Skip to contentSkip to conversation

iavf: Add devlink and devlink rate support

PATCH Superseded Wenjun Wu 8 participants 64 messages Jul 27, 2023 – Dec 18, 2023

Subsystems (4)
lore
Share & tools
lore

Wenjun Wu

author
#0 lore raw

cover iavf: Add devlink and devlink rate support

To allow user to configure queue bandwidth, devlink port support is added to support devlink port rate API. [1]

Add devlink framework registration/unregistration on iavf driver initialization and remove, and devlink port of DEVLINK_PORT_FLAVOUR_VIRTUAL is created to be associated iavf netdevice.

iavf rate tree with root node, queue nodes, and leaf node is created and registered with devlink rate when iavf adapter is configured, and if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node parent iavf_root pci/0000:af:01.0/iavf_root: type node

                         +---------+
                         |   root  |
                         +----+----+
                              |
            |-----------------|-----------------|
       +----v----+       +----v----+       +----v----+
       |  txq_0  |       |  txq_1  |       |  txq_x  |
       +----+----+       +----+----+       +----+----+

User can configure the tx_max and tx_share of each queue. Once any one of the queues are fully configured, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF

Example:

1.To Set the queue tx_share:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_share 100 MBps

2.To Set the queue tx_max:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_max 200 MBps

3.To Show Current devlink port rate info:
devlink port function rate function show
[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node tx_share 800Mbit tx_max 1600Mbit parent iavf_root pci/0000:af:01.0/iavf_root: type node

[1]https://lore.kernel.org/netdev/[email protected]/

Jun Zhang (3):
iavf: Add devlink and devlink port support
iavf: Add devlink port function rate API support
iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

Wenjun Wu (2):
virtchnl: support queue rate limit and quanta size configuration
ice: Support VF queue rate limit and quanta size configuration

 drivers/net/ethernet/intel/Kconfig            |   1 +
 drivers/net/ethernet/intel/iavf/Makefile      |   2 +-
 drivers/net/ethernet/intel/iavf/iavf.h        |  20 +
 .../net/ethernet/intel/iavf/iavf_devlink.c    | 388 ++++++++++++++++++
 .../net/ethernet/intel/iavf/iavf_devlink.h    |  39 ++
 drivers/net/ethernet/intel/iavf/iavf_main.c   |  60 ++-
 .../net/ethernet/intel/iavf/iavf_virtchnl.c   | 228 +++++++++-
 drivers/net/ethernet/intel/ice/ice.h          |   2 +
 drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
 drivers/net/ethernet/intel/ice/ice_common.c   |  19 +
 .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
 drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
 drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
 drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
 drivers/net/ethernet/intel/ice/ice_virtchnl.c | 317 ++++++++++++++
 drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
 .../intel/ice/ice_virtchnl_allowlist.c        |   6 +
 include/linux/avf/virtchnl.h                  | 113 +++++
 18 files changed, 1225 insertions(+), 3 deletions(-)
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.c
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.h
Signature · 6 lines
2.34.1

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
author
#1 lore raw

PATCH 1/5 virtchnl: support queue rate limit and quanta size configuration

This patch adds new virtchnl opcodes and structures for rate limit and quanta size configuration, which include:
1. VIRTCHNL_OP_CONFIG_QUEUE_BW, to configure max bandwidth for each VF per queue.
2. VIRTCHNL_OP_CONFIG_QUANTA, to configure quanta size per queue.
3. VIRTCHNL_OP_GET_QOS_CAPS, VF queries current QoS configuration, such as enabled TCs, arbiter type, up2tc and bandwidth of VSI node. The configuration is previously set by DCB and PF, and now is the potential QoS capability of VF. VF can take it as reference to configure queue TC mapping.

---
 include/linux/avf/virtchnl.h | 113 +++++++++++++++++++++++++++++++++++
 1 file changed, 113 insertions(+)
include/linux/avf/virtchnl.h +113 -0
index c15221dcb75e..f1250ddd063d 100644
@@ -84,6 +84,9 @@ enum virtchnl_rx_hsplit {
 	VIRTCHNL_RX_HSPLIT_SPLIT_SCTP    = 8,
 };
 
+enum virtchnl_bw_limit_type {
+	VIRTCHNL_BW_SHAPER = 0,
+};
 /* END GENERIC DEFINES */
 
 /* Opcodes for VF-PF communication. These are placed in the v_opcode field
@@ -145,6 +148,11 @@ enum virtchnl_ops {
 	VIRTCHNL_OP_DISABLE_VLAN_STRIPPING_V2 = 55,
 	VIRTCHNL_OP_ENABLE_VLAN_INSERTION_V2 = 56,
 	VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2 = 57,
+	/* opcode 57 - 65 are reserved */
+	VIRTCHNL_OP_GET_QOS_CAPS = 66,
+	/* opcode 68 through 111 are reserved */
+	VIRTCHNL_OP_CONFIG_QUEUE_BW = 112,
+	VIRTCHNL_OP_CONFIG_QUANTA = 113,
 	VIRTCHNL_OP_MAX,
 };
 
@@ -253,6 +261,7 @@ VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_vsi_resource);
 #define VIRTCHNL_VF_OFFLOAD_RX_FLEX_DESC	BIT(26)
 #define VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF		BIT(27)
 #define VIRTCHNL_VF_OFFLOAD_FDIR_PF		BIT(28)
+#define VIRTCHNL_VF_OFFLOAD_QOS			BIT(29)
 
 #define VF_BASE_MODE_OFFLOADS (VIRTCHNL_VF_OFFLOAD_L2 | \
 			       VIRTCHNL_VF_OFFLOAD_VLAN | \
@@ -1367,6 +1376,83 @@ struct virtchnl_fdir_del {
 
 VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_fdir_del);
 
+struct virtchnl_shaper_bw {
+	/* Unit is Kbps */
+	u32 committed;
+	u32 peak;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_shaper_bw);
+
+/* VIRTCHNL_OP_GET_QOS_CAPS
+ * VF sends this message to get its QoS Caps, such as
+ * TC number, Arbiter and Bandwidth.
+ */
+struct virtchnl_qos_cap_elem {
+	u8 tc_num;
+	u8 tc_prio;
+#define VIRTCHNL_ABITER_STRICT      0
+#define VIRTCHNL_ABITER_ETS         2
+	u8 arbiter;
+#define VIRTCHNL_STRICT_WEIGHT      1
+	u8 weight;
+	enum virtchnl_bw_limit_type type;
+	union {
+		struct virtchnl_shaper_bw shaper;
+		u8 pad2[32];
+	};
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(40, virtchnl_qos_cap_elem);
+
+struct virtchnl_qos_cap_list {
+	u16 vsi_id;
+	u16 num_elem;
+	struct virtchnl_qos_cap_elem cap[1];
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(44, virtchnl_qos_cap_list);
+
+/* VIRTCHNL_OP_CONFIG_QUEUE_BW */
+struct virtchnl_queue_bw {
+	u16 queue_id;
+	u8 tc;
+	u8 pad;
+	struct virtchnl_shaper_bw shaper;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_queue_bw);
+
+struct virtchnl_queues_bw_cfg {
+	u16 vsi_id;
+	u16 num_queues;
+	struct virtchnl_queue_bw cfg[1];
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_queues_bw_cfg);
+
+enum virtchnl_queue_type {
+	VIRTCHNL_QUEUE_TYPE_TX			= 0,
+	VIRTCHNL_QUEUE_TYPE_RX			= 1,
+};
+
+/* structure to specify a chunk of contiguous queues */
+struct virtchnl_queue_chunk {
+	/* see enum virtchnl_queue_type */
+	s32 type;
+	u16 start_queue_id;
+	u16 num_queues;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_queue_chunk);
+
+struct virtchnl_quanta_cfg {
+	u16 quanta_size;
+	struct virtchnl_queue_chunk queue_select;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_quanta_cfg);
+
 /**
  * virtchnl_vc_validate_vf_msg
  * @ver: Virtchnl version info
@@ -1558,6 +1644,33 @@ virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode,
 	case VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2:
 		valid_len = sizeof(struct virtchnl_vlan_setting);
 		break;
+	case VIRTCHNL_OP_GET_QOS_CAPS:
+		break;
+	case VIRTCHNL_OP_CONFIG_QUEUE_BW:
+		valid_len = sizeof(struct virtchnl_queues_bw_cfg);
+		if (msglen >= valid_len) {
+			struct virtchnl_queues_bw_cfg *q_bw =
+				(struct virtchnl_queues_bw_cfg *)msg;
+			if (q_bw->num_queues == 0) {
+				err_msg_format = true;
+				break;
+			}
+			valid_len += (q_bw->num_queues - 1) *
+					 sizeof(q_bw->cfg[0]);
+		}
+		break;
+	case VIRTCHNL_OP_CONFIG_QUANTA:
+		valid_len = sizeof(struct virtchnl_quanta_cfg);
+		if (msglen >= valid_len) {
+			struct virtchnl_quanta_cfg *q_quanta =
+				(struct virtchnl_quanta_cfg *)msg;
+			if (q_quanta->quanta_size == 0 ||
+			    q_quanta->queue_select.num_queues == 0) {
+				err_msg_format = true;
+				break;
+			}
+		}
+		break;
 	/* These are always errors coming from the VF. */
 	case VIRTCHNL_OP_EVENT:
 	case VIRTCHNL_OP_UNKNOWN:
Signature · 6 lines
2.34.1

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
maintainer
#2 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v1 1/5] virtchnl: support queue rate limit and quanta size configuration

Quoted 78 lines · Wenjun Wu

This patch adds new virtchnl opcodes and structures for rate limit and quanta size configuration, which include:
1. VIRTCHNL_OP_CONFIG_QUEUE_BW, to configure max bandwidth for each VF per queue.
2. VIRTCHNL_OP_CONFIG_QUANTA, to configure quanta size per queue.
3. VIRTCHNL_OP_GET_QOS_CAPS, VF queries current QoS configuration, such as enabled TCs, arbiter type, up2tc and bandwidth of VSI node. The configuration is previously set by DCB and PF, and now is the potential QoS capability of VF. VF can take it as reference to configure queue TC mapping.

---
  include/linux/avf/virtchnl.h | 113 +++++++++++++++++++++++++++++++++++
  1 file changed, 113 insertions(+)
include/linux/avf/virtchnl.h +33 -0
index c15221dcb75e..f1250ddd063d 100644
@@ -84,6 +84,9 @@ enum virtchnl_rx_hsplit {
  	VIRTCHNL_RX_HSPLIT_SPLIT_SCTP    = 8,
  };
 
+enum virtchnl_bw_limit_type {
+	VIRTCHNL_BW_SHAPER = 0,
+};
  /* END GENERIC DEFINES */
 
  /* Opcodes for VF-PF communication. These are placed in the v_opcode field
@@ -145,6 +148,11 @@ enum virtchnl_ops {
  	VIRTCHNL_OP_DISABLE_VLAN_STRIPPING_V2 = 55,
  	VIRTCHNL_OP_ENABLE_VLAN_INSERTION_V2 = 56,
  	VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2 = 57,
+	/* opcode 57 - 65 are reserved */
+	VIRTCHNL_OP_GET_QOS_CAPS = 66,
+	/* opcode 68 through 111 are reserved */
+	VIRTCHNL_OP_CONFIG_QUEUE_BW = 112,
+	VIRTCHNL_OP_CONFIG_QUANTA = 113,
  	VIRTCHNL_OP_MAX,
  };
 
@@ -253,6 +261,7 @@ VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_vsi_resource);
  #define VIRTCHNL_VF_OFFLOAD_RX_FLEX_DESC	BIT(26)
  #define VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF		BIT(27)
  #define VIRTCHNL_VF_OFFLOAD_FDIR_PF		BIT(28)
+#define VIRTCHNL_VF_OFFLOAD_QOS			BIT(29)
 
  #define VF_BASE_MODE_OFFLOADS (VIRTCHNL_VF_OFFLOAD_L2 | \
  			       VIRTCHNL_VF_OFFLOAD_VLAN | \
@@ -1367,6 +1376,83 @@ struct virtchnl_fdir_del {
 
  VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_fdir_del);
 
+struct virtchnl_shaper_bw {
+	/* Unit is Kbps */
+	u32 committed;
+	u32 peak;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_shaper_bw);
+
+/* VIRTCHNL_OP_GET_QOS_CAPS
+ * VF sends this message to get its QoS Caps, such as
+ * TC number, Arbiter and Bandwidth.
+ */
+struct virtchnl_qos_cap_elem {
+	u8 tc_num;
+	u8 tc_prio;
+#define VIRTCHNL_ABITER_STRICT      0
+#define VIRTCHNL_ABITER_ETS         2
+	u8 arbiter;
+#define VIRTCHNL_STRICT_WEIGHT      1
+	u8 weight;
+	enum virtchnl_bw_limit_type type;
+	union {
+		struct virtchnl_shaper_bw shaper;
+		u8 pad2[32];

include/linux/avf/virtchnl.h struct virtchnl_fdir_del { · line 1412

Quoted 10 lines
include/linux/avf/virtchnl.h +10 -0
+	};
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(40, virtchnl_qos_cap_elem);
+
+struct virtchnl_qos_cap_list {
+	u16 vsi_id;
+	u16 num_elem;
+	struct virtchnl_qos_cap_elem cap[1];
+};

If it's not too late to use a flex arrays, we should. Otherwise, this should model after Olek's work [1].

Adding Olek in case he has input.

Quoted 7 lines
include/linux/avf/virtchnl.h +7 -0
+
+VIRTCHNL_CHECK_STRUCT_LEN(44, virtchnl_qos_cap_list);
+
+/* VIRTCHNL_OP_CONFIG_QUEUE_BW */
+struct virtchnl_queue_bw {
+	u16 queue_id;
+	u8 tc;

include/linux/avf/virtchnl.h struct virtchnl_fdir_del { · line 1429

Quoted 10 lines
include/linux/avf/virtchnl.h +10 -0
+	u8 pad;
+	struct virtchnl_shaper_bw shaper;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_queue_bw);
+
+struct virtchnl_queues_bw_cfg {
+	u16 vsi_id;
+	u16 num_queues;
+	struct virtchnl_queue_bw cfg[1];

same here

Quoted 30 lines
include/linux/avf/virtchnl.h +26 -0
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_queues_bw_cfg);
+
+enum virtchnl_queue_type {
+	VIRTCHNL_QUEUE_TYPE_TX			= 0,
+	VIRTCHNL_QUEUE_TYPE_RX			= 1,
+};
+
+/* structure to specify a chunk of contiguous queues */
+struct virtchnl_queue_chunk {
+	/* see enum virtchnl_queue_type */
+	s32 type;
+	u16 start_queue_id;
+	u16 num_queues;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_queue_chunk);
+
+struct virtchnl_quanta_cfg {
+	u16 quanta_size;
+	struct virtchnl_queue_chunk queue_select;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_quanta_cfg);
+
  /**
   * virtchnl_vc_validate_vf_msg
   * @ver: Virtchnl version info

include/linux/avf/virtchnl.h virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode, · line 1653

Quoted 10 lines
include/linux/avf/virtchnl.h +7 -0
@@ -1558,6 +1644,33 @@ virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode,
  	case VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2:
  		valid_len = sizeof(struct virtchnl_vlan_setting);
  		break;
+	case VIRTCHNL_OP_GET_QOS_CAPS:
+		break;
+	case VIRTCHNL_OP_CONFIG_QUEUE_BW:
+		valid_len = sizeof(struct virtchnl_queues_bw_cfg);
+		if (msglen >= valid_len) {
+			struct virtchnl_queues_bw_cfg *q_bw =
+				(struct virtchnl_queues_bw_cfg *)msg;

missing newline here.

include/linux/avf/virtchnl.h virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode, · line 1659

Quoted 6 lines
include/linux/avf/virtchnl.h +6 -0
+			if (q_bw->num_queues == 0) {
+				err_msg_format = true;
+				break;
+			}
+			valid_len += (q_bw->num_queues - 1) *
+					 sizeof(q_bw->cfg[0]);

See referenced series for changes to this too.

include/linux/avf/virtchnl.h virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode, · line 1666

Quoted 7 lines
include/linux/avf/virtchnl.h +7 -0
+		}
+		break;
+	case VIRTCHNL_OP_CONFIG_QUANTA:
+		valid_len = sizeof(struct virtchnl_quanta_cfg);
+		if (msglen >= valid_len) {
+			struct virtchnl_quanta_cfg *q_quanta =
+				(struct virtchnl_quanta_cfg *)msg;

need newline

include/linux/avf/virtchnl.h virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode, · line 1676

Quoted 10 lines
include/linux/avf/virtchnl.h +7 -0
+			if (q_quanta->quanta_size == 0 ||
+			    q_quanta->queue_select.num_queues == 0) {
+				err_msg_format = true;
+				break;
+			}
+		}
+		break;
  	/* These are always errors coming from the VF. */
  	case VIRTCHNL_OP_EVENT:
  	case VIRTCHNL_OP_UNKNOWN:

[1] https://lore.kernel.org/netdev/[email protected]/#t
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

#3 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v1 1/5] virtchnl: support queue rate limit and quanta size configuration

Thanks for the careful review!

Quoted 39 lines

-----Original Message-----
From: Nguyen, Anthony L <[email protected]>
Sent: Tuesday, August 1, 2023 6:23 AM
To: Wu, Wenjun1 <[email protected]>; intel-wired- [email protected]
Cc: Aggarwal, Mitu <[email protected]>; Zhang, Qi Z <[email protected]>; Lobakin, Aleksander <[email protected]>
Subject: Re: [Intel-wired-lan] [PATCH iwl-next v1 1/5] virtchnl: support queue rate limit and quanta size configuration

Quoted 3 lines · Wenjun Wu

This patch adds new virtchnl opcodes and structures for rate limit and quanta size configuration, which include:
1. VIRTCHNL_OP_CONFIG_QUEUE_BW, to configure max bandwidth for

each VF

Quoted 11 lines

per queue.
2. VIRTCHNL_OP_CONFIG_QUANTA, to configure quanta size per queue.
3. VIRTCHNL_OP_GET_QOS_CAPS, VF queries current QoS configuration, such as enabled TCs, arbiter type, up2tc and bandwidth of VSI node. The configuration is previously set by DCB and PF, and now is the potential QoS capability of VF. VF can take it as reference to configure queue TC mapping.

---
  include/linux/avf/virtchnl.h | 113

+++++++++++++++++++++++++++++++++++

include/linux/avf/virtchnl.h struct virtchnl_fdir_del {

Quoted 7 lines
  1 file changed, 113 insertions(+)
include/linux/avf/virtchnl.h +6 -0
+VIRTCHNL_CHECK_STRUCT_LEN(40, virtchnl_qos_cap_elem);
+
+struct virtchnl_qos_cap_list {
+	u16 vsi_id;
+	u16 num_elem;
+	struct virtchnl_qos_cap_elem cap[1]; };

If it's not too late to use a flex arrays, we should. Otherwise, this should model after Olek's work [1].

Quoted 1 line

Adding Olek in case he has input.

Yes, I will change it to flex array in the next version.

Quoted 12 lines
Quoted 12 lines
include/linux/avf/virtchnl.h +11 -0
+
+VIRTCHNL_CHECK_STRUCT_LEN(44, virtchnl_qos_cap_list);
+
+/* VIRTCHNL_OP_CONFIG_QUEUE_BW */
+struct virtchnl_queue_bw {
+	u16 queue_id;
+	u8 tc;
+	u8 pad;
+	struct virtchnl_shaper_bw shaper;
+};
+
Quoted 8 lines

include/linux/avf/virtchnl.h struct virtchnl_fdir_del { · line 1429

Quoted 6 lines
include/linux/avf/virtchnl.h +6 -0
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_queue_bw);
+
+struct virtchnl_queues_bw_cfg {
+	u16 vsi_id;
+	u16 num_queues;
+	struct virtchnl_queue_bw cfg[1];

same here

Got it.

Quoted 31 lines

include/linux/avf/virtchnl.h virtchnl_vc_validate_vf_msg(struct

Quoted 30 lines
include/linux/avf/virtchnl.h +25 -0
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_queues_bw_cfg);
+
+enum virtchnl_queue_type {
+	VIRTCHNL_QUEUE_TYPE_TX			= 0,
+	VIRTCHNL_QUEUE_TYPE_RX			= 1,
+};
+
+/* structure to specify a chunk of contiguous queues */ struct
+virtchnl_queue_chunk {
+	/* see enum virtchnl_queue_type */
+	s32 type;
+	u16 start_queue_id;
+	u16 num_queues;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_queue_chunk);
+
+struct virtchnl_quanta_cfg {
+	u16 quanta_size;
+	struct virtchnl_queue_chunk queue_select; };
+
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_quanta_cfg);
+
  /**
   * virtchnl_vc_validate_vf_msg
   * @ver: Virtchnl version info
@@ -1558,6 +1644,33 @@ virtchnl_vc_validate_vf_msg(struct

virtchnl_version_info *ver, u32 v_opcode,

Quoted 12 lines

include/linux/avf/virtchnl.h virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode, · line 1653

Quoted 10 lines
include/linux/avf/virtchnl.h +7 -0
  	case VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2:
  		valid_len = sizeof(struct virtchnl_vlan_setting);
  		break;
+	case VIRTCHNL_OP_GET_QOS_CAPS:
+		break;
+	case VIRTCHNL_OP_CONFIG_QUEUE_BW:
+		valid_len = sizeof(struct virtchnl_queues_bw_cfg);
+		if (msglen >= valid_len) {
+			struct virtchnl_queues_bw_cfg *q_bw =
+				(struct virtchnl_queues_bw_cfg *)msg;

missing newline here.

I will add a new line here in the next version.

Quoted 9 lines

include/linux/avf/virtchnl.h virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode, · line 1659

Quoted 7 lines
include/linux/avf/virtchnl.h +6 -0
+			if (q_bw->num_queues == 0) {
+				err_msg_format = true;
+				break;
+			}
+			valid_len += (q_bw->num_queues - 1) *
+					 sizeof(q_bw->cfg[0]);

See referenced series for changes to this too.

I will use struct_size here in the next version.

Quoted 10 lines

include/linux/avf/virtchnl.h virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode, · line 1666

Quoted 8 lines
include/linux/avf/virtchnl.h +7 -0
+		}
+		break;
+	case VIRTCHNL_OP_CONFIG_QUANTA:
+		valid_len = sizeof(struct virtchnl_quanta_cfg);
+		if (msglen >= valid_len) {
+			struct virtchnl_quanta_cfg *q_quanta =
+				(struct virtchnl_quanta_cfg *)msg;

need newline

Got it.

Quoted 15 lines

include/linux/avf/virtchnl.h virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode, · line 1676

Quoted 11 lines
include/linux/avf/virtchnl.h +7 -0
+			if (q_quanta->quanta_size == 0 ||
+			    q_quanta->queue_select.num_queues == 0) {
+				err_msg_format = true;
+				break;
+			}
+		}
+		break;
  	/* These are always errors coming from the VF. */
  	case VIRTCHNL_OP_EVENT:
  	case VIRTCHNL_OP_UNKNOWN:

[1]
https://lore.kernel.org/netdev/20230728155207.10042-1-
[email protected]/#t

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

author
#4 lore raw

PATCH 2/5 ice: Support VF queue rate limit and quanta size configuration

In reply to Wenjun Wu

Add support to configure VF queue rate limit and quanta size.

For quanta size configuration, the quanta profiles are divided evenly by PF numbers. For each port, the first quanta profile is reserved for default. When VF is asked to set queue quanta size, PF will search for an available profile, change the fields and assigned this profile to the queue.

---
 drivers/net/ethernet/intel/ice/ice.h          |   2 +
 drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
 drivers/net/ethernet/intel/ice/ice_common.c   |  19 ++
 .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
 drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
 drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
 drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
 drivers/net/ethernet/intel/ice/ice_virtchnl.c | 317 ++++++++++++++++++
 drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
 .../intel/ice/ice_virtchnl_allowlist.c        |   6 +
 10 files changed, 377 insertions(+)
drivers/net/ethernet/intel/ice/ice.h +2 -0
index 125a2e753e29..25267ae6ab62 100644
@@ -637,6 +637,8 @@ struct ice_pf {
 #define ICE_VF_AGG_NODE_ID_START	65
 #define ICE_MAX_VF_AGG_NODES		32
 	struct ice_agg_node vf_agg_node[ICE_MAX_VF_AGG_NODES];
+
+	u8 n_quanta_prof_used;
 };
 
 extern struct workqueue_struct *ice_lag_wq;
drivers/net/ethernet/intel/ice/ice_base.c +2 -0
index 4a12316f7b46..c5274d1eb5bf 100644
@@ -377,6 +377,8 @@ ice_setup_tx_ctx(struct ice_tx_ring *ring, struct ice_tlan_ctx *tlan_ctx, u16 pf
 		break;
 	}
 
+	tlan_ctx->quanta_prof_idx = ring->quanta_prof_id;
+
 	tlan_ctx->tso_ena = ICE_TX_LEGACY;
 	tlan_ctx->tso_qnum = pf_q;
 
drivers/net/ethernet/intel/ice/ice_common.c +19 -0
index 6899f6af1866..606823ed68e8 100644
@@ -2470,6 +2470,23 @@ ice_parse_func_caps(struct ice_hw *hw, struct ice_hw_func_caps *func_p,
 	ice_recalc_port_limited_caps(hw, &func_p->common_cap);
 }
 
+/**
+ * ice_func_id_to_logical_id - map from function id to logical pf id
+ * @active_function_bitmap: active function bitmap
+ * @pf_id: function number of device
+ */
+static int ice_func_id_to_logical_id(u32 active_function_bitmap, u8 pf_id)
+{
+	u8 logical_id = 0;
+	u8 i;
+
+	for (i = 0; i < pf_id; i++)
+		if (active_function_bitmap & BIT(i))
+			logical_id++;
+
+	return logical_id;
+}
+
 /**
  * ice_parse_valid_functions_cap - Parse ICE_AQC_CAPS_VALID_FUNCTIONS caps
  * @hw: pointer to the HW struct
@@ -2487,6 +2504,8 @@ ice_parse_valid_functions_cap(struct ice_hw *hw, struct ice_hw_dev_caps *dev_p,
 	dev_p->num_funcs = hweight32(number);
 	ice_debug(hw, ICE_DBG_INIT, "dev caps: num_funcs = %d\n",
 		  dev_p->num_funcs);
+
+	hw->logical_pf_id = ice_func_id_to_logical_id(number, hw->pf_id);
 }
 
 /**
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +8 -0
index 20f40dfeb761..999bd4633d4f 100644
@@ -500,5 +500,13 @@
 #define PFPM_WUS_FW_RST_WK_M			BIT(31)
 #define VFINT_DYN_CTLN(_i)			(0x00003800 + ((_i) * 4))
 #define VFINT_DYN_CTLN_CLEARPBA_M		BIT(1)
+#define GLCOMM_QUANTA_PROF(_i)			(0x002D2D68 + ((_i) * 4))
+#define GLCOMM_QUANTA_PROF_MAX_INDEX		15
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_S	0
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_M	ICE_M(0x3FFF, 0)
+#define GLCOMM_QUANTA_PROF_MAX_CMD_S		16
+#define GLCOMM_QUANTA_PROF_MAX_CMD_M		ICE_M(0xFF, 16)
+#define GLCOMM_QUANTA_PROF_MAX_DESC_S		24
+#define GLCOMM_QUANTA_PROF_MAX_DESC_M		ICE_M(0x3F, 24)
 
 #endif /* _ICE_HW_AUTOGEN_H_ */
drivers/net/ethernet/intel/ice/ice_txrx.h +2 -0
index 166413fc33f4..7e152ab5b727 100644
@@ -381,6 +381,8 @@ struct ice_tx_ring {
 	u8 flags;
 	u8 dcb_tc;			/* Traffic class of ring */
 	u8 ptp_tx;
+
+	u16 quanta_prof_id;
 } ____cacheline_internodealigned_in_smp;
 
 static inline bool ice_ring_uses_build_skb(struct ice_rx_ring *ring)
drivers/net/ethernet/intel/ice/ice_type.h +1 -0
index bf47936e396a..3e17a1e7c6be 100644
@@ -843,6 +843,7 @@ struct ice_hw {
 	u8 revision_id;
 
 	u8 pf_id;		/* device profile info */
+	u8 logical_pf_id;
 
 	u16 max_burst_size;	/* driver sets this value */
 
drivers/net/ethernet/intel/ice/ice_vf_lib.h +9 -0
index 67172fdd9bc2..6499d83cc706 100644
@@ -52,6 +52,13 @@ struct ice_mdd_vf_events {
 	u16 last_printed;
 };
 
+struct ice_vf_qs_bw {
+	u16 queue_id;
+	u32 committed;
+	u32 peak;
+	u8 tc;
+};
+
 /* VF operations */
 struct ice_vf_ops {
 	enum ice_disq_rst_src reset_type;
@@ -133,6 +140,8 @@ struct ice_vf {
 
 	/* devlink port data */
 	struct devlink_port devlink_port;
+
+	struct ice_vf_qs_bw qs_bw[ICE_MAX_RSS_QS_PER_VF];
 };
 
 /* Flags for controlling behavior of ice_reset_vf */
drivers/net/ethernet/intel/ice/ice_virtchnl.c +317 -0

353 lines not on this page · raw message

drivers/net/ethernet/intel/ice/ice_virtchnl.h +11 -0
index cd747718de73..0efb9c0f669a 100644
@@ -13,6 +13,13 @@
 /* Restrict number of MAC Addr and VLAN that non-trusted VF can programmed */
 #define ICE_MAX_VLAN_PER_VF		8
 
+#define ICE_DFLT_QUANTA 1024
+#define ICE_MAX_QUANTA_SIZE 4096
+#define ICE_MIN_QUANTA_SIZE 256
+
+#define calc_quanta_desc(x)	\
+	max_t(u16, 12, min_t(u16, 63, (((x) + 66) / 132) * 2 + 4))
+
 /* MAC filters: 1 is reserved for the VF's default/perm_addr/LAA MAC, 1 for
  * broadcast, and 16 for additional unicast/multicast filters
  */
@@ -51,6 +58,10 @@ struct ice_virtchnl_ops {
 	int (*dis_vlan_stripping_v2_msg)(struct ice_vf *vf, u8 *msg);
 	int (*ena_vlan_insertion_v2_msg)(struct ice_vf *vf, u8 *msg);
 	int (*dis_vlan_insertion_v2_msg)(struct ice_vf *vf, u8 *msg);
+	int (*get_qos_caps)(struct ice_vf *vf);
+	int (*cfg_q_tc_map)(struct ice_vf *vf, u8 *msg);
+	int (*cfg_q_bw)(struct ice_vf *vf, u8 *msg);
+	int (*cfg_q_quanta)(struct ice_vf *vf, u8 *msg);
 };
 
 #ifdef CONFIG_PCI_IOV
drivers/net/ethernet/intel/ice/ice_virtchnl_allowlist.c +6 -0
index 7d547fa616fa..2e3f63a429cd 100644
@@ -85,6 +85,11 @@ static const u32 fdir_pf_allowlist_opcodes[] = {
 	VIRTCHNL_OP_ADD_FDIR_FILTER, VIRTCHNL_OP_DEL_FDIR_FILTER,
 };
 
+static const u32 tc_allowlist_opcodes[] = {
+	VIRTCHNL_OP_GET_QOS_CAPS, VIRTCHNL_OP_CONFIG_QUEUE_BW,
+	VIRTCHNL_OP_CONFIG_QUANTA,
+};
+
 struct allowlist_opcode_info {
 	const u32 *opcodes;
 	size_t size;
@@ -105,6 +110,7 @@ static const struct allowlist_opcode_info allowlist_opcodes[] = {
 	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF, adv_rss_pf_allowlist_opcodes),
 	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_FDIR_PF, fdir_pf_allowlist_opcodes),
 	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_VLAN_V2, vlan_v2_allowlist_opcodes),
+	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_QOS, tc_allowlist_opcodes),
 };
 
 /**
Signature · 6 lines
2.34.1

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
maintainer
#5 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v1 2/5] ice: Support VF queue rate limit and quanta size configuration

Quoted 91 lines · Wenjun Wu

Add support to configure VF queue rate limit and quanta size.

For quanta size configuration, the quanta profiles are divided evenly by PF numbers. For each port, the first quanta profile is reserved for default. When VF is asked to set queue quanta size, PF will search for an available profile, change the fields and assigned this profile to the queue.

---
  drivers/net/ethernet/intel/ice/ice.h          |   2 +
  drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
  drivers/net/ethernet/intel/ice/ice_common.c   |  19 ++
  .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
  drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
  drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
  drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
  drivers/net/ethernet/intel/ice/ice_virtchnl.c | 317 ++++++++++++++++++
  drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
  .../intel/ice/ice_virtchnl_allowlist.c        |   6 +
  10 files changed, 377 insertions(+)
drivers/net/ethernet/intel/ice/ice.h +2 -0
index 125a2e753e29..25267ae6ab62 100644
@@ -637,6 +637,8 @@ struct ice_pf {
  #define ICE_VF_AGG_NODE_ID_START	65
  #define ICE_MAX_VF_AGG_NODES		32
  	struct ice_agg_node vf_agg_node[ICE_MAX_VF_AGG_NODES];
+
+	u8 n_quanta_prof_used;
  };
 
  extern struct workqueue_struct *ice_lag_wq;
drivers/net/ethernet/intel/ice/ice_base.c +2 -0
index 4a12316f7b46..c5274d1eb5bf 100644
@@ -377,6 +377,8 @@ ice_setup_tx_ctx(struct ice_tx_ring *ring, struct ice_tlan_ctx *tlan_ctx, u16 pf
  		break;
  	}
 
+	tlan_ctx->quanta_prof_idx = ring->quanta_prof_id;
+
  	tlan_ctx->tso_ena = ICE_TX_LEGACY;
  	tlan_ctx->tso_qnum = pf_q;
 
drivers/net/ethernet/intel/ice/ice_common.c +19 -0
index 6899f6af1866..606823ed68e8 100644
@@ -2470,6 +2470,23 @@ ice_parse_func_caps(struct ice_hw *hw, struct ice_hw_func_caps *func_p,
  	ice_recalc_port_limited_caps(hw, &func_p->common_cap);
  }
 
+/**
+ * ice_func_id_to_logical_id - map from function id to logical pf id
+ * @active_function_bitmap: active function bitmap
+ * @pf_id: function number of device
+ */
+static int ice_func_id_to_logical_id(u32 active_function_bitmap, u8 pf_id)
+{
+	u8 logical_id = 0;
+	u8 i;
+
+	for (i = 0; i < pf_id; i++)
+		if (active_function_bitmap & BIT(i))
+			logical_id++;
+
+	return logical_id;
+}
+
  /**
   * ice_parse_valid_functions_cap - Parse ICE_AQC_CAPS_VALID_FUNCTIONS caps
   * @hw: pointer to the HW struct
@@ -2487,6 +2504,8 @@ ice_parse_valid_functions_cap(struct ice_hw *hw, struct ice_hw_dev_caps *dev_p,
  	dev_p->num_funcs = hweight32(number);
  	ice_debug(hw, ICE_DBG_INIT, "dev caps: num_funcs = %d\n",
  		  dev_p->num_funcs);
+
+	hw->logical_pf_id = ice_func_id_to_logical_id(number, hw->pf_id);
  }
 
  /**
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +0 -0
index 20f40dfeb761..999bd4633d4f 100644
@@ -500,5 +500,13 @@
  #define PFPM_WUS_FW_RST_WK_M			BIT(31)

drivers/net/ethernet/intel/ice/ice_hw_autogen.h line 510

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +8 -0
  #define VFINT_DYN_CTLN(_i)			(0x00003800 + ((_i) * 4))
  #define VFINT_DYN_CTLN_CLEARPBA_M		BIT(1)
+#define GLCOMM_QUANTA_PROF(_i)			(0x002D2D68 + ((_i) * 4))
+#define GLCOMM_QUANTA_PROF_MAX_INDEX		15
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_S	0
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_M	ICE_M(0x3FFF, 0)
+#define GLCOMM_QUANTA_PROF_MAX_CMD_S		16
+#define GLCOMM_QUANTA_PROF_MAX_CMD_M		ICE_M(0xFF, 16)
+#define GLCOMM_QUANTA_PROF_MAX_DESC_S		24
+#define GLCOMM_QUANTA_PROF_MAX_DESC_M		ICE_M(0x3F, 24)

These don't look like the right placement within the file. Please check/correct.

