes in QoS,
+for example creation of the new traffic class. The driver will prevent DCB
+or ADQ configuration if user started making any changes to the nodes using
+devlink-rate API. To configure those features a driver reload is necessary.
+Correspondingly if ADQ or DCB will get configured the driver won't export
+hierarchy at all, or will remove the untouched hierarchy if those
+features are enabled after the hierarchy is exported, but before any
+changes are made.
+
+This feature is also dependent on switchdev being enabled in the system.
+It's required bacause devlink-rate requires devlink-port objects to be
+present, and those objects are only created in switchdev mode.
+
+If the driver is set to the switchdev mode, it will export internal
+hierarchy the moment VF's are created. Root of the tree is always
+represented by the node_0. This node can't be deleted by the user. Leaf
+nodes and nodes with children also can't be deleted.
+
+.. list-table:: Attributes supported
+    :widths: 15 85
+
+    * - Name
+      - Description
+    * - ``tx_max``
+      - maximum bandwidth to be consumed by the tree Node. Rate Limit is
+        an absolute number specifying a maximum amount of bytes a Node may
+        consume during the course of one second. Rate limit guarantees
+        that a link will not oversaturate the receiver on the remote end
+        and also enforces an SLA between the subscriber and network
+        provider.
+    * - ``tx_share``
+      - minimum bandwidth allocated to a tree node when it is not blocked.
+        It specifies an absolute BW. While tx_max defines the maximum
+        bandwidth the node may consume, the tx_share marks committed BW
+        for the Node.
+    * - ``tx_priority``
+      - allows for usage of strict priority arbiter among siblings. This
+        arbitration scheme attempts to schedule nodes based on their
+        priority as long as the nodes remain within their bandwidth limit.
+        Range 0-7. Nodes with priority 7 have the highest priority and are
+        selected first, while nodes with priority 0 have the lowest
+        priority. Nodes that have the same priority are treated equally.
+    * - ``tx_weight``
+      - allows for usage of Weighted Fair Queuing arbitration scheme among
+        siblings. This arbitration scheme can be used simultaneously with
+        the strict priority. Range 1-200. Only relative values mater for
+        arbitration.
+
+``tx_priority`` and ``tx_weight`` can be used simultaneously. In that case
+nodes with the same priority form a WFQ subgroup in the sibling group
+and arbitration among them is based on assigned weights.
+
+.. code:: shell
+
+    # enable switchdev
+    $ devlink dev eswitch set pci/0000:4b:00.0 mode switchdev
+
+    # at this point driver should export internal hierarchy
+    $ echo 2 > /sys/class/net/ens785np0/device/sriov_numvfs
+
+    $ devlink port function rate show
+    pci/0000:4b:00.0/node_25: type node parent node_24
+    pci/0000:4b:00.0/node_24: type node parent node_0
+    pci/0000:4b:00.0/node_32: type node parent node_31
+    pci/0000:4b:00.0/node_31: type node parent node_30
+    pci/0000:4b:00.0/node_30: type node parent node_16
+    pci/0000:4b:00.0/node_19: type node parent node_18
+    pci/0000:4b:00.0/node_18: type node parent node_17
+    pci/0000:4b:00.0/node_17: type node parent node_16
+    pci/0000:4b:00.0/node_14: type node parent node_5
+    pci/0000:4b:00.0/node_5: type node parent node_3
+    pci/0000:4b:00.0/node_13: type node parent node_4
+    pci/0000:4b:00.0/node_12: type node parent node_4
+    pci/0000:4b:00.0/node_11: type node parent node_4
+    pci/0000:4b:00.0/node_10: type node parent node_4
+    pci/0000:4b:00.0/node_9: type node parent node_4
+    pci/0000:4b:00.0/node_8: type node parent node_4
+    pci/0000:4b:00.0/node_7: type node parent node_4
+    pci/0000:4b:00.0/node_6: type node parent node_4
+    pci/0000:4b:00.0/node_4: type node parent node_3
+    pci/0000:4b:00.0/node_3: type node parent node_16
+    pci/0000:4b:00.0/node_16: type node parent node_15
+    pci/0000:4b:00.0/node_15: type node parent node_0
+    pci/0000:4b:00.0/node_2: type node parent node_1
+    pci/0000:4b:00.0/node_1: type node parent node_0
+    pci/0000:4b:00.0/node_0: type node
+    pci/0000:4b:00.0/1: type leaf parent node_25
+    pci/0000:4b:00.0/2: type leaf parent node_25
+
+    # let's create some custom node
+    $ devlink port function rate add pci/0000:4b:00.0/node_custom parent node_0
+
+    # second custom node
+    $ devlink port function rate add pci/0000:4b:00.0/node_custom_1 parent node_custom
+
+    # reassign second VF to newly created branch
+    $ devlink port function rate set pci/0000:4b:00.0/2 parent node_custom_1
+
+    # assign tx_weight to the VF
+    $ devlink port function rate set pci/0000:4b:00.0/2 tx_weight 5
+
+    # assign tx_share to the VF
+    $ devlink port function rate set pci/0000:4b:00.0/2 tx_share 500Mbps
-- 
2.37.2

[PATCH net-next v11 11/11] Documentation: Add documentation for new devlink-rate attributes

[Michal Wilczynski]

Provide documentation for newly introduced netlink attributes for
devlink-rate: tx_priority and tx_weight.