Quoted 74 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +0 -0
  #endif /* _ICE_HW_AUTOGEN_H_ */
drivers/net/ethernet/intel/ice/ice_txrx.h +2 -0
index 166413fc33f4..7e152ab5b727 100644
@@ -381,6 +381,8 @@ struct ice_tx_ring {
  	u8 flags;
  	u8 dcb_tc;			/* Traffic class of ring */
  	u8 ptp_tx;
+
+	u16 quanta_prof_id;
  } ____cacheline_internodealigned_in_smp;
 
  static inline bool ice_ring_uses_build_skb(struct ice_rx_ring *ring)
drivers/net/ethernet/intel/ice/ice_type.h +1 -0
index bf47936e396a..3e17a1e7c6be 100644
@@ -843,6 +843,7 @@ struct ice_hw {
  	u8 revision_id;
 
  	u8 pf_id;		/* device profile info */
+	u8 logical_pf_id;
 
  	u16 max_burst_size;	/* driver sets this value */
 
drivers/net/ethernet/intel/ice/ice_vf_lib.h +9 -0
index 67172fdd9bc2..6499d83cc706 100644
@@ -52,6 +52,13 @@ struct ice_mdd_vf_events {
  	u16 last_printed;
  };
 
+struct ice_vf_qs_bw {
+	u16 queue_id;
+	u32 committed;
+	u32 peak;
+	u8 tc;
+};
+
  /* VF operations */
  struct ice_vf_ops {
  	enum ice_disq_rst_src reset_type;
@@ -133,6 +140,8 @@ struct ice_vf {
 
  	/* devlink port data */
  	struct devlink_port devlink_port;
+
+	struct ice_vf_qs_bw qs_bw[ICE_MAX_RSS_QS_PER_VF];
  };
 
  /* Flags for controlling behavior of ice_reset_vf */
drivers/net/ethernet/intel/ice/ice_virtchnl.c +5 -0
index 85d996531502..016b7e1d6e91 100644
@@ -495,6 +495,9 @@ static int ice_vc_get_vf_res_msg(struct ice_vf *vf, u8 *msg)
  	if (vf->driver_caps & VIRTCHNL_VF_OFFLOAD_USO)
  		vfres->vf_cap_flags |= VIRTCHNL_VF_OFFLOAD_USO;
 
+	if (vf->driver_caps & VIRTCHNL_VF_OFFLOAD_QOS)
+		vfres->vf_cap_flags |= VIRTCHNL_VF_OFFLOAD_QOS;
+
  	vfres->num_vsis = 1;
  	/* Tx and Rx queue are equal for VF */
  	vfres->num_queue_pairs = vsi->num_txq;
@@ -985,6 +988,174 @@ static int ice_vc_config_rss_lut(struct ice_vf *vf, u8 *msg)
  				     NULL, 0);
  }
 
+/**
+ * ice_vc_get_qos_caps - Get current QoS caps from PF

drivers/net/ethernet/intel/ice/ice_virtchnl.c static int ice_vc_config_rss_lut(struct ice_vf *vf, u8 *msg) · line 1002

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +10 -0
+ * @vf: pointer to the VF info
+ *
+ * Get VF's QoS capabilities, such as TC number, arbiter and
+ * bandwidth from PF.
+ */
+static int ice_vc_get_qos_caps(struct ice_vf *vf)
+{
+	enum virtchnl_status_code v_ret = VIRTCHNL_STATUS_SUCCESS;
+	struct virtchnl_qos_cap_list *cap_list = NULL;
+	u8 tc_prio[ICE_MAX_TRAFFIC_CLASS] = {0};

init with { } is preferred

Quoted 20 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +20 -0
+	struct virtchnl_qos_cap_elem *cfg = NULL;
+	struct ice_vsi_ctx *vsi_ctx;
+	struct ice_pf *pf = vf->pf;
+	struct ice_port_info *pi;
+	struct ice_vsi *vsi;
+	u8 numtc, tc;
+	u16 len = 0;
+	int ret, i;
+
+	if (!test_bit(ICE_VF_STATE_ACTIVE, vf->vf_states)) {
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	vsi = ice_get_vf_vsi(vf);
+	if (!vsi) {
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+

drivers/net/ethernet/intel/ice/ice_virtchnl.c static int ice_vc_config_rss_lut(struct ice_vf *vf, u8 *msg) · line 1032

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +10 -0
+	pi = pf->hw.port_info;
+	numtc = vsi->tc_cfg.numtc;
+
+	vsi_ctx = ice_get_vsi_ctx(pi->hw, vf->lan_vsi_idx);
+	if (!vsi_ctx) {
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	len = sizeof(*cap_list) + sizeof(cap_list->cap[0]) * (numtc - 1);

I believe struct_size helper can be used here

Quoted 58 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +58 -0
+	cap_list = kzalloc(len, GFP_KERNEL);
+	if (!cap_list) {
+		v_ret = VIRTCHNL_STATUS_ERR_NO_MEMORY;
+		len = 0;
+		goto err;
+	}
+
+	cap_list->vsi_id = vsi->vsi_num;
+	cap_list->num_elem = numtc;
+
+	/* Store the UP2TC configuration from DCB to a user priority bitmap
+	 * of each TC. Each element of prio_of_tc represents one TC. Each
+	 * bitmap indicates the user priorities belong to this TC.
+	 */
+	for (i = 0; i < ICE_MAX_USER_PRIORITY; i++) {
+		tc = pi->qos_cfg.local_dcbx_cfg.etscfg.prio_table[i];
+		tc_prio[tc] |= BIT(i);
+	}
+
+	for (i = 0; i < numtc; i++) {
+		cfg = &cap_list->cap[i];
+		cfg->tc_num = i;
+		cfg->tc_prio = tc_prio[i];
+		cfg->arbiter = pi->qos_cfg.local_dcbx_cfg.etscfg.tsatable[i];
+		cfg->weight = VIRTCHNL_STRICT_WEIGHT;
+		cfg->type = VIRTCHNL_BW_SHAPER;
+		cfg->shaper.committed = vsi_ctx->sched.bw_t_info[i].cir_bw.bw;
+		cfg->shaper.peak = vsi_ctx->sched.bw_t_info[i].eir_bw.bw;
+	}
+
+err:
+	ret = ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_GET_QOS_CAPS, v_ret,
+				    (u8 *)cap_list, len);
+	kfree(cap_list);
+	return ret;
+}
+
+/**
+ * ice_vf_cfg_qs_bw - Configure per queue bandwidth
+ * @vf: pointer to the VF info
+ * @num_queues: number of queues to be configured
+ *
+ * Configure per queue bandwidth.
+ */
+static int ice_vf_cfg_qs_bw(struct ice_vf *vf, u16 num_queues)
+{
+	struct ice_hw *hw = &vf->pf->hw;
+	struct ice_vsi *vsi;
+	u32 p_rate;
+	int ret;
+	u16 i;
+	u8 tc;
+
+	vsi = ice_get_vf_vsi(vf);
+	if (!vsi)
+		return VIRTCHNL_STATUS_ERR_PARAM;
+
+	for (i = 0; i < num_queues; i++) {

drivers/net/ethernet/intel/ice/ice_virtchnl.c static int ice_vc_config_rss_lut(struct ice_vf *vf, u8 *msg) · line 1100

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +10 -0
+		p_rate = vf->qs_bw[i].peak;
+		tc = vf->qs_bw[i].tc;
+		if (p_rate) {
+			ret = ice_cfg_q_bw_lmt(hw->port_info, vsi->idx, tc,
+					       vf->qs_bw[i].queue_id,
+					       ICE_MAX_BW, p_rate);
+		} else {
+			ret = ice_cfg_q_bw_dflt_lmt(hw->port_info, vsi->idx, tc,
+						    vf->qs_bw[i].queue_id,
+						    ICE_MAX_BW);

These functions return kernel error codes...

drivers/net/ethernet/intel/ice/ice_virtchnl.c static int ice_vc_config_rss_lut(struct ice_vf *vf, u8 *msg)

Quoted 5 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +5 -0
+		}
+		if (ret)
+			return ret;
+	}
+ > +	return VIRTCHNL_STATUS_SUCCESS;

... this and the error above are returning VIRTCHNL errors. These are not returning consistent types.

Quoted 40 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +40 -0
+}
+
+/**
+ * ice_vf_cfg_q_quanta_profile
+ * @vf: pointer to the VF info
+ * @quanta_prof_idx: pointer to the quanta profile index
+ * @quanta_size: quanta size to be set
+ *
+ * This function chooses available quanta profile and configures the register.
+ * The quanta profile is evenly divided by the number of device ports, and then
+ * available to the specific PF and VFs. The first profile for each PF is a
+ * reserved default profile. Only quanta size of the rest unused profile can be
+ * modified.
+ */
+static int ice_vf_cfg_q_quanta_profile(struct ice_vf *vf, u16 quanta_size,
+				       u16 *quanta_prof_idx)
+{
+	const u16 n_desc = calc_quanta_desc(quanta_size);
+	struct ice_hw *hw = &vf->pf->hw;
+	const u16 n_cmd = 2 * n_desc;
+	struct ice_pf *pf = vf->pf;
+	u16 per_pf, begin_id;
+	u8 n_used;
+	u32 reg;
+
+	per_pf = (GLCOMM_QUANTA_PROF_MAX_INDEX + 1) / hw->dev_caps.num_funcs;
+	begin_id = hw->logical_pf_id * per_pf;
+	n_used = pf->n_quanta_prof_used;
+
+	if (quanta_size == ICE_DFLT_QUANTA) {
+		*quanta_prof_idx = begin_id;
+	} else {
+		if (n_used < per_pf) {
+			*quanta_prof_idx = begin_id + 1 + n_used;
+			pf->n_quanta_prof_used++;
+		} else {
+			return VIRTCHNL_STATUS_ERR_NOT_SUPPORTED;
+		}
+	}
+

drivers/net/ethernet/intel/ice/ice_virtchnl.c static int ice_vc_config_rss_lut(struct ice_vf *vf, u8 *msg) · line 1156

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +10 -0
+	reg = rd32(hw, GLCOMM_QUANTA_PROF(*quanta_prof_idx));
+	reg &= ~GLCOMM_QUANTA_PROF_QUANTA_SIZE_M;
+	reg |= quanta_size << GLCOMM_QUANTA_PROF_QUANTA_SIZE_S;
+	reg &= ~GLCOMM_QUANTA_PROF_MAX_CMD_M;
+	reg |= n_cmd << GLCOMM_QUANTA_PROF_MAX_CMD_S;
+	reg &= ~GLCOMM_QUANTA_PROF_MAX_DESC_M;
+	reg |= n_desc << GLCOMM_QUANTA_PROF_MAX_DESC_S;
+	wr32(hw, GLCOMM_QUANTA_PROF(*quanta_prof_idx), reg);
+
+	return VIRTCHNL_STATUS_SUCCESS;

Is this really supposed to return VIRTCHNL codes? That's not a standard convention the driver is doing.

Quoted 128 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +121 -0
+}
+
  /**
   * ice_vc_cfg_promiscuous_mode_msg
   * @vf: pointer to the VF info
@@ -1587,6 +1758,137 @@ static int ice_vc_cfg_irq_map_msg(struct ice_vf *vf, u8 *msg)
  				     NULL, 0);
  }
 
+/**
+ * ice_vc_cfg_q_bw - Configure per queue bandwidth
+ * @vf: pointer to the VF info
+ * @msg: pointer to the msg buffer which holds the command descriptor
+ *
+ * Configure VF queues bandwidth.
+ */
+static int ice_vc_cfg_q_bw(struct ice_vf *vf, u8 *msg)
+{
+	enum virtchnl_status_code v_ret = VIRTCHNL_STATUS_SUCCESS;
+	struct virtchnl_queues_bw_cfg *qbw =
+		(struct virtchnl_queues_bw_cfg *)msg;
+	struct ice_vf_qs_bw *qs_bw;
+	struct ice_vsi *vsi;
+	size_t len;
+	u16 i;
+
+	if (!test_bit(ICE_VF_STATE_ACTIVE, vf->vf_states) ||
+	    !ice_vc_isvalid_vsi_id(vf, qbw->vsi_id)) {
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	vsi = ice_get_vf_vsi(vf);
+	if (!vsi || vsi->vsi_num != qbw->vsi_id) {
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	if (qbw->num_queues > ICE_MAX_RSS_QS_PER_VF ||
+	    qbw->num_queues > min_t(u16, vsi->alloc_txq, vsi->alloc_rxq)) {
+		dev_err(ice_pf_to_dev(vf->pf), "VF-%d trying to configure more than allocated number of queues: %d\n",
+			vf->vf_id, min_t(u16, vsi->alloc_txq, vsi->alloc_rxq));
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	len = sizeof(struct ice_vf_qs_bw) * qbw->num_queues;
+	qs_bw = kzalloc(len, GFP_KERNEL);
+	if (!qs_bw) {
+		v_ret = VIRTCHNL_STATUS_ERR_NO_MEMORY;
+		goto err_bw;
+	}
+
+	for (i = 0; i < qbw->num_queues; i++) {
+		qs_bw[i].queue_id = qbw->cfg[i].queue_id;
+		qs_bw[i].peak = qbw->cfg[i].shaper.peak;
+		qs_bw[i].committed = qbw->cfg[i].shaper.committed;
+		qs_bw[i].tc = qbw->cfg[i].tc;
+	}
+
+	memcpy(vf->qs_bw, qs_bw, len);
+
+err_bw:
+	kfree(qs_bw);
+
+err:
+	/* send the response to the VF */
+	return ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_CONFIG_QUEUE_BW,
+				    v_ret, NULL, 0);
+}
+
+/**
+ * ice_vc_cfg_q_quanta - Configure per queue quanta
+ * @vf: pointer to the VF info
+ * @msg: pointer to the msg buffer which holds the command descriptor
+ *
+ * Configure VF queues quanta.
+ */
+static int ice_vc_cfg_q_quanta(struct ice_vf *vf, u8 *msg)
+{
+	u16 quanta_prof_id, quanta_size, start_qid, num_queues, end_qid, i;
+	enum virtchnl_status_code v_ret = VIRTCHNL_STATUS_SUCCESS;
+	struct virtchnl_quanta_cfg *qquanta =
+		(struct virtchnl_quanta_cfg *)msg;
+	struct ice_vsi *vsi;
+	int ret;
+
+	start_qid = qquanta->queue_select.start_queue_id;
+	num_queues = qquanta->queue_select.num_queues;
+	quanta_size = qquanta->quanta_size;
+	end_qid = start_qid + num_queues;
+
+	if (!test_bit(ICE_VF_STATE_ACTIVE, vf->vf_states)) {
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	vsi = ice_get_vf_vsi(vf);
+	if (!vsi) {
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	if (end_qid > ICE_MAX_RSS_QS_PER_VF ||
+	    end_qid > min_t(u16, vsi->alloc_txq, vsi->alloc_rxq)) {
+		dev_err(ice_pf_to_dev(vf->pf), "VF-%d trying to configure more than allocated number of queues: %d\n",
+			vf->vf_id, min_t(u16, vsi->alloc_txq, vsi->alloc_rxq));
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	if (quanta_size > ICE_MAX_QUANTA_SIZE ||
+	    quanta_size < ICE_MIN_QUANTA_SIZE) {
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	if (quanta_size % 64) {
+		dev_err(ice_pf_to_dev(vf->pf), "quanta size should be the product of 64\n");
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	ret = ice_vf_cfg_q_quanta_profile(vf, quanta_size,
+					  &quanta_prof_id);
+	if (ret) {
+		v_ret = VIRTCHNL_STATUS_ERR_NOT_SUPPORTED;
+		goto err;

drivers/net/ethernet/intel/ice/ice_virtchnl.c static int ice_vc_cfg_irq_map_msg(struct ice_vf *vf, u8 *msg) · line 1889

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +10 -0
+	}
+
+	for (i = start_qid; i < end_qid; i++)
+		vsi->tx_rings[i]->quanta_prof_id = quanta_prof_id;
+
+err:
+	/* send the response to the VF */
+	ret = ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_CONFIG_QUANTA,
+				    v_ret, NULL, 0);
+	return ret;

return ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_CONFIG_QUANTA,

			    v_ret, NULL, 0);
Quoted 87 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +17 -0
+}
+
  /**
   * ice_vc_cfg_qs_msg
   * @vf: pointer to the VF info
@@ -1710,6 +2012,9 @@ static int ice_vc_cfg_qs_msg(struct ice_vf *vf, u8 *msg)
  		}
  	}
 
+	if (ice_vf_cfg_qs_bw(vf, qci->num_queue_pairs))
+		goto error_param;
+
  	/* send the response to the VF */
  	return ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_CONFIG_VSI_QUEUES,
  				     VIRTCHNL_STATUS_SUCCESS, NULL, 0);
@@ -3687,6 +3992,9 @@ static const struct ice_virtchnl_ops ice_virtchnl_dflt_ops = {
  	.dis_vlan_stripping_v2_msg = ice_vc_dis_vlan_stripping_v2_msg,
  	.ena_vlan_insertion_v2_msg = ice_vc_ena_vlan_insertion_v2_msg,
  	.dis_vlan_insertion_v2_msg = ice_vc_dis_vlan_insertion_v2_msg,
+	.get_qos_caps = ice_vc_get_qos_caps,
+	.cfg_q_bw = ice_vc_cfg_q_bw,
+	.cfg_q_quanta = ice_vc_cfg_q_quanta,
  };
 
  /**
@@ -4040,6 +4348,15 @@ void ice_vc_process_vf_msg(struct ice_pf *pf, struct ice_rq_event_info *event,
  	case VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2:
  		err = ops->dis_vlan_insertion_v2_msg(vf, msg);
  		break;
+	case VIRTCHNL_OP_GET_QOS_CAPS:
+		err = ops->get_qos_caps(vf);
+		break;
+	case VIRTCHNL_OP_CONFIG_QUEUE_BW:
+		err = ops->cfg_q_bw(vf, msg);
+		break;
+	case VIRTCHNL_OP_CONFIG_QUANTA:
+		err = ops->cfg_q_quanta(vf, msg);
+		break;
  	case VIRTCHNL_OP_UNKNOWN:
  	default:
  		dev_err(dev, "Unsupported opcode %d from VF %d\n", v_opcode,
drivers/net/ethernet/intel/ice/ice_virtchnl.h +11 -0
index cd747718de73..0efb9c0f669a 100644
@@ -13,6 +13,13 @@
  /* Restrict number of MAC Addr and VLAN that non-trusted VF can programmed */
  #define ICE_MAX_VLAN_PER_VF		8
 
+#define ICE_DFLT_QUANTA 1024
+#define ICE_MAX_QUANTA_SIZE 4096
+#define ICE_MIN_QUANTA_SIZE 256
+
+#define calc_quanta_desc(x)	\
+	max_t(u16, 12, min_t(u16, 63, (((x) + 66) / 132) * 2 + 4))
+
  /* MAC filters: 1 is reserved for the VF's default/perm_addr/LAA MAC, 1 for
   * broadcast, and 16 for additional unicast/multicast filters
   */
@@ -51,6 +58,10 @@ struct ice_virtchnl_ops {
  	int (*dis_vlan_stripping_v2_msg)(struct ice_vf *vf, u8 *msg);
  	int (*ena_vlan_insertion_v2_msg)(struct ice_vf *vf, u8 *msg);
  	int (*dis_vlan_insertion_v2_msg)(struct ice_vf *vf, u8 *msg);
+	int (*get_qos_caps)(struct ice_vf *vf);
+	int (*cfg_q_tc_map)(struct ice_vf *vf, u8 *msg);
+	int (*cfg_q_bw)(struct ice_vf *vf, u8 *msg);
+	int (*cfg_q_quanta)(struct ice_vf *vf, u8 *msg);
  };
 
  #ifdef CONFIG_PCI_IOV
drivers/net/ethernet/intel/ice/ice_virtchnl_allowlist.c +5 -0
index 7d547fa616fa..2e3f63a429cd 100644
@@ -85,6 +85,11 @@ static const u32 fdir_pf_allowlist_opcodes[] = {
  	VIRTCHNL_OP_ADD_FDIR_FILTER, VIRTCHNL_OP_DEL_FDIR_FILTER,
  };
 
+static const u32 tc_allowlist_opcodes[] = {
+	VIRTCHNL_OP_GET_QOS_CAPS, VIRTCHNL_OP_CONFIG_QUEUE_BW,
+	VIRTCHNL_OP_CONFIG_QUANTA,
+};
+
  struct allowlist_opcode_info {
  	const u32 *opcodes;
  	size_t size;

drivers/net/ethernet/intel/ice/ice_virtchnl_allowlist.c static const struct allowlist_opcode_info allowlist_opcodes[] = { · line 116

Quoted 7 lines
drivers/net/ethernet/intel/ice/ice_virtchnl_allowlist.c +1 -0
@@ -105,6 +110,7 @@ static const struct allowlist_opcode_info allowlist_opcodes[] = {
  	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF, adv_rss_pf_allowlist_opcodes),
  	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_FDIR_PF, fdir_pf_allowlist_opcodes),
  	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_VLAN_V2, vlan_v2_allowlist_opcodes),
+	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_QOS, tc_allowlist_opcodes),
  };
 
  /**

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

#6 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v1 2/5] ice: Support VF queue rate limit and quanta size configuration

Thanks for the careful review!

Quoted 57 lines

-----Original Message-----
From: Nguyen, Anthony L <[email protected]>
Sent: Tuesday, August 1, 2023 6:23 AM
To: Wu, Wenjun1 <[email protected]>; intel-wired- [email protected]
Cc: Aggarwal, Mitu <[email protected]>; Zhang, Qi Z <[email protected]>
Subject: Re: [Intel-wired-lan] [PATCH iwl-next v1 2/5] ice: Support VF queue rate limit and quanta size configuration

Quoted 34 lines · Wenjun Wu

Add support to configure VF queue rate limit and quanta size.

For quanta size configuration, the quanta profiles are divided evenly by PF numbers. For each port, the first quanta profile is reserved for default. When VF is asked to set queue quanta size, PF will search for an available profile, change the fields and assigned this profile to the queue.

---
  drivers/net/ethernet/intel/ice/ice.h          |   2 +
  drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
  drivers/net/ethernet/intel/ice/ice_common.c   |  19 ++
  .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
  drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
  drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
  drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
  drivers/net/ethernet/intel/ice/ice_virtchnl.c | 317 ++++++++++++++++++
  drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
  .../intel/ice/ice_virtchnl_allowlist.c        |   6 +
  10 files changed, 377 insertions(+)
drivers/net/ethernet/intel/ice/ice.h +0 -0
index 125a2e753e29..25267ae6ab62 100644
@@ -637,6 +637,8 @@ struct ice_pf {
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +0 -0
index 20f40dfeb761..999bd4633d4f 100644

drivers/net/ethernet/intel/ice/ice_hw_autogen.h line 501

Quoted 2 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +0 -0
@@ -500,5 +500,13 @@
  #define PFPM_WUS_FW_RST_WK_M			BIT(31)
  #define VFINT_DYN_CTLN(_i)			(0x00003800 + ((_i) *

4))

drivers/net/ethernet/intel/ice/ice_hw_autogen.h line 503

Quoted 2 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +1 -0
  #define VFINT_DYN_CTLN_CLEARPBA_M		BIT(1)
+#define GLCOMM_QUANTA_PROF(_i)			(0x002D2D68

+ ((_i) * 4))

drivers/net/ethernet/intel/ice/ice_hw_autogen.h

Quoted 3 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +3 -0
+#define GLCOMM_QUANTA_PROF_MAX_INDEX		15
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_S	0
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_M	ICE_M(0x3FFF,

0)

Quoted 8 lines

drivers/net/ethernet/intel/ice/ice_hw_autogen.h line 508

Quoted 2 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +2 -0
+#define GLCOMM_QUANTA_PROF_MAX_CMD_S		16
+#define GLCOMM_QUANTA_PROF_MAX_CMD_M		ICE_M(0xFF,

16)

drivers/net/ethernet/intel/ice/ice_hw_autogen.h line 510

Quoted 2 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +2 -0
+#define GLCOMM_QUANTA_PROF_MAX_DESC_S		24
+#define GLCOMM_QUANTA_PROF_MAX_DESC_M		ICE_M(0x3F,

24)

These don't look like the right placement within the file. Please check/correct.

Yes, I will fix it.

Quoted 16 lines
Quoted 4 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +1 -0
  #endif /* _ICE_HW_AUTOGEN_H_ */
+/**

drivers/net/ethernet/intel/ice/ice_hw_autogen.h

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +10 -0
+ * ice_vc_get_qos_caps - Get current QoS caps from PF
+ * @vf: pointer to the VF info
+ *
+ * Get VF's QoS capabilities, such as TC number, arbiter and
+ * bandwidth from PF.
+ */
+static int ice_vc_get_qos_caps(struct ice_vf *vf) {
+	enum virtchnl_status_code v_ret = VIRTCHNL_STATUS_SUCCESS;
+	struct virtchnl_qos_cap_list *cap_list = NULL;
+	u8 tc_prio[ICE_MAX_TRAFFIC_CLASS] = {0};

init with { } is preferred

Perhaps you mean u8 tc_prio[ICE_MAX_TRAFFIC_CLASS] = { 0 }; I will change it in the next version.

Quoted 25 lines
Quoted 25 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +24 -0
+	struct virtchnl_qos_cap_elem *cfg = NULL;
+	struct ice_vsi_ctx *vsi_ctx;
+	struct ice_pf *pf = vf->pf;
+	struct ice_port_info *pi;
+	struct ice_vsi *vsi;
+	u8 numtc, tc;
+	u16 len = 0;
+	int ret, i;
+
+	if (!test_bit(ICE_VF_STATE_ACTIVE, vf->vf_states)) {
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	vsi = ice_get_vf_vsi(vf);
+	if (!vsi) {
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	pi = pf->hw.port_info;
+	numtc = vsi->tc_cfg.numtc;
+
+	vsi_ctx = ice_get_vsi_ctx(pi->hw, vf->lan_vsi_idx);
Quoted 8 lines

drivers/net/ethernet/intel/ice/ice_hw_autogen.h

Quoted 6 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +6 -0
+	if (!vsi_ctx) {
+		v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+		goto err;
+	}
+
+	len = sizeof(*cap_list) + sizeof(cap_list->cap[0]) * (numtc - 1);

I believe struct_size helper can be used here

Yes, I agree.

Quoted 74 lines
Quoted 29 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +27 -0
+	cap_list = kzalloc(len, GFP_KERNEL);
+	if (!cap_list) {
+		v_ret = VIRTCHNL_STATUS_ERR_NO_MEMORY;
+		len = 0;
+		goto err;
+	}
+
+	cap_list->vsi_id = vsi->vsi_num;
+	cap_list->num_elem = numtc;
+
+	/* Store the UP2TC configuration from DCB to a user priority bitmap
+	 * of each TC. Each element of prio_of_tc represents one TC. Each
+	 * bitmap indicates the user priorities belong to this TC.
+	 */
+	for (i = 0; i < ICE_MAX_USER_PRIORITY; i++) {
+		tc = pi->qos_cfg.local_dcbx_cfg.etscfg.prio_table[i];
+		tc_prio[tc] |= BIT(i);
+	}
+
+	for (i = 0; i < numtc; i++) {
+		cfg = &cap_list->cap[i];
+		cfg->tc_num = i;
+		cfg->tc_prio = tc_prio[i];
+		cfg->arbiter = pi->qos_cfg.local_dcbx_cfg.etscfg.tsatable[i];
+		cfg->weight = VIRTCHNL_STRICT_WEIGHT;
+		cfg->type = VIRTCHNL_BW_SHAPER;
+		cfg->shaper.committed = vsi_ctx-

sched.bw_t_info[i].cir_bw.bw;

drivers/net/ethernet/intel/ice/ice_hw_autogen.h

Quoted 5 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +5 -0
+		cfg->shaper.peak = vsi_ctx->sched.bw_t_info[i].eir_bw.bw;
+	}
+
+err:
+	ret = ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_GET_QOS_CAPS,

v_ret,

Quoted 23 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +23 -0
+				    (u8 *)cap_list, len);
+	kfree(cap_list);
+	return ret;
+}
+
+/**
+ * ice_vf_cfg_qs_bw - Configure per queue bandwidth
+ * @vf: pointer to the VF info
+ * @num_queues: number of queues to be configured
+ *
+ * Configure per queue bandwidth.
+ */
+static int ice_vf_cfg_qs_bw(struct ice_vf *vf, u16 num_queues) {
+	struct ice_hw *hw = &vf->pf->hw;
+	struct ice_vsi *vsi;
+	u32 p_rate;
+	int ret;
+	u16 i;
+	u8 tc;
+
+	vsi = ice_get_vf_vsi(vf);
+	if (!vsi)
+		return VIRTCHNL_STATUS_ERR_PARAM;

drivers/net/ethernet/intel/ice/ice_hw_autogen.h

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +10 -0
+
+	for (i = 0; i < num_queues; i++) {
+		p_rate = vf->qs_bw[i].peak;
+		tc = vf->qs_bw[i].tc;
+		if (p_rate) {
+			ret = ice_cfg_q_bw_lmt(hw->port_info, vsi->idx, tc,
+					       vf->qs_bw[i].queue_id,
+					       ICE_MAX_BW, p_rate);
+		} else {
+			ret = ice_cfg_q_bw_dflt_lmt(hw->port_info, vsi->idx,

tc,

drivers/net/ethernet/intel/ice/ice_hw_autogen.h

Quoted 2 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +2 -0
+						    vf->qs_bw[i].queue_id,
+						    ICE_MAX_BW);

These functions return kernel error codes...

Quoted 8 lines

drivers/net/ethernet/intel/ice/ice_hw_autogen.h

Quoted 5 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +5 -0
+		}
+		if (ret)
+			return ret;
+	}
+ > +	return VIRTCHNL_STATUS_SUCCESS;

... this and the error above are returning VIRTCHNL errors. These are not returning consistent types.

I will change it to return 0; to align to the kernel error codes.

Quoted 51 lines

drivers/net/ethernet/intel/ice/ice_hw_autogen.h

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +9 -0
+}
+
+/**
+ * ice_vf_cfg_q_quanta_profile
+ * @vf: pointer to the VF info
+ * @quanta_prof_idx: pointer to the quanta profile index
+ * @quanta_size: quanta size to be set
+ *
+ * This function chooses available quanta profile and configures the

register.

Quoted 40 lines

+ * The quanta profile is evenly divided by the number of device

drivers/net/ethernet/intel/ice/ice_hw_autogen.h +19 -0
+ports, and then
+ * available to the specific PF and VFs. The first profile for each
+PF is a
+ * reserved default profile. Only quanta size of the rest unused
+profile can be
+ * modified.
+ */
+static int ice_vf_cfg_q_quanta_profile(struct ice_vf *vf, u16 quanta_size,
+				       u16 *quanta_prof_idx)
+{
+	const u16 n_desc = calc_quanta_desc(quanta_size);
+	struct ice_hw *hw = &vf->pf->hw;
+	const u16 n_cmd = 2 * n_desc;
+	struct ice_pf *pf = vf->pf;
+	u16 per_pf, begin_id;
+	u8 n_used;
+	u32 reg;
+
+	per_pf = (GLCOMM_QUANTA_PROF_MAX_INDEX + 1) / hw-

dev_caps.num_funcs;

drivers/net/ethernet/intel/ice/ice_hw_autogen.h +19 -0
+	begin_id = hw->logical_pf_id * per_pf;
+	n_used = pf->n_quanta_prof_used;
+
+	if (quanta_size == ICE_DFLT_QUANTA) {
+		*quanta_prof_idx = begin_id;
+	} else {
+		if (n_used < per_pf) {
+			*quanta_prof_idx = begin_id + 1 + n_used;
+			pf->n_quanta_prof_used++;
+		} else {
+			return VIRTCHNL_STATUS_ERR_NOT_SUPPORTED;
+		}
+	}
+
+	reg = rd32(hw, GLCOMM_QUANTA_PROF(*quanta_prof_idx));
+	reg &= ~GLCOMM_QUANTA_PROF_QUANTA_SIZE_M;
+	reg |= quanta_size << GLCOMM_QUANTA_PROF_QUANTA_SIZE_S;
+	reg &= ~GLCOMM_QUANTA_PROF_MAX_CMD_M;
+	reg |= n_cmd << GLCOMM_QUANTA_PROF_MAX_CMD_S;
Quoted 8 lines

drivers/net/ethernet/intel/ice/ice_hw_autogen.h

Quoted 5 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +5 -0
+	reg &= ~GLCOMM_QUANTA_PROF_MAX_DESC_M;
+	reg |= n_desc << GLCOMM_QUANTA_PROF_MAX_DESC_S;
+	wr32(hw, GLCOMM_QUANTA_PROF(*quanta_prof_idx), reg);
+
+	return VIRTCHNL_STATUS_SUCCESS;

Is this really supposed to return VIRTCHNL codes? That's not a standard convention the driver is doing.

It seems better to use kernel error code in this function, I will change it in the next version.

Quoted 9 lines

drivers/net/ethernet/intel/ice/ice_hw_autogen.h

Quoted 6 lines
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +5 -0
+err:
+	/* send the response to the VF */
+	ret = ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_CONFIG_QUANTA,
+				    v_ret, NULL, 0);
+	return ret;

return ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_CONFIG_QUANTA,

			    v_ret, NULL, 0);

Got it.

Quoted 3 lines
Quoted 3 lines

/**

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

author
#7 lore raw

PATCH 3/5 iavf: Add devlink and devlink port support

In reply to Wenjun Wu

From: Jun Zhang <[email protected]>

To allow user to configure queue bandwidth, devlink port support is added to support devlink port rate API.

Add devlink framework registration/unregistration on iavf driver initialization and remove, and devlink port of DEVLINK_PORT_FLAVOUR_VIRTUAL is created to be associated iavf net device.

---
 drivers/net/ethernet/intel/Kconfig            |  1 +
 drivers/net/ethernet/intel/iavf/Makefile      |  2 +-
 drivers/net/ethernet/intel/iavf/iavf.h        |  6 ++
 .../net/ethernet/intel/iavf/iavf_devlink.c    | 93 +++++++++++++++++++
 .../net/ethernet/intel/iavf/iavf_devlink.h    | 17 ++++
 drivers/net/ethernet/intel/iavf/iavf_main.c   | 14 +++
 6 files changed, 132 insertions(+), 1 deletion(-)
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.c
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.h
drivers/net/ethernet/intel/Kconfig +1 -0
index 9bc0a9519899..f916b8ef6acb 100644
@@ -256,6 +256,7 @@ config I40EVF
 	tristate "Intel(R) Ethernet Adaptive Virtual Function support"
 	select IAVF
 	depends on PCI_MSI
+	select NET_DEVLINK
 	help
 	  This driver supports virtual functions for Intel XL710,
 	  X710, X722, XXV710, and all devices advertising support for
drivers/net/ethernet/intel/iavf/Makefile +1 -1
index 9c3e45c54d01..b5d7db97ab8b 100644
@@ -12,5 +12,5 @@ subdir-ccflags-y += -I$(src)
 obj-$(CONFIG_IAVF) += iavf.o
 
 iavf-objs := iavf_main.o iavf_ethtool.o iavf_virtchnl.o iavf_fdir.o \
-	     iavf_adv_rss.o \
+	     iavf_adv_rss.o iavf_devlink.o \
 	     iavf_txrx.o iavf_common.o iavf_adminq.o iavf_client.o
drivers/net/ethernet/intel/iavf/iavf.h +6 -0
index 8cbdebc5b698..519aeaec793c 100644
@@ -33,9 +33,11 @@
 #include <net/udp.h>
 #include <net/tc_act/tc_gact.h>
 #include <net/tc_act/tc_mirred.h>
+#include <net/devlink.h>
 
 #include "iavf_type.h"
 #include <linux/avf/virtchnl.h>
+#include "iavf_devlink.h"
 #include "iavf_txrx.h"
 #include "iavf_fdir.h"
 #include "iavf_adv_rss.h"
@@ -369,6 +371,10 @@ struct iavf_adapter {
 	struct net_device *netdev;
 	struct pci_dev *pdev;
 
+	/* devlink & port data */
+	struct devlink *devlink;
+	struct devlink_port devlink_port;
+
 	struct iavf_hw hw; /* defined in iavf_type.h */
 
 	enum iavf_state_t state;
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +93 -0

93 lines not on this page · raw message

drivers/net/ethernet/intel/iavf/iavf_devlink.h new file +17 -0
new file mode 100644
index 000000000000..5c122278611a
@@ -0,0 +1,17 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright (C) 2023 Intel Corporation */
+
+#ifndef _IAVF_DEVLINK_H_
+#define _IAVF_DEVLINK_H_
+
+/* iavf devlink structure pointing to iavf adapter */
+struct iavf_devlink {
+	struct iavf_adapter *devlink_ref;	/* ref to iavf adapter */
+};
+
+int iavf_devlink_register(struct iavf_adapter *adapter);
+void iavf_devlink_unregister(struct iavf_adapter *adapter);
+int iavf_devlink_port_register(struct iavf_adapter *adapter);
+void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
+
+#endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +14 -0
index 06ea61f30b6f..c9ee1e8712a8 100644
@@ -2036,6 +2036,7 @@ static void iavf_finish_config(struct work_struct *work)
 				iavf_free_rss(adapter);
 				iavf_free_misc_irq(adapter);
 				iavf_reset_interrupt_capability(adapter);
+				iavf_devlink_port_unregister(adapter);
 				iavf_change_state(adapter,
 						  __IAVF_INIT_CONFIG_ADAPTER);
 				goto out;
@@ -2707,6 +2708,9 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 	if (err)
 		goto err_sw_init;
 
+	if (!adapter->netdev_registered)
+		iavf_devlink_port_register(adapter);
+
 	netif_carrier_off(netdev);
 	adapter->link_up = false;
 	netif_tx_stop_all_queues(netdev);
@@ -2748,6 +2752,7 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 err_mem:
 	iavf_free_rss(adapter);
 	iavf_free_misc_irq(adapter);
+	iavf_devlink_port_unregister(adapter);
 err_sw_init:
 	iavf_reset_interrupt_capability(adapter);
 err:
@@ -4994,6 +4999,12 @@ static int iavf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	/* Setup the wait queue for indicating virtchannel events */
 	init_waitqueue_head(&adapter->vc_waitqueue);
 
+	/* Register iavf adapter with devlink */
+	err = iavf_devlink_register(adapter);
+	if (err)
+		dev_err(&pdev->dev, "devlink registration failed: %d\n", err);
+
+	/* Keep driver interface even on devlink registration failure */
 	return 0;
 
 err_ioremap:
@@ -5138,6 +5149,9 @@ static void iavf_remove(struct pci_dev *pdev)
 				 err);
 	}
 
+	iavf_devlink_port_unregister(adapter);
+	iavf_devlink_unregister(adapter);
+
 	mutex_lock(&adapter->crit_lock);
 	dev_info(&adapter->pdev->dev, "Removing device\n");
 	iavf_change_state(adapter, __IAVF_REMOVE);
Signature · 6 lines
2.34.1

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
author
#8 lore raw

PATCH 4/5 iavf: Add devlink port function rate API support

In reply to Wenjun Wu

From: Jun Zhang <[email protected]>

To allow user to configure queue based parameters, devlink port function rate api functions are added for setting node tx_max and tx_share parameters.

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

---
 .../net/ethernet/intel/iavf/iavf_devlink.c    | 270 +++++++++++++++++-
 .../net/ethernet/intel/iavf/iavf_devlink.h    |  21 ++
 drivers/net/ethernet/intel/iavf/iavf_main.c   |   7 +-
 3 files changed, 295 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/iavf/iavf_devlink.c +268 -2
index 991d041e5922..e8469fda054d 100644
@@ -4,7 +4,273 @@
 #include "iavf.h"
 #include "iavf_devlink.h"
 
-static const struct devlink_ops iavf_devlink_ops = {};
+/**
+ * iavf_devlink_rate_init_rate_tree - export rate tree to devlink rate
+ * @adapter: iavf adapter struct instance
+ *
+ * This function builds Rate Tree based on iavf adapter configuration
+ * and exports it's contents to devlink rate.
+ */
+void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+	struct iavf_dev_rate_node *iavf_r_node;
+	struct iavf_dev_rate_node *iavf_q_node;
+	struct devlink_rate *dl_root_node;
+	struct devlink_rate *dl_tmp_node;
+	int q_num, size, i;
+
+	if (!adapter->devlink_port.registered)
+		return;
+
+	iavf_r_node = &dl_priv->root_node;
+	memset(iavf_r_node, 0, sizeof(*iavf_r_node));
+	iavf_r_node->tx_max = adapter->link_speed;
+	strncpy(iavf_r_node->name, "iavf_root", IAVF_RATE_NODE_NAME);
+
+	devl_lock(adapter->devlink);
+	dl_root_node = devl_rate_node_create(adapter->devlink, iavf_r_node,
+					     iavf_r_node->name, NULL);
+	if (!dl_root_node || IS_ERR(dl_root_node))
+		goto err_node;
+
+	iavf_r_node->rate_node = dl_root_node;
+
+	/* Allocate queue nodes, and chain them under root */
+	q_num = adapter->num_active_queues;
+	if (q_num > 0) {
+		size = q_num * sizeof(struct iavf_dev_rate_node);
+		dl_priv->queue_nodes = kzalloc(size, GFP_KERNEL);
+		if (!dl_priv->queue_nodes)
+			goto err_node;
+
+		memset(dl_priv->queue_nodes, 0, size);
+
+		for (i = 0; i < q_num; ++i) {
+			iavf_q_node = &dl_priv->queue_nodes[i];
+			snprintf(iavf_q_node->name, IAVF_RATE_NODE_NAME,
+				 "txq_%d", i);
+			dl_tmp_node = devl_rate_node_create(adapter->devlink,
+							    iavf_q_node,
+							    iavf_q_node->name,
+							    dl_root_node);
+			if (!dl_tmp_node || IS_ERR(dl_tmp_node)) {
+				kfree(dl_priv->queue_nodes);
+				goto err_node;
+			}
+
+			iavf_q_node->rate_node = dl_tmp_node;
+			iavf_q_node->tx_max = IAVF_TX_DEFAULT;
+			iavf_q_node->tx_share = 0;
+		}
+	}
+
+	dl_priv->update_in_progress = false;
+	dl_priv->iavf_dev_rate_initialized = true;
+	devl_unlock(adapter->devlink);
+	return;
+err_node:
+	devl_rate_nodes_destroy(adapter->devlink);
+	dl_priv->iavf_dev_rate_initialized = false;
+	devl_unlock(adapter->devlink);
+}
+
+/**
+ * iavf_devlink_rate_deinit_rate_tree - Unregister rate tree with devlink rate
+ * @adapter: iavf adapter struct instance
+ *
+ * This function unregisters the current iavf rate tree registered with devlink
+ * rate and frees resources.
+ */
+void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+
+	if (!dl_priv->iavf_dev_rate_initialized)
+		return;
+
+	devl_lock(adapter->devlink);
+	devl_rate_leaf_destroy(&adapter->devlink_port);
+	devl_rate_nodes_destroy(adapter->devlink);
+	kfree(dl_priv->queue_nodes);
+	devl_unlock(adapter->devlink);
+}
+
+/**
+ * iavf_check_update_config - check if updating queue parameters needed
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ *
+ * This function sets queue bw & quanta size configuration if all
+ * queue parameters are set
+ */
+static int iavf_check_update_config(struct iavf_adapter *adapter,
+				    struct iavf_dev_rate_node *node)
+{
+	/* Update queue bw if any one of the queues have been fully updated by
+	 * user, the other queues either use the default value or the last
+	 * fully updated value
+	 */
+	if (node->tx_update_flag ==
+	    (IAVF_FLAG_TX_MAX_UPDATED | IAVF_FLAG_TX_SHARE_UPDATED)) {
+		node->tx_max = node->tx_max_temp;
+		node->tx_share = node->tx_share_temp;
+	} else {
+		return 0;
+	}
+
+	/* Reconfig queue bw only when iavf driver on running state */
+	if (adapter->state != __IAVF_RUNNING)
+		return -EBUSY;
+
+	return 0;
+}
+
+/**
+ * iavf_update_queue_tx_share - sets tx min parameter
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ * @bw: bandwidth in bytes per second
+ * @extack: extended netdev ack structure
+ *
+ * This function sets min BW limit.
+ */
+static int iavf_update_queue_tx_share(struct iavf_adapter *adapter,
+				      struct iavf_dev_rate_node *node,
+				      u64 bw, struct netlink_ext_ack *extack)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+	u64 tx_share_sum = 0;
+
+	/* Keep in kbps */
+	node->tx_share_temp = div_u64(bw, IAVF_RATE_DIV_FACTOR);
+
+	if (ADV_LINK_SUPPORT(adapter)) {
+		int i;
+
+		for (i = 0; i < adapter->num_active_queues; ++i) {
+			if (node != &dl_priv->queue_nodes[i])
+				tx_share_sum +=
+					dl_priv->queue_nodes[i].tx_share;
+			else
+				tx_share_sum += node->tx_share_temp;
+		}
+
+		if (tx_share_sum / 1000  > adapter->link_speed_mbps)
+			return -EINVAL;
+	}
+
+	node->tx_update_flag |= IAVF_FLAG_TX_SHARE_UPDATED;
+	return iavf_check_update_config(adapter, node);
+}
+
+/**
+ * iavf_update_queue_tx_max - sets tx max parameter
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ * @bw: bandwidth in bytes per second
+ * @extack: extended netdev ack structure
+ *
+ * This function sets max BW limit.
+ */
+static int iavf_update_queue_tx_max(struct iavf_adapter *adapter,
+				    struct iavf_dev_rate_node *node,
+				    u64 bw, struct netlink_ext_ack *extack)
+{
+	/* Keep in kbps */
+	node->tx_max_temp = div_u64(bw, IAVF_RATE_DIV_FACTOR);
+	if (ADV_LINK_SUPPORT(adapter)) {
+		if (node->tx_max_temp / 1000 > adapter->link_speed_mbps)
+			return -EINVAL;
+	}
+
+	node->tx_update_flag |= IAVF_FLAG_TX_MAX_UPDATED;
+
+	return iavf_check_update_config(adapter, node);
+}
+
+/**
+ * iavf_devlink_rate_node_tx_max_set - devlink_rate API for setting tx max
+ * @rate_node: devlink rate struct instance
+ *
+ * This function implements rate_node_tx_max_set function of devlink_ops
+ */
+static int iavf_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node,
+					     void *priv, u64 tx_max,
+					     struct netlink_ext_ack *extack)
+{
+	struct iavf_dev_rate_node *node = priv;
+	struct iavf_devlink *dl_priv;
+	struct iavf_adapter *adapter;
+
+	if (!node)
+		return 0;
+
+	dl_priv = devlink_priv(rate_node->devlink);
+	adapter = dl_priv->devlink_ref;
+
+	/* Check if last update is in progress */
+	if (dl_priv->update_in_progress)
+		return -EBUSY;
+
+	if (node == &dl_priv->root_node)
+		return 0;
+
+	return iavf_update_queue_tx_max(adapter, node, tx_max, extack);
+}
+
+/**
+ * iavf_devlink_rate_node_tx_share_set - devlink_rate API for setting tx share
+ * @rate_node: devlink rate struct instance
+ *
+ * This function implements rate_node_tx_share_set function of devlink_ops
+ */
+static int iavf_devlink_rate_node_tx_share_set(struct devlink_rate *rate_node,
+					       void *priv, u64 tx_share,
+					       struct netlink_ext_ack *extack)
+{
+	struct iavf_dev_rate_node *node = priv;
+	struct iavf_devlink *dl_priv;
+	struct iavf_adapter *adapter;
+
+	if (!node)
+		return 0;
+
+	dl_priv = devlink_priv(rate_node->devlink);
+	adapter = dl_priv->devlink_ref;
+
+	/* Check if last update is in progress */
+	if (dl_priv->update_in_progress)
+		return -EBUSY;
+
+	if (node == &dl_priv->root_node)
+		return 0;
+
+	return iavf_update_queue_tx_share(adapter, node, tx_share, extack);
+}
+
+static int iavf_devlink_rate_node_del(struct devlink_rate *rate_node,
+				      void *priv,
+				      struct netlink_ext_ack *extack)
+{
+	return -EINVAL;
+}
+
+static int iavf_devlink_set_parent(struct devlink_rate *devlink_rate,
+				   struct devlink_rate *parent,
+				   void *priv, void *parent_priv,
+				   struct netlink_ext_ack *extack)
+{
+	return -EINVAL;
+}
+
+static const struct devlink_ops iavf_devlink_ops = {
+	.rate_node_tx_share_set = iavf_devlink_rate_node_tx_share_set,
+	.rate_node_tx_max_set = iavf_devlink_rate_node_tx_max_set,
+	.rate_node_del = iavf_devlink_rate_node_del,
+	.rate_leaf_parent_set = iavf_devlink_set_parent,
+	.rate_node_parent_set = iavf_devlink_set_parent,
+};
 
 /**
  * iavf_devlink_register - Register allocated devlink instance for iavf adapter
@@ -30,7 +296,7 @@ int iavf_devlink_register(struct iavf_adapter *adapter)
 	adapter->devlink = devlink;
 	ref = devlink_priv(devlink);
 	ref->devlink_ref = adapter;
-
+	ref->iavf_dev_rate_initialized = false;
 	devlink_register(devlink);
 
 	return 0;
drivers/net/ethernet/intel/iavf/iavf_devlink.h +21 -0
index 5c122278611a..897ff5fc87af 100644
@@ -4,14 +4,35 @@
 #ifndef _IAVF_DEVLINK_H_
 #define _IAVF_DEVLINK_H_
 
+#define IAVF_RATE_NODE_NAME			12
+struct iavf_dev_rate_node {
+	char name[IAVF_RATE_NODE_NAME];
+	struct devlink_rate *rate_node;
+	u8 tx_update_flag;
+#define IAVF_FLAG_TX_SHARE_UPDATED		BIT(0)
+#define IAVF_FLAG_TX_MAX_UPDATED		BIT(1)
+	u64 tx_max;
+	u64 tx_share;
+	u64 tx_max_temp;
+	u64 tx_share_temp;
+#define IAVF_RATE_DIV_FACTOR			125
+#define IAVF_TX_DEFAULT				100000
+};
+
 /* iavf devlink structure pointing to iavf adapter */
 struct iavf_devlink {
 	struct iavf_adapter *devlink_ref;	/* ref to iavf adapter */
+	struct iavf_dev_rate_node root_node;
+	struct iavf_dev_rate_node *queue_nodes;
+	bool iavf_dev_rate_initialized;
+	bool update_in_progress;
 };
 
 int iavf_devlink_register(struct iavf_adapter *adapter);
 void iavf_devlink_unregister(struct iavf_adapter *adapter);
 int iavf_devlink_port_register(struct iavf_adapter *adapter);
 void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
+void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter);
+void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter);
 
 #endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +6 -1
index c9ee1e8712a8..b621e44e8890 100644
@@ -2036,6 +2036,7 @@ static void iavf_finish_config(struct work_struct *work)
 				iavf_free_rss(adapter);
 				iavf_free_misc_irq(adapter);
 				iavf_reset_interrupt_capability(adapter);
+				iavf_devlink_rate_deinit_rate_tree(adapter);
 				iavf_devlink_port_unregister(adapter);
 				iavf_change_state(adapter,
 						  __IAVF_INIT_CONFIG_ADAPTER);
@@ -2708,8 +2709,10 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 	if (err)
 		goto err_sw_init;
 
-	if (!adapter->netdev_registered)
+	if (!adapter->netdev_registered) {
 		iavf_devlink_port_register(adapter);
+		iavf_devlink_rate_init_rate_tree(adapter);
+	}
 
 	netif_carrier_off(netdev);
 	adapter->link_up = false;
@@ -2752,6 +2755,7 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 err_mem:
 	iavf_free_rss(adapter);
 	iavf_free_misc_irq(adapter);
+	iavf_devlink_rate_deinit_rate_tree(adapter);
 	iavf_devlink_port_unregister(adapter);
 err_sw_init:
 	iavf_reset_interrupt_capability(adapter);
@@ -5149,6 +5153,7 @@ static void iavf_remove(struct pci_dev *pdev)
 				 err);
 	}
 
+	iavf_devlink_rate_deinit_rate_tree(adapter);
 	iavf_devlink_port_unregister(adapter);
 	iavf_devlink_unregister(adapter);
 
Signature · 6 lines
2.34.1

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
author
#9 lore raw

PATCH 5/5 iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

In reply to Wenjun Wu

From: Jun Zhang <[email protected]>

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

User can configure the tx_max and tx_share of each queue. If any one of the queues have been fully updated by user, i.e. both tx_max and tx_share have been updated for that queue, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

---
 drivers/net/ethernet/intel/iavf/iavf.h        |  14 ++
 .../net/ethernet/intel/iavf/iavf_devlink.c    |  29 +++
 .../net/ethernet/intel/iavf/iavf_devlink.h    |   1 +
 drivers/net/ethernet/intel/iavf/iavf_main.c   |  45 +++-
 .../net/ethernet/intel/iavf/iavf_virtchnl.c   | 228 +++++++++++++++++-
 5 files changed, 313 insertions(+), 4 deletions(-)
drivers/net/ethernet/intel/iavf/iavf.h +14 -0
index 519aeaec793c..e9b781cacffa 100644
@@ -252,6 +252,9 @@ struct iavf_cloud_filter {
 #define IAVF_RESET_WAIT_DETECTED_COUNT 500
 #define IAVF_RESET_WAIT_COMPLETE_COUNT 2000
 
+#define IAVF_MAX_QOS_TC_NUM		8
+#define IAVF_DEFAULT_QUANTA_SIZE	1024
+
 /* board specific private data structure */
 struct iavf_adapter {
 	struct workqueue_struct *wq;
@@ -351,6 +354,9 @@ struct iavf_adapter {
 #define IAVF_FLAG_AQ_DISABLE_CTAG_VLAN_INSERTION	BIT_ULL(36)
 #define IAVF_FLAG_AQ_ENABLE_STAG_VLAN_INSERTION		BIT_ULL(37)
 #define IAVF_FLAG_AQ_DISABLE_STAG_VLAN_INSERTION	BIT_ULL(38)
+#define IAVF_FLAG_AQ_CONFIGURE_QUEUES_BW		BIT_ULL(39)
+#define IAVF_FLAG_AQ_CONFIGURE_QUEUES_QUANTA_SIZE	BIT_ULL(40)
+#define IAVF_FLAG_AQ_GET_QOS_CAPS			BIT_ULL(41)
 
 	/* flags for processing extended capability messages during
 	 * __IAVF_INIT_EXTENDED_CAPS. Each capability exchange requires
@@ -374,6 +380,7 @@ struct iavf_adapter {
 	/* devlink & port data */
 	struct devlink *devlink;
 	struct devlink_port devlink_port;
+	bool devlink_update;
 
 	struct iavf_hw hw; /* defined in iavf_type.h */
 
@@ -423,6 +430,8 @@ struct iavf_adapter {
 			       VIRTCHNL_VF_OFFLOAD_FDIR_PF)
 #define ADV_RSS_SUPPORT(_a) ((_a)->vf_res->vf_cap_flags & \
 			     VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF)
+#define QOS_ALLOWED(_a) ((_a)->vf_res->vf_cap_flags & \
+			 VIRTCHNL_VF_OFFLOAD_QOS)
 	struct virtchnl_vf_resource *vf_res; /* incl. all VSIs */
 	struct virtchnl_vsi_resource *vsi_res; /* our LAN VSI */
 	struct virtchnl_version_info pf_version;
@@ -431,6 +440,7 @@ struct iavf_adapter {
 	struct virtchnl_vlan_caps vlan_v2_caps;
 	u16 msg_enable;
 	struct iavf_eth_stats current_stats;
+	struct virtchnl_qos_cap_list *qos_caps;
 	struct iavf_vsi vsi;
 	u32 aq_wait_count;
 	/* RSS stuff */
@@ -577,6 +587,10 @@ void iavf_notify_client_message(struct iavf_vsi *vsi, u8 *msg, u16 len);
 void iavf_notify_client_l2_params(struct iavf_vsi *vsi);
 void iavf_notify_client_open(struct iavf_vsi *vsi);
 void iavf_notify_client_close(struct iavf_vsi *vsi, bool reset);
+void iavf_update_queue_config(struct iavf_adapter *adapter);
+void iavf_configure_queues_bw(struct iavf_adapter *adapter);
+void iavf_configure_queues_quanta_size(struct iavf_adapter *adapter);
+void iavf_get_qos_caps(struct iavf_adapter *adapter);
 void iavf_enable_channels(struct iavf_adapter *adapter);
 void iavf_disable_channels(struct iavf_adapter *adapter);
 void iavf_add_cloud_filter(struct iavf_adapter *adapter);
drivers/net/ethernet/intel/iavf/iavf_devlink.c +29 -0
index e8469fda054d..40af7f6e0a86 100644
@@ -96,6 +96,30 @@ void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
 	devl_unlock(adapter->devlink);
 }
 
+/**
+ * iavf_notify_queue_config_complete - notify updating queue completion
+ * @adapter: iavf adapter struct instance
+ *
+ * This function sets the queue configuration update status when all
+ * queue parameters have been sent to PF
+ */
+void iavf_notify_queue_config_complete(struct iavf_adapter *adapter)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+	int q_num = adapter->num_active_queues;
+	int i;
+
+	/* clean up rate tree update flags*/
+	for (i = 0; i < q_num; i++)
+		if (dl_priv->queue_nodes[i].tx_update_flag ==
+		    (IAVF_FLAG_TX_MAX_UPDATED | IAVF_FLAG_TX_SHARE_UPDATED)) {
+			dl_priv->queue_nodes[i].tx_update_flag = 0;
+			break;
+		}
+
+	dl_priv->update_in_progress = false;
+}
+
 /**
  * iavf_check_update_config - check if updating queue parameters needed
  * @adapter: iavf adapter struct instance
@@ -107,6 +131,8 @@ void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
 static int iavf_check_update_config(struct iavf_adapter *adapter,
 				    struct iavf_dev_rate_node *node)
 {
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+
 	/* Update queue bw if any one of the queues have been fully updated by
 	 * user, the other queues either use the default value or the last
 	 * fully updated value
@@ -123,6 +149,8 @@ static int iavf_check_update_config(struct iavf_adapter *adapter,
 	if (adapter->state != __IAVF_RUNNING)
 		return -EBUSY;
 
+	dl_priv->update_in_progress = true;
+	iavf_update_queue_config(adapter);
 	return 0;
 }
 
@@ -294,6 +322,7 @@ int iavf_devlink_register(struct iavf_adapter *adapter)
 
 	/* Init iavf adapter devlink */
 	adapter->devlink = devlink;
+	adapter->devlink_update = false;
 	ref = devlink_priv(devlink);
 	ref->devlink_ref = adapter;
 	ref->iavf_dev_rate_initialized = false;
drivers/net/ethernet/intel/iavf/iavf_devlink.h +1 -0
index 897ff5fc87af..a8a41f343f56 100644
@@ -34,5 +34,6 @@ int iavf_devlink_port_register(struct iavf_adapter *adapter);
 void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
 void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter);
 void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter);
+void iavf_notify_queue_config_complete(struct iavf_adapter *adapter);
 
 #endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +42 -3

84 lines not on this page · raw message

drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +227 -1

252 lines not on this page · raw message

Signature · 6 lines
2.34.1

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan
maintainer
#10 lore raw
In reply to Wenjun Wu
Quoted 79 lines · Wenjun Wu

To allow user to configure queue bandwidth, devlink port support is added to support devlink port rate API. [1]

Add devlink framework registration/unregistration on iavf driver initialization and remove, and devlink port of DEVLINK_PORT_FLAVOUR_VIRTUAL is created to be associated iavf netdevice.

iavf rate tree with root node, queue nodes, and leaf node is created and registered with devlink rate when iavf adapter is configured, and if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node parent iavf_root pci/0000:af:01.0/iavf_root: type node

                          +---------+
                          |   root  |
                          +----+----+
                               |
             |-----------------|-----------------|
        +----v----+       +----v----+       +----v----+
        |  txq_0  |       |  txq_1  |       |  txq_x  |
        +----+----+       +----+----+       +----+----+

User can configure the tx_max and tx_share of each queue. Once any one of the queues are fully configured, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF

Example:

1.To Set the queue tx_share:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_share 100 MBps

2.To Set the queue tx_max:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_max 200 MBps

3.To Show Current devlink port rate info:
devlink port function rate function show
[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node tx_share 800Mbit tx_max 1600Mbit parent iavf_root pci/0000:af:01.0/iavf_root: type node

[1]https://lore.kernel.org/netdev/[email protected]/

Quoted 8 lines

Jun Zhang (3):
iavf: Add devlink and devlink port support
iavf: Add devlink port function rate API support
iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

Wenjun Wu (2):
virtchnl: support queue rate limit and quanta size configuration
ice: Support VF queue rate limit and quanta size configuration

This series does not apply.

Quoted 19 lines
  drivers/net/ethernet/intel/Kconfig            |   1 +
  drivers/net/ethernet/intel/iavf/Makefile      |   2 +-
  drivers/net/ethernet/intel/iavf/iavf.h        |  20 +
  .../net/ethernet/intel/iavf/iavf_devlink.c    | 388 ++++++++++++++++++
  .../net/ethernet/intel/iavf/iavf_devlink.h    |  39 ++
  drivers/net/ethernet/intel/iavf/iavf_main.c   |  60 ++-
  .../net/ethernet/intel/iavf/iavf_virtchnl.c   | 228 +++++++++-
  drivers/net/ethernet/intel/ice/ice.h          |   2 +
  drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
  drivers/net/ethernet/intel/ice/ice_common.c   |  19 +
  .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
  drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
  drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
  drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
  drivers/net/ethernet/intel/ice/ice_virtchnl.c | 317 ++++++++++++++
  drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
  .../intel/ice/ice_virtchnl_allowlist.c        |   6 +
  include/linux/avf/virtchnl.h                  | 113 +++++
  18 files changed, 1225 insertions(+), 3 deletions(-)
Quoted 3 lines

create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.c create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.h

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

#11 lore raw
Quoted 88 lines · Tony Nguyen
Quoted 85 lines · Wenjun Wu

To allow user to configure queue bandwidth, devlink port support is added to support devlink port rate API. [1]

Add devlink framework registration/unregistration on iavf driver initialization and remove, and devlink port of DEVLINK_PORT_FLAVOUR_VIRTUAL is created to be associated iavf netdevice.

iavf rate tree with root node, queue nodes, and leaf node is created and registered with devlink rate when iavf adapter is configured, and if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node parent iavf_root pci/0000:af:01.0/iavf_root: type node

                          +---------+
                          |   root  |
                          +----+----+
                               |
             |-----------------|-----------------|
        +----v----+       +----v----+       +----v----+
        |  txq_0  |       |  txq_1  |       |  txq_x  |
        +----+----+       +----+----+       +----+----+

User can configure the tx_max and tx_share of each queue. Once any one of the queues are fully configured, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF

Example:

1.To Set the queue tx_share:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_share 100 MBps

2.To Set the queue tx_max:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_max 200 MBps

3.To Show Current devlink port rate info:
devlink port function rate function show
[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node tx_share 800Mbit tx_max 1600Mbit parent iavf_root pci/0000:af:01.0/iavf_root: type node

[1]https://lore.kernel.org/netdev/[email protected]/

Jun Zhang (3):
   iavf: Add devlink and devlink port support
   iavf: Add devlink port function rate API support
   iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

Quoted 6 lines
Quoted 3 lines

Wenjun Wu (2):
   virtchnl: support queue rate limit and quanta size configuration
   ice: Support VF queue rate limit and quanta size configuration

This series does not apply.

Nice find. Will do a rebase on V2

Quoted 23 lines
Quoted 19 lines

drivers/net/ethernet/intel/Kconfig            |   1 + drivers/net/ethernet/intel/iavf/Makefile      |   2 +- drivers/net/ethernet/intel/iavf/iavf.h        |  20 + .../net/ethernet/intel/iavf/iavf_devlink.c    | 388 ++++++++++++++++++ .../net/ethernet/intel/iavf/iavf_devlink.h    |  39 ++ drivers/net/ethernet/intel/iavf/iavf_main.c   |  60 ++- .../net/ethernet/intel/iavf/iavf_virtchnl.c   | 228 +++++++++- drivers/net/ethernet/intel/ice/ice.h          |   2 + drivers/net/ethernet/intel/ice/ice_base.c     |   2 + drivers/net/ethernet/intel/ice/ice_common.c   |  19 + .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 + drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 + drivers/net/ethernet/intel/ice/ice_type.h     |   1 + drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 + drivers/net/ethernet/intel/ice/ice_virtchnl.c | 317 ++++++++++++++ drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 + .../intel/ice/ice_virtchnl_allowlist.c        |   6 + include/linux/avf/virtchnl.h                  | 113 +++++

Quoted 4 lines
  18 files changed, 1225 insertions(+), 3 deletions(-)

  create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.c create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.h

Quoted 4 lines

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

author
#12 lore raw

v2 cover iavf: Add devlink and devlink rate support

In reply to Wenjun Wu

To allow user to configure queue bandwidth, devlink port support is added to support devlink port rate API. [1]

Add devlink framework registration/unregistration on iavf driver initialization and remove, and devlink port of DEVLINK_PORT_FLAVOUR_VIRTUAL is created to be associated iavf netdevice.

iavf rate tree with root node, queue nodes, and leaf node is created and registered with devlink rate when iavf adapter is configured, and if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node parent iavf_root pci/0000:af:01.0/iavf_root: type node

                         +---------+
                         |   root  |
                         +----+----+
                              |
            |-----------------|-----------------|
       +----v----+       +----v----+       +----v----+
       |  txq_0  |       |  txq_1  |       |  txq_x  |
       +----+----+       +----+----+       +----+----+

User can configure the tx_max and tx_share of each queue. Once any one of the queues are fully configured, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF

Example:

1.To Set the queue tx_share:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_share 100 MBps

2.To Set the queue tx_max:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_max 200 MBps

3.To Show Current devlink port rate info:
devlink port function rate function show
[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node tx_share 800Mbit tx_max 1600Mbit parent iavf_root pci/0000:af:01.0/iavf_root: type node

[1]https://lore.kernel.org/netdev/[email protected]/

Change log:

v2:
- Change static array to flex array
- Use struct_size helper
- Align all the error code types in the function
- Move the register field definitions to the right place in the file
- Fix coding style
- Adapted to queue bw cfg and qos cap list virtchnl message with flex array fields

---
Jun Zhang (3):
iavf: Add devlink and devlink port support
iavf: Add devlink port function rate API support
iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

Wenjun Wu (2):
virtchnl: support queue rate limit and quanta size configuration
ice: Support VF queue rate limit and quanta size configuration

 drivers/net/ethernet/intel/Kconfig            |   1 +
 drivers/net/ethernet/intel/iavf/Makefile      |   2 +-
 drivers/net/ethernet/intel/iavf/iavf.h        |  20 +
 .../net/ethernet/intel/iavf/iavf_devlink.c    | 388 ++++++++++++++++++
 .../net/ethernet/intel/iavf/iavf_devlink.h    |  39 ++
 drivers/net/ethernet/intel/iavf/iavf_main.c   |  60 ++-
 .../net/ethernet/intel/iavf/iavf_virtchnl.c   | 228 +++++++++-
 drivers/net/ethernet/intel/ice/ice.h          |   2 +
 drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
 drivers/net/ethernet/intel/ice/ice_common.c   |  19 +
 .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
 drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
 drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
 drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
 drivers/net/ethernet/intel/ice/ice_virtchnl.c | 312 ++++++++++++++
 drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
 .../intel/ice/ice_virtchnl_allowlist.c        |   6 +
 include/linux/avf/virtchnl.h                  | 114 +++++
 18 files changed, 1221 insertions(+), 3 deletions(-)
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.c
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.h
Signature · 1 line
2.34.1
author
#13 lore raw

PATCH v2 1/5 virtchnl: support queue rate limit and quanta size configuration

This patch adds new virtchnl opcodes and structures for rate limit and quanta size configuration, which include:
1. VIRTCHNL_OP_CONFIG_QUEUE_BW, to configure max bandwidth for each VF per queue.
2. VIRTCHNL_OP_CONFIG_QUANTA, to configure quanta size per queue.
3. VIRTCHNL_OP_GET_QOS_CAPS, VF queries current QoS configuration, such as enabled TCs, arbiter type, up2tc and bandwidth of VSI node. The configuration is previously set by DCB and PF, and now is the potential QoS capability of VF. VF can take it as reference to configure queue TC mapping.

---
 include/linux/avf/virtchnl.h | 114 +++++++++++++++++++++++++++++++++++
 1 file changed, 114 insertions(+)
include/linux/avf/virtchnl.h +114 -0
index c15221dcb75e..10566a1458bb 100644
@@ -84,6 +84,9 @@ enum virtchnl_rx_hsplit {
 	VIRTCHNL_RX_HSPLIT_SPLIT_SCTP    = 8,
 };
 
+enum virtchnl_bw_limit_type {
+	VIRTCHNL_BW_SHAPER = 0,
+};
 /* END GENERIC DEFINES */
 
 /* Opcodes for VF-PF communication. These are placed in the v_opcode field
@@ -145,6 +148,11 @@ enum virtchnl_ops {
 	VIRTCHNL_OP_DISABLE_VLAN_STRIPPING_V2 = 55,
 	VIRTCHNL_OP_ENABLE_VLAN_INSERTION_V2 = 56,
 	VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2 = 57,
+	/* opcode 57 - 65 are reserved */
+	VIRTCHNL_OP_GET_QOS_CAPS = 66,
+	/* opcode 68 through 111 are reserved */
+	VIRTCHNL_OP_CONFIG_QUEUE_BW = 112,
+	VIRTCHNL_OP_CONFIG_QUANTA = 113,
 	VIRTCHNL_OP_MAX,
 };
 
@@ -253,6 +261,7 @@ VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_vsi_resource);
 #define VIRTCHNL_VF_OFFLOAD_RX_FLEX_DESC	BIT(26)
 #define VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF		BIT(27)
 #define VIRTCHNL_VF_OFFLOAD_FDIR_PF		BIT(28)
+#define VIRTCHNL_VF_OFFLOAD_QOS			BIT(29)
 
 #define VF_BASE_MODE_OFFLOADS (VIRTCHNL_VF_OFFLOAD_L2 | \
 			       VIRTCHNL_VF_OFFLOAD_VLAN | \
@@ -1367,6 +1376,83 @@ struct virtchnl_fdir_del {
 
 VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_fdir_del);
 
+struct virtchnl_shaper_bw {
+	/* Unit is Kbps */
+	u32 committed;
+	u32 peak;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_shaper_bw);
+
+/* VIRTCHNL_OP_GET_QOS_CAPS
+ * VF sends this message to get its QoS Caps, such as
+ * TC number, Arbiter and Bandwidth.
+ */
+struct virtchnl_qos_cap_elem {
+	u8 tc_num;
+	u8 tc_prio;
+#define VIRTCHNL_ABITER_STRICT      0
+#define VIRTCHNL_ABITER_ETS         2
+	u8 arbiter;
+#define VIRTCHNL_STRICT_WEIGHT      1
+	u8 weight;
+	enum virtchnl_bw_limit_type type;
+	union {
+		struct virtchnl_shaper_bw shaper;
+		u8 pad2[32];
+	};
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(40, virtchnl_qos_cap_elem);
+
+struct virtchnl_qos_cap_list {
+	u16 vsi_id;
+	u16 num_elem;
+	struct virtchnl_qos_cap_elem cap[];
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(4, virtchnl_qos_cap_list);
+
+/* VIRTCHNL_OP_CONFIG_QUEUE_BW */
+struct virtchnl_queue_bw {
+	u16 queue_id;
+	u8 tc;
+	u8 pad;
+	struct virtchnl_shaper_bw shaper;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_queue_bw);
+
+struct virtchnl_queues_bw_cfg {
+	u16 vsi_id;
+	u16 num_queues;
+	struct virtchnl_queue_bw cfg[];
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(4, virtchnl_queues_bw_cfg);
+
+enum virtchnl_queue_type {
+	VIRTCHNL_QUEUE_TYPE_TX			= 0,
+	VIRTCHNL_QUEUE_TYPE_RX			= 1,
+};
+
+/* structure to specify a chunk of contiguous queues */
+struct virtchnl_queue_chunk {
+	/* see enum virtchnl_queue_type */
+	s32 type;
+	u16 start_queue_id;
+	u16 num_queues;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_queue_chunk);
+
+struct virtchnl_quanta_cfg {
+	u16 quanta_size;
+	struct virtchnl_queue_chunk queue_select;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_quanta_cfg);
+
 /**
  * virtchnl_vc_validate_vf_msg
  * @ver: Virtchnl version info
@@ -1558,6 +1644,34 @@ virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode,
 	case VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2:
 		valid_len = sizeof(struct virtchnl_vlan_setting);
 		break;
+	case VIRTCHNL_OP_GET_QOS_CAPS:
+		break;
+	case VIRTCHNL_OP_CONFIG_QUEUE_BW:
+		valid_len = sizeof(struct virtchnl_queues_bw_cfg);
+		if (msglen >= valid_len) {
+			struct virtchnl_queues_bw_cfg *q_bw =
+				(struct virtchnl_queues_bw_cfg *)msg;
+
+			if (q_bw->num_queues == 0) {
+				err_msg_format = true;
+				break;
+			}
+			valid_len = struct_size(q_bw, cfg, q_bw->num_queues);
+		}
+		break;
+	case VIRTCHNL_OP_CONFIG_QUANTA:
+		valid_len = sizeof(struct virtchnl_quanta_cfg);
+		if (msglen >= valid_len) {
+			struct virtchnl_quanta_cfg *q_quanta =
+				(struct virtchnl_quanta_cfg *)msg;
+
+			if (q_quanta->quanta_size == 0 ||
+			    q_quanta->queue_select.num_queues == 0) {
+				err_msg_format = true;
+				break;
+			}
+		}
+		break;
 	/* These are always errors coming from the VF. */
 	case VIRTCHNL_OP_EVENT:
 	case VIRTCHNL_OP_UNKNOWN:
Signature · 1 line
2.34.1
author
#14 lore raw

PATCH v2 2/5 ice: Support VF queue rate limit and quanta size configuration

In reply to Wenjun Wu

Add support to configure VF queue rate limit and quanta size.

For quanta size configuration, the quanta profiles are divided evenly by PF numbers. For each port, the first quanta profile is reserved for default. When VF is asked to set queue quanta size, PF will search for an available profile, change the fields and assigned this profile to the queue.