Mention the possibility to export tree from the driver.

Signed-off-by: Michal Wilczynski <michal.wilczynski@intel.com>
---
 .../networking/devlink/devlink-port.rst       | 33 ++++++++++++++++++-
 1 file changed, 32 insertions(+), 1 deletion(-)

diff --git a/Documentation/networking/devlink/devlink-port.rst b/Documentation/networking/devlink/devlink-port.rst
index 7627b1da01f2..643f5903d1d8 100644
--- a/Documentation/networking/devlink/devlink-port.rst
+++ b/Documentation/networking/devlink/devlink-port.rst
@@ -191,13 +191,44 @@ API allows to configure following rate object's parameters:
 ``tx_max``
   Maximum TX rate value.
 
+``tx_priority``
+  Allows for usage of strict priority arbiter among siblings. This
+  arbitration scheme attempts to schedule nodes based on their priority
+  as long as the nodes remain within their bandwidth limit. The higher the
+  priority the higher the probability that the node will get selected for
+  scheduling.
+
+``tx_weight``
+  Allows for usage of Weighted Fair Queuing arbitration scheme among
+  siblings. This arbitration scheme can be used simultaneously with the
+  strict priority. As a node is configured with a higher rate it gets more
+  BW relative to it's siblings. Values are relative like a percentage
+  points, they basically tell how much BW should node take relative to
+  it's siblings.
+
 ``parent``
   Parent node name. Parent node rate limits are considered as additional limits
   to all node children limits. ``tx_max`` is an upper limit for children.
   ``tx_share`` is a total bandwidth distributed among children.
 
+``tx_priority`` and ``tx_weight`` can be used simultaneously. In that case
+nodes with the same priority form a WFQ subgroup in the sibling group
+and arbitration among them is based on assigned weights.
+
+Arbitration flow from the high level:
+#. Choose a node, or group of nodes with the highest priority that stays
+   within the BW limit and are not blocked. Use ``tx_priority`` as a
+   parameter for this arbitration.
+#. If group of nodes have the same priority perform WFQ arbitration on
+   that subgroup. Use ``tx_weight`` as a parameter for this arbitration.
+#. Select the winner node, and continue arbitration flow among it's children,
+   until leaf node is reached, and the winner is established.
+#. If all the nodes from the highest priority sub-group are satisfied, or
+   overused their assigned BW, move to the lower priority nodes.
+
 Driver implementations are allowed to support both or either rate object types
-and setting methods of their parameters.
+and setting methods of their parameters. Additionally driver implementation
+may export nodes/leafs and their child-parent relationships.
 
 Terms and Definitions
 =====================
-- 
2.37.2

Re: [PATCH net-next v11 08/11] ice: Implement devlink-rate API

[Jakub Kicinski]

On Mon, 14 Nov 2022 18:31:35 +0100 Michal Wilczynski wrote:
> There is a need to support modification of Tx scheduler tree, in the
> ice driver. This will allow user to control Tx settings of each node in
> the internal hierarchy of nodes. As a result user will be able to use
> Hierarchy QoS implemented entirely in the hardware.
> 
> This patch implemenents devlink-rate API. It also exports initial
> default hierarchy. It's mostly dictated by the fact that the tree
> can't be removed entirely, all we can do is enable the user to modify
> it. For example root node shouldn't ever be removed, also nodes that
> have children are off-limits.

drivers/net/ethernet/intel/ice/ice_devlink.c:794: warning: Function parameter or member 'bw' not described in 'ice_set_object_tx_share'
drivers/net/ethernet/intel/ice/ice_devlink.c:818: warning: Function parameter or member 'bw' not described in 'ice_set_object_tx_max'
drivers/net/ethernet/intel/ice/ice_devlink.c:842: warning: Function parameter or member 'priority' not described in 'ice_set_object_tx_priority'
drivers/net/ethernet/intel/ice/ice_devlink.c:871: warning: Function parameter or member 'weight' not described in 'ice_set_object_tx_weight'

Feel free to post a v12 with these fixed, I'll review the latest
tomorrow.