---
 drivers/net/ethernet/intel/ice/ice.h          |   2 +
 drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
 drivers/net/ethernet/intel/ice/ice_common.c   |  19 ++
 .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
 drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
 drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
 drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
 drivers/net/ethernet/intel/ice/ice_virtchnl.c | 312 ++++++++++++++++++
 drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
 .../intel/ice/ice_virtchnl_allowlist.c        |   6 +
 10 files changed, 372 insertions(+)
drivers/net/ethernet/intel/ice/ice.h +2 -0
index 34be1cb1e28f..677ab9571b3f 100644
@@ -642,6 +642,8 @@ struct ice_pf {
 #define ICE_VF_AGG_NODE_ID_START	65
 #define ICE_MAX_VF_AGG_NODES		32
 	struct ice_agg_node vf_agg_node[ICE_MAX_VF_AGG_NODES];
+
+	u8 num_quanta_prof_used;
 };
 
 extern struct workqueue_struct *ice_lag_wq;
drivers/net/ethernet/intel/ice/ice_base.c +2 -0
index 9ab9fb558b5e..efd01874434f 100644
@@ -377,6 +377,8 @@ ice_setup_tx_ctx(struct ice_tx_ring *ring, struct ice_tlan_ctx *tlan_ctx, u16 pf
 		break;
 	}
 
+	tlan_ctx->quanta_prof_idx = ring->quanta_prof_id;
+
 	tlan_ctx->tso_ena = ICE_TX_LEGACY;
 	tlan_ctx->tso_qnum = pf_q;
 
drivers/net/ethernet/intel/ice/ice_common.c +19 -0
index a0e43599eb55..910525a19a4c 100644
@@ -2463,6 +2463,23 @@ ice_parse_func_caps(struct ice_hw *hw, struct ice_hw_func_caps *func_p,
 	ice_recalc_port_limited_caps(hw, &func_p->common_cap);
 }
 
+/**
+ * ice_func_id_to_logical_id - map from function id to logical pf id
+ * @active_function_bitmap: active function bitmap
+ * @pf_id: function number of device
+ */
+static int ice_func_id_to_logical_id(u32 active_function_bitmap, u8 pf_id)
+{
+	u8 logical_id = 0;
+	u8 i;
+
+	for (i = 0; i < pf_id; i++)
+		if (active_function_bitmap & BIT(i))
+			logical_id++;
+
+	return logical_id;
+}
+
 /**
  * ice_parse_valid_functions_cap - Parse ICE_AQC_CAPS_VALID_FUNCTIONS caps
  * @hw: pointer to the HW struct
@@ -2480,6 +2497,8 @@ ice_parse_valid_functions_cap(struct ice_hw *hw, struct ice_hw_dev_caps *dev_p,
 	dev_p->num_funcs = hweight32(number);
 	ice_debug(hw, ICE_DBG_INIT, "dev caps: num_funcs = %d\n",
 		  dev_p->num_funcs);
+
+	hw->logical_pf_id = ice_func_id_to_logical_id(number, hw->pf_id);
 }
 
 /**
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +8 -0
index 6756f3d51d14..9da94e000394 100644
@@ -6,6 +6,14 @@
 #ifndef _ICE_HW_AUTOGEN_H_
 #define _ICE_HW_AUTOGEN_H_
 
+#define GLCOMM_QUANTA_PROF(_i)			(0x002D2D68 + ((_i) * 4))
+#define GLCOMM_QUANTA_PROF_MAX_INDEX		15
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_S	0
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_M	ICE_M(0x3FFF, 0)
+#define GLCOMM_QUANTA_PROF_MAX_CMD_S		16
+#define GLCOMM_QUANTA_PROF_MAX_CMD_M		ICE_M(0xFF, 16)
+#define GLCOMM_QUANTA_PROF_MAX_DESC_S		24
+#define GLCOMM_QUANTA_PROF_MAX_DESC_M		ICE_M(0x3F, 24)
 #define QTX_COMM_DBELL(_DBQM)			(0x002C0000 + ((_DBQM) * 4))
 #define QTX_COMM_HEAD(_DBQM)			(0x000E0000 + ((_DBQM) * 4))
 #define QTX_COMM_HEAD_HEAD_S			0
drivers/net/ethernet/intel/ice/ice_txrx.h +2 -0
index 166413fc33f4..7e152ab5b727 100644
@@ -381,6 +381,8 @@ struct ice_tx_ring {
 	u8 flags;
 	u8 dcb_tc;			/* Traffic class of ring */
 	u8 ptp_tx;
+
+	u16 quanta_prof_id;
 } ____cacheline_internodealigned_in_smp;
 
 static inline bool ice_ring_uses_build_skb(struct ice_rx_ring *ring)
drivers/net/ethernet/intel/ice/ice_type.h +1 -0
index a5429eca4350..504b367f1c77 100644
@@ -850,6 +850,7 @@ struct ice_hw {
 	u8 revision_id;
 
 	u8 pf_id;		/* device profile info */
+	u8 logical_pf_id;
 	enum ice_phy_model phy_model;
 
 	u16 max_burst_size;	/* driver sets this value */
drivers/net/ethernet/intel/ice/ice_vf_lib.h +9 -0
index 67172fdd9bc2..6499d83cc706 100644
@@ -52,6 +52,13 @@ struct ice_mdd_vf_events {
 	u16 last_printed;
 };
 
+struct ice_vf_qs_bw {
+	u16 queue_id;
+	u32 committed;
+	u32 peak;
+	u8 tc;
+};
+
 /* VF operations */
 struct ice_vf_ops {
 	enum ice_disq_rst_src reset_type;
@@ -133,6 +140,8 @@ struct ice_vf {
 
 	/* devlink port data */
 	struct devlink_port devlink_port;
+
+	struct ice_vf_qs_bw qs_bw[ICE_MAX_RSS_QS_PER_VF];
 };
 
 /* Flags for controlling behavior of ice_reset_vf */
drivers/net/ethernet/intel/ice/ice_virtchnl.c +312 -0

348 lines not on this page · raw message

drivers/net/ethernet/intel/ice/ice_virtchnl.h +11 -0
index cd747718de73..0efb9c0f669a 100644
@@ -13,6 +13,13 @@
 /* Restrict number of MAC Addr and VLAN that non-trusted VF can programmed */
 #define ICE_MAX_VLAN_PER_VF		8
 
+#define ICE_DFLT_QUANTA 1024
+#define ICE_MAX_QUANTA_SIZE 4096
+#define ICE_MIN_QUANTA_SIZE 256
+
+#define calc_quanta_desc(x)	\
+	max_t(u16, 12, min_t(u16, 63, (((x) + 66) / 132) * 2 + 4))
+
 /* MAC filters: 1 is reserved for the VF's default/perm_addr/LAA MAC, 1 for
  * broadcast, and 16 for additional unicast/multicast filters
  */
@@ -51,6 +58,10 @@ struct ice_virtchnl_ops {
 	int (*dis_vlan_stripping_v2_msg)(struct ice_vf *vf, u8 *msg);
 	int (*ena_vlan_insertion_v2_msg)(struct ice_vf *vf, u8 *msg);
 	int (*dis_vlan_insertion_v2_msg)(struct ice_vf *vf, u8 *msg);
+	int (*get_qos_caps)(struct ice_vf *vf);
+	int (*cfg_q_tc_map)(struct ice_vf *vf, u8 *msg);
+	int (*cfg_q_bw)(struct ice_vf *vf, u8 *msg);
+	int (*cfg_q_quanta)(struct ice_vf *vf, u8 *msg);
 };
 
 #ifdef CONFIG_PCI_IOV
drivers/net/ethernet/intel/ice/ice_virtchnl_allowlist.c +6 -0
index 7d547fa616fa..2e3f63a429cd 100644
@@ -85,6 +85,11 @@ static const u32 fdir_pf_allowlist_opcodes[] = {
 	VIRTCHNL_OP_ADD_FDIR_FILTER, VIRTCHNL_OP_DEL_FDIR_FILTER,
 };
 
+static const u32 tc_allowlist_opcodes[] = {
+	VIRTCHNL_OP_GET_QOS_CAPS, VIRTCHNL_OP_CONFIG_QUEUE_BW,
+	VIRTCHNL_OP_CONFIG_QUANTA,
+};
+
 struct allowlist_opcode_info {
 	const u32 *opcodes;
 	size_t size;
@@ -105,6 +110,7 @@ static const struct allowlist_opcode_info allowlist_opcodes[] = {
 	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF, adv_rss_pf_allowlist_opcodes),
 	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_FDIR_PF, fdir_pf_allowlist_opcodes),
 	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_VLAN_V2, vlan_v2_allowlist_opcodes),
+	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_QOS, tc_allowlist_opcodes),
 };
 
 /**
Signature · 1 line
2.34.1
#15 lore raw

Re: [PATCH iwl-next v2 2/5] ice: Support VF queue rate limit and quanta size configuration

Quoted 137 lines · Wenjun Wu

Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.

Add support to configure VF queue rate limit and quanta size.

For quanta size configuration, the quanta profiles are divided evenly by PF numbers. For each port, the first quanta profile is reserved for default. When VF is asked to set queue quanta size, PF will search for an available profile, change the fields and assigned this profile to the queue.

---
  drivers/net/ethernet/intel/ice/ice.h          |   2 +
  drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
  drivers/net/ethernet/intel/ice/ice_common.c   |  19 ++
  .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
  drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
  drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
  drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
  drivers/net/ethernet/intel/ice/ice_virtchnl.c | 312 ++++++++++++++++++
  drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
  .../intel/ice/ice_virtchnl_allowlist.c        |   6 +
  10 files changed, 372 insertions(+)
drivers/net/ethernet/intel/ice/ice.h +2 -0
index 34be1cb1e28f..677ab9571b3f 100644
@@ -642,6 +642,8 @@ struct ice_pf {
  #define ICE_VF_AGG_NODE_ID_START       65
  #define ICE_MAX_VF_AGG_NODES           32
         struct ice_agg_node vf_agg_node[ICE_MAX_VF_AGG_NODES];
+
+       u8 num_quanta_prof_used;
  };
 
  extern struct workqueue_struct *ice_lag_wq;
drivers/net/ethernet/intel/ice/ice_base.c +2 -0
index 9ab9fb558b5e..efd01874434f 100644
@@ -377,6 +377,8 @@ ice_setup_tx_ctx(struct ice_tx_ring *ring, struct ice_tlan_ctx *tlan_ctx, u16 pf
                 break;
         }
 
+       tlan_ctx->quanta_prof_idx = ring->quanta_prof_id;
+
         tlan_ctx->tso_ena = ICE_TX_LEGACY;
         tlan_ctx->tso_qnum = pf_q;
 
drivers/net/ethernet/intel/ice/ice_common.c +19 -0
index a0e43599eb55..910525a19a4c 100644
@@ -2463,6 +2463,23 @@ ice_parse_func_caps(struct ice_hw *hw, struct ice_hw_func_caps *func_p,
         ice_recalc_port_limited_caps(hw, &func_p->common_cap);
  }
 
+/**
+ * ice_func_id_to_logical_id - map from function id to logical pf id
+ * @active_function_bitmap: active function bitmap
+ * @pf_id: function number of device
+ */
+static int ice_func_id_to_logical_id(u32 active_function_bitmap, u8 pf_id)
+{
+       u8 logical_id = 0;
+       u8 i;
+
+       for (i = 0; i < pf_id; i++)
+               if (active_function_bitmap & BIT(i))
+                       logical_id++;
+
+       return logical_id;
+}
+
  /**
   * ice_parse_valid_functions_cap - Parse ICE_AQC_CAPS_VALID_FUNCTIONS caps
   * @hw: pointer to the HW struct
@@ -2480,6 +2497,8 @@ ice_parse_valid_functions_cap(struct ice_hw *hw, struct ice_hw_dev_caps *dev_p,
         dev_p->num_funcs = hweight32(number);
         ice_debug(hw, ICE_DBG_INIT, "dev caps: num_funcs = %d\n",
                   dev_p->num_funcs);
+
+       hw->logical_pf_id = ice_func_id_to_logical_id(number, hw->pf_id);
  }
 
  /**
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +8 -0
index 6756f3d51d14..9da94e000394 100644
@@ -6,6 +6,14 @@
  #ifndef _ICE_HW_AUTOGEN_H_
  #define _ICE_HW_AUTOGEN_H_
 
+#define GLCOMM_QUANTA_PROF(_i)                 (0x002D2D68 + ((_i) * 4))
+#define GLCOMM_QUANTA_PROF_MAX_INDEX           15
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_S       0
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_M       ICE_M(0x3FFF, 0)
+#define GLCOMM_QUANTA_PROF_MAX_CMD_S           16
+#define GLCOMM_QUANTA_PROF_MAX_CMD_M           ICE_M(0xFF, 16)
+#define GLCOMM_QUANTA_PROF_MAX_DESC_S          24
+#define GLCOMM_QUANTA_PROF_MAX_DESC_M          ICE_M(0x3F, 24)
  #define QTX_COMM_DBELL(_DBQM)                  (0x002C0000 + ((_DBQM) * 4))
  #define QTX_COMM_HEAD(_DBQM)                   (0x000E0000 + ((_DBQM) * 4))
  #define QTX_COMM_HEAD_HEAD_S                   0
drivers/net/ethernet/intel/ice/ice_txrx.h +2 -0
index 166413fc33f4..7e152ab5b727 100644
@@ -381,6 +381,8 @@ struct ice_tx_ring {
         u8 flags;
         u8 dcb_tc;                      /* Traffic class of ring */
         u8 ptp_tx;
+
+       u16 quanta_prof_id;
  } ____cacheline_internodealigned_in_smp;
 
  static inline bool ice_ring_uses_build_skb(struct ice_rx_ring *ring)
drivers/net/ethernet/intel/ice/ice_type.h +1 -0
index a5429eca4350..504b367f1c77 100644
@@ -850,6 +850,7 @@ struct ice_hw {
         u8 revision_id;
 
         u8 pf_id;               /* device profile info */
+       u8 logical_pf_id;
         enum ice_phy_model phy_model;
 
         u16 max_burst_size;     /* driver sets this value */
drivers/net/ethernet/intel/ice/ice_vf_lib.h +0 -0
index 67172fdd9bc2..6499d83cc706 100644

drivers/net/ethernet/intel/ice/ice_vf_lib.h struct ice_mdd_vf_events { · line 60

Quoted 9 lines
drivers/net/ethernet/intel/ice/ice_vf_lib.h +6 -0
@@ -52,6 +52,13 @@ struct ice_mdd_vf_events {
         u16 last_printed;
  };
 
+struct ice_vf_qs_bw {
+       u16 queue_id;
+       u32 committed;
+       u32 peak;
+       u8 tc;
+};

Nit, but you should re-arrange this struct to have the largest numbers first.

Quoted 113 lines
drivers/net/ethernet/intel/ice/ice_vf_lib.h +3 -0
+
  /* VF operations */
  struct ice_vf_ops {
         enum ice_disq_rst_src reset_type;
@@ -133,6 +140,8 @@ struct ice_vf {
 
         /* devlink port data */
         struct devlink_port devlink_port;
+
+       struct ice_vf_qs_bw qs_bw[ICE_MAX_RSS_QS_PER_VF];
  };
 
  /* Flags for controlling behavior of ice_reset_vf */
drivers/net/ethernet/intel/ice/ice_virtchnl.c +85 -0
index 85d996531502..9fc1a9d1bcd4 100644
@@ -495,6 +495,9 @@ static int ice_vc_get_vf_res_msg(struct ice_vf *vf, u8 *msg)
         if (vf->driver_caps & VIRTCHNL_VF_OFFLOAD_USO)
                 vfres->vf_cap_flags |= VIRTCHNL_VF_OFFLOAD_USO;
 
+       if (vf->driver_caps & VIRTCHNL_VF_OFFLOAD_QOS)
+               vfres->vf_cap_flags |= VIRTCHNL_VF_OFFLOAD_QOS;
+
         vfres->num_vsis = 1;
         /* Tx and Rx queue are equal for VF */
         vfres->num_queue_pairs = vsi->num_txq;
@@ -985,6 +988,170 @@ static int ice_vc_config_rss_lut(struct ice_vf *vf, u8 *msg)
                                      NULL, 0);
  }
 
+/**
+ * ice_vc_get_qos_caps - Get current QoS caps from PF
+ * @vf: pointer to the VF info
+ *
+ * Get VF's QoS capabilities, such as TC number, arbiter and
+ * bandwidth from PF.
+ */
+static int ice_vc_get_qos_caps(struct ice_vf *vf)
+{
+       enum virtchnl_status_code v_ret = VIRTCHNL_STATUS_SUCCESS; > +       struct virtchnl_qos_cap_list *cap_list = NULL;
+       u8 tc_prio[ICE_MAX_TRAFFIC_CLASS] = { 0 };
+       struct virtchnl_qos_cap_elem *cfg = NULL;
+       struct ice_vsi_ctx *vsi_ctx;
+       struct ice_pf *pf = vf->pf;
+       struct ice_port_info *pi;
+       struct ice_vsi *vsi;
+       u8 numtc, tc;
+       u16 len = 0;
+       int ret, i;
+
+       if (!test_bit(ICE_VF_STATE_ACTIVE, vf->vf_states)) {
+               v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+               goto err;
+       }
+
+       vsi = ice_get_vf_vsi(vf);
+       if (!vsi) {
+               v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+               goto err;
+       }
+
+       pi = pf->hw.port_info;
+       numtc = vsi->tc_cfg.numtc;
+
+       vsi_ctx = ice_get_vsi_ctx(pi->hw, vf->lan_vsi_idx);
+       if (!vsi_ctx) {
+               v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+               goto err;
+       }
+
+       len = struct_size(cap_list, cap, numtc);
+       cap_list = kzalloc(len, GFP_KERNEL);
+       if (!cap_list) {
+               v_ret = VIRTCHNL_STATUS_ERR_NO_MEMORY;
+               len = 0;
+               goto err;
+       }
+
+       cap_list->vsi_id = vsi->vsi_num;
+       cap_list->num_elem = numtc;
+
+       /* Store the UP2TC configuration from DCB to a user priority bitmap
+        * of each TC. Each element of prio_of_tc represents one TC. Each
+        * bitmap indicates the user priorities belong to this TC.
+        */
+       for (i = 0; i < ICE_MAX_USER_PRIORITY; i++) {
+               tc = pi->qos_cfg.local_dcbx_cfg.etscfg.prio_table[i];
+               tc_prio[tc] |= BIT(i);
+       }
+
+       for (i = 0; i < numtc; i++) {
+               cfg = &cap_list->cap[i];
+               cfg->tc_num = i;
+               cfg->tc_prio = tc_prio[i];
+               cfg->arbiter = pi->qos_cfg.local_dcbx_cfg.etscfg.tsatable[i];
+               cfg->weight = VIRTCHNL_STRICT_WEIGHT;
+               cfg->type = VIRTCHNL_BW_SHAPER;
+               cfg->shaper.committed = vsi_ctx->sched.bw_t_info[i].cir_bw.bw;
+               cfg->shaper.peak = vsi_ctx->sched.bw_t_info[i].eir_bw.bw;
+       }
+
+err:
+       ret = ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_GET_QOS_CAPS, v_ret,
+                                   (u8 *)cap_list, len);
+       kfree(cap_list);
+       return ret;
+}
+
+/**
+ * ice_vf_cfg_qs_bw - Configure per queue bandwidth
+ * @vf: pointer to the VF info
+ * @num_queues: number of queues to be configured

drivers/net/ethernet/intel/ice/ice_virtchnl.c static int ice_vc_config_rss_lut(struct ice_vf *vf, u8 *msg) · line 1082

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +10 -0
+ *
+ * Configure per queue bandwidth.
+ */
+static int ice_vf_cfg_qs_bw(struct ice_vf *vf, u16 num_queues)
+{
+       struct ice_hw *hw = &vf->pf->hw;
+       struct ice_vsi *vsi;
+       u32 p_rate;
+       int ret; > +       u16 i;
+       u8 tc;

Nit, the scope of p_rate and tc can be reduced to the for loop below.

Quoted 7 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +7 -0
+
+       vsi = ice_get_vf_vsi(vf);
+       if (!vsi)
+               return -EINVAL;
+
+       for (i = 0; i < num_queues; i++) {
+               p_rate = vf->qs_bw[i].peak;

drivers/net/ethernet/intel/ice/ice_virtchnl.c static int ice_vc_config_rss_lut(struct ice_vf *vf, u8 *msg) · line 1101

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +10 -0
+               tc = vf->qs_bw[i].tc;
+               if (p_rate) {
+                       ret = ice_cfg_q_bw_lmt(hw->port_info, vsi->idx, tc,
+                                              vf->qs_bw[i].queue_id,
+                                              ICE_MAX_BW, p_rate);
+               } else {
+                       ret = ice_cfg_q_bw_dflt_lmt(hw->port_info, vsi->idx, tc,
+                                                   vf->qs_bw[i].queue_id,
+                                                   ICE_MAX_BW);
+               }

Nit, brackets not needed for single statement in the if/else blocks.

Quoted 23 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +23 -0
+               if (ret)
+                       return ret;
+       }
+
+       return 0;
+}
+
+/**
+ * ice_vf_cfg_q_quanta_profile
+ * @vf: pointer to the VF info
+ * @quanta_prof_idx: pointer to the quanta profile index
+ * @quanta_size: quanta size to be set
+ *
+ * This function chooses available quanta profile and configures the register.
+ * The quanta profile is evenly divided by the number of device ports, and then
+ * available to the specific PF and VFs. The first profile for each PF is a
+ * reserved default profile. Only quanta size of the rest unused profile can be
+ * modified.
+ */
+static int ice_vf_cfg_q_quanta_profile(struct ice_vf *vf, u16 quanta_size,
+                                      u16 *quanta_prof_idx)
+{
+       const u16 n_desc = calc_quanta_desc(quanta_size);

drivers/net/ethernet/intel/ice/ice_virtchnl.c static int ice_vc_config_rss_lut(struct ice_vf *vf, u8 *msg) · line 1134

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +10 -0
+       struct ice_hw *hw = &vf->pf->hw;
+       const u16 n_cmd = 2 * n_desc;
+       struct ice_pf *pf = vf->pf;
+       u16 per_pf, begin_id;
+       u8 n_used;
+       u32 reg;
+
+       per_pf = (GLCOMM_QUANTA_PROF_MAX_INDEX + 1) / hw->dev_caps.num_funcs;
+       begin_id = hw->logical_pf_id * per_pf;
+       n_used = pf->num_quanta_prof_used;

Nit, per_pf and n_used can be local to the first else block below.

Quoted 59 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +52 -0
+
+       if (quanta_size == ICE_DFLT_QUANTA) {
+               *quanta_prof_idx = begin_id;
+       } else {
+               if (n_used < per_pf) {
+                       *quanta_prof_idx = begin_id + 1 + n_used;
+                       pf->num_quanta_prof_used++;
+               } else {
+                       return -EINVAL;
+               }
+       }
+
+       reg = FIELD_PREP(GLCOMM_QUANTA_PROF_QUANTA_SIZE_M, quanta_size) |
+             FIELD_PREP(GLCOMM_QUANTA_PROF_MAX_CMD_M, n_cmd) |
+             FIELD_PREP(GLCOMM_QUANTA_PROF_MAX_DESC_M, n_desc);
+       wr32(hw, GLCOMM_QUANTA_PROF(*quanta_prof_idx), reg);
+
+       return 0;
+}
+
  /**
   * ice_vc_cfg_promiscuous_mode_msg
   * @vf: pointer to the VF info
@@ -1587,6 +1754,136 @@ static int ice_vc_cfg_irq_map_msg(struct ice_vf *vf, u8 *msg)
                                      NULL, 0);
  }
 
+/**
+ * ice_vc_cfg_q_bw - Configure per queue bandwidth
+ * @vf: pointer to the VF info
+ * @msg: pointer to the msg buffer which holds the command descriptor
+ *
+ * Configure VF queues bandwidth.
+ */
+static int ice_vc_cfg_q_bw(struct ice_vf *vf, u8 *msg)
+{
+       enum virtchnl_status_code v_ret = VIRTCHNL_STATUS_SUCCESS;
+       struct virtchnl_queues_bw_cfg *qbw =
+               (struct virtchnl_queues_bw_cfg *)msg;
+       struct ice_vf_qs_bw *qs_bw;
+       struct ice_vsi *vsi;
+       size_t len;
+       u16 i;
+
+       if (!test_bit(ICE_VF_STATE_ACTIVE, vf->vf_states) ||
+           !ice_vc_isvalid_vsi_id(vf, qbw->vsi_id)) {
+               v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+               goto err;
+       }
+
+       vsi = ice_get_vf_vsi(vf);
+       if (!vsi || vsi->vsi_num != qbw->vsi_id) {
+               v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+               goto err;
+       }
+
+       if (qbw->num_queues > ICE_MAX_RSS_QS_PER_VF ||
+           qbw->num_queues > min_t(u16, vsi->alloc_txq, vsi->alloc_rxq)) {
+               dev_err(ice_pf_to_dev(vf->pf), "VF-%d trying to configure more than allocated number of queues: %d\n",

drivers/net/ethernet/intel/ice/ice_virtchnl.c static int ice_vc_cfg_irq_map_msg(struct ice_vf *vf, u8 *msg) · line 1798

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +10 -0
+                       vf->vf_id, min_t(u16, vsi->alloc_txq, vsi->alloc_rxq));
+               v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+               goto err;
+       }
+
+       len = sizeof(struct ice_vf_qs_bw) * qbw->num_queues;
+       qs_bw = kzalloc(len, GFP_KERNEL);
+       if (!qs_bw) {
+               v_ret = VIRTCHNL_STATUS_ERR_NO_MEMORY;
+               goto err_bw;

If there's a reason you need to allocate this, the goto label should be "err" instead of "err_bw".

drivers/net/ethernet/intel/ice/ice_virtchnl.c static int ice_vc_cfg_irq_map_msg(struct ice_vf *vf, u8 *msg) · line 1806

Quoted 8 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +8 -0
+       }
+
+       for (i = 0; i < qbw->num_queues; i++) {
+               qs_bw[i].queue_id = qbw->cfg[i].queue_id;
+               qs_bw[i].peak = qbw->cfg[i].shaper.peak;
+               qs_bw[i].committed = qbw->cfg[i].shaper.committed;
+               qs_bw[i].tc = qbw->cfg[i].tc;
+       }

Do you need to allocate qs_bw? What's stopping you from setting vf->qs_bw directly here? This would remove the kzalloc above and memcpy below. It would also get rid of the "err_bw" label.

Also, does the virtchnl_queues_bw_cfg need to be validated in any way?

Quoted 22 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +22 -0
+
+       memcpy(vf->qs_bw, qs_bw, len);
+
+err_bw:
+       kfree(qs_bw);
+
+err:
+       /* send the response to the VF */
+       return ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_CONFIG_QUEUE_BW,
+                                   v_ret, NULL, 0);
+}
+
+/**
+ * ice_vc_cfg_q_quanta - Configure per queue quanta
+ * @vf: pointer to the VF info
+ * @msg: pointer to the msg buffer which holds the command descriptor
+ *
+ * Configure VF queues quanta.
+ */
+static int ice_vc_cfg_q_quanta(struct ice_vf *vf, u8 *msg)
+{
+       u16 quanta_prof_id, quanta_size, start_qid, num_queues, end_qid, i;

drivers/net/ethernet/intel/ice/ice_virtchnl.c static int ice_vc_cfg_irq_map_msg(struct ice_vf *vf, u8 *msg) · line 1838

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +10 -0
+       enum virtchnl_status_code v_ret = VIRTCHNL_STATUS_SUCCESS;
+       struct virtchnl_quanta_cfg *qquanta =
+               (struct virtchnl_quanta_cfg *)msg;
+       struct ice_vsi *vsi;
+       int ret;
+
+       start_qid = qquanta->queue_select.start_queue_id;
+       num_queues = qquanta->queue_select.num_queues;
+       quanta_size = qquanta->quanta_size;
+       end_qid = start_qid + num_queues;

Does it make sense to set these right before they are used instead?

Quoted 144 lines
drivers/net/ethernet/intel/ice/ice_virtchnl.c +63 -0
+
+       if (!test_bit(ICE_VF_STATE_ACTIVE, vf->vf_states)) {
+               v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+               goto err;
+       }
+
+       vsi = ice_get_vf_vsi(vf);
+       if (!vsi) {
+               v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+               goto err;
+       }
+
+       if (end_qid > ICE_MAX_RSS_QS_PER_VF ||
+           end_qid > min_t(u16, vsi->alloc_txq, vsi->alloc_rxq)) {
+               dev_err(ice_pf_to_dev(vf->pf), "VF-%d trying to configure more than allocated number of queues: %d\n",
+                       vf->vf_id, min_t(u16, vsi->alloc_txq, vsi->alloc_rxq));
+               v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+               goto err;
+       }
+
+       if (quanta_size > ICE_MAX_QUANTA_SIZE ||
+           quanta_size < ICE_MIN_QUANTA_SIZE) {
+               v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+               goto err;
+       }
+
+       if (quanta_size % 64) {
+               dev_err(ice_pf_to_dev(vf->pf), "quanta size should be the product of 64\n");
+               v_ret = VIRTCHNL_STATUS_ERR_PARAM;
+               goto err;
+       }
+
+       ret = ice_vf_cfg_q_quanta_profile(vf, quanta_size,
+                                         &quanta_prof_id);
+       if (ret) {
+               v_ret = VIRTCHNL_STATUS_ERR_NOT_SUPPORTED;
+               goto err;
+       }
+
+       for (i = start_qid; i < end_qid; i++)
+               vsi->tx_rings[i]->quanta_prof_id = quanta_prof_id;
+
+err:
+       /* send the response to the VF */
+       return ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_CONFIG_QUANTA,
+                                    v_ret, NULL, 0);
+}
+
  /**
   * ice_vc_cfg_qs_msg
   * @vf: pointer to the VF info
@@ -1710,6 +2007,9 @@ static int ice_vc_cfg_qs_msg(struct ice_vf *vf, u8 *msg)
                 }
         }
 
+       if (ice_vf_cfg_qs_bw(vf, qci->num_queue_pairs))
+               goto error_param;
+
         /* send the response to the VF */
         return ice_vc_send_msg_to_vf(vf, VIRTCHNL_OP_CONFIG_VSI_QUEUES,
                                      VIRTCHNL_STATUS_SUCCESS, NULL, 0);
@@ -3687,6 +3987,9 @@ static const struct ice_virtchnl_ops ice_virtchnl_dflt_ops = {
         .dis_vlan_stripping_v2_msg = ice_vc_dis_vlan_stripping_v2_msg,
         .ena_vlan_insertion_v2_msg = ice_vc_ena_vlan_insertion_v2_msg,
         .dis_vlan_insertion_v2_msg = ice_vc_dis_vlan_insertion_v2_msg,
+       .get_qos_caps = ice_vc_get_qos_caps,
+       .cfg_q_bw = ice_vc_cfg_q_bw,
+       .cfg_q_quanta = ice_vc_cfg_q_quanta,
  };
 
  /**
@@ -4040,6 +4343,15 @@ void ice_vc_process_vf_msg(struct ice_pf *pf, struct ice_rq_event_info *event,
         case VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2:
                 err = ops->dis_vlan_insertion_v2_msg(vf, msg);
                 break;
+       case VIRTCHNL_OP_GET_QOS_CAPS:
+               err = ops->get_qos_caps(vf);
+               break;
+       case VIRTCHNL_OP_CONFIG_QUEUE_BW:
+               err = ops->cfg_q_bw(vf, msg);
+               break;
+       case VIRTCHNL_OP_CONFIG_QUANTA:
+               err = ops->cfg_q_quanta(vf, msg);
+               break;
         case VIRTCHNL_OP_UNKNOWN:
         default:
                 dev_err(dev, "Unsupported opcode %d from VF %d\n", v_opcode,
drivers/net/ethernet/intel/ice/ice_virtchnl.h +11 -0
index cd747718de73..0efb9c0f669a 100644
@@ -13,6 +13,13 @@
  /* Restrict number of MAC Addr and VLAN that non-trusted VF can programmed */
  #define ICE_MAX_VLAN_PER_VF            8
 
+#define ICE_DFLT_QUANTA 1024
+#define ICE_MAX_QUANTA_SIZE 4096
+#define ICE_MIN_QUANTA_SIZE 256
+
+#define calc_quanta_desc(x)    \
+       max_t(u16, 12, min_t(u16, 63, (((x) + 66) / 132) * 2 + 4))
+
  /* MAC filters: 1 is reserved for the VF's default/perm_addr/LAA MAC, 1 for
   * broadcast, and 16 for additional unicast/multicast filters
   */
@@ -51,6 +58,10 @@ struct ice_virtchnl_ops {
         int (*dis_vlan_stripping_v2_msg)(struct ice_vf *vf, u8 *msg);
         int (*ena_vlan_insertion_v2_msg)(struct ice_vf *vf, u8 *msg);
         int (*dis_vlan_insertion_v2_msg)(struct ice_vf *vf, u8 *msg);
+       int (*get_qos_caps)(struct ice_vf *vf);
+       int (*cfg_q_tc_map)(struct ice_vf *vf, u8 *msg);
+       int (*cfg_q_bw)(struct ice_vf *vf, u8 *msg);
+       int (*cfg_q_quanta)(struct ice_vf *vf, u8 *msg);
  };
 
  #ifdef CONFIG_PCI_IOV
drivers/net/ethernet/intel/ice/ice_virtchnl_allowlist.c +6 -1
index 7d547fa616fa..2e3f63a429cd 100644
@@ -85,6 +85,11 @@ static const u32 fdir_pf_allowlist_opcodes[] = {
         VIRTCHNL_OP_ADD_FDIR_FILTER, VIRTCHNL_OP_DEL_FDIR_FILTER,
  };
 
+static const u32 tc_allowlist_opcodes[] = {
+       VIRTCHNL_OP_GET_QOS_CAPS, VIRTCHNL_OP_CONFIG_QUEUE_BW,
+       VIRTCHNL_OP_CONFIG_QUANTA,
+};
+
  struct allowlist_opcode_info {
         const u32 *opcodes;
         size_t size;
@@ -105,6 +110,7 @@ static const struct allowlist_opcode_info allowlist_opcodes[] = {
         ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF, adv_rss_pf_allowlist_opcodes),
         ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_FDIR_PF, fdir_pf_allowlist_opcodes),
         ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_VLAN_V2, vlan_v2_allowlist_opcodes),
+       ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_QOS, tc_allowlist_opcodes),
  };
 
  /**
--

2.34.1

author
#16 lore raw

PATCH v2 3/5 iavf: Add devlink and devlink port support

In reply to Wenjun Wu

From: Jun Zhang <[email protected]>

To allow user to configure queue bandwidth, devlink port support is added to support devlink port rate API.

Add devlink framework registration/unregistration on iavf driver initialization and remove, and devlink port of DEVLINK_PORT_FLAVOUR_VIRTUAL is created to be associated iavf net device.

---
 drivers/net/ethernet/intel/Kconfig            |  1 +
 drivers/net/ethernet/intel/iavf/Makefile      |  2 +-
 drivers/net/ethernet/intel/iavf/iavf.h        |  6 ++
 .../net/ethernet/intel/iavf/iavf_devlink.c    | 93 +++++++++++++++++++
 .../net/ethernet/intel/iavf/iavf_devlink.h    | 17 ++++
 drivers/net/ethernet/intel/iavf/iavf_main.c   | 14 +++
 6 files changed, 132 insertions(+), 1 deletion(-)
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.c
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.h
drivers/net/ethernet/intel/Kconfig +1 -0
index 9bc0a9519899..f916b8ef6acb 100644
@@ -256,6 +256,7 @@ config I40EVF
 	tristate "Intel(R) Ethernet Adaptive Virtual Function support"
 	select IAVF
 	depends on PCI_MSI
+	select NET_DEVLINK
 	help
 	  This driver supports virtual functions for Intel XL710,
 	  X710, X722, XXV710, and all devices advertising support for
drivers/net/ethernet/intel/iavf/Makefile +1 -1
index 9c3e45c54d01..b5d7db97ab8b 100644
@@ -12,5 +12,5 @@ subdir-ccflags-y += -I$(src)
 obj-$(CONFIG_IAVF) += iavf.o
 
 iavf-objs := iavf_main.o iavf_ethtool.o iavf_virtchnl.o iavf_fdir.o \
-	     iavf_adv_rss.o \
+	     iavf_adv_rss.o iavf_devlink.o \
 	     iavf_txrx.o iavf_common.o iavf_adminq.o iavf_client.o
drivers/net/ethernet/intel/iavf/iavf.h +6 -0
index 8cbdebc5b698..519aeaec793c 100644
@@ -33,9 +33,11 @@
 #include <net/udp.h>
 #include <net/tc_act/tc_gact.h>
 #include <net/tc_act/tc_mirred.h>
+#include <net/devlink.h>
 
 #include "iavf_type.h"
 #include <linux/avf/virtchnl.h>
+#include "iavf_devlink.h"
 #include "iavf_txrx.h"
 #include "iavf_fdir.h"
 #include "iavf_adv_rss.h"
@@ -369,6 +371,10 @@ struct iavf_adapter {
 	struct net_device *netdev;
 	struct pci_dev *pdev;
 
+	/* devlink & port data */
+	struct devlink *devlink;
+	struct devlink_port devlink_port;
+
 	struct iavf_hw hw; /* defined in iavf_type.h */
 
 	enum iavf_state_t state;
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +93 -0

93 lines not on this page · raw message

drivers/net/ethernet/intel/iavf/iavf_devlink.h new file +17 -0
new file mode 100644
index 000000000000..5c122278611a
@@ -0,0 +1,17 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright (C) 2023 Intel Corporation */
+
+#ifndef _IAVF_DEVLINK_H_
+#define _IAVF_DEVLINK_H_
+
+/* iavf devlink structure pointing to iavf adapter */
+struct iavf_devlink {
+	struct iavf_adapter *devlink_ref;	/* ref to iavf adapter */
+};
+
+int iavf_devlink_register(struct iavf_adapter *adapter);
+void iavf_devlink_unregister(struct iavf_adapter *adapter);
+int iavf_devlink_port_register(struct iavf_adapter *adapter);
+void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
+
+#endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +14 -0
index 7b300c86ceda..db010e68d5d2 100644
@@ -2037,6 +2037,7 @@ static void iavf_finish_config(struct work_struct *work)
 				iavf_free_rss(adapter);
 				iavf_free_misc_irq(adapter);
 				iavf_reset_interrupt_capability(adapter);
+				iavf_devlink_port_unregister(adapter);
 				iavf_change_state(adapter,
 						  __IAVF_INIT_CONFIG_ADAPTER);
 				goto out;
@@ -2708,6 +2709,9 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 	if (err)
 		goto err_sw_init;
 
+	if (!adapter->netdev_registered)
+		iavf_devlink_port_register(adapter);
+
 	netif_carrier_off(netdev);
 	adapter->link_up = false;
 	netif_tx_stop_all_queues(netdev);
@@ -2749,6 +2753,7 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 err_mem:
 	iavf_free_rss(adapter);
 	iavf_free_misc_irq(adapter);
+	iavf_devlink_port_unregister(adapter);
 err_sw_init:
 	iavf_reset_interrupt_capability(adapter);
 err:
@@ -4995,6 +5000,12 @@ static int iavf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	/* Setup the wait queue for indicating virtchannel events */
 	init_waitqueue_head(&adapter->vc_waitqueue);
 
+	/* Register iavf adapter with devlink */
+	err = iavf_devlink_register(adapter);
+	if (err)
+		dev_err(&pdev->dev, "devlink registration failed: %d\n", err);
+
+	/* Keep driver interface even on devlink registration failure */
 	return 0;
 
 err_ioremap:
@@ -5139,6 +5150,9 @@ static void iavf_remove(struct pci_dev *pdev)
 				 err);
 	}
 
+	iavf_devlink_port_unregister(adapter);
+	iavf_devlink_unregister(adapter);
+
 	mutex_lock(&adapter->crit_lock);
 	dev_info(&adapter->pdev->dev, "Removing device\n");
 	iavf_change_state(adapter, __IAVF_REMOVE);
Signature · 1 line
2.34.1
#17 lore raw

Re: [PATCH iwl-next v2 3/5] iavf: Add devlink and devlink port support

Quoted 64 lines · Wenjun Wu

Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.

From: Jun Zhang <[email protected]>

To allow user to configure queue bandwidth, devlink port support is added to support devlink port rate API.

Add devlink framework registration/unregistration on iavf driver initialization and remove, and devlink port of DEVLINK_PORT_FLAVOUR_VIRTUAL is created to be associated iavf net device.

---
  drivers/net/ethernet/intel/Kconfig            |  1 +
  drivers/net/ethernet/intel/iavf/Makefile      |  2 +-
  drivers/net/ethernet/intel/iavf/iavf.h        |  6 ++
  .../net/ethernet/intel/iavf/iavf_devlink.c    | 93 +++++++++++++++++++
  .../net/ethernet/intel/iavf/iavf_devlink.h    | 17 ++++
  drivers/net/ethernet/intel/iavf/iavf_main.c   | 14 +++
  6 files changed, 132 insertions(+), 1 deletion(-)

create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.c create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.h

drivers/net/ethernet/intel/Kconfig +1 -0
index 9bc0a9519899..f916b8ef6acb 100644
@@ -256,6 +256,7 @@ config I40EVF
         tristate "Intel(R) Ethernet Adaptive Virtual Function support"
         select IAVF
         depends on PCI_MSI
+       select NET_DEVLINK
         help
           This driver supports virtual functions for Intel XL710,
           X710, X722, XXV710, and all devices advertising support for
drivers/net/ethernet/intel/iavf/Makefile +1 -1
index 9c3e45c54d01..b5d7db97ab8b 100644
@@ -12,5 +12,5 @@ subdir-ccflags-y += -I$(src)
  obj-$(CONFIG_IAVF) += iavf.o
 
  iavf-objs := iavf_main.o iavf_ethtool.o iavf_virtchnl.o iavf_fdir.o \
-            iavf_adv_rss.o \
+            iavf_adv_rss.o iavf_devlink.o \
              iavf_txrx.o iavf_common.o iavf_adminq.o iavf_client.o
drivers/net/ethernet/intel/iavf/iavf.h +2 -0
index 8cbdebc5b698..519aeaec793c 100644
@@ -33,9 +33,11 @@
  #include <net/udp.h>
  #include <net/tc_act/tc_gact.h>
  #include <net/tc_act/tc_mirred.h>
+#include <net/devlink.h>
 
  #include "iavf_type.h"
  #include <linux/avf/virtchnl.h>
+#include "iavf_devlink.h"
  #include "iavf_txrx.h"
  #include "iavf_fdir.h"
  #include "iavf_adv_rss.h"

drivers/net/ethernet/intel/iavf/iavf.h struct iavf_adapter { · line 374

Quoted 4 lines
drivers/net/ethernet/intel/iavf/iavf.h +1 -0
@@ -369,6 +371,10 @@ struct iavf_adapter {
         struct net_device *netdev;
         struct pci_dev *pdev;
 
+       /* devlink & port data */

Nit, this comment doesn't add anything to the reader.

Quoted 15 lines
drivers/net/ethernet/intel/iavf/iavf.h +3 -0
+       struct devlink *devlink;
+       struct devlink_port devlink_port;
+
         struct iavf_hw hw; /* defined in iavf_type.h */
 
         enum iavf_state_t state;
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +3 -0
new file mode 100644
index 000000000000..991d041e5922
@@ -0,0 +1,93 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright (C) 2023 Intel Corporation */
+

drivers/net/ethernet/intel/iavf/iavf_devlink.c line 13

Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +10 -0
+#include "iavf.h"
+#include "iavf_devlink.h"
+
+static const struct devlink_ops iavf_devlink_ops = {};
+
+/**
+ * iavf_devlink_register - Register allocated devlink instance for iavf adapter
+ * @adapter: the iavf adapter to register the devlink for.
+ *
+ * Register the devlink instance associated with this iavf adapter

Nit, seems like this is a duplicate of what you above.

drivers/net/ethernet/intel/iavf/iavf_devlink.c line 23

Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +10 -0
+ *
+ * Return: zero on success or an error code on failure.
+ */
+int iavf_devlink_register(struct iavf_adapter *adapter)
+{
+       struct device *dev = &adapter->pdev->dev;
+       struct iavf_devlink *ref;
+       struct devlink *devlink;
+
+       /* Allocate devlink instance */

Nit, unnecessary comment.

drivers/net/ethernet/intel/iavf/iavf_devlink.c line 29

Quoted 6 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +6 -0
+       devlink = devlink_alloc(&iavf_devlink_ops, sizeof(struct iavf_devlink),
+                               dev);
+       if (!devlink)
+               return -ENOMEM;
+
+       /* Init iavf adapter devlink */

Nit, unnecessary comment.

Quoted 9 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +9 -0
+       adapter->devlink = devlink;
+       ref = devlink_priv(devlink);
+       ref->devlink_ref = adapter;
+
+       devlink_register(devlink);
+
+       return 0;
+}
+

drivers/net/ethernet/intel/iavf/iavf_devlink.c line 48

Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +10 -0
+/**
+ * iavf_devlink_unregister - Unregister devlink resources for iavf adapter.
+ * @adapter: the iavf adapter structure
+ *
+ * Releases resources used by devlink and cleans up associated memory.
+ */
+void iavf_devlink_unregister(struct iavf_adapter *adapter)
+{
+       devlink_unregister(adapter->devlink);
+       devlink_free(adapter->devlink);

Seems like you should bail out if (!adapter->devlink) since allocation of adapter->devlink can fail in iavf_devlink_register().

Quoted 8 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +8 -0
+}
+
+/**
+ * iavf_devlink_port_register - Register devlink port for iavf adapter
+ * @adapter: the iavf adapter to register the devlink port for.
+ *
+ * Register the devlink port instance associated with this iavf adapter
+ * before iavf adapter registers with netdevice

drivers/net/ethernet/intel/iavf/iavf_devlink.c line 66

Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +10 -0
+ *
+ * Return: zero on success or an error code on failure.
+ */
+int iavf_devlink_port_register(struct iavf_adapter *adapter)
+{
+       struct device *dev = &adapter->pdev->dev;
+       struct devlink_port_attrs attrs = {};
+       int err;
+
+       /* Create devlink port: attr/port flavour, port index */

Nit, unnecessary comment.

Quoted 33 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +27 -0
+       SET_NETDEV_DEVLINK_PORT(adapter->netdev, &adapter->devlink_port);
+       attrs.flavour = DEVLINK_PORT_FLAVOUR_VIRTUAL;
+       memset(&adapter->devlink_port, 0, sizeof(adapter->devlink_port));
+       devlink_port_attrs_set(&adapter->devlink_port, &attrs);
+
+       /* Register with driver specific index (device id) */
+       err = devlink_port_register(adapter->devlink, &adapter->devlink_port,
+                                   adapter->hw.bus.device);
+       if (err)
+               dev_err(dev, "devlink port registration failed: %d\n", err);
+
+       return err;
+}
+
+/**
+ * iavf_devlink_port_unregister - Unregister devlink port for iavf adapter.
+ * @adapter: the iavf adapter structure
+ *
+ * Releases resources used by devlink port and registration with devlink.
+ */
+void iavf_devlink_port_unregister(struct iavf_adapter *adapter)
+{
+       if (!adapter->devlink_port.registered)
+               return;
+
+       devlink_port_unregister(&adapter->devlink_port);
+}
drivers/net/ethernet/intel/iavf/iavf_devlink.h new file +0 -0
new file mode 100644
index 000000000000..5c122278611a

drivers/net/ethernet/intel/iavf/iavf_devlink.h line 7

Quoted 7 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.h new file +7 -0
@@ -0,0 +1,17 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright (C) 2023 Intel Corporation */
+
+#ifndef _IAVF_DEVLINK_H_
+#define _IAVF_DEVLINK_H_
+
+/* iavf devlink structure pointing to iavf adapter */

Nit, unnecessary comment.

Quoted 64 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.h new file +10 -0
+struct iavf_devlink {
+       struct iavf_adapter *devlink_ref;       /* ref to iavf adapter */
+};
+
+int iavf_devlink_register(struct iavf_adapter *adapter);
+void iavf_devlink_unregister(struct iavf_adapter *adapter);
+int iavf_devlink_port_register(struct iavf_adapter *adapter);
+void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
+
+#endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +14 -1
index 7b300c86ceda..db010e68d5d2 100644
@@ -2037,6 +2037,7 @@ static void iavf_finish_config(struct work_struct *work)
                                 iavf_free_rss(adapter);
                                 iavf_free_misc_irq(adapter);
                                 iavf_reset_interrupt_capability(adapter);
+                               iavf_devlink_port_unregister(adapter);
                                 iavf_change_state(adapter,
                                                   __IAVF_INIT_CONFIG_ADAPTER);
                                 goto out;
@@ -2708,6 +2709,9 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
         if (err)
                 goto err_sw_init;
 
+       if (!adapter->netdev_registered)
+               iavf_devlink_port_register(adapter);
+
         netif_carrier_off(netdev);
         adapter->link_up = false;
         netif_tx_stop_all_queues(netdev);
@@ -2749,6 +2753,7 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
  err_mem:
         iavf_free_rss(adapter);
         iavf_free_misc_irq(adapter);
+       iavf_devlink_port_unregister(adapter);
  err_sw_init:
         iavf_reset_interrupt_capability(adapter);
  err:
@@ -4995,6 +5000,12 @@ static int iavf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
         /* Setup the wait queue for indicating virtchannel events */
         init_waitqueue_head(&adapter->vc_waitqueue);
 
+       /* Register iavf adapter with devlink */
+       err = iavf_devlink_register(adapter);
+       if (err)
+               dev_err(&pdev->dev, "devlink registration failed: %d\n", err);
+
+       /* Keep driver interface even on devlink registration failure */
         return 0;
 
  err_ioremap:
@@ -5139,6 +5150,9 @@ static void iavf_remove(struct pci_dev *pdev)
                                  err);
         }
 
+       iavf_devlink_port_unregister(adapter);
+       iavf_devlink_unregister(adapter);
+
         mutex_lock(&adapter->crit_lock);
         dev_info(&adapter->pdev->dev, "Removing device\n");
         iavf_change_state(adapter, __IAVF_REMOVE);
--
Quoted 3 lines

2.34.1

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

author
#18 lore raw

PATCH v2 4/5 iavf: Add devlink port function rate API support

In reply to Wenjun Wu

From: Jun Zhang <[email protected]>

To allow user to configure queue based parameters, devlink port function rate api functions are added for setting node tx_max and tx_share parameters.

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

---
 .../net/ethernet/intel/iavf/iavf_devlink.c    | 270 +++++++++++++++++-
 .../net/ethernet/intel/iavf/iavf_devlink.h    |  21 ++
 drivers/net/ethernet/intel/iavf/iavf_main.c   |   7 +-
 3 files changed, 295 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/iavf/iavf_devlink.c +268 -2
index 991d041e5922..a2bd5295c216 100644
@@ -4,7 +4,273 @@
 #include "iavf.h"
 #include "iavf_devlink.h"
 
-static const struct devlink_ops iavf_devlink_ops = {};
+/**
+ * iavf_devlink_rate_init_rate_tree - export rate tree to devlink rate
+ * @adapter: iavf adapter struct instance
+ *
+ * This function builds Rate Tree based on iavf adapter configuration
+ * and exports it's contents to devlink rate.
+ */
+void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+	struct iavf_dev_rate_node *iavf_r_node;
+	struct iavf_dev_rate_node *iavf_q_node;
+	struct devlink_rate *dl_root_node;
+	struct devlink_rate *dl_tmp_node;
+	int q_num, size, i;
+
+	if (!adapter->devlink_port.registered)
+		return;
+
+	iavf_r_node = &dl_priv->root_node;
+	memset(iavf_r_node, 0, sizeof(*iavf_r_node));
+	iavf_r_node->tx_max = adapter->link_speed;
+	strscpy(iavf_r_node->name, "iavf_root", IAVF_RATE_NODE_NAME);
+
+	devl_lock(adapter->devlink);
+	dl_root_node = devl_rate_node_create(adapter->devlink, iavf_r_node,
+					     iavf_r_node->name, NULL);
+	if (!dl_root_node || IS_ERR(dl_root_node))
+		goto err_node;
+
+	iavf_r_node->rate_node = dl_root_node;
+
+	/* Allocate queue nodes, and chain them under root */
+	q_num = adapter->num_active_queues;
+	if (q_num > 0) {
+		size = q_num * sizeof(struct iavf_dev_rate_node);
+		dl_priv->queue_nodes = kzalloc(size, GFP_KERNEL);
+		if (!dl_priv->queue_nodes)
+			goto err_node;
+
+		memset(dl_priv->queue_nodes, 0, size);
+
+		for (i = 0; i < q_num; ++i) {
+			iavf_q_node = &dl_priv->queue_nodes[i];
+			snprintf(iavf_q_node->name, IAVF_RATE_NODE_NAME,
+				 "txq_%d", i);
+			dl_tmp_node = devl_rate_node_create(adapter->devlink,
+							    iavf_q_node,
+							    iavf_q_node->name,
+							    dl_root_node);
+			if (!dl_tmp_node || IS_ERR(dl_tmp_node)) {
+				kfree(dl_priv->queue_nodes);
+				goto err_node;
+			}
+
+			iavf_q_node->rate_node = dl_tmp_node;
+			iavf_q_node->tx_max = IAVF_TX_DEFAULT;
+			iavf_q_node->tx_share = 0;
+		}
+	}
+
+	dl_priv->update_in_progress = false;
+	dl_priv->iavf_dev_rate_initialized = true;
+	devl_unlock(adapter->devlink);
+	return;
+err_node:
+	devl_rate_nodes_destroy(adapter->devlink);
+	dl_priv->iavf_dev_rate_initialized = false;
+	devl_unlock(adapter->devlink);
+}
+
+/**
+ * iavf_devlink_rate_deinit_rate_tree - Unregister rate tree with devlink rate
+ * @adapter: iavf adapter struct instance
+ *
+ * This function unregisters the current iavf rate tree registered with devlink
+ * rate and frees resources.
+ */
+void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+
+	if (!dl_priv->iavf_dev_rate_initialized)
+		return;
+
+	devl_lock(adapter->devlink);
+	devl_rate_leaf_destroy(&adapter->devlink_port);
+	devl_rate_nodes_destroy(adapter->devlink);
+	kfree(dl_priv->queue_nodes);
+	devl_unlock(adapter->devlink);
+}
+
+/**
+ * iavf_check_update_config - check if updating queue parameters needed
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ *
+ * This function sets queue bw & quanta size configuration if all
+ * queue parameters are set
+ */
+static int iavf_check_update_config(struct iavf_adapter *adapter,
+				    struct iavf_dev_rate_node *node)
+{
+	/* Update queue bw if any one of the queues have been fully updated by
+	 * user, the other queues either use the default value or the last
+	 * fully updated value
+	 */
+	if (node->tx_update_flag ==
+	    (IAVF_FLAG_TX_MAX_UPDATED | IAVF_FLAG_TX_SHARE_UPDATED)) {
+		node->tx_max = node->tx_max_temp;
+		node->tx_share = node->tx_share_temp;
+	} else {
+		return 0;
+	}
+
+	/* Reconfig queue bw only when iavf driver on running state */
+	if (adapter->state != __IAVF_RUNNING)
+		return -EBUSY;
+
+	return 0;
+}
+
+/**
+ * iavf_update_queue_tx_share - sets tx min parameter
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ * @bw: bandwidth in bytes per second
+ * @extack: extended netdev ack structure
+ *
+ * This function sets min BW limit.
+ */
+static int iavf_update_queue_tx_share(struct iavf_adapter *adapter,
+				      struct iavf_dev_rate_node *node,
+				      u64 bw, struct netlink_ext_ack *extack)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+	u64 tx_share_sum = 0;
+
+	/* Keep in kbps */
+	node->tx_share_temp = div_u64(bw, IAVF_RATE_DIV_FACTOR);
+
+	if (ADV_LINK_SUPPORT(adapter)) {
+		int i;
+
+		for (i = 0; i < adapter->num_active_queues; ++i) {
+			if (node != &dl_priv->queue_nodes[i])
+				tx_share_sum +=
+					dl_priv->queue_nodes[i].tx_share;
+			else
+				tx_share_sum += node->tx_share_temp;
+		}
+
+		if (tx_share_sum / 1000  > adapter->link_speed_mbps)
+			return -EINVAL;
+	}
+
+	node->tx_update_flag |= IAVF_FLAG_TX_SHARE_UPDATED;
+	return iavf_check_update_config(adapter, node);
+}
+
+/**
+ * iavf_update_queue_tx_max - sets tx max parameter
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ * @bw: bandwidth in bytes per second
+ * @extack: extended netdev ack structure
+ *
+ * This function sets max BW limit.
+ */
+static int iavf_update_queue_tx_max(struct iavf_adapter *adapter,
+				    struct iavf_dev_rate_node *node,
+				    u64 bw, struct netlink_ext_ack *extack)
+{
+	/* Keep in kbps */
+	node->tx_max_temp = div_u64(bw, IAVF_RATE_DIV_FACTOR);
+	if (ADV_LINK_SUPPORT(adapter)) {
+		if (node->tx_max_temp / 1000 > adapter->link_speed_mbps)
+			return -EINVAL;
+	}
+
+	node->tx_update_flag |= IAVF_FLAG_TX_MAX_UPDATED;
+
+	return iavf_check_update_config(adapter, node);
+}
+
+/**
+ * iavf_devlink_rate_node_tx_max_set - devlink_rate API for setting tx max
+ * @rate_node: devlink rate struct instance
+ *
+ * This function implements rate_node_tx_max_set function of devlink_ops
+ */
+static int iavf_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node,
+					     void *priv, u64 tx_max,
+					     struct netlink_ext_ack *extack)
+{
+	struct iavf_dev_rate_node *node = priv;
+	struct iavf_devlink *dl_priv;
+	struct iavf_adapter *adapter;
+
+	if (!node)
+		return 0;
+
+	dl_priv = devlink_priv(rate_node->devlink);
+	adapter = dl_priv->devlink_ref;
+
+	/* Check if last update is in progress */
+	if (dl_priv->update_in_progress)
+		return -EBUSY;
+
+	if (node == &dl_priv->root_node)
+		return 0;
+
+	return iavf_update_queue_tx_max(adapter, node, tx_max, extack);
+}
+
+/**
+ * iavf_devlink_rate_node_tx_share_set - devlink_rate API for setting tx share
+ * @rate_node: devlink rate struct instance
+ *
+ * This function implements rate_node_tx_share_set function of devlink_ops
+ */
+static int iavf_devlink_rate_node_tx_share_set(struct devlink_rate *rate_node,
+					       void *priv, u64 tx_share,
+					       struct netlink_ext_ack *extack)
+{
+	struct iavf_dev_rate_node *node = priv;
+	struct iavf_devlink *dl_priv;
+	struct iavf_adapter *adapter;
+
+	if (!node)
+		return 0;
+
+	dl_priv = devlink_priv(rate_node->devlink);
+	adapter = dl_priv->devlink_ref;
+
+	/* Check if last update is in progress */
+	if (dl_priv->update_in_progress)
+		return -EBUSY;
+
+	if (node == &dl_priv->root_node)
+		return 0;
+
+	return iavf_update_queue_tx_share(adapter, node, tx_share, extack);
+}
+
+static int iavf_devlink_rate_node_del(struct devlink_rate *rate_node,
+				      void *priv,
+				      struct netlink_ext_ack *extack)
+{
+	return -EINVAL;
+}
+
+static int iavf_devlink_set_parent(struct devlink_rate *devlink_rate,
+				   struct devlink_rate *parent,
+				   void *priv, void *parent_priv,
+				   struct netlink_ext_ack *extack)
+{
+	return -EINVAL;
+}
+
+static const struct devlink_ops iavf_devlink_ops = {
+	.rate_node_tx_share_set = iavf_devlink_rate_node_tx_share_set,
+	.rate_node_tx_max_set = iavf_devlink_rate_node_tx_max_set,
+	.rate_node_del = iavf_devlink_rate_node_del,
+	.rate_leaf_parent_set = iavf_devlink_set_parent,
+	.rate_node_parent_set = iavf_devlink_set_parent,
+};
 
 /**
  * iavf_devlink_register - Register allocated devlink instance for iavf adapter
@@ -30,7 +296,7 @@ int iavf_devlink_register(struct iavf_adapter *adapter)
 	adapter->devlink = devlink;
 	ref = devlink_priv(devlink);
 	ref->devlink_ref = adapter;
-
+	ref->iavf_dev_rate_initialized = false;
 	devlink_register(devlink);
 
 	return 0;
drivers/net/ethernet/intel/iavf/iavf_devlink.h +21 -0
index 5c122278611a..897ff5fc87af 100644
@@ -4,14 +4,35 @@
 #ifndef _IAVF_DEVLINK_H_
 #define _IAVF_DEVLINK_H_
 
+#define IAVF_RATE_NODE_NAME			12
+struct iavf_dev_rate_node {
+	char name[IAVF_RATE_NODE_NAME];
+	struct devlink_rate *rate_node;
+	u8 tx_update_flag;
+#define IAVF_FLAG_TX_SHARE_UPDATED		BIT(0)
+#define IAVF_FLAG_TX_MAX_UPDATED		BIT(1)
+	u64 tx_max;
+	u64 tx_share;
+	u64 tx_max_temp;
+	u64 tx_share_temp;
+#define IAVF_RATE_DIV_FACTOR			125
+#define IAVF_TX_DEFAULT				100000
+};
+
 /* iavf devlink structure pointing to iavf adapter */
 struct iavf_devlink {
 	struct iavf_adapter *devlink_ref;	/* ref to iavf adapter */
+	struct iavf_dev_rate_node root_node;
+	struct iavf_dev_rate_node *queue_nodes;
+	bool iavf_dev_rate_initialized;
+	bool update_in_progress;
 };
 
 int iavf_devlink_register(struct iavf_adapter *adapter);
 void iavf_devlink_unregister(struct iavf_adapter *adapter);
 int iavf_devlink_port_register(struct iavf_adapter *adapter);
 void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
+void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter);
+void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter);
 
 #endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +6 -1
index db010e68d5d2..7348b65f9f19 100644
@@ -2037,6 +2037,7 @@ static void iavf_finish_config(struct work_struct *work)
 				iavf_free_rss(adapter);
 				iavf_free_misc_irq(adapter);
 				iavf_reset_interrupt_capability(adapter);
+				iavf_devlink_rate_deinit_rate_tree(adapter);
 				iavf_devlink_port_unregister(adapter);
 				iavf_change_state(adapter,
 						  __IAVF_INIT_CONFIG_ADAPTER);
@@ -2709,8 +2710,10 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 	if (err)
 		goto err_sw_init;
 
-	if (!adapter->netdev_registered)
+	if (!adapter->netdev_registered) {
 		iavf_devlink_port_register(adapter);
+		iavf_devlink_rate_init_rate_tree(adapter);
+	}
 
 	netif_carrier_off(netdev);
 	adapter->link_up = false;
@@ -2753,6 +2756,7 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 err_mem:
 	iavf_free_rss(adapter);
 	iavf_free_misc_irq(adapter);
+	iavf_devlink_rate_deinit_rate_tree(adapter);
 	iavf_devlink_port_unregister(adapter);
 err_sw_init:
 	iavf_reset_interrupt_capability(adapter);
@@ -5150,6 +5154,7 @@ static void iavf_remove(struct pci_dev *pdev)
 				 err);
 	}
 
+	iavf_devlink_rate_deinit_rate_tree(adapter);
 	iavf_devlink_port_unregister(adapter);
 	iavf_devlink_unregister(adapter);
 
Signature · 1 line
2.34.1
reviewer
#19 lore raw

Re: [PATCH iwl-next v2 4/5] iavf: Add devlink port function rate API support

Quoted 10 lines · Wenjun Wu

From: Jun Zhang <[email protected]>

To allow user to configure queue based parameters, devlink port function rate api functions are added for setting node tx_max and tx_share parameters.

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

...

Quoted 18 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c +18 -0
+/**
+ * iavf_update_queue_tx_max - sets tx max parameter
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ * @bw: bandwidth in bytes per second
+ * @extack: extended netdev ack structure
+ *
+ * This function sets max BW limit.
+ */
+static int iavf_update_queue_tx_max(struct iavf_adapter *adapter,
+				    struct iavf_dev_rate_node *node,
+				    u64 bw, struct netlink_ext_ack *extack)
+{
+	/* Keep in kbps */
+	node->tx_max_temp = div_u64(bw, IAVF_RATE_DIV_FACTOR);
+	if (ADV_LINK_SUPPORT(adapter)) {
+		if (node->tx_max_temp / 1000 > adapter->link_speed_mbps)
+			return -EINVAL;

drivers/net/ethernet/intel/iavf/iavf_devlink.c line 194

Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c +10 -0
+	}
+
+	node->tx_update_flag |= IAVF_FLAG_TX_MAX_UPDATED;
+
+	return iavf_check_update_config(adapter, node);
+}
+
+/**
+ * iavf_devlink_rate_node_tx_max_set - devlink_rate API for setting tx max
+ * @rate_node: devlink rate struct instance

Hi Jun Zhang,

Please describe all the parameters of iavf_devlink_rate_node_tx_max_set in it's kernel doc.

./scripts/kernel-doc -none is your friend here.

Quoted 6 lines

+ *
+ * This function implements rate_node_tx_max_set function of devlink_ops
+ */

drivers/net/ethernet/intel/iavf/iavf_devlink.c +3 -0
+static int iavf_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node,
+					     void *priv, u64 tx_max,
+					     struct netlink_ext_ack *extack)

...

Signature · 1 line
pw-bot: changes-requested
#20 lore raw

Re: [PATCH iwl-next v2 4/5] iavf: Add devlink port function rate API support

Quoted 43 lines · Simon Horman
Quoted 10 lines · Wenjun Wu

From: Jun Zhang <[email protected]>

To allow user to configure queue based parameters, devlink port function rate api functions are added for setting node tx_max and tx_share parameters.

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

...

Quoted 18 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c +18 -0
+/**
+ * iavf_update_queue_tx_max - sets tx max parameter
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ * @bw: bandwidth in bytes per second
+ * @extack: extended netdev ack structure
+ *
+ * This function sets max BW limit.
+ */
+static int iavf_update_queue_tx_max(struct iavf_adapter *adapter,
+				    struct iavf_dev_rate_node *node,
+				    u64 bw, struct netlink_ext_ack *extack)
+{
+	/* Keep in kbps */
+	node->tx_max_temp = div_u64(bw, IAVF_RATE_DIV_FACTOR);
+	if (ADV_LINK_SUPPORT(adapter)) {
+		if (node->tx_max_temp / 1000 > adapter->link_speed_mbps)
+			return -EINVAL;

drivers/net/ethernet/intel/iavf/iavf_devlink.c line 194

Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c +10 -0
+	}
+
+	node->tx_update_flag |= IAVF_FLAG_TX_MAX_UPDATED;
+
+	return iavf_check_update_config(adapter, node);
+}
+
+/**
+ * iavf_devlink_rate_node_tx_max_set - devlink_rate API for setting tx max
+ * @rate_node: devlink rate struct instance

Hi Jun Zhang,

Quoted 4 lines

Please describe all the parameters of iavf_devlink_rate_node_tx_max_set in it's kernel doc.

./scripts/kernel-doc -none is your friend here.

Thanks.

As this function is an implementation of Kernel API. To be in sync w/ kernel api definition documentation, will omit the function description here.

Quoted 8 lines
Quoted 6 lines

+ *
+ * This function implements rate_node_tx_max_set function of devlink_ops
+ */

drivers/net/ethernet/intel/iavf/iavf_devlink.c +3 -0
+static int iavf_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node,
+					     void *priv, u64 tx_max,
+					     struct netlink_ext_ack *extack)

...

#21 lore raw

Re: [PATCH iwl-next v2 4/5] iavf: Add devlink port function rate API support

In reply to Wenjun Wu
Quoted 59 lines · Wenjun Wu

Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.

From: Jun Zhang <[email protected]>

To allow user to configure queue based parameters, devlink port function rate api functions are added for setting node tx_max and tx_share parameters.

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

---
  .../net/ethernet/intel/iavf/iavf_devlink.c    | 270 +++++++++++++++++-
  .../net/ethernet/intel/iavf/iavf_devlink.h    |  21 ++
  drivers/net/ethernet/intel/iavf/iavf_main.c   |   7 +-
  3 files changed, 295 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/iavf/iavf_devlink.c +31 -1
index 991d041e5922..a2bd5295c216 100644
@@ -4,7 +4,273 @@
  #include "iavf.h"
  #include "iavf_devlink.h"
 
-static const struct devlink_ops iavf_devlink_ops = {};
+/**
+ * iavf_devlink_rate_init_rate_tree - export rate tree to devlink rate
+ * @adapter: iavf adapter struct instance
+ *
+ * This function builds Rate Tree based on iavf adapter configuration
+ * and exports it's contents to devlink rate.
+ */
+void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter)
+{
+       struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+       struct iavf_dev_rate_node *iavf_r_node;
+       struct iavf_dev_rate_node *iavf_q_node;
+       struct devlink_rate *dl_root_node;
+       struct devlink_rate *dl_tmp_node;
+       int q_num, size, i;
+
+       if (!adapter->devlink_port.registered)
+               return;
+
+       iavf_r_node = &dl_priv->root_node;
+       memset(iavf_r_node, 0, sizeof(*iavf_r_node));
+       iavf_r_node->tx_max = adapter->link_speed;
+       strscpy(iavf_r_node->name, "iavf_root", IAVF_RATE_NODE_NAME);
+
+       devl_lock(adapter->devlink);
+       dl_root_node = devl_rate_node_create(adapter->devlink, iavf_r_node,
+                                            iavf_r_node->name, NULL);
+       if (!dl_root_node || IS_ERR(dl_root_node))
+               goto err_node;
+
+       iavf_r_node->rate_node = dl_root_node;

drivers/net/ethernet/intel/iavf/iavf_devlink.c line 47

Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c +10 -0
+
+       /* Allocate queue nodes, and chain them under root */
+       q_num = adapter->num_active_queues;
+       if (q_num > 0) {
+               size = q_num * sizeof(struct iavf_dev_rate_node);
+               dl_priv->queue_nodes = kzalloc(size, GFP_KERNEL);
+               if (!dl_priv->queue_nodes)
+                       goto err_node;
+
+               memset(dl_priv->queue_nodes, 0, size);

Why not just use kcalloc() here? Also, it seems like there's no need to zero the memory here.

Quoted 62 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c +62 -0
+
+               for (i = 0; i < q_num; ++i) {
+                       iavf_q_node = &dl_priv->queue_nodes[i];
+                       snprintf(iavf_q_node->name, IAVF_RATE_NODE_NAME,
+                                "txq_%d", i);
+                       dl_tmp_node = devl_rate_node_create(adapter->devlink,
+                                                           iavf_q_node,
+                                                           iavf_q_node->name,
+                                                           dl_root_node);
+                       if (!dl_tmp_node || IS_ERR(dl_tmp_node)) {
+                               kfree(dl_priv->queue_nodes); > +                               goto err_node;
+                       }
+
+                       iavf_q_node->rate_node = dl_tmp_node;
+                       iavf_q_node->tx_max = IAVF_TX_DEFAULT;
+                       iavf_q_node->tx_share = 0;
+               }
+       }
+
+       dl_priv->update_in_progress = false;
+       dl_priv->iavf_dev_rate_initialized = true;
+       devl_unlock(adapter->devlink);
+       return;
+err_node:
+       devl_rate_nodes_destroy(adapter->devlink);
+       dl_priv->iavf_dev_rate_initialized = false;
+       devl_unlock(adapter->devlink);
+}
+
+/**
+ * iavf_devlink_rate_deinit_rate_tree - Unregister rate tree with devlink rate
+ * @adapter: iavf adapter struct instance
+ *
+ * This function unregisters the current iavf rate tree registered with devlink
+ * rate and frees resources.
+ */
+void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
+{
+       struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+
+       if (!dl_priv->iavf_dev_rate_initialized)
+               return;
+
+       devl_lock(adapter->devlink);
+       devl_rate_leaf_destroy(&adapter->devlink_port);
+       devl_rate_nodes_destroy(adapter->devlink);
+       kfree(dl_priv->queue_nodes);
+       devl_unlock(adapter->devlink);
+}
+
+/**
+ * iavf_check_update_config - check if updating queue parameters needed
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ *
+ * This function sets queue bw & quanta size configuration if all
+ * queue parameters are set
+ */
+static int iavf_check_update_config(struct iavf_adapter *adapter,
+                                   struct iavf_dev_rate_node *node)
+{
+       /* Update queue bw if any one of the queues have been fully updated by

drivers/net/ethernet/intel/iavf/iavf_devlink.c line 120

Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c +10 -0
+        * user, the other queues either use the default value or the last
+        * fully updated value
+        */
+       if (node->tx_update_flag ==
+           (IAVF_FLAG_TX_MAX_UPDATED | IAVF_FLAG_TX_SHARE_UPDATED)) {
+               node->tx_max = node->tx_max_temp;
+               node->tx_share = node->tx_share_temp;
+       } else {
+               return 0;
+       }

I think it would more readable to do the following:

if (node->tx_update_flag !=
(IAVE_FLAG_TX_MAX_UPDATED | IAVF_FLAG_TX_SHARE_UPDATED))
return 0;

/* rest of function */

Quoted 185 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.c +154 -1
+
+       /* Reconfig queue bw only when iavf driver on running state */
+       if (adapter->state != __IAVF_RUNNING)
+               return -EBUSY;
+
+       return 0;
+}
+
+/**
+ * iavf_update_queue_tx_share - sets tx min parameter
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ * @bw: bandwidth in bytes per second
+ * @extack: extended netdev ack structure
+ *
+ * This function sets min BW limit.
+ */
+static int iavf_update_queue_tx_share(struct iavf_adapter *adapter,
+                                     struct iavf_dev_rate_node *node,
+                                     u64 bw, struct netlink_ext_ack *extack)
+{
+       struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+       u64 tx_share_sum = 0;
+
+       /* Keep in kbps */
+       node->tx_share_temp = div_u64(bw, IAVF_RATE_DIV_FACTOR);
+
+       if (ADV_LINK_SUPPORT(adapter)) {
+               int i;
+
+               for (i = 0; i < adapter->num_active_queues; ++i) {
+                       if (node != &dl_priv->queue_nodes[i])
+                               tx_share_sum +=
+                                       dl_priv->queue_nodes[i].tx_share;
+                       else
+                               tx_share_sum += node->tx_share_temp;
+               }
+
+               if (tx_share_sum / 1000  > adapter->link_speed_mbps)
+                       return -EINVAL;
+       }
+
+       node->tx_update_flag |= IAVF_FLAG_TX_SHARE_UPDATED;
+       return iavf_check_update_config(adapter, node);
+}
+
+/**
+ * iavf_update_queue_tx_max - sets tx max parameter
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ * @bw: bandwidth in bytes per second
+ * @extack: extended netdev ack structure
+ *
+ * This function sets max BW limit.
+ */
+static int iavf_update_queue_tx_max(struct iavf_adapter *adapter,
+                                   struct iavf_dev_rate_node *node,
+                                   u64 bw, struct netlink_ext_ack *extack)
+{
+       /* Keep in kbps */
+       node->tx_max_temp = div_u64(bw, IAVF_RATE_DIV_FACTOR);
+       if (ADV_LINK_SUPPORT(adapter)) {
+               if (node->tx_max_temp / 1000 > adapter->link_speed_mbps)
+                       return -EINVAL;
+       }
+
+       node->tx_update_flag |= IAVF_FLAG_TX_MAX_UPDATED;
+
+       return iavf_check_update_config(adapter, node);
+}
+
+/**
+ * iavf_devlink_rate_node_tx_max_set - devlink_rate API for setting tx max
+ * @rate_node: devlink rate struct instance
+ *
+ * This function implements rate_node_tx_max_set function of devlink_ops
+ */
+static int iavf_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node,
+                                            void *priv, u64 tx_max,
+                                            struct netlink_ext_ack *extack)
+{
+       struct iavf_dev_rate_node *node = priv;
+       struct iavf_devlink *dl_priv;
+       struct iavf_adapter *adapter;
+
+       if (!node)
+               return 0;
+
+       dl_priv = devlink_priv(rate_node->devlink);
+       adapter = dl_priv->devlink_ref;
+
+       /* Check if last update is in progress */
+       if (dl_priv->update_in_progress)
+               return -EBUSY;
+
+       if (node == &dl_priv->root_node)
+               return 0;
+
+       return iavf_update_queue_tx_max(adapter, node, tx_max, extack);
+}
+
+/**
+ * iavf_devlink_rate_node_tx_share_set - devlink_rate API for setting tx share
+ * @rate_node: devlink rate struct instance
+ *
+ * This function implements rate_node_tx_share_set function of devlink_ops
+ */
+static int iavf_devlink_rate_node_tx_share_set(struct devlink_rate *rate_node,
+                                              void *priv, u64 tx_share,
+                                              struct netlink_ext_ack *extack)
+{
+       struct iavf_dev_rate_node *node = priv;
+       struct iavf_devlink *dl_priv;
+       struct iavf_adapter *adapter;
+
+       if (!node)
+               return 0;
+
+       dl_priv = devlink_priv(rate_node->devlink);
+       adapter = dl_priv->devlink_ref;
+
+       /* Check if last update is in progress */
+       if (dl_priv->update_in_progress)
+               return -EBUSY;
+
+       if (node == &dl_priv->root_node)
+               return 0;
+
+       return iavf_update_queue_tx_share(adapter, node, tx_share, extack);
+}
+
+static int iavf_devlink_rate_node_del(struct devlink_rate *rate_node,
+                                     void *priv,
+                                     struct netlink_ext_ack *extack)
+{
+       return -EINVAL;
+}
+
+static int iavf_devlink_set_parent(struct devlink_rate *devlink_rate,
+                                  struct devlink_rate *parent,
+                                  void *priv, void *parent_priv,
+                                  struct netlink_ext_ack *extack)
+{
+       return -EINVAL;
+}
+
+static const struct devlink_ops iavf_devlink_ops = {
+       .rate_node_tx_share_set = iavf_devlink_rate_node_tx_share_set,
+       .rate_node_tx_max_set = iavf_devlink_rate_node_tx_max_set,
+       .rate_node_del = iavf_devlink_rate_node_del,
+       .rate_leaf_parent_set = iavf_devlink_set_parent,
+       .rate_node_parent_set = iavf_devlink_set_parent,
+};
 
  /**
   * iavf_devlink_register - Register allocated devlink instance for iavf adapter
@@ -30,7 +296,7 @@ int iavf_devlink_register(struct iavf_adapter *adapter)
         adapter->devlink = devlink;
         ref = devlink_priv(devlink);
         ref->devlink_ref = adapter;
-
+       ref->iavf_dev_rate_initialized = false;
         devlink_register(devlink);
 
         return 0;
drivers/net/ethernet/intel/iavf/iavf_devlink.h +12 -0
index 5c122278611a..897ff5fc87af 100644
@@ -4,14 +4,35 @@
  #ifndef _IAVF_DEVLINK_H_
  #define _IAVF_DEVLINK_H_
 
+#define IAVF_RATE_NODE_NAME                    12
+struct iavf_dev_rate_node {
+       char name[IAVF_RATE_NODE_NAME];
+       struct devlink_rate *rate_node;
+       u8 tx_update_flag;
+#define IAVF_FLAG_TX_SHARE_UPDATED             BIT(0)
+#define IAVF_FLAG_TX_MAX_UPDATED               BIT(1)
+       u64 tx_max;
+       u64 tx_share;
+       u64 tx_max_temp;
+       u64 tx_share_temp;
+#define IAVF_RATE_DIV_FACTOR                   125

drivers/net/ethernet/intel/iavf/iavf_devlink.h line 28

Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.h +7 -0
+#define IAVF_TX_DEFAULT                                100000
+};
+
  /* iavf devlink structure pointing to iavf adapter */
  struct iavf_devlink {
         struct iavf_adapter *devlink_ref;       /* ref to iavf adapter */
+       struct iavf_dev_rate_node root_node;
+       struct iavf_dev_rate_node *queue_nodes;
+       bool iavf_dev_rate_initialized;
+       bool update_in_progress;

It seems like this is never true until patch 5/5, so IMO it makes sense to add it there. Also, why is this bool needed? Is there not another flag or lock that can be used instead of adding this new flag/bit?

Quoted 54 lines
drivers/net/ethernet/intel/iavf/iavf_devlink.h +2 -0
  };
 
  int iavf_devlink_register(struct iavf_adapter *adapter);
  void iavf_devlink_unregister(struct iavf_adapter *adapter);
  int iavf_devlink_port_register(struct iavf_adapter *adapter);
  void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
+void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter);
+void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter);
 
  #endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +6 -2
index db010e68d5d2..7348b65f9f19 100644
@@ -2037,6 +2037,7 @@ static void iavf_finish_config(struct work_struct *work)
                                 iavf_free_rss(adapter);
                                 iavf_free_misc_irq(adapter);
                                 iavf_reset_interrupt_capability(adapter);
+                               iavf_devlink_rate_deinit_rate_tree(adapter);
                                 iavf_devlink_port_unregister(adapter);
                                 iavf_change_state(adapter,
                                                   __IAVF_INIT_CONFIG_ADAPTER);
@@ -2709,8 +2710,10 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
         if (err)
                 goto err_sw_init;
 
-       if (!adapter->netdev_registered)
+       if (!adapter->netdev_registered) {
                 iavf_devlink_port_register(adapter);
+               iavf_devlink_rate_init_rate_tree(adapter);
+       }
 
         netif_carrier_off(netdev);
         adapter->link_up = false;
@@ -2753,6 +2756,7 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
  err_mem:
         iavf_free_rss(adapter);
         iavf_free_misc_irq(adapter);
+       iavf_devlink_rate_deinit_rate_tree(adapter);
         iavf_devlink_port_unregister(adapter);
  err_sw_init:
         iavf_reset_interrupt_capability(adapter);
@@ -5150,6 +5154,7 @@ static void iavf_remove(struct pci_dev *pdev)
                                  err);
         }
 
+       iavf_devlink_rate_deinit_rate_tree(adapter);
         iavf_devlink_port_unregister(adapter);
         iavf_devlink_unregister(adapter);
 
--

2.34.1

author
#22 lore raw

PATCH v2 5/5 iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

In reply to Wenjun Wu

From: Jun Zhang <[email protected]>

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

User can configure the tx_max and tx_share of each queue. If any one of the queues have been fully updated by user, i.e. both tx_max and tx_share have been updated for that queue, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

---
 drivers/net/ethernet/intel/iavf/iavf.h        |  14 ++
 .../net/ethernet/intel/iavf/iavf_devlink.c    |  29 +++
 .../net/ethernet/intel/iavf/iavf_devlink.h    |   1 +
 drivers/net/ethernet/intel/iavf/iavf_main.c   |  45 +++-
 .../net/ethernet/intel/iavf/iavf_virtchnl.c   | 228 +++++++++++++++++-
 5 files changed, 313 insertions(+), 4 deletions(-)
drivers/net/ethernet/intel/iavf/iavf.h +14 -0
index 519aeaec793c..e9b781cacffa 100644
@@ -252,6 +252,9 @@ struct iavf_cloud_filter {
 #define IAVF_RESET_WAIT_DETECTED_COUNT 500
 #define IAVF_RESET_WAIT_COMPLETE_COUNT 2000
 
+#define IAVF_MAX_QOS_TC_NUM		8
+#define IAVF_DEFAULT_QUANTA_SIZE	1024
+
 /* board specific private data structure */
 struct iavf_adapter {
 	struct workqueue_struct *wq;
@@ -351,6 +354,9 @@ struct iavf_adapter {
 #define IAVF_FLAG_AQ_DISABLE_CTAG_VLAN_INSERTION	BIT_ULL(36)
 #define IAVF_FLAG_AQ_ENABLE_STAG_VLAN_INSERTION		BIT_ULL(37)
 #define IAVF_FLAG_AQ_DISABLE_STAG_VLAN_INSERTION	BIT_ULL(38)
+#define IAVF_FLAG_AQ_CONFIGURE_QUEUES_BW		BIT_ULL(39)
+#define IAVF_FLAG_AQ_CONFIGURE_QUEUES_QUANTA_SIZE	BIT_ULL(40)
+#define IAVF_FLAG_AQ_GET_QOS_CAPS			BIT_ULL(41)
 
 	/* flags for processing extended capability messages during
 	 * __IAVF_INIT_EXTENDED_CAPS. Each capability exchange requires
@@ -374,6 +380,7 @@ struct iavf_adapter {
 	/* devlink & port data */
 	struct devlink *devlink;
 	struct devlink_port devlink_port;
+	bool devlink_update;
 
 	struct iavf_hw hw; /* defined in iavf_type.h */
 
@@ -423,6 +430,8 @@ struct iavf_adapter {
 			       VIRTCHNL_VF_OFFLOAD_FDIR_PF)
 #define ADV_RSS_SUPPORT(_a) ((_a)->vf_res->vf_cap_flags & \
 			     VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF)
+#define QOS_ALLOWED(_a) ((_a)->vf_res->vf_cap_flags & \
+			 VIRTCHNL_VF_OFFLOAD_QOS)
 	struct virtchnl_vf_resource *vf_res; /* incl. all VSIs */
 	struct virtchnl_vsi_resource *vsi_res; /* our LAN VSI */
 	struct virtchnl_version_info pf_version;
@@ -431,6 +440,7 @@ struct iavf_adapter {
 	struct virtchnl_vlan_caps vlan_v2_caps;
 	u16 msg_enable;
 	struct iavf_eth_stats current_stats;
+	struct virtchnl_qos_cap_list *qos_caps;
 	struct iavf_vsi vsi;
 	u32 aq_wait_count;
 	/* RSS stuff */
@@ -577,6 +587,10 @@ void iavf_notify_client_message(struct iavf_vsi *vsi, u8 *msg, u16 len);
 void iavf_notify_client_l2_params(struct iavf_vsi *vsi);
 void iavf_notify_client_open(struct iavf_vsi *vsi);
 void iavf_notify_client_close(struct iavf_vsi *vsi, bool reset);
+void iavf_update_queue_config(struct iavf_adapter *adapter);
+void iavf_configure_queues_bw(struct iavf_adapter *adapter);
+void iavf_configure_queues_quanta_size(struct iavf_adapter *adapter);
+void iavf_get_qos_caps(struct iavf_adapter *adapter);
 void iavf_enable_channels(struct iavf_adapter *adapter);
 void iavf_disable_channels(struct iavf_adapter *adapter);
 void iavf_add_cloud_filter(struct iavf_adapter *adapter);
drivers/net/ethernet/intel/iavf/iavf_devlink.c +29 -0
index a2bd5295c216..dbe88eb538a8 100644
@@ -96,6 +96,30 @@ void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
 	devl_unlock(adapter->devlink);
 }
 
+/**
+ * iavf_notify_queue_config_complete - notify updating queue completion
+ * @adapter: iavf adapter struct instance
+ *
+ * This function sets the queue configuration update status when all
+ * queue parameters have been sent to PF
+ */
+void iavf_notify_queue_config_complete(struct iavf_adapter *adapter)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+	int q_num = adapter->num_active_queues;
+	int i;
+
+	/* clean up rate tree update flags*/
+	for (i = 0; i < q_num; i++)
+		if (dl_priv->queue_nodes[i].tx_update_flag ==
+		    (IAVF_FLAG_TX_MAX_UPDATED | IAVF_FLAG_TX_SHARE_UPDATED)) {
+			dl_priv->queue_nodes[i].tx_update_flag = 0;
+			break;
+		}
+
+	dl_priv->update_in_progress = false;
+}
+
 /**
  * iavf_check_update_config - check if updating queue parameters needed
  * @adapter: iavf adapter struct instance
@@ -107,6 +131,8 @@ void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
 static int iavf_check_update_config(struct iavf_adapter *adapter,
 				    struct iavf_dev_rate_node *node)
 {
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+
 	/* Update queue bw if any one of the queues have been fully updated by
 	 * user, the other queues either use the default value or the last
 	 * fully updated value
@@ -123,6 +149,8 @@ static int iavf_check_update_config(struct iavf_adapter *adapter,
 	if (adapter->state != __IAVF_RUNNING)
 		return -EBUSY;
 
+	dl_priv->update_in_progress = true;
+	iavf_update_queue_config(adapter);
 	return 0;
 }
 
@@ -294,6 +322,7 @@ int iavf_devlink_register(struct iavf_adapter *adapter)
 
 	/* Init iavf adapter devlink */
 	adapter->devlink = devlink;
+	adapter->devlink_update = false;
 	ref = devlink_priv(devlink);
 	ref->devlink_ref = adapter;
 	ref->iavf_dev_rate_initialized = false;
drivers/net/ethernet/intel/iavf/iavf_devlink.h +1 -0
index 897ff5fc87af..a8a41f343f56 100644
@@ -34,5 +34,6 @@ int iavf_devlink_port_register(struct iavf_adapter *adapter);
 void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
 void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter);
 void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter);
+void iavf_notify_queue_config_complete(struct iavf_adapter *adapter);
 
 #endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +42 -3

84 lines not on this page · raw message

drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +227 -1

252 lines not on this page · raw message

Signature · 1 line
2.34.1
reviewer
#23 lore raw

Re: [PATCH iwl-next v2 5/5] iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

Quoted 1 line · Wenjun Wu

From: Jun Zhang <[email protected]>

...

drivers/net/ethernet/intel/iavf/iavf_virtchnl.c void iavf_virtchnl_completion(struct iavf_adapter *adapter, · line 2692

Quoted 7 lines
drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +3 -0
@@ -2471,6 +2687,16 @@ void iavf_virtchnl_completion(struct iavf_adapter *adapter,
 		if (!v_retval)
 			iavf_netdev_features_vlan_strip_set(netdev, false);
 		break;
+	case VIRTCHNL_OP_GET_QOS_CAPS:
+		u16 len = struct_size(adapter->qos_caps, cap,
+				      IAVF_MAX_QOS_TC_NUM);

Hi Jun Zhang and Wenju Wu,

clang-16 complains about this quite a lot.
I think it is because it wants the declaration of len - and thus the rest of this case - inside a block ({}).

.../iavf_virtchnl.c:2691:3: error: expected expression

                 u16 len = struct_size(adapter->qos_caps, cap,
                 ^
 .../iavf_virtchnl.c:2693:46: error: use of undeclared identifier 'len'
                 memcpy(adapter->qos_caps, msg, min(msglen, len));
                                                            ^

.../iavf_virtchnl.c:2693:46: error: use of undeclared identifier 'len'

Quoted 14 lines
drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +7 -0
+		memcpy(adapter->qos_caps, msg, min(msglen, len));
+		break;
+	case VIRTCHNL_OP_CONFIG_QUANTA:
+		iavf_notify_queue_config_complete(adapter);
+		break;
+	case VIRTCHNL_OP_CONFIG_QUEUE_BW:
+		break;
 	default:
 		if (adapter->current_op && (v_opcode != adapter->current_op))
 			dev_warn(&adapter->pdev->dev, "Expected response %d from PF, received %d\n",
Signature · 1 line
2.34.1
#24 lore raw

Re: [PATCH iwl-next v2 5/5] iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

Quoted 17 lines · Simon Horman
Quoted 1 line · Wenjun Wu

From: Jun Zhang <[email protected]>

...

drivers/net/ethernet/intel/iavf/iavf_virtchnl.c void iavf_virtchnl_completion(struct iavf_adapter *adapter, · line 2692

Quoted 7 lines
drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +3 -0
@@ -2471,6 +2687,16 @@ void iavf_virtchnl_completion(struct iavf_adapter *adapter,
  		if (!v_retval)
  			iavf_netdev_features_vlan_strip_set(netdev, false);
  		break;
+	case VIRTCHNL_OP_GET_QOS_CAPS:
+		u16 len = struct_size(adapter->qos_caps, cap,
+				      IAVF_MAX_QOS_TC_NUM);

Hi Jun Zhang and Wenju Wu,

clang-16 complains about this quite a lot.
I think it is because it wants the declaration of len - and thus the rest of this case - inside a block ({}).

Quoted 7 lines

.../iavf_virtchnl.c:2691:3: error: expected expression

                  u16 len = struct_size(adapter->qos_caps, cap,
                  ^
  .../iavf_virtchnl.c:2693:46: error: use of undeclared identifier 'len'
                  memcpy(adapter->qos_caps, msg, min(msglen, len));
                                                             ^

.../iavf_virtchnl.c:2693:46: error: use of undeclared identifier 'len'

Thanks for the solution.

Quoted 14 lines
Quoted 14 lines
drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +7 -0
+		memcpy(adapter->qos_caps, msg, min(msglen, len));
+		break;
+	case VIRTCHNL_OP_CONFIG_QUANTA:
+		iavf_notify_queue_config_complete(adapter);
+		break;
+	case VIRTCHNL_OP_CONFIG_QUEUE_BW:
+		break;
  	default:
  		if (adapter->current_op && (v_opcode != adapter->current_op))
  			dev_warn(&adapter->pdev->dev, "Expected response %d from PF, received %d\n",
Signature · 1 line
2.34.1
#25 lore raw

Re: [PATCH iwl-next v2 5/5] iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

In reply to Wenjun Wu
Quoted 24 lines · Wenjun Wu

Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding.

From: Jun Zhang <[email protected]>

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

User can configure the tx_max and tx_share of each queue. If any one of the queues have been fully updated by user, i.e. both tx_max and tx_share have been updated for that queue, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

---
  drivers/net/ethernet/intel/iavf/iavf.h        |  14 ++
  .../net/ethernet/intel/iavf/iavf_devlink.c    |  29 +++
  .../net/ethernet/intel/iavf/iavf_devlink.h    |   1 +
  drivers/net/ethernet/intel/iavf/iavf_main.c   |  45 +++-
  .../net/ethernet/intel/iavf/iavf_virtchnl.c   | 228 +++++++++++++++++-
  5 files changed, 313 insertions(+), 4 deletions(-)

[...]

Quoted 28 lines
drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +26 -0
+/**
+ * iavf_set_tc_queue_bw - set bw of allocated tc/queues
+ * @adapter: iavf adapter struct instance
+ *
+ * This function requests PF to set queue bw of multiple tc(s)
+ */
+static void iavf_set_tc_queue_bw(struct iavf_adapter *adapter)
+{
+       struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+       struct virtchnl_queues_bw_cfg *queues_bw_cfg;
+       struct iavf_dev_rate_node *queue_rate;
+       u16 queue_to_tc[256];
+       size_t len;
+       int q_idx;
+       int i, j;
 
 
+       u16 tc;
+
+       if (adapter->current_op != VIRTCHNL_OP_UNKNOWN) {
+               /* bail because we already have a command pending */
+               dev_err(&adapter->pdev->dev,
+                       "Cannot set tc queue bw, command %d pending\n",
+                       adapter->current_op);
+               return;
+       }
+
+       len = struct_size(queues_bw_cfg, cfg, adapter->num_active_queues);

drivers/net/ethernet/intel/iavf/iavf_virtchnl.c iavf_set_adapter_link_speed_from_vpe(struct iavf_adapter *adapter, · line 1619

Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +10 -0
+       queues_bw_cfg = kzalloc(len, GFP_KERNEL);
+       if (!queues_bw_cfg)
+               return;
+
+       queue_rate = dl_priv->queue_nodes;
+       queues_bw_cfg->vsi_id = adapter->vsi.id;
+       queues_bw_cfg->num_queues = adapter->ch_config.total_qps;
+
+       /* build tc[queue] */
+       for (i = 0; i < adapter->num_tc; i++) {

Nit, the scope of q_idx and j can have their scope reduced to this for loop.

Quoted 12 lines
drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +12 -0
+               for (j = 0; j < adapter->ch_config.ch_info[i].count; ++j) {
+                       q_idx = j + adapter->ch_config.ch_info[i].offset;
+                       queue_to_tc[q_idx] = i;
+               }
+       }
+
+       for (i = 0; i < queues_bw_cfg->num_queues; i++) {
+               tc = queue_to_tc[i];
+               queues_bw_cfg->cfg[i].queue_id = i;
+               queues_bw_cfg->cfg[i].shaper.peak = queue_rate[i].tx_max;
+               queues_bw_cfg->cfg[i].shaper.committed =
+                                                   queue_rate[i].tx_share;

drivers/net/ethernet/intel/iavf/iavf_virtchnl.c iavf_set_adapter_link_speed_from_vpe(struct iavf_adapter *adapter, · line 1641

Quoted 10 lines
drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +10 -0
+               queues_bw_cfg->cfg[i].tc = tc;
+       }
+
+       adapter->current_op = VIRTCHNL_OP_CONFIG_QUEUE_BW;
+       adapter->aq_required &= ~IAVF_FLAG_AQ_CONFIGURE_QUEUES_BW;
+       iavf_send_pf_msg(adapter, VIRTCHNL_OP_CONFIG_QUEUE_BW,
+                        (u8 *)queues_bw_cfg, len);
+       kfree(queues_bw_cfg);
+}
+

[...]
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

author
#26 lore raw

v3 cover iavf: Add devlink and devlink rate support

In reply to Wenjun Wu

To allow user to configure queue bandwidth, devlink port support is added to support devlink port rate API. [1]

Add devlink framework registration/unregistration on iavf driver initialization and remove, and devlink port of DEVLINK_PORT_FLAVOUR_VIRTUAL is created to be associated iavf netdevice.

iavf rate tree with root node, queue nodes, and leaf node is created and registered with devlink rate when iavf adapter is configured, and if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node parent iavf_root pci/0000:af:01.0/iavf_root: type node

                         +---------+
                         |   root  |
                         +----+----+
                              |
            |-----------------|-----------------|
       +----v----+       +----v----+       +----v----+
       |  txq_0  |       |  txq_1  |       |  txq_x  |
       +----+----+       +----+----+       +----+----+

User can configure the tx_max and tx_share of each queue. Once any one of the queues are fully configured, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF

Example:

1.To Set the queue tx_share:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_share 100 MBps

2.To Set the queue tx_max:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_max 200 MBps

3.To Show Current devlink port rate info:
devlink port function rate function show
[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node tx_share 800Mbit tx_max 1600Mbit parent iavf_root pci/0000:af:01.0/iavf_root: type node

[1]https://lore.kernel.org/netdev/[email protected]/

Change log:

v3:
- Rebase the code
- Changed rate node max/share set function description
- Put variable in local scope

v2:
- Change static array to flex array
- Use struct_size helper
- Align all the error code types in the function
- Move the register field definitions to the right place in the file
- Fix coding style
- Adapted to queue bw cfg and qos cap list virtchnl message with flex array fields

---
Jun Zhang (3):
iavf: Add devlink and devlink port support
iavf: Add devlink port function rate API support
iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

Wenjun Wu (2):
virtchnl: support queue rate limit and quanta size configuration
ice: Support VF queue rate limit and quanta size configuration

 drivers/net/ethernet/intel/Kconfig            |   1 +
 drivers/net/ethernet/intel/iavf/Makefile      |   2 +-
 drivers/net/ethernet/intel/iavf/iavf.h        |  20 +
 .../net/ethernet/intel/iavf/iavf_devlink.c    | 376 ++++++++++++++++++
 .../net/ethernet/intel/iavf/iavf_devlink.h    |  39 ++
 drivers/net/ethernet/intel/iavf/iavf_main.c   |  60 ++-
 .../net/ethernet/intel/iavf/iavf_virtchnl.c   | 230 ++++++++++-
 drivers/net/ethernet/intel/ice/ice.h          |   2 +
 drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
 drivers/net/ethernet/intel/ice/ice_common.c   |  19 +
 .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
 drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
 drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
 drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
 drivers/net/ethernet/intel/ice/ice_virtchnl.c | 312 +++++++++++++++
 drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
 .../intel/ice/ice_virtchnl_allowlist.c        |   6 +
 include/linux/avf/virtchnl.h                  | 119 ++++++
 18 files changed, 1216 insertions(+), 3 deletions(-)
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.c
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.h
Signature · 1 line
2.34.1
author
#27 lore raw

PATCH v3 1/5 virtchnl: support queue rate limit and quanta size configuration

This patch adds new virtchnl opcodes and structures for rate limit and quanta size configuration, which include:
1. VIRTCHNL_OP_CONFIG_QUEUE_BW, to configure max bandwidth for each VF per queue.
2. VIRTCHNL_OP_CONFIG_QUANTA, to configure quanta size per queue.
3. VIRTCHNL_OP_GET_QOS_CAPS, VF queries current QoS configuration, such as enabled TCs, arbiter type, up2tc and bandwidth of VSI node. The configuration is previously set by DCB and PF, and now is the potential QoS capability of VF. VF can take it as reference to configure queue TC mapping.

---
 include/linux/avf/virtchnl.h | 119 +++++++++++++++++++++++++++++++++++
 1 file changed, 119 insertions(+)
include/linux/avf/virtchnl.h +119 -0
index d0807ad43f93..0132c002ca06 100644
@@ -84,6 +84,9 @@ enum virtchnl_rx_hsplit {
 	VIRTCHNL_RX_HSPLIT_SPLIT_SCTP    = 8,
 };
 
+enum virtchnl_bw_limit_type {
+	VIRTCHNL_BW_SHAPER = 0,
+};
 /* END GENERIC DEFINES */
 
 /* Opcodes for VF-PF communication. These are placed in the v_opcode field
@@ -145,6 +148,11 @@ enum virtchnl_ops {
 	VIRTCHNL_OP_DISABLE_VLAN_STRIPPING_V2 = 55,
 	VIRTCHNL_OP_ENABLE_VLAN_INSERTION_V2 = 56,
 	VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2 = 57,
+	/* opcode 57 - 65 are reserved */
+	VIRTCHNL_OP_GET_QOS_CAPS = 66,
+	/* opcode 68 through 111 are reserved */
+	VIRTCHNL_OP_CONFIG_QUEUE_BW = 112,
+	VIRTCHNL_OP_CONFIG_QUANTA = 113,
 	VIRTCHNL_OP_MAX,
 };
 
@@ -253,6 +261,7 @@ VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_vsi_resource);
 #define VIRTCHNL_VF_OFFLOAD_RX_FLEX_DESC	BIT(26)
 #define VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF		BIT(27)
 #define VIRTCHNL_VF_OFFLOAD_FDIR_PF		BIT(28)
+#define VIRTCHNL_VF_OFFLOAD_QOS			BIT(29)
 
 #define VF_BASE_MODE_OFFLOADS (VIRTCHNL_VF_OFFLOAD_L2 | \
 			       VIRTCHNL_VF_OFFLOAD_VLAN | \
@@ -1377,6 +1386,85 @@ struct virtchnl_fdir_del {
 
 VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_fdir_del);
 
+struct virtchnl_shaper_bw {
+	/* Unit is Kbps */
+	u32 committed;
+	u32 peak;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_shaper_bw);
+
+/* VIRTCHNL_OP_GET_QOS_CAPS
+ * VF sends this message to get its QoS Caps, such as
+ * TC number, Arbiter and Bandwidth.
+ */
+struct virtchnl_qos_cap_elem {
+	u8 tc_num;
+	u8 tc_prio;
+#define VIRTCHNL_ABITER_STRICT      0
+#define VIRTCHNL_ABITER_ETS         2
+	u8 arbiter;
+#define VIRTCHNL_STRICT_WEIGHT      1
+	u8 weight;
+	enum virtchnl_bw_limit_type type;
+	union {
+		struct virtchnl_shaper_bw shaper;
+		u8 pad2[32];
+	};
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(40, virtchnl_qos_cap_elem);
+
+struct virtchnl_qos_cap_list {
+	u16 vsi_id;
+	u16 num_elem;
+	struct virtchnl_qos_cap_elem cap[];
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(4, virtchnl_qos_cap_list);
+#define virtchnl_qos_cap_list_LEGACY_SIZEOF	44
+
+/* VIRTCHNL_OP_CONFIG_QUEUE_BW */
+struct virtchnl_queue_bw {
+	u16 queue_id;
+	u8 tc;
+	u8 pad;
+	struct virtchnl_shaper_bw shaper;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_queue_bw);
+
+struct virtchnl_queues_bw_cfg {
+	u16 vsi_id;
+	u16 num_queues;
+	struct virtchnl_queue_bw cfg[];
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(4, virtchnl_queues_bw_cfg);
+#define virtchnl_queues_bw_cfg_LEGACY_SIZEOF	16
+
+enum virtchnl_queue_type {
+	VIRTCHNL_QUEUE_TYPE_TX			= 0,
+	VIRTCHNL_QUEUE_TYPE_RX			= 1,
+};
+
+/* structure to specify a chunk of contiguous queues */
+struct virtchnl_queue_chunk {
+	/* see enum virtchnl_queue_type */
+	s32 type;
+	u16 start_queue_id;
+	u16 num_queues;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_queue_chunk);
+
+struct virtchnl_quanta_cfg {
+	u16 quanta_size;
+	struct virtchnl_queue_chunk queue_select;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_quanta_cfg);
+
 #define __vss_byone(p, member, count, old)				      \
 	(struct_size(p, member, count) + (old - 1 - struct_size(p, member, 0)))
 
@@ -1399,6 +1487,8 @@ VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_fdir_del);
 		 __vss(virtchnl_vlan_filter_list_v2, __vss_byelem, p, m, c),  \
 		 __vss(virtchnl_tc_info, __vss_byelem, p, m, c),	      \
 		 __vss(virtchnl_rdma_qvlist_info, __vss_byelem, p, m, c),     \
+		 __vss(virtchnl_qos_cap_list, __vss_byelem, p, m, c),	      \
+		 __vss(virtchnl_queues_bw_cfg, __vss_byelem, p, m, c),	      \
 		 __vss(virtchnl_rss_key, __vss_byone, p, m, c),		      \
 		 __vss(virtchnl_rss_lut, __vss_byone, p, m, c))
 
@@ -1595,6 +1685,35 @@ virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode,
 	case VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2:
 		valid_len = sizeof(struct virtchnl_vlan_setting);
 		break;
+	case VIRTCHNL_OP_GET_QOS_CAPS:
+		break;
+	case VIRTCHNL_OP_CONFIG_QUEUE_BW:
+		valid_len = virtchnl_queues_bw_cfg_LEGACY_SIZEOF;
+		if (msglen >= valid_len) {
+			struct virtchnl_queues_bw_cfg *q_bw =
+				(struct virtchnl_queues_bw_cfg *)msg;
+
+			valid_len = virtchnl_struct_size(q_bw, cfg,
+							 q_bw->num_queues);
+			if (q_bw->num_queues == 0) {
+				err_msg_format = true;
+				break;
+			}
+		}
+		break;
+	case VIRTCHNL_OP_CONFIG_QUANTA:
+		valid_len = sizeof(struct virtchnl_quanta_cfg);
+		if (msglen >= valid_len) {
+			struct virtchnl_quanta_cfg *q_quanta =
+				(struct virtchnl_quanta_cfg *)msg;
+
+			if (q_quanta->quanta_size == 0 ||
+			    q_quanta->queue_select.num_queues == 0) {
+				err_msg_format = true;
+				break;
+			}
+		}
+		break;
 	/* These are always errors coming from the VF. */
 	case VIRTCHNL_OP_EVENT:
 	case VIRTCHNL_OP_UNKNOWN:
Signature · 1 line
2.34.1
author
#28 lore raw

PATCH v3 2/5 ice: Support VF queue rate limit and quanta size configuration

In reply to Wenjun Wu

Add support to configure VF queue rate limit and quanta size.

For quanta size configuration, the quanta profiles are divided evenly by PF numbers. For each port, the first quanta profile is reserved for default. When VF is asked to set queue quanta size, PF will search for an available profile, change the fields and assigned this profile to the queue.

---
 drivers/net/ethernet/intel/ice/ice.h          |   2 +
 drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
 drivers/net/ethernet/intel/ice/ice_common.c   |  19 ++
 .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
 drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
 drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
 drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
 drivers/net/ethernet/intel/ice/ice_virtchnl.c | 312 ++++++++++++++++++
 drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
 .../intel/ice/ice_virtchnl_allowlist.c        |   6 +
 10 files changed, 372 insertions(+)
drivers/net/ethernet/intel/ice/ice.h +2 -0
index 5d307bacf7c6..a4c9e6523fba 100644
@@ -644,6 +644,8 @@ struct ice_pf {
 #define ICE_VF_AGG_NODE_ID_START	65
 #define ICE_MAX_VF_AGG_NODES		32
 	struct ice_agg_node vf_agg_node[ICE_MAX_VF_AGG_NODES];
+
+	u8 num_quanta_prof_used;
 };
 
 extern struct workqueue_struct *ice_lag_wq;
drivers/net/ethernet/intel/ice/ice_base.c +2 -0
index 7fa43827a3f0..2b9319801dc3 100644
@@ -377,6 +377,8 @@ ice_setup_tx_ctx(struct ice_tx_ring *ring, struct ice_tlan_ctx *tlan_ctx, u16 pf
 		break;
 	}
 
+	tlan_ctx->quanta_prof_idx = ring->quanta_prof_id;
+
 	tlan_ctx->tso_ena = ICE_TX_LEGACY;
 	tlan_ctx->tso_qnum = pf_q;
 
drivers/net/ethernet/intel/ice/ice_common.c +19 -0
index 2652e4f5c4a2..7076bc1d85ab 100644
@@ -2251,6 +2251,23 @@ ice_parse_func_caps(struct ice_hw *hw, struct ice_hw_func_caps *func_p,
 	ice_recalc_port_limited_caps(hw, &func_p->common_cap);
 }
 
+/**
+ * ice_func_id_to_logical_id - map from function id to logical pf id
+ * @active_function_bitmap: active function bitmap
+ * @pf_id: function number of device
+ */
+static int ice_func_id_to_logical_id(u32 active_function_bitmap, u8 pf_id)
+{
+	u8 logical_id = 0;
+	u8 i;
+
+	for (i = 0; i < pf_id; i++)
+		if (active_function_bitmap & BIT(i))
+			logical_id++;
+
+	return logical_id;
+}
+
 /**
  * ice_parse_valid_functions_cap - Parse ICE_AQC_CAPS_VALID_FUNCTIONS caps
  * @hw: pointer to the HW struct
@@ -2268,6 +2285,8 @@ ice_parse_valid_functions_cap(struct ice_hw *hw, struct ice_hw_dev_caps *dev_p,
 	dev_p->num_funcs = hweight32(number);
 	ice_debug(hw, ICE_DBG_INIT, "dev caps: num_funcs = %d\n",
 		  dev_p->num_funcs);
+
+	hw->logical_pf_id = ice_func_id_to_logical_id(number, hw->pf_id);
 }
 
 /**
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +8 -0
index 6756f3d51d14..9da94e000394 100644
@@ -6,6 +6,14 @@
 #ifndef _ICE_HW_AUTOGEN_H_
 #define _ICE_HW_AUTOGEN_H_
 
+#define GLCOMM_QUANTA_PROF(_i)			(0x002D2D68 + ((_i) * 4))
+#define GLCOMM_QUANTA_PROF_MAX_INDEX		15
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_S	0
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_M	ICE_M(0x3FFF, 0)
+#define GLCOMM_QUANTA_PROF_MAX_CMD_S		16
+#define GLCOMM_QUANTA_PROF_MAX_CMD_M		ICE_M(0xFF, 16)
+#define GLCOMM_QUANTA_PROF_MAX_DESC_S		24
+#define GLCOMM_QUANTA_PROF_MAX_DESC_M		ICE_M(0x3F, 24)
 #define QTX_COMM_DBELL(_DBQM)			(0x002C0000 + ((_DBQM) * 4))
 #define QTX_COMM_HEAD(_DBQM)			(0x000E0000 + ((_DBQM) * 4))
 #define QTX_COMM_HEAD_HEAD_S			0
drivers/net/ethernet/intel/ice/ice_txrx.h +2 -0
index 166413fc33f4..7e152ab5b727 100644
@@ -381,6 +381,8 @@ struct ice_tx_ring {
 	u8 flags;
 	u8 dcb_tc;			/* Traffic class of ring */
 	u8 ptp_tx;
+
+	u16 quanta_prof_id;
 } ____cacheline_internodealigned_in_smp;
 
 static inline bool ice_ring_uses_build_skb(struct ice_rx_ring *ring)
drivers/net/ethernet/intel/ice/ice_type.h +1 -0
index f6061b508857..e9164c866315 100644
@@ -833,6 +833,7 @@ struct ice_hw {
 	u8 revision_id;
 
 	u8 pf_id;		/* device profile info */
+	u8 logical_pf_id;
 	enum ice_phy_model phy_model;
 
 	u16 max_burst_size;	/* driver sets this value */
drivers/net/ethernet/intel/ice/ice_vf_lib.h +9 -0
index 48fea6fa0362..a6078b583b79 100644
@@ -52,6 +52,13 @@ struct ice_mdd_vf_events {
 	u16 last_printed;
 };
 
+struct ice_vf_qs_bw {
+	u16 queue_id;
+	u32 committed;
+	u32 peak;
+	u8 tc;
+};
+
 /* VF operations */
 struct ice_vf_ops {
 	enum ice_disq_rst_src reset_type;
@@ -133,6 +140,8 @@ struct ice_vf {
 
 	/* devlink port data */
 	struct devlink_port devlink_port;
+
+	struct ice_vf_qs_bw qs_bw[ICE_MAX_RSS_QS_PER_VF];
 };
 
 /* Flags for controlling behavior of ice_reset_vf */
drivers/net/ethernet/intel/ice/ice_virtchnl.c +312 -0

348 lines not on this page · raw message

drivers/net/ethernet/intel/ice/ice_virtchnl.h +11 -0
index cd747718de73..0efb9c0f669a 100644
@@ -13,6 +13,13 @@
 /* Restrict number of MAC Addr and VLAN that non-trusted VF can programmed */
 #define ICE_MAX_VLAN_PER_VF		8
 
+#define ICE_DFLT_QUANTA 1024
+#define ICE_MAX_QUANTA_SIZE 4096
+#define ICE_MIN_QUANTA_SIZE 256
+
+#define calc_quanta_desc(x)	\
+	max_t(u16, 12, min_t(u16, 63, (((x) + 66) / 132) * 2 + 4))
+
 /* MAC filters: 1 is reserved for the VF's default/perm_addr/LAA MAC, 1 for
  * broadcast, and 16 for additional unicast/multicast filters
  */
@@ -51,6 +58,10 @@ struct ice_virtchnl_ops {
 	int (*dis_vlan_stripping_v2_msg)(struct ice_vf *vf, u8 *msg);
 	int (*ena_vlan_insertion_v2_msg)(struct ice_vf *vf, u8 *msg);
 	int (*dis_vlan_insertion_v2_msg)(struct ice_vf *vf, u8 *msg);
+	int (*get_qos_caps)(struct ice_vf *vf);
+	int (*cfg_q_tc_map)(struct ice_vf *vf, u8 *msg);
+	int (*cfg_q_bw)(struct ice_vf *vf, u8 *msg);
+	int (*cfg_q_quanta)(struct ice_vf *vf, u8 *msg);
 };
 
 #ifdef CONFIG_PCI_IOV
drivers/net/ethernet/intel/ice/ice_virtchnl_allowlist.c +6 -0
index 7d547fa616fa..2e3f63a429cd 100644
@@ -85,6 +85,11 @@ static const u32 fdir_pf_allowlist_opcodes[] = {
 	VIRTCHNL_OP_ADD_FDIR_FILTER, VIRTCHNL_OP_DEL_FDIR_FILTER,
 };
 
+static const u32 tc_allowlist_opcodes[] = {
+	VIRTCHNL_OP_GET_QOS_CAPS, VIRTCHNL_OP_CONFIG_QUEUE_BW,
+	VIRTCHNL_OP_CONFIG_QUANTA,
+};
+
 struct allowlist_opcode_info {
 	const u32 *opcodes;
 	size_t size;
@@ -105,6 +110,7 @@ static const struct allowlist_opcode_info allowlist_opcodes[] = {
 	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF, adv_rss_pf_allowlist_opcodes),
 	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_FDIR_PF, fdir_pf_allowlist_opcodes),
 	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_VLAN_V2, vlan_v2_allowlist_opcodes),
+	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_QOS, tc_allowlist_opcodes),
 };
 
 /**
Signature · 1 line
2.34.1
author
#29 lore raw

PATCH v3 3/5 iavf: Add devlink and devlink port support

In reply to Wenjun Wu

From: Jun Zhang <[email protected]>

To allow user to configure queue bandwidth, devlink port support is added to support devlink port rate API.

Add devlink framework registration/unregistration on iavf driver initialization and remove, and devlink port of DEVLINK_PORT_FLAVOUR_VIRTUAL is created to be associated iavf net device.

---
 drivers/net/ethernet/intel/Kconfig            |  1 +
 drivers/net/ethernet/intel/iavf/Makefile      |  2 +-
 drivers/net/ethernet/intel/iavf/iavf.h        |  6 ++
 .../net/ethernet/intel/iavf/iavf_devlink.c    | 93 +++++++++++++++++++
 .../net/ethernet/intel/iavf/iavf_devlink.h    | 17 ++++
 drivers/net/ethernet/intel/iavf/iavf_main.c   | 14 +++
 6 files changed, 132 insertions(+), 1 deletion(-)
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.c
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.h
drivers/net/ethernet/intel/Kconfig +1 -0
index d57f70d6e4d4..5bda31c4c652 100644
@@ -256,6 +256,7 @@ config I40EVF
 	tristate "Intel(R) Ethernet Adaptive Virtual Function support"
 	select IAVF
 	depends on PCI_MSI
+	select NET_DEVLINK
 	help
 	  This driver supports virtual functions for Intel XL710,
 	  X710, X722, XXV710, and all devices advertising support for
drivers/net/ethernet/intel/iavf/Makefile +1 -1
index 9c3e45c54d01..b5d7db97ab8b 100644
@@ -12,5 +12,5 @@ subdir-ccflags-y += -I$(src)
 obj-$(CONFIG_IAVF) += iavf.o
 
 iavf-objs := iavf_main.o iavf_ethtool.o iavf_virtchnl.o iavf_fdir.o \
-	     iavf_adv_rss.o \
+	     iavf_adv_rss.o iavf_devlink.o \
 	     iavf_txrx.o iavf_common.o iavf_adminq.o iavf_client.o
drivers/net/ethernet/intel/iavf/iavf.h +6 -0
index 85fba85fbb23..eec294b5a426 100644
@@ -33,9 +33,11 @@
 #include <net/udp.h>
 #include <net/tc_act/tc_gact.h>
 #include <net/tc_act/tc_mirred.h>
+#include <net/devlink.h>
 
 #include "iavf_type.h"
 #include <linux/avf/virtchnl.h>
+#include "iavf_devlink.h"
 #include "iavf_txrx.h"
 #include "iavf_fdir.h"
 #include "iavf_adv_rss.h"
@@ -369,6 +371,10 @@ struct iavf_adapter {
 	struct net_device *netdev;
 	struct pci_dev *pdev;
 
+	/* devlink & port data */
+	struct devlink *devlink;
+	struct devlink_port devlink_port;
+
 	struct iavf_hw hw; /* defined in iavf_type.h */
 
 	enum iavf_state_t state;
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +93 -0

93 lines not on this page · raw message

drivers/net/ethernet/intel/iavf/iavf_devlink.h new file +17 -0
new file mode 100644
index 000000000000..5c122278611a
@@ -0,0 +1,17 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright (C) 2023 Intel Corporation */
+
+#ifndef _IAVF_DEVLINK_H_
+#define _IAVF_DEVLINK_H_
+
+/* iavf devlink structure pointing to iavf adapter */
+struct iavf_devlink {
+	struct iavf_adapter *devlink_ref;	/* ref to iavf adapter */
+};
+
+int iavf_devlink_register(struct iavf_adapter *adapter);
+void iavf_devlink_unregister(struct iavf_adapter *adapter);
+int iavf_devlink_port_register(struct iavf_adapter *adapter);
+void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
+
+#endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +14 -0
index b23ca9d80189..1fb14f3f1ad0 100644
@@ -2038,6 +2038,7 @@ static void iavf_finish_config(struct work_struct *work)
 				iavf_free_rss(adapter);
 				iavf_free_misc_irq(adapter);
 				iavf_reset_interrupt_capability(adapter);
+				iavf_devlink_port_unregister(adapter);
 				iavf_change_state(adapter,
 						  __IAVF_INIT_CONFIG_ADAPTER);
 				goto out;
@@ -2709,6 +2710,9 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 	if (err)
 		goto err_sw_init;
 
+	if (!adapter->netdev_registered)
+		iavf_devlink_port_register(adapter);
+
 	netif_carrier_off(netdev);
 	adapter->link_up = false;
 	netif_tx_stop_all_queues(netdev);
@@ -2750,6 +2754,7 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 err_mem:
 	iavf_free_rss(adapter);
 	iavf_free_misc_irq(adapter);
+	iavf_devlink_port_unregister(adapter);
 err_sw_init:
 	iavf_reset_interrupt_capability(adapter);
 err:
@@ -4996,6 +5001,12 @@ static int iavf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	/* Setup the wait queue for indicating virtchannel events */
 	init_waitqueue_head(&adapter->vc_waitqueue);
 
+	/* Register iavf adapter with devlink */
+	err = iavf_devlink_register(adapter);
+	if (err)
+		dev_err(&pdev->dev, "devlink registration failed: %d\n", err);
+
+	/* Keep driver interface even on devlink registration failure */
 	return 0;
 
 err_ioremap:
@@ -5140,6 +5151,9 @@ static void iavf_remove(struct pci_dev *pdev)
 				 err);
 	}
 
+	iavf_devlink_port_unregister(adapter);
+	iavf_devlink_unregister(adapter);
+
 	mutex_lock(&adapter->crit_lock);
 	dev_info(&adapter->pdev->dev, "Removing device\n");
 	iavf_change_state(adapter, __IAVF_REMOVE);
Signature · 1 line
2.34.1
author
#30 lore raw

PATCH v3 4/5 iavf: Add devlink port function rate API support

In reply to Wenjun Wu

From: Jun Zhang <[email protected]>

To allow user to configure queue based parameters, devlink port function rate api functions are added for setting node tx_max and tx_share parameters.

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

---
 .../net/ethernet/intel/iavf/iavf_devlink.c    | 258 +++++++++++++++++-
 .../net/ethernet/intel/iavf/iavf_devlink.h    |  21 ++
 drivers/net/ethernet/intel/iavf/iavf_main.c   |   7 +-
 3 files changed, 283 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/iavf/iavf_devlink.c +256 -2
index 991d041e5922..24ba3744859a 100644
@@ -4,7 +4,261 @@
 #include "iavf.h"
 #include "iavf_devlink.h"
 
-static const struct devlink_ops iavf_devlink_ops = {};
+/**
+ * iavf_devlink_rate_init_rate_tree - export rate tree to devlink rate
+ * @adapter: iavf adapter struct instance
+ *
+ * This function builds Rate Tree based on iavf adapter configuration
+ * and exports it's contents to devlink rate.
+ */
+void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+	struct iavf_dev_rate_node *iavf_r_node;
+	struct iavf_dev_rate_node *iavf_q_node;
+	struct devlink_rate *dl_root_node;
+	struct devlink_rate *dl_tmp_node;
+	int q_num, size, i;
+
+	if (!adapter->devlink_port.registered)
+		return;
+
+	iavf_r_node = &dl_priv->root_node;
+	memset(iavf_r_node, 0, sizeof(*iavf_r_node));
+	iavf_r_node->tx_max = adapter->link_speed;
+	strscpy(iavf_r_node->name, "iavf_root", IAVF_RATE_NODE_NAME);
+
+	devl_lock(adapter->devlink);
+	dl_root_node = devl_rate_node_create(adapter->devlink, iavf_r_node,
+					     iavf_r_node->name, NULL);
+	if (!dl_root_node || IS_ERR(dl_root_node))
+		goto err_node;
+
+	iavf_r_node->rate_node = dl_root_node;
+
+	/* Allocate queue nodes, and chain them under root */
+	q_num = adapter->num_active_queues;
+	if (q_num > 0) {
+		size = q_num * sizeof(struct iavf_dev_rate_node);
+		dl_priv->queue_nodes = kzalloc(size, GFP_KERNEL);
+		if (!dl_priv->queue_nodes)
+			goto err_node;
+
+		memset(dl_priv->queue_nodes, 0, size);
+
+		for (i = 0; i < q_num; ++i) {
+			iavf_q_node = &dl_priv->queue_nodes[i];
+			snprintf(iavf_q_node->name, IAVF_RATE_NODE_NAME,
+				 "txq_%d", i);
+			dl_tmp_node = devl_rate_node_create(adapter->devlink,
+							    iavf_q_node,
+							    iavf_q_node->name,
+							    dl_root_node);
+			if (!dl_tmp_node || IS_ERR(dl_tmp_node)) {
+				kfree(dl_priv->queue_nodes);
+				goto err_node;
+			}
+
+			iavf_q_node->rate_node = dl_tmp_node;
+			iavf_q_node->tx_max = IAVF_TX_DEFAULT;
+			iavf_q_node->tx_share = 0;
+		}
+	}
+
+	dl_priv->update_in_progress = false;
+	dl_priv->iavf_dev_rate_initialized = true;
+	devl_unlock(adapter->devlink);
+	return;
+err_node:
+	devl_rate_nodes_destroy(adapter->devlink);
+	dl_priv->iavf_dev_rate_initialized = false;
+	devl_unlock(adapter->devlink);
+}
+
+/**
+ * iavf_devlink_rate_deinit_rate_tree - Unregister rate tree with devlink rate
+ * @adapter: iavf adapter struct instance
+ *
+ * This function unregisters the current iavf rate tree registered with devlink
+ * rate and frees resources.
+ */
+void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+
+	if (!dl_priv->iavf_dev_rate_initialized)
+		return;
+
+	devl_lock(adapter->devlink);
+	devl_rate_leaf_destroy(&adapter->devlink_port);
+	devl_rate_nodes_destroy(adapter->devlink);
+	kfree(dl_priv->queue_nodes);
+	devl_unlock(adapter->devlink);
+}
+
+/**
+ * iavf_check_update_config - check if updating queue parameters needed
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ *
+ * This function sets queue bw & quanta size configuration if all
+ * queue parameters are set
+ */
+static int iavf_check_update_config(struct iavf_adapter *adapter,
+				    struct iavf_dev_rate_node *node)
+{
+	/* Update queue bw if any one of the queues have been fully updated by
+	 * user, the other queues either use the default value or the last
+	 * fully updated value
+	 */
+	if (node->tx_update_flag ==
+	    (IAVF_FLAG_TX_MAX_UPDATED | IAVF_FLAG_TX_SHARE_UPDATED)) {
+		node->tx_max = node->tx_max_temp;
+		node->tx_share = node->tx_share_temp;
+	} else {
+		return 0;
+	}
+
+	/* Reconfig queue bw only when iavf driver on running state */
+	if (adapter->state != __IAVF_RUNNING)
+		return -EBUSY;
+
+	return 0;
+}
+
+/**
+ * iavf_update_queue_tx_share - sets tx min parameter
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ * @bw: bandwidth in bytes per second
+ * @extack: extended netdev ack structure
+ *
+ * This function sets min BW limit.
+ */
+static int iavf_update_queue_tx_share(struct iavf_adapter *adapter,
+				      struct iavf_dev_rate_node *node,
+				      u64 bw, struct netlink_ext_ack *extack)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+	u64 tx_share_sum = 0;
+
+	/* Keep in kbps */
+	node->tx_share_temp = div_u64(bw, IAVF_RATE_DIV_FACTOR);
+
+	if (ADV_LINK_SUPPORT(adapter)) {
+		int i;
+
+		for (i = 0; i < adapter->num_active_queues; ++i) {
+			if (node != &dl_priv->queue_nodes[i])
+				tx_share_sum +=
+					dl_priv->queue_nodes[i].tx_share;
+			else
+				tx_share_sum += node->tx_share_temp;
+		}
+
+		if (tx_share_sum / 1000  > adapter->link_speed_mbps)
+			return -EINVAL;
+	}
+
+	node->tx_update_flag |= IAVF_FLAG_TX_SHARE_UPDATED;
+	return iavf_check_update_config(adapter, node);
+}
+
+/**
+ * iavf_update_queue_tx_max - sets tx max parameter
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ * @bw: bandwidth in bytes per second
+ * @extack: extended netdev ack structure
+ *
+ * This function sets max BW limit.
+ */
+static int iavf_update_queue_tx_max(struct iavf_adapter *adapter,
+				    struct iavf_dev_rate_node *node,
+				    u64 bw, struct netlink_ext_ack *extack)
+{
+	/* Keep in kbps */
+	node->tx_max_temp = div_u64(bw, IAVF_RATE_DIV_FACTOR);
+	if (ADV_LINK_SUPPORT(adapter)) {
+		if (node->tx_max_temp / 1000 > adapter->link_speed_mbps)
+			return -EINVAL;
+	}
+
+	node->tx_update_flag |= IAVF_FLAG_TX_MAX_UPDATED;
+
+	return iavf_check_update_config(adapter, node);
+}
+
+static int iavf_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node,
+					     void *priv, u64 tx_max,
+					     struct netlink_ext_ack *extack)
+{
+	struct iavf_dev_rate_node *node = priv;
+	struct iavf_devlink *dl_priv;
+	struct iavf_adapter *adapter;
+
+	if (!node)
+		return 0;
+
+	dl_priv = devlink_priv(rate_node->devlink);
+	adapter = dl_priv->devlink_ref;
+
+	/* Check if last update is in progress */
+	if (dl_priv->update_in_progress)
+		return -EBUSY;
+
+	if (node == &dl_priv->root_node)
+		return 0;
+
+	return iavf_update_queue_tx_max(adapter, node, tx_max, extack);
+}
+
+static int iavf_devlink_rate_node_tx_share_set(struct devlink_rate *rate_node,
+					       void *priv, u64 tx_share,
+					       struct netlink_ext_ack *extack)
+{
+	struct iavf_dev_rate_node *node = priv;
+	struct iavf_devlink *dl_priv;
+	struct iavf_adapter *adapter;
+
+	if (!node)
+		return 0;
+
+	dl_priv = devlink_priv(rate_node->devlink);
+	adapter = dl_priv->devlink_ref;
+
+	/* Check if last update is in progress */
+	if (dl_priv->update_in_progress)
+		return -EBUSY;
+
+	if (node == &dl_priv->root_node)
+		return 0;
+
+	return iavf_update_queue_tx_share(adapter, node, tx_share, extack);
+}
+
+static int iavf_devlink_rate_node_del(struct devlink_rate *rate_node,
+				      void *priv,
+				      struct netlink_ext_ack *extack)
+{
+	return -EINVAL;
+}
+
+static int iavf_devlink_set_parent(struct devlink_rate *devlink_rate,
+				   struct devlink_rate *parent,
+				   void *priv, void *parent_priv,
+				   struct netlink_ext_ack *extack)
+{
+	return -EINVAL;
+}
+
+static const struct devlink_ops iavf_devlink_ops = {
+	.rate_node_tx_share_set = iavf_devlink_rate_node_tx_share_set,
+	.rate_node_tx_max_set = iavf_devlink_rate_node_tx_max_set,
+	.rate_node_del = iavf_devlink_rate_node_del,
+	.rate_leaf_parent_set = iavf_devlink_set_parent,
+	.rate_node_parent_set = iavf_devlink_set_parent,
+};
 
 /**
  * iavf_devlink_register - Register allocated devlink instance for iavf adapter
@@ -30,7 +284,7 @@ int iavf_devlink_register(struct iavf_adapter *adapter)
 	adapter->devlink = devlink;
 	ref = devlink_priv(devlink);
 	ref->devlink_ref = adapter;
-
+	ref->iavf_dev_rate_initialized = false;
 	devlink_register(devlink);
 
 	return 0;
drivers/net/ethernet/intel/iavf/iavf_devlink.h +21 -0
index 5c122278611a..897ff5fc87af 100644
@@ -4,14 +4,35 @@
 #ifndef _IAVF_DEVLINK_H_
 #define _IAVF_DEVLINK_H_
 
+#define IAVF_RATE_NODE_NAME			12
+struct iavf_dev_rate_node {
+	char name[IAVF_RATE_NODE_NAME];
+	struct devlink_rate *rate_node;
+	u8 tx_update_flag;
+#define IAVF_FLAG_TX_SHARE_UPDATED		BIT(0)
+#define IAVF_FLAG_TX_MAX_UPDATED		BIT(1)
+	u64 tx_max;
+	u64 tx_share;
+	u64 tx_max_temp;
+	u64 tx_share_temp;
+#define IAVF_RATE_DIV_FACTOR			125
+#define IAVF_TX_DEFAULT				100000
+};
+
 /* iavf devlink structure pointing to iavf adapter */
 struct iavf_devlink {
 	struct iavf_adapter *devlink_ref;	/* ref to iavf adapter */
+	struct iavf_dev_rate_node root_node;
+	struct iavf_dev_rate_node *queue_nodes;
+	bool iavf_dev_rate_initialized;
+	bool update_in_progress;
 };
 
 int iavf_devlink_register(struct iavf_adapter *adapter);
 void iavf_devlink_unregister(struct iavf_adapter *adapter);
 int iavf_devlink_port_register(struct iavf_adapter *adapter);
 void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
+void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter);
+void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter);
 
 #endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +6 -1
index 1fb14f3f1ad0..2aec6427d5e2 100644
@@ -2038,6 +2038,7 @@ static void iavf_finish_config(struct work_struct *work)
 				iavf_free_rss(adapter);
 				iavf_free_misc_irq(adapter);
 				iavf_reset_interrupt_capability(adapter);
+				iavf_devlink_rate_deinit_rate_tree(adapter);
 				iavf_devlink_port_unregister(adapter);
 				iavf_change_state(adapter,
 						  __IAVF_INIT_CONFIG_ADAPTER);
@@ -2710,8 +2711,10 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 	if (err)
 		goto err_sw_init;
 
-	if (!adapter->netdev_registered)
+	if (!adapter->netdev_registered) {
 		iavf_devlink_port_register(adapter);
+		iavf_devlink_rate_init_rate_tree(adapter);
+	}
 
 	netif_carrier_off(netdev);
 	adapter->link_up = false;
@@ -2754,6 +2757,7 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 err_mem:
 	iavf_free_rss(adapter);
 	iavf_free_misc_irq(adapter);
+	iavf_devlink_rate_deinit_rate_tree(adapter);
 	iavf_devlink_port_unregister(adapter);
 err_sw_init:
 	iavf_reset_interrupt_capability(adapter);
@@ -5151,6 +5155,7 @@ static void iavf_remove(struct pci_dev *pdev)
 				 err);
 	}
 
+	iavf_devlink_rate_deinit_rate_tree(adapter);
 	iavf_devlink_port_unregister(adapter);
 	iavf_devlink_unregister(adapter);
 
Signature · 1 line
2.34.1
author
#31 lore raw

PATCH v3 5/5 iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

In reply to Wenjun Wu

From: Jun Zhang <[email protected]>

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

User can configure the tx_max and tx_share of each queue. If any one of the queues have been fully updated by user, i.e. both tx_max and tx_share have been updated for that queue, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

---
 drivers/net/ethernet/intel/iavf/iavf.h        |  14 ++
 .../net/ethernet/intel/iavf/iavf_devlink.c    |  29 +++
 .../net/ethernet/intel/iavf/iavf_devlink.h    |   1 +
 drivers/net/ethernet/intel/iavf/iavf_main.c   |  45 +++-
 .../net/ethernet/intel/iavf/iavf_virtchnl.c   | 230 +++++++++++++++++-
 5 files changed, 315 insertions(+), 4 deletions(-)
drivers/net/ethernet/intel/iavf/iavf.h +14 -0
index eec294b5a426..27a230f58816 100644
@@ -252,6 +252,9 @@ struct iavf_cloud_filter {
 #define IAVF_RESET_WAIT_DETECTED_COUNT 500
 #define IAVF_RESET_WAIT_COMPLETE_COUNT 2000
 
+#define IAVF_MAX_QOS_TC_NUM		8
+#define IAVF_DEFAULT_QUANTA_SIZE	1024
+
 /* board specific private data structure */
 struct iavf_adapter {
 	struct workqueue_struct *wq;
@@ -351,6 +354,9 @@ struct iavf_adapter {
 #define IAVF_FLAG_AQ_DISABLE_CTAG_VLAN_INSERTION	BIT_ULL(36)
 #define IAVF_FLAG_AQ_ENABLE_STAG_VLAN_INSERTION		BIT_ULL(37)
 #define IAVF_FLAG_AQ_DISABLE_STAG_VLAN_INSERTION	BIT_ULL(38)
+#define IAVF_FLAG_AQ_CONFIGURE_QUEUES_BW		BIT_ULL(39)
+#define IAVF_FLAG_AQ_CONFIGURE_QUEUES_QUANTA_SIZE	BIT_ULL(40)
+#define IAVF_FLAG_AQ_GET_QOS_CAPS			BIT_ULL(41)
 
 	/* flags for processing extended capability messages during
 	 * __IAVF_INIT_EXTENDED_CAPS. Each capability exchange requires
@@ -374,6 +380,7 @@ struct iavf_adapter {
 	/* devlink & port data */
 	struct devlink *devlink;
 	struct devlink_port devlink_port;
+	bool devlink_update;
 
 	struct iavf_hw hw; /* defined in iavf_type.h */
 
@@ -423,6 +430,8 @@ struct iavf_adapter {
 			       VIRTCHNL_VF_OFFLOAD_FDIR_PF)
 #define ADV_RSS_SUPPORT(_a) ((_a)->vf_res->vf_cap_flags & \
 			     VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF)
+#define QOS_ALLOWED(_a) ((_a)->vf_res->vf_cap_flags & \
+			 VIRTCHNL_VF_OFFLOAD_QOS)
 	struct virtchnl_vf_resource *vf_res; /* incl. all VSIs */
 	struct virtchnl_vsi_resource *vsi_res; /* our LAN VSI */
 	struct virtchnl_version_info pf_version;
@@ -431,6 +440,7 @@ struct iavf_adapter {
 	struct virtchnl_vlan_caps vlan_v2_caps;
 	u16 msg_enable;
 	struct iavf_eth_stats current_stats;
+	struct virtchnl_qos_cap_list *qos_caps;
 	struct iavf_vsi vsi;
 	u32 aq_wait_count;
 	/* RSS stuff */
@@ -577,6 +587,10 @@ void iavf_notify_client_message(struct iavf_vsi *vsi, u8 *msg, u16 len);
 void iavf_notify_client_l2_params(struct iavf_vsi *vsi);
 void iavf_notify_client_open(struct iavf_vsi *vsi);
 void iavf_notify_client_close(struct iavf_vsi *vsi, bool reset);
+void iavf_update_queue_config(struct iavf_adapter *adapter);
+void iavf_configure_queues_bw(struct iavf_adapter *adapter);
+void iavf_configure_queues_quanta_size(struct iavf_adapter *adapter);
+void iavf_get_qos_caps(struct iavf_adapter *adapter);
 void iavf_enable_channels(struct iavf_adapter *adapter);
 void iavf_disable_channels(struct iavf_adapter *adapter);
 void iavf_add_cloud_filter(struct iavf_adapter *adapter);
drivers/net/ethernet/intel/iavf/iavf_devlink.c +29 -0
index 24ba3744859a..0ab9a0a9823e 100644
@@ -96,6 +96,30 @@ void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
 	devl_unlock(adapter->devlink);
 }
 
+/**
+ * iavf_notify_queue_config_complete - notify updating queue completion
+ * @adapter: iavf adapter struct instance
+ *
+ * This function sets the queue configuration update status when all
+ * queue parameters have been sent to PF
+ */
+void iavf_notify_queue_config_complete(struct iavf_adapter *adapter)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+	int q_num = adapter->num_active_queues;
+	int i;
+
+	/* clean up rate tree update flags*/
+	for (i = 0; i < q_num; i++)
+		if (dl_priv->queue_nodes[i].tx_update_flag ==
+		    (IAVF_FLAG_TX_MAX_UPDATED | IAVF_FLAG_TX_SHARE_UPDATED)) {
+			dl_priv->queue_nodes[i].tx_update_flag = 0;
+			break;
+		}
+
+	dl_priv->update_in_progress = false;
+}
+
 /**
  * iavf_check_update_config - check if updating queue parameters needed
  * @adapter: iavf adapter struct instance
@@ -107,6 +131,8 @@ void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
 static int iavf_check_update_config(struct iavf_adapter *adapter,
 				    struct iavf_dev_rate_node *node)
 {
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+
 	/* Update queue bw if any one of the queues have been fully updated by
 	 * user, the other queues either use the default value or the last
 	 * fully updated value
@@ -123,6 +149,8 @@ static int iavf_check_update_config(struct iavf_adapter *adapter,
 	if (adapter->state != __IAVF_RUNNING)
 		return -EBUSY;
 
+	dl_priv->update_in_progress = true;
+	iavf_update_queue_config(adapter);
 	return 0;
 }
 
@@ -282,6 +310,7 @@ int iavf_devlink_register(struct iavf_adapter *adapter)
 
 	/* Init iavf adapter devlink */
 	adapter->devlink = devlink;
+	adapter->devlink_update = false;
 	ref = devlink_priv(devlink);
 	ref->devlink_ref = adapter;
 	ref->iavf_dev_rate_initialized = false;
drivers/net/ethernet/intel/iavf/iavf_devlink.h +1 -0
index 897ff5fc87af..a8a41f343f56 100644
@@ -34,5 +34,6 @@ int iavf_devlink_port_register(struct iavf_adapter *adapter);
 void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
 void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter);
 void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter);
+void iavf_notify_queue_config_complete(struct iavf_adapter *adapter);
 
 #endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +42 -3

84 lines not on this page · raw message

drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +229 -1

254 lines not on this page · raw message

Signature · 1 line
2.34.1
reviewer
#32 lore raw

Re: [PATCH iwl-next v3 5/5] iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

Quoted 13 lines · Wenjun Wu

From: Jun Zhang <[email protected]>

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

User can configure the tx_max and tx_share of each queue. If any one of the queues have been fully updated by user, i.e. both tx_max and tx_share have been updated for that queue, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

...

Quoted 1 line
drivers/net/ethernet/intel/iavf/iavf_main.c +0 -0

...

drivers/net/ethernet/intel/iavf/iavf_main.c static int iavf_probe(struct pci_dev *pdev, const struct pci_device_id *ent) · line 5040

Quoted 7 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +3 -0
@@ -5005,10 +5035,18 @@ static int iavf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	/* Setup the wait queue for indicating virtchannel events */
 	init_waitqueue_head(&adapter->vc_waitqueue);
 
+	len = struct_size(adapter->qos_caps, cap, IAVF_MAX_QOS_TC_NUM);
+	adapter->qos_caps = kzalloc(len, GFP_KERNEL);
+	if (!adapter->qos_caps)

Hi Jun Zhang and Wenjun Wu,

The goto below leads to the function returning err. Should err be set to an error value here?

As flagged by Smatch and Coccinelle.

Quoted 13 lines
drivers/net/ethernet/intel/iavf/iavf_main.c +6 -1
+		goto err_ioremap;
+
 	/* Register iavf adapter with devlink */
 	err = iavf_devlink_register(adapter);
-	if (err)
+	if (err) {
 		dev_err(&pdev->dev, "devlink registration failed: %d\n", err);
+		kfree(adapter->qos_caps);
+		goto err_ioremap;
+	}
 
 	/* Keep driver interface even on devlink registration failure */
 	return 0;

...

author
#33 lore raw

v4 cover iavf: Add devlink and devlink rate support

In reply to Wenjun Wu

To allow user to configure queue bandwidth, devlink port support is added to support devlink port rate API. [1]

Add devlink framework registration/unregistration on iavf driver initialization and remove, and devlink port of DEVLINK_PORT_FLAVOUR_VIRTUAL is created to be associated iavf netdevice.

iavf rate tree with root node, queue nodes, and leaf node is created and registered with devlink rate when iavf adapter is configured, and if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node parent iavf_root pci/0000:af:01.0/iavf_root: type node

                         +---------+
                         |   root  |
                         +----+----+
                              |
            |-----------------|-----------------|
       +----v----+       +----v----+       +----v----+
       |  txq_0  |       |  txq_1  |       |  txq_x  |
       +----+----+       +----+----+       +----+----+

User can configure the tx_max and tx_share of each queue. Once any one of the queues are fully configured, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF

Example:

1.To Set the queue tx_share:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_share 100 MBps

2.To Set the queue tx_max:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_max 200 MBps

3.To Show Current devlink port rate info:
devlink port function rate function show
[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node tx_share 800Mbit tx_max 1600Mbit parent iavf_root pci/0000:af:01.0/iavf_root: type node

[1]https://lore.kernel.org/netdev/[email protected]/

Change log:

v4:
- Rearrange the ice_vf_qs_bw structure, put the largest number first
- Minimize the scope of values
- Remove the unnecessary brackets
- Remove the unnecessary memory allocation.
- Added Error Code and moved devlink registration before aq lock initialization
- Changed devlink registration for error handling in case of allocation failure
- Used kcalloc for object array memory allocation and initialization
- Changed functions & comments for readability

v3:
- Rebase the code
- Changed rate node max/share set function description
- Put variable in local scope

v2:
- Change static array to flex array
- Use struct_size helper
- Align all the error code types in the function
- Move the register field definitions to the right place in the file
- Fix coding style
- Adapted to queue bw cfg and qos cap list virtchnl message with flex array fields
---

Jun Zhang (3):
iavf: Add devlink and devlink port support
iavf: Add devlink port function rate API support
iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

Wenjun Wu (2):
virtchnl: support queue rate limit and quanta size configuration
ice: Support VF queue rate limit and quanta size configuration

 drivers/net/ethernet/intel/Kconfig            |   1 +
 drivers/net/ethernet/intel/iavf/Makefile      |   2 +-
 drivers/net/ethernet/intel/iavf/iavf.h        |  19 +
 .../net/ethernet/intel/iavf/iavf_devlink.c    | 377 ++++++++++++++++++
 .../net/ethernet/intel/iavf/iavf_devlink.h    |  38 ++
 drivers/net/ethernet/intel/iavf/iavf_main.c   |  64 ++-
 .../net/ethernet/intel/iavf/iavf_virtchnl.c   | 231 ++++++++++-
 drivers/net/ethernet/intel/ice/ice.h          |   2 +
 drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
 drivers/net/ethernet/intel/ice/ice_common.c   |  19 +
 .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
 drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
 drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
 drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
 drivers/net/ethernet/intel/ice/ice_virtchnl.c | 310 ++++++++++++++
 drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
 .../intel/ice/ice_virtchnl_allowlist.c        |   6 +
 include/linux/avf/virtchnl.h                  | 119 ++++++
 18 files changed, 1218 insertions(+), 3 deletions(-)
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.c
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.h
Signature · 1 line
2.34.1
author
#34 lore raw

PATCH v4 1/5 virtchnl: support queue rate limit and quanta size configuration

This patch adds new virtchnl opcodes and structures for rate limit and quanta size configuration, which include:
1. VIRTCHNL_OP_CONFIG_QUEUE_BW, to configure max bandwidth for each VF per queue.
2. VIRTCHNL_OP_CONFIG_QUANTA, to configure quanta size per queue.
3. VIRTCHNL_OP_GET_QOS_CAPS, VF queries current QoS configuration, such as enabled TCs, arbiter type, up2tc and bandwidth of VSI node. The configuration is previously set by DCB and PF, and now is the potential QoS capability of VF. VF can take it as reference to configure queue TC mapping.

---
 include/linux/avf/virtchnl.h | 119 +++++++++++++++++++++++++++++++++++
 1 file changed, 119 insertions(+)
include/linux/avf/virtchnl.h +119 -0
index d0807ad43f93..0132c002ca06 100644
@@ -84,6 +84,9 @@ enum virtchnl_rx_hsplit {
 	VIRTCHNL_RX_HSPLIT_SPLIT_SCTP    = 8,
 };
 
+enum virtchnl_bw_limit_type {
+	VIRTCHNL_BW_SHAPER = 0,
+};
 /* END GENERIC DEFINES */
 
 /* Opcodes for VF-PF communication. These are placed in the v_opcode field
@@ -145,6 +148,11 @@ enum virtchnl_ops {
 	VIRTCHNL_OP_DISABLE_VLAN_STRIPPING_V2 = 55,
 	VIRTCHNL_OP_ENABLE_VLAN_INSERTION_V2 = 56,
 	VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2 = 57,
+	/* opcode 57 - 65 are reserved */
+	VIRTCHNL_OP_GET_QOS_CAPS = 66,
+	/* opcode 68 through 111 are reserved */
+	VIRTCHNL_OP_CONFIG_QUEUE_BW = 112,
+	VIRTCHNL_OP_CONFIG_QUANTA = 113,
 	VIRTCHNL_OP_MAX,
 };
 
@@ -253,6 +261,7 @@ VIRTCHNL_CHECK_STRUCT_LEN(16, virtchnl_vsi_resource);
 #define VIRTCHNL_VF_OFFLOAD_RX_FLEX_DESC	BIT(26)
 #define VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF		BIT(27)
 #define VIRTCHNL_VF_OFFLOAD_FDIR_PF		BIT(28)
+#define VIRTCHNL_VF_OFFLOAD_QOS			BIT(29)
 
 #define VF_BASE_MODE_OFFLOADS (VIRTCHNL_VF_OFFLOAD_L2 | \
 			       VIRTCHNL_VF_OFFLOAD_VLAN | \
@@ -1377,6 +1386,85 @@ struct virtchnl_fdir_del {
 
 VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_fdir_del);
 
+struct virtchnl_shaper_bw {
+	/* Unit is Kbps */
+	u32 committed;
+	u32 peak;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_shaper_bw);
+
+/* VIRTCHNL_OP_GET_QOS_CAPS
+ * VF sends this message to get its QoS Caps, such as
+ * TC number, Arbiter and Bandwidth.
+ */
+struct virtchnl_qos_cap_elem {
+	u8 tc_num;
+	u8 tc_prio;
+#define VIRTCHNL_ABITER_STRICT      0
+#define VIRTCHNL_ABITER_ETS         2
+	u8 arbiter;
+#define VIRTCHNL_STRICT_WEIGHT      1
+	u8 weight;
+	enum virtchnl_bw_limit_type type;
+	union {
+		struct virtchnl_shaper_bw shaper;
+		u8 pad2[32];
+	};
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(40, virtchnl_qos_cap_elem);
+
+struct virtchnl_qos_cap_list {
+	u16 vsi_id;
+	u16 num_elem;
+	struct virtchnl_qos_cap_elem cap[];
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(4, virtchnl_qos_cap_list);
+#define virtchnl_qos_cap_list_LEGACY_SIZEOF	44
+
+/* VIRTCHNL_OP_CONFIG_QUEUE_BW */
+struct virtchnl_queue_bw {
+	u16 queue_id;
+	u8 tc;
+	u8 pad;
+	struct virtchnl_shaper_bw shaper;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_queue_bw);
+
+struct virtchnl_queues_bw_cfg {
+	u16 vsi_id;
+	u16 num_queues;
+	struct virtchnl_queue_bw cfg[];
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(4, virtchnl_queues_bw_cfg);
+#define virtchnl_queues_bw_cfg_LEGACY_SIZEOF	16
+
+enum virtchnl_queue_type {
+	VIRTCHNL_QUEUE_TYPE_TX			= 0,
+	VIRTCHNL_QUEUE_TYPE_RX			= 1,
+};
+
+/* structure to specify a chunk of contiguous queues */
+struct virtchnl_queue_chunk {
+	/* see enum virtchnl_queue_type */
+	s32 type;
+	u16 start_queue_id;
+	u16 num_queues;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(8, virtchnl_queue_chunk);
+
+struct virtchnl_quanta_cfg {
+	u16 quanta_size;
+	struct virtchnl_queue_chunk queue_select;
+};
+
+VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_quanta_cfg);
+
 #define __vss_byone(p, member, count, old)				      \
 	(struct_size(p, member, count) + (old - 1 - struct_size(p, member, 0)))
 
@@ -1399,6 +1487,8 @@ VIRTCHNL_CHECK_STRUCT_LEN(12, virtchnl_fdir_del);
 		 __vss(virtchnl_vlan_filter_list_v2, __vss_byelem, p, m, c),  \
 		 __vss(virtchnl_tc_info, __vss_byelem, p, m, c),	      \
 		 __vss(virtchnl_rdma_qvlist_info, __vss_byelem, p, m, c),     \
+		 __vss(virtchnl_qos_cap_list, __vss_byelem, p, m, c),	      \
+		 __vss(virtchnl_queues_bw_cfg, __vss_byelem, p, m, c),	      \
 		 __vss(virtchnl_rss_key, __vss_byone, p, m, c),		      \
 		 __vss(virtchnl_rss_lut, __vss_byone, p, m, c))
 
@@ -1595,6 +1685,35 @@ virtchnl_vc_validate_vf_msg(struct virtchnl_version_info *ver, u32 v_opcode,
 	case VIRTCHNL_OP_DISABLE_VLAN_INSERTION_V2:
 		valid_len = sizeof(struct virtchnl_vlan_setting);
 		break;
+	case VIRTCHNL_OP_GET_QOS_CAPS:
+		break;
+	case VIRTCHNL_OP_CONFIG_QUEUE_BW:
+		valid_len = virtchnl_queues_bw_cfg_LEGACY_SIZEOF;
+		if (msglen >= valid_len) {
+			struct virtchnl_queues_bw_cfg *q_bw =
+				(struct virtchnl_queues_bw_cfg *)msg;
+
+			valid_len = virtchnl_struct_size(q_bw, cfg,
+							 q_bw->num_queues);
+			if (q_bw->num_queues == 0) {
+				err_msg_format = true;
+				break;
+			}
+		}
+		break;
+	case VIRTCHNL_OP_CONFIG_QUANTA:
+		valid_len = sizeof(struct virtchnl_quanta_cfg);
+		if (msglen >= valid_len) {
+			struct virtchnl_quanta_cfg *q_quanta =
+				(struct virtchnl_quanta_cfg *)msg;
+
+			if (q_quanta->quanta_size == 0 ||
+			    q_quanta->queue_select.num_queues == 0) {
+				err_msg_format = true;
+				break;
+			}
+		}
+		break;
 	/* These are always errors coming from the VF. */
 	case VIRTCHNL_OP_EVENT:
 	case VIRTCHNL_OP_UNKNOWN:
Signature · 1 line
2.34.1
author
#35 lore raw

PATCH v4 2/5 ice: Support VF queue rate limit and quanta size configuration

In reply to Wenjun Wu

Add support to configure VF queue rate limit and quanta size.

For quanta size configuration, the quanta profiles are divided evenly by PF numbers. For each port, the first quanta profile is reserved for default. When VF is asked to set queue quanta size, PF will search for an available profile, change the fields and assigned this profile to the queue.

---
 drivers/net/ethernet/intel/ice/ice.h          |   2 +
 drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
 drivers/net/ethernet/intel/ice/ice_common.c   |  19 ++
 .../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
 drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
 drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
 drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
 drivers/net/ethernet/intel/ice/ice_virtchnl.c | 310 ++++++++++++++++++
 drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
 .../intel/ice/ice_virtchnl_allowlist.c        |   6 +
 10 files changed, 370 insertions(+)
drivers/net/ethernet/intel/ice/ice.h +2 -0
index cf6c961e8d9b..25cdf8623063 100644
@@ -641,6 +641,8 @@ struct ice_pf {
 #define ICE_VF_AGG_NODE_ID_START	65
 #define ICE_MAX_VF_AGG_NODES		32
 	struct ice_agg_node vf_agg_node[ICE_MAX_VF_AGG_NODES];
+
+	u8 num_quanta_prof_used;
 };
 
 extern struct workqueue_struct *ice_lag_wq;
drivers/net/ethernet/intel/ice/ice_base.c +2 -0
index 7fa43827a3f0..2b9319801dc3 100644
@@ -377,6 +377,8 @@ ice_setup_tx_ctx(struct ice_tx_ring *ring, struct ice_tlan_ctx *tlan_ctx, u16 pf
 		break;
 	}
 
+	tlan_ctx->quanta_prof_idx = ring->quanta_prof_id;
+
 	tlan_ctx->tso_ena = ICE_TX_LEGACY;
 	tlan_ctx->tso_qnum = pf_q;
 
drivers/net/ethernet/intel/ice/ice_common.c +19 -0
index 2a19802847a5..86128ca1b7a5 100644
@@ -2463,6 +2463,23 @@ ice_parse_func_caps(struct ice_hw *hw, struct ice_hw_func_caps *func_p,
 	ice_recalc_port_limited_caps(hw, &func_p->common_cap);
 }
 
+/**
+ * ice_func_id_to_logical_id - map from function id to logical pf id
+ * @active_function_bitmap: active function bitmap
+ * @pf_id: function number of device
+ */
+static int ice_func_id_to_logical_id(u32 active_function_bitmap, u8 pf_id)
+{
+	u8 logical_id = 0;
+	u8 i;
+
+	for (i = 0; i < pf_id; i++)
+		if (active_function_bitmap & BIT(i))
+			logical_id++;
+
+	return logical_id;
+}
+
 /**
  * ice_parse_valid_functions_cap - Parse ICE_AQC_CAPS_VALID_FUNCTIONS caps
  * @hw: pointer to the HW struct
@@ -2480,6 +2497,8 @@ ice_parse_valid_functions_cap(struct ice_hw *hw, struct ice_hw_dev_caps *dev_p,
 	dev_p->num_funcs = hweight32(number);
 	ice_debug(hw, ICE_DBG_INIT, "dev caps: num_funcs = %d\n",
 		  dev_p->num_funcs);
+
+	hw->logical_pf_id = ice_func_id_to_logical_id(number, hw->pf_id);
 }
 
 /**
drivers/net/ethernet/intel/ice/ice_hw_autogen.h +8 -0
index 6756f3d51d14..9da94e000394 100644
@@ -6,6 +6,14 @@
 #ifndef _ICE_HW_AUTOGEN_H_
 #define _ICE_HW_AUTOGEN_H_
 
+#define GLCOMM_QUANTA_PROF(_i)			(0x002D2D68 + ((_i) * 4))
+#define GLCOMM_QUANTA_PROF_MAX_INDEX		15
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_S	0
+#define GLCOMM_QUANTA_PROF_QUANTA_SIZE_M	ICE_M(0x3FFF, 0)
+#define GLCOMM_QUANTA_PROF_MAX_CMD_S		16
+#define GLCOMM_QUANTA_PROF_MAX_CMD_M		ICE_M(0xFF, 16)
+#define GLCOMM_QUANTA_PROF_MAX_DESC_S		24
+#define GLCOMM_QUANTA_PROF_MAX_DESC_M		ICE_M(0x3F, 24)
 #define QTX_COMM_DBELL(_DBQM)			(0x002C0000 + ((_DBQM) * 4))
 #define QTX_COMM_HEAD(_DBQM)			(0x000E0000 + ((_DBQM) * 4))
 #define QTX_COMM_HEAD_HEAD_S			0
drivers/net/ethernet/intel/ice/ice_txrx.h +2 -0
index 166413fc33f4..7e152ab5b727 100644
@@ -381,6 +381,8 @@ struct ice_tx_ring {
 	u8 flags;
 	u8 dcb_tc;			/* Traffic class of ring */
 	u8 ptp_tx;
+
+	u16 quanta_prof_id;
 } ____cacheline_internodealigned_in_smp;
 
 static inline bool ice_ring_uses_build_skb(struct ice_rx_ring *ring)
drivers/net/ethernet/intel/ice/ice_type.h +1 -0
index a5429eca4350..504b367f1c77 100644
@@ -850,6 +850,7 @@ struct ice_hw {
 	u8 revision_id;
 
 	u8 pf_id;		/* device profile info */
+	u8 logical_pf_id;
 	enum ice_phy_model phy_model;
 
 	u16 max_burst_size;	/* driver sets this value */
drivers/net/ethernet/intel/ice/ice_vf_lib.h +9 -0
index 48fea6fa0362..7fe81208c62c 100644
@@ -52,6 +52,13 @@ struct ice_mdd_vf_events {
 	u16 last_printed;
 };
 
+struct ice_vf_qs_bw {
+	u32 committed;
+	u32 peak;
+	u16 queue_id;
+	u8 tc;
+};
+
 /* VF operations */
 struct ice_vf_ops {
 	enum ice_disq_rst_src reset_type;
@@ -133,6 +140,8 @@ struct ice_vf {
 
 	/* devlink port data */
 	struct devlink_port devlink_port;
+
+	struct ice_vf_qs_bw qs_bw[ICE_MAX_RSS_QS_PER_VF];
 };
 
 /* Flags for controlling behavior of ice_reset_vf */
drivers/net/ethernet/intel/ice/ice_virtchnl.c +310 -0

346 lines not on this page · raw message

drivers/net/ethernet/intel/ice/ice_virtchnl.h +11 -0
index cd747718de73..0efb9c0f669a 100644
@@ -13,6 +13,13 @@
 /* Restrict number of MAC Addr and VLAN that non-trusted VF can programmed */
 #define ICE_MAX_VLAN_PER_VF		8
 
+#define ICE_DFLT_QUANTA 1024
+#define ICE_MAX_QUANTA_SIZE 4096
+#define ICE_MIN_QUANTA_SIZE 256
+
+#define calc_quanta_desc(x)	\
+	max_t(u16, 12, min_t(u16, 63, (((x) + 66) / 132) * 2 + 4))
+
 /* MAC filters: 1 is reserved for the VF's default/perm_addr/LAA MAC, 1 for
  * broadcast, and 16 for additional unicast/multicast filters
  */
@@ -51,6 +58,10 @@ struct ice_virtchnl_ops {
 	int (*dis_vlan_stripping_v2_msg)(struct ice_vf *vf, u8 *msg);
 	int (*ena_vlan_insertion_v2_msg)(struct ice_vf *vf, u8 *msg);
 	int (*dis_vlan_insertion_v2_msg)(struct ice_vf *vf, u8 *msg);
+	int (*get_qos_caps)(struct ice_vf *vf);
+	int (*cfg_q_tc_map)(struct ice_vf *vf, u8 *msg);
+	int (*cfg_q_bw)(struct ice_vf *vf, u8 *msg);
+	int (*cfg_q_quanta)(struct ice_vf *vf, u8 *msg);
 };
 
 #ifdef CONFIG_PCI_IOV
drivers/net/ethernet/intel/ice/ice_virtchnl_allowlist.c +6 -0
index 7d547fa616fa..2e3f63a429cd 100644
@@ -85,6 +85,11 @@ static const u32 fdir_pf_allowlist_opcodes[] = {
 	VIRTCHNL_OP_ADD_FDIR_FILTER, VIRTCHNL_OP_DEL_FDIR_FILTER,
 };
 
+static const u32 tc_allowlist_opcodes[] = {
+	VIRTCHNL_OP_GET_QOS_CAPS, VIRTCHNL_OP_CONFIG_QUEUE_BW,
+	VIRTCHNL_OP_CONFIG_QUANTA,
+};
+
 struct allowlist_opcode_info {
 	const u32 *opcodes;
 	size_t size;
@@ -105,6 +110,7 @@ static const struct allowlist_opcode_info allowlist_opcodes[] = {
 	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF, adv_rss_pf_allowlist_opcodes),
 	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_FDIR_PF, fdir_pf_allowlist_opcodes),
 	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_VLAN_V2, vlan_v2_allowlist_opcodes),
+	ALLOW_ITEM(VIRTCHNL_VF_OFFLOAD_QOS, tc_allowlist_opcodes),
 };
 
 /**
Signature · 1 line
2.34.1
author
#36 lore raw

PATCH v4 3/5 iavf: Add devlink and devlink port support

In reply to Wenjun Wu

From: Jun Zhang <[email protected]>

To allow user to configure queue bandwidth, devlink port support is added to support devlink port rate API.

Add devlink framework registration/unregistration on iavf driver initialization and remove, and devlink port of DEVLINK_PORT_FLAVOUR_VIRTUAL is created to be associated iavf net device.

---
 drivers/net/ethernet/intel/Kconfig            |  1 +
 drivers/net/ethernet/intel/iavf/Makefile      |  2 +-
 drivers/net/ethernet/intel/iavf/iavf.h        |  5 +
 .../net/ethernet/intel/iavf/iavf_devlink.c    | 94 +++++++++++++++++++
 .../net/ethernet/intel/iavf/iavf_devlink.h    | 16 ++++
 drivers/net/ethernet/intel/iavf/iavf_main.c   | 17 ++++
 6 files changed, 134 insertions(+), 1 deletion(-)
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.c
 create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.h
drivers/net/ethernet/intel/Kconfig +1 -0
index 9bc0a9519899..f916b8ef6acb 100644
@@ -256,6 +256,7 @@ config I40EVF
 	tristate "Intel(R) Ethernet Adaptive Virtual Function support"
 	select IAVF
 	depends on PCI_MSI
+	select NET_DEVLINK
 	help
 	  This driver supports virtual functions for Intel XL710,
 	  X710, X722, XXV710, and all devices advertising support for
drivers/net/ethernet/intel/iavf/Makefile +1 -1
index 9c3e45c54d01..b5d7db97ab8b 100644
@@ -12,5 +12,5 @@ subdir-ccflags-y += -I$(src)
 obj-$(CONFIG_IAVF) += iavf.o
 
 iavf-objs := iavf_main.o iavf_ethtool.o iavf_virtchnl.o iavf_fdir.o \
-	     iavf_adv_rss.o \
+	     iavf_adv_rss.o iavf_devlink.o \
 	     iavf_txrx.o iavf_common.o iavf_adminq.o iavf_client.o
drivers/net/ethernet/intel/iavf/iavf.h +5 -0
index 85fba85fbb23..72a68061e396 100644
@@ -33,9 +33,11 @@
 #include <net/udp.h>
 #include <net/tc_act/tc_gact.h>
 #include <net/tc_act/tc_mirred.h>
+#include <net/devlink.h>
 
 #include "iavf_type.h"
 #include <linux/avf/virtchnl.h>
+#include "iavf_devlink.h"
 #include "iavf_txrx.h"
 #include "iavf_fdir.h"
 #include "iavf_adv_rss.h"
@@ -369,6 +371,9 @@ struct iavf_adapter {
 	struct net_device *netdev;
 	struct pci_dev *pdev;
 
+	struct devlink *devlink;
+	struct devlink_port devlink_port;
+
 	struct iavf_hw hw; /* defined in iavf_type.h */
 
 	enum iavf_state_t state;
drivers/net/ethernet/intel/iavf/iavf_devlink.c new file +94 -0

94 lines not on this page · raw message

drivers/net/ethernet/intel/iavf/iavf_devlink.h new file +16 -0
new file mode 100644
index 000000000000..65e453bbd1a8
@@ -0,0 +1,16 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright (C) 2023 Intel Corporation */
+
+#ifndef _IAVF_DEVLINK_H_
+#define _IAVF_DEVLINK_H_
+
+struct iavf_devlink {
+	struct iavf_adapter *devlink_ref;	/* ref to iavf adapter */
+};
+
+int iavf_devlink_register(struct iavf_adapter *adapter);
+void iavf_devlink_unregister(struct iavf_adapter *adapter);
+int iavf_devlink_port_register(struct iavf_adapter *adapter);
+void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
+
+#endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +17 -0
index b23ca9d80189..3a93d0cac60c 100644
@@ -2038,6 +2038,7 @@ static void iavf_finish_config(struct work_struct *work)
 				iavf_free_rss(adapter);
 				iavf_free_misc_irq(adapter);
 				iavf_reset_interrupt_capability(adapter);
+				iavf_devlink_port_unregister(adapter);
 				iavf_change_state(adapter,
 						  __IAVF_INIT_CONFIG_ADAPTER);
 				goto out;
@@ -2709,6 +2710,9 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 	if (err)
 		goto err_sw_init;
 
+	if (!adapter->netdev_registered)
+		iavf_devlink_port_register(adapter);
+
 	netif_carrier_off(netdev);
 	adapter->link_up = false;
 	netif_tx_stop_all_queues(netdev);
@@ -2750,6 +2754,7 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 err_mem:
 	iavf_free_rss(adapter);
 	iavf_free_misc_irq(adapter);
+	iavf_devlink_port_unregister(adapter);
 err_sw_init:
 	iavf_reset_interrupt_capability(adapter);
 err:
@@ -4960,6 +4965,13 @@ static int iavf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	hw->bus.func = PCI_FUNC(pdev->devfn);
 	hw->bus.bus_id = pdev->bus->number;
 
+	/* Register iavf adapter with devlink */
+	err = iavf_devlink_register(adapter);
+	if (err) {
+		dev_err(&pdev->dev, "devlink registration failed: %d\n", err);
+		goto err_devlink_reg;
+	}
+
 	/* set up the locks for the AQ, do this only once in probe
 	 * and destroy them only once in remove
 	 */
@@ -4998,6 +5010,8 @@ static int iavf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 
 	return 0;
 
+err_devlink_reg:
+	iounmap(hw->hw_addr);
 err_ioremap:
 	destroy_workqueue(adapter->wq);
 err_alloc_wq:
@@ -5140,6 +5154,9 @@ static void iavf_remove(struct pci_dev *pdev)
 				 err);
 	}
 
+	iavf_devlink_port_unregister(adapter);
+	iavf_devlink_unregister(adapter);
+
 	mutex_lock(&adapter->crit_lock);
 	dev_info(&adapter->pdev->dev, "Removing device\n");
 	iavf_change_state(adapter, __IAVF_REMOVE);
Signature · 1 line
2.34.1
author
#37 lore raw

PATCH v4 4/5 iavf: Add devlink port function rate API support

In reply to Wenjun Wu

From: Jun Zhang <[email protected]>

To allow user to configure queue based parameters, devlink port function rate api functions are added for setting node tx_max and tx_share parameters.

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

---
 .../net/ethernet/intel/iavf/iavf_devlink.c    | 258 +++++++++++++++++-
 .../net/ethernet/intel/iavf/iavf_devlink.h    |  21 ++
 drivers/net/ethernet/intel/iavf/iavf_main.c   |   7 +-
 3 files changed, 283 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/iavf/iavf_devlink.c +256 -2
index 1cace56e3f56..732076c2126f 100644
@@ -4,7 +4,261 @@
 #include "iavf.h"
 #include "iavf_devlink.h"
 
-static const struct devlink_ops iavf_devlink_ops = {};
+/**
+ * iavf_devlink_rate_init_rate_tree - export rate tree to devlink rate
+ * @adapter: iavf adapter struct instance
+ *
+ * This function builds Rate Tree based on iavf adapter configuration
+ * and exports it's contents to devlink rate.
+ */
+void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+	struct iavf_dev_rate_node *iavf_r_node;
+	struct iavf_dev_rate_node *iavf_q_node;
+	struct devlink_rate *dl_root_node;
+	struct devlink_rate *dl_tmp_node;
+	int q_num;
+
+	if (!adapter->devlink_port.registered)
+		return;
+
+	iavf_r_node = &dl_priv->root_node;
+	memset(iavf_r_node, 0, sizeof(*iavf_r_node));
+	iavf_r_node->tx_max = adapter->link_speed;
+	strscpy(iavf_r_node->name, "iavf_root", IAVF_RATE_NODE_NAME);
+
+	devl_lock(adapter->devlink);
+	dl_root_node = devl_rate_node_create(adapter->devlink, iavf_r_node,
+					     iavf_r_node->name, NULL);
+	if (!dl_root_node || IS_ERR(dl_root_node))
+		goto err_node;
+
+	iavf_r_node->rate_node = dl_root_node;
+
+	/* Allocate queue nodes, and chain them under root */
+	q_num = adapter->num_active_queues;
+	if (q_num > 0) {
+		int i;
+
+		dl_priv->queue_nodes =
+			kcalloc(q_num, sizeof(struct iavf_dev_rate_node),
+				GFP_KERNEL);
+		if (!dl_priv->queue_nodes)
+			goto err_node;
+
+		for (i = 0; i < q_num; ++i) {
+			iavf_q_node = &dl_priv->queue_nodes[i];
+			snprintf(iavf_q_node->name, IAVF_RATE_NODE_NAME,
+				 "txq_%d", i);
+			dl_tmp_node = devl_rate_node_create(adapter->devlink,
+							    iavf_q_node,
+							    iavf_q_node->name,
+							    dl_root_node);
+			if (!dl_tmp_node || IS_ERR(dl_tmp_node)) {
+				kfree(dl_priv->queue_nodes);
+				goto err_node;
+			}
+
+			iavf_q_node->rate_node = dl_tmp_node;
+			iavf_q_node->tx_max = IAVF_TX_DEFAULT;
+			iavf_q_node->tx_share = 0;
+		}
+	}
+
+	dl_priv->update_in_progress = false;
+	dl_priv->iavf_dev_rate_initialized = true;
+	devl_unlock(adapter->devlink);
+	return;
+err_node:
+	devl_rate_nodes_destroy(adapter->devlink);
+	dl_priv->iavf_dev_rate_initialized = false;
+	devl_unlock(adapter->devlink);
+}
+
+/**
+ * iavf_devlink_rate_deinit_rate_tree - Unregister rate tree with devlink rate
+ * @adapter: iavf adapter struct instance
+ *
+ * This function unregisters the current iavf rate tree registered with devlink
+ * rate and frees resources.
+ */
+void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+
+	if (!dl_priv->iavf_dev_rate_initialized)
+		return;
+
+	devl_lock(adapter->devlink);
+	devl_rate_leaf_destroy(&adapter->devlink_port);
+	devl_rate_nodes_destroy(adapter->devlink);
+	kfree(dl_priv->queue_nodes);
+	devl_unlock(adapter->devlink);
+}
+
+/**
+ * iavf_check_update_config - check if updating queue parameters needed
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ *
+ * This function sets queue bw & quanta size configuration if all
+ * queue parameters are set
+ */
+static int iavf_check_update_config(struct iavf_adapter *adapter,
+				    struct iavf_dev_rate_node *node)
+{
+	/* Update queue bw if any one of the queues have been fully updated by
+	 * user, the other queues either use the default value or the last
+	 * fully updated value
+	 */
+	if (node->tx_update_flag !=
+	    (IAVF_FLAG_TX_MAX_UPDATED | IAVF_FLAG_TX_SHARE_UPDATED))
+		return 0;
+
+	node->tx_max = node->tx_max_temp;
+	node->tx_share = node->tx_share_temp;
+
+	/* Reconfig queue bw only when iavf driver on running state */
+	if (adapter->state != __IAVF_RUNNING)
+		return -EBUSY;
+
+	return 0;
+}
+
+/**
+ * iavf_update_queue_tx_share - sets tx min parameter
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ * @bw: bandwidth in bytes per second
+ * @extack: extended netdev ack structure
+ *
+ * This function sets min BW limit.
+ */
+static int iavf_update_queue_tx_share(struct iavf_adapter *adapter,
+				      struct iavf_dev_rate_node *node,
+				      u64 bw, struct netlink_ext_ack *extack)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+	u64 tx_share_sum = 0;
+
+	/* Keep in kbps */
+	node->tx_share_temp = div_u64(bw, IAVF_RATE_DIV_FACTOR);
+
+	if (ADV_LINK_SUPPORT(adapter)) {
+		int i;
+
+		for (i = 0; i < adapter->num_active_queues; ++i) {
+			if (node != &dl_priv->queue_nodes[i])
+				tx_share_sum +=
+					dl_priv->queue_nodes[i].tx_share;
+			else
+				tx_share_sum += node->tx_share_temp;
+		}
+
+		if (tx_share_sum / 1000  > adapter->link_speed_mbps)
+			return -EINVAL;
+	}
+
+	node->tx_update_flag |= IAVF_FLAG_TX_SHARE_UPDATED;
+	return iavf_check_update_config(adapter, node);
+}
+
+/**
+ * iavf_update_queue_tx_max - sets tx max parameter
+ * @adapter: iavf adapter struct instance
+ * @node: iavf rate node struct instance
+ * @bw: bandwidth in bytes per second
+ * @extack: extended netdev ack structure
+ *
+ * This function sets max BW limit.
+ */
+static int iavf_update_queue_tx_max(struct iavf_adapter *adapter,
+				    struct iavf_dev_rate_node *node,
+				    u64 bw, struct netlink_ext_ack *extack)
+{
+	/* Keep in kbps */
+	node->tx_max_temp = div_u64(bw, IAVF_RATE_DIV_FACTOR);
+	if (ADV_LINK_SUPPORT(adapter)) {
+		if (node->tx_max_temp / 1000 > adapter->link_speed_mbps)
+			return -EINVAL;
+	}
+
+	node->tx_update_flag |= IAVF_FLAG_TX_MAX_UPDATED;
+
+	return iavf_check_update_config(adapter, node);
+}
+
+static int iavf_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node,
+					     void *priv, u64 tx_max,
+					     struct netlink_ext_ack *extack)
+{
+	struct iavf_dev_rate_node *node = priv;
+	struct iavf_devlink *dl_priv;
+	struct iavf_adapter *adapter;
+
+	if (!node)
+		return 0;
+
+	dl_priv = devlink_priv(rate_node->devlink);
+	adapter = dl_priv->devlink_ref;
+
+	/* Check if last update is in progress */
+	if (dl_priv->update_in_progress)
+		return -EBUSY;
+
+	if (node == &dl_priv->root_node)
+		return 0;
+
+	return iavf_update_queue_tx_max(adapter, node, tx_max, extack);
+}
+
+static int iavf_devlink_rate_node_tx_share_set(struct devlink_rate *rate_node,
+					       void *priv, u64 tx_share,
+					       struct netlink_ext_ack *extack)
+{
+	struct iavf_dev_rate_node *node = priv;
+	struct iavf_devlink *dl_priv;
+	struct iavf_adapter *adapter;
+
+	if (!node)
+		return 0;
+
+	dl_priv = devlink_priv(rate_node->devlink);
+	adapter = dl_priv->devlink_ref;
+
+	/* Check if last update is in progress */
+	if (dl_priv->update_in_progress)
+		return -EBUSY;
+
+	if (node == &dl_priv->root_node)
+		return 0;
+
+	return iavf_update_queue_tx_share(adapter, node, tx_share, extack);
+}
+
+static int iavf_devlink_rate_node_del(struct devlink_rate *rate_node,
+				      void *priv,
+				      struct netlink_ext_ack *extack)
+{
+	return -EINVAL;
+}
+
+static int iavf_devlink_set_parent(struct devlink_rate *devlink_rate,
+				   struct devlink_rate *parent,
+				   void *priv, void *parent_priv,
+				   struct netlink_ext_ack *extack)
+{
+	return -EINVAL;
+}
+
+static const struct devlink_ops iavf_devlink_ops = {
+	.rate_node_tx_share_set = iavf_devlink_rate_node_tx_share_set,
+	.rate_node_tx_max_set = iavf_devlink_rate_node_tx_max_set,
+	.rate_node_del = iavf_devlink_rate_node_del,
+	.rate_leaf_parent_set = iavf_devlink_set_parent,
+	.rate_node_parent_set = iavf_devlink_set_parent,
+};
 
 /**
  * iavf_devlink_register - Register devlink interface for this VF adapter
@@ -29,7 +283,7 @@ int iavf_devlink_register(struct iavf_adapter *adapter)
 
 	ref = devlink_priv(devlink);
 	ref->devlink_ref = adapter;
-
+	ref->iavf_dev_rate_initialized = false;
 	devlink_register(devlink);
 
 	return 0;
drivers/net/ethernet/intel/iavf/iavf_devlink.h +21 -0
index 65e453bbd1a8..751e9e093ab1 100644
@@ -4,13 +4,34 @@
 #ifndef _IAVF_DEVLINK_H_
 #define _IAVF_DEVLINK_H_
 
+#define IAVF_RATE_NODE_NAME			12
+struct iavf_dev_rate_node {
+	char name[IAVF_RATE_NODE_NAME];
+	struct devlink_rate *rate_node;
+	u8 tx_update_flag;
+#define IAVF_FLAG_TX_SHARE_UPDATED		BIT(0)
+#define IAVF_FLAG_TX_MAX_UPDATED		BIT(1)
+	u64 tx_max;
+	u64 tx_share;
+	u64 tx_max_temp;
+	u64 tx_share_temp;
+#define IAVF_RATE_DIV_FACTOR			125
+#define IAVF_TX_DEFAULT				100000
+};
+
 struct iavf_devlink {
 	struct iavf_adapter *devlink_ref;	/* ref to iavf adapter */
+	struct iavf_dev_rate_node root_node;
+	struct iavf_dev_rate_node *queue_nodes;
+	bool iavf_dev_rate_initialized;
+	bool update_in_progress;
 };
 
 int iavf_devlink_register(struct iavf_adapter *adapter);
 void iavf_devlink_unregister(struct iavf_adapter *adapter);
 int iavf_devlink_port_register(struct iavf_adapter *adapter);
 void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
+void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter);
+void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter);
 
 #endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +6 -1
index 3a93d0cac60c..699c6375200a 100644
@@ -2038,6 +2038,7 @@ static void iavf_finish_config(struct work_struct *work)
 				iavf_free_rss(adapter);
 				iavf_free_misc_irq(adapter);
 				iavf_reset_interrupt_capability(adapter);
+				iavf_devlink_rate_deinit_rate_tree(adapter);
 				iavf_devlink_port_unregister(adapter);
 				iavf_change_state(adapter,
 						  __IAVF_INIT_CONFIG_ADAPTER);
@@ -2710,8 +2711,10 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 	if (err)
 		goto err_sw_init;
 
-	if (!adapter->netdev_registered)
+	if (!adapter->netdev_registered) {
 		iavf_devlink_port_register(adapter);
+		iavf_devlink_rate_init_rate_tree(adapter);
+	}
 
 	netif_carrier_off(netdev);
 	adapter->link_up = false;
@@ -2754,6 +2757,7 @@ static void iavf_init_config_adapter(struct iavf_adapter *adapter)
 err_mem:
 	iavf_free_rss(adapter);
 	iavf_free_misc_irq(adapter);
+	iavf_devlink_rate_deinit_rate_tree(adapter);
 	iavf_devlink_port_unregister(adapter);
 err_sw_init:
 	iavf_reset_interrupt_capability(adapter);
@@ -5154,6 +5158,7 @@ static void iavf_remove(struct pci_dev *pdev)
 				 err);
 	}
 
+	iavf_devlink_rate_deinit_rate_tree(adapter);
 	iavf_devlink_port_unregister(adapter);
 	iavf_devlink_unregister(adapter);
 
Signature · 1 line
2.34.1
author
#38 lore raw

PATCH v4 5/5 iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

In reply to Wenjun Wu

From: Jun Zhang <[email protected]>

iavf rate tree with root node and queue nodes is created and registered with devlink rate when iavf adapter is configured.

User can configure the tx_max and tx_share of each queue. If any one of the queues have been fully updated by user, i.e. both tx_max and tx_share have been updated for that queue, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

---
 drivers/net/ethernet/intel/iavf/iavf.h        |  14 ++
 .../net/ethernet/intel/iavf/iavf_devlink.c    |  29 +++
 .../net/ethernet/intel/iavf/iavf_devlink.h    |   1 +
 drivers/net/ethernet/intel/iavf/iavf_main.c   |  46 +++-
 .../net/ethernet/intel/iavf/iavf_virtchnl.c   | 231 +++++++++++++++++-
 5 files changed, 317 insertions(+), 4 deletions(-)
drivers/net/ethernet/intel/iavf/iavf.h +14 -0
index 72a68061e396..c04cd1d45be7 100644
@@ -252,6 +252,9 @@ struct iavf_cloud_filter {
 #define IAVF_RESET_WAIT_DETECTED_COUNT 500
 #define IAVF_RESET_WAIT_COMPLETE_COUNT 2000
 
+#define IAVF_MAX_QOS_TC_NUM		8
+#define IAVF_DEFAULT_QUANTA_SIZE	1024
+
 /* board specific private data structure */
 struct iavf_adapter {
 	struct workqueue_struct *wq;
@@ -351,6 +354,9 @@ struct iavf_adapter {
 #define IAVF_FLAG_AQ_DISABLE_CTAG_VLAN_INSERTION	BIT_ULL(36)
 #define IAVF_FLAG_AQ_ENABLE_STAG_VLAN_INSERTION		BIT_ULL(37)
 #define IAVF_FLAG_AQ_DISABLE_STAG_VLAN_INSERTION	BIT_ULL(38)
+#define IAVF_FLAG_AQ_CONFIGURE_QUEUES_BW		BIT_ULL(39)
+#define IAVF_FLAG_AQ_CONFIGURE_QUEUES_QUANTA_SIZE	BIT_ULL(40)
+#define IAVF_FLAG_AQ_GET_QOS_CAPS			BIT_ULL(41)
 
 	/* flags for processing extended capability messages during
 	 * __IAVF_INIT_EXTENDED_CAPS. Each capability exchange requires
@@ -373,6 +379,7 @@ struct iavf_adapter {
 
 	struct devlink *devlink;
 	struct devlink_port devlink_port;
+	bool devlink_update;
 
 	struct iavf_hw hw; /* defined in iavf_type.h */
 
@@ -422,6 +429,8 @@ struct iavf_adapter {
 			       VIRTCHNL_VF_OFFLOAD_FDIR_PF)
 #define ADV_RSS_SUPPORT(_a) ((_a)->vf_res->vf_cap_flags & \
 			     VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF)
+#define QOS_ALLOWED(_a) ((_a)->vf_res->vf_cap_flags & \
+			 VIRTCHNL_VF_OFFLOAD_QOS)
 	struct virtchnl_vf_resource *vf_res; /* incl. all VSIs */
 	struct virtchnl_vsi_resource *vsi_res; /* our LAN VSI */
 	struct virtchnl_version_info pf_version;
@@ -430,6 +439,7 @@ struct iavf_adapter {
 	struct virtchnl_vlan_caps vlan_v2_caps;
 	u16 msg_enable;
 	struct iavf_eth_stats current_stats;
+	struct virtchnl_qos_cap_list *qos_caps;
 	struct iavf_vsi vsi;
 	u32 aq_wait_count;
 	/* RSS stuff */
@@ -576,6 +586,10 @@ void iavf_notify_client_message(struct iavf_vsi *vsi, u8 *msg, u16 len);
 void iavf_notify_client_l2_params(struct iavf_vsi *vsi);
 void iavf_notify_client_open(struct iavf_vsi *vsi);
 void iavf_notify_client_close(struct iavf_vsi *vsi, bool reset);
+void iavf_update_queue_config(struct iavf_adapter *adapter);
+void iavf_configure_queues_bw(struct iavf_adapter *adapter);
+void iavf_configure_queues_quanta_size(struct iavf_adapter *adapter);
+void iavf_get_qos_caps(struct iavf_adapter *adapter);
 void iavf_enable_channels(struct iavf_adapter *adapter);
 void iavf_disable_channels(struct iavf_adapter *adapter);
 void iavf_add_cloud_filter(struct iavf_adapter *adapter);
drivers/net/ethernet/intel/iavf/iavf_devlink.c +29 -0
index 732076c2126f..aefe707aafbc 100644
@@ -97,6 +97,30 @@ void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
 	devl_unlock(adapter->devlink);
 }
 
+/**
+ * iavf_notify_queue_config_complete - notify updating queue completion
+ * @adapter: iavf adapter struct instance
+ *
+ * This function sets the queue configuration update status when all
+ * queue parameters have been sent to PF
+ */
+void iavf_notify_queue_config_complete(struct iavf_adapter *adapter)
+{
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+	int q_num = adapter->num_active_queues;
+	int i;
+
+	/* clean up rate tree update flags*/
+	for (i = 0; i < q_num; i++)
+		if (dl_priv->queue_nodes[i].tx_update_flag ==
+		    (IAVF_FLAG_TX_MAX_UPDATED | IAVF_FLAG_TX_SHARE_UPDATED)) {
+			dl_priv->queue_nodes[i].tx_update_flag = 0;
+			break;
+		}
+
+	dl_priv->update_in_progress = false;
+}
+
 /**
  * iavf_check_update_config - check if updating queue parameters needed
  * @adapter: iavf adapter struct instance
@@ -108,6 +132,8 @@ void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter)
 static int iavf_check_update_config(struct iavf_adapter *adapter,
 				    struct iavf_dev_rate_node *node)
 {
+	struct iavf_devlink *dl_priv = devlink_priv(adapter->devlink);
+
 	/* Update queue bw if any one of the queues have been fully updated by
 	 * user, the other queues either use the default value or the last
 	 * fully updated value
@@ -123,6 +149,8 @@ static int iavf_check_update_config(struct iavf_adapter *adapter,
 	if (adapter->state != __IAVF_RUNNING)
 		return -EBUSY;
 
+	dl_priv->update_in_progress = true;
+	iavf_update_queue_config(adapter);
 	return 0;
 }
 
@@ -281,6 +309,7 @@ int iavf_devlink_register(struct iavf_adapter *adapter)
 	if (!devlink)
 		return -ENOMEM;
 
+	adapter->devlink_update = false;
 	ref = devlink_priv(devlink);
 	ref->devlink_ref = adapter;
 	ref->iavf_dev_rate_initialized = false;
drivers/net/ethernet/intel/iavf/iavf_devlink.h +1 -0
index 751e9e093ab1..4709aa1a0341 100644
@@ -33,5 +33,6 @@ int iavf_devlink_port_register(struct iavf_adapter *adapter);
 void iavf_devlink_port_unregister(struct iavf_adapter *adapter);
 void iavf_devlink_rate_init_rate_tree(struct iavf_adapter *adapter);
 void iavf_devlink_rate_deinit_rate_tree(struct iavf_adapter *adapter);
+void iavf_notify_queue_config_complete(struct iavf_adapter *adapter);
 
 #endif /* _IAVF_DEVLINK_H_ */
drivers/net/ethernet/intel/iavf/iavf_main.c +43 -3

89 lines not on this page · raw message

drivers/net/ethernet/intel/iavf/iavf_virtchnl.c +230 -1

255 lines not on this page · raw message

Signature · 1 line
2.34.1
#39 lore raw
In reply to Wenjun Wu
Quoted 11 lines · Tue, Aug 22, 2023 at 05:39:58AM CEST, [email protected] wrote:

To allow user to configure queue bandwidth, devlink port support is added to support devlink port rate API. [1]

Add devlink framework registration/unregistration on iavf driver initialization and remove, and devlink port of DEVLINK_PORT_FLAVOUR_VIRTUAL is created to be associated iavf netdevice.

iavf rate tree with root node, queue nodes, and leaf node is created and registered with devlink rate when iavf adapter is configured, and if PF indicates support of VIRTCHNL_VF_OFFLOAD_QOS through VF Resource / Capability Exchange.

NACK! Port function is there to configure the VF/SF from the eswitch side. Yet you use it for the configureation of the actual VF, which is clear misuse. Please don't

Quoted 128 lines

[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node parent iavf_root pci/0000:af:01.0/iavf_root: type node

                        +---------+
                        |   root  |
                        +----+----+
                             |
           |-----------------|-----------------|
      +----v----+       +----v----+       +----v----+
      |  txq_0  |       |  txq_1  |       |  txq_x  |
      +----+----+       +----+----+       +----+----+

User can configure the tx_max and tx_share of each queue. Once any one of the queues are fully configured, VIRTCHNL opcodes of VIRTCHNL_OP_CONFIG_QUEUE_BW and VIRTCHNL_OP_CONFIG_QUANTA will be sent to PF to configure queues allocated to VF

Example:

1.To Set the queue tx_share:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_share 100 MBps

2.To Set the queue tx_max:
devlink port function rate set pci/0000:af:01.0 txq_0 tx_max 200 MBps

3.To Show Current devlink port rate info:
devlink port function rate function show
[root@localhost ~]# devlink port function rate show pci/0000:af:01.0/txq_15: type node parent iavf_root pci/0000:af:01.0/txq_14: type node parent iavf_root pci/0000:af:01.0/txq_13: type node parent iavf_root pci/0000:af:01.0/txq_12: type node parent iavf_root pci/0000:af:01.0/txq_11: type node parent iavf_root pci/0000:af:01.0/txq_10: type node parent iavf_root pci/0000:af:01.0/txq_9: type node parent iavf_root pci/0000:af:01.0/txq_8: type node parent iavf_root pci/0000:af:01.0/txq_7: type node parent iavf_root pci/0000:af:01.0/txq_6: type node parent iavf_root pci/0000:af:01.0/txq_5: type node parent iavf_root pci/0000:af:01.0/txq_4: type node parent iavf_root pci/0000:af:01.0/txq_3: type node parent iavf_root pci/0000:af:01.0/txq_2: type node parent iavf_root pci/0000:af:01.0/txq_1: type node parent iavf_root pci/0000:af:01.0/txq_0: type node tx_share 800Mbit tx_max 1600Mbit parent iavf_root pci/0000:af:01.0/iavf_root: type node

[1]https://lore.kernel.org/netdev/[email protected]/

Change log:

v4:
- Rearrange the ice_vf_qs_bw structure, put the largest number first
- Minimize the scope of values
- Remove the unnecessary brackets
- Remove the unnecessary memory allocation.
- Added Error Code and moved devlink registration before aq lock initialization
- Changed devlink registration for error handling in case of allocation failure
- Used kcalloc for object array memory allocation and initialization
- Changed functions & comments for readability

v3:
- Rebase the code
- Changed rate node max/share set function description
- Put variable in local scope

v2:
- Change static array to flex array
- Use struct_size helper
- Align all the error code types in the function
- Move the register field definitions to the right place in the file
- Fix coding style
- Adapted to queue bw cfg and qos cap list virtchnl message with flex array fields
---

Jun Zhang (3):
iavf: Add devlink and devlink port support
iavf: Add devlink port function rate API support
iavf: Add VIRTCHNL Opcodes Support for Queue bw Setting

Wenjun Wu (2):
virtchnl: support queue rate limit and quanta size configuration
ice: Support VF queue rate limit and quanta size configuration

drivers/net/ethernet/intel/Kconfig            |   1 +
drivers/net/ethernet/intel/iavf/Makefile      |   2 +-
drivers/net/ethernet/intel/iavf/iavf.h        |  19 +
.../net/ethernet/intel/iavf/iavf_devlink.c    | 377 ++++++++++++++++++
.../net/ethernet/intel/iavf/iavf_devlink.h    |  38 ++
drivers/net/ethernet/intel/iavf/iavf_main.c   |  64 ++-
.../net/ethernet/intel/iavf/iavf_virtchnl.c   | 231 ++++++++++-
drivers/net/ethernet/intel/ice/ice.h          |   2 +
drivers/net/ethernet/intel/ice/ice_base.c     |   2 +
drivers/net/ethernet/intel/ice/ice_common.c   |  19 +
.../net/ethernet/intel/ice/ice_hw_autogen.h   |   8 +
drivers/net/ethernet/intel/ice/ice_txrx.h     |   2 +
drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
drivers/net/ethernet/intel/ice/ice_vf_lib.h   |   9 +
drivers/net/ethernet/intel/ice/ice_virtchnl.c | 310 ++++++++++++++
drivers/net/ethernet/intel/ice/ice_virtchnl.h |  11 +
.../intel/ice/ice_virtchnl_allowlist.c        |   6 +
include/linux/avf/virtchnl.h                  | 119 ++++++
18 files changed, 1218 insertions(+), 3 deletions(-)

create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.c create mode 100644 drivers/net/ethernet/intel/iavf/iavf_devlink.h

Signature · 1 line
2.34.1
maintainer
#40 lore raw
Quoted 3 lines · On Tue, 22 Aug 2023 08:12:28 +0200 Jiri Pirko wrote:

NACK! Port function is there to configure the VF/SF from the eswitch side. Yet you use it for the configureation of the actual VF, which is clear misuse. Please don't

Stating where they are supposed to configure the rate would be helpful.

#41 lore raw

Re: [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 6 lines · Tue, Aug 22, 2023 at 05:12:55PM CEST, [email protected] wrote:
Quoted 3 lines · On Tue, 22 Aug 2023 08:12:28 +0200 Jiri Pirko wrote:

NACK! Port function is there to configure the VF/SF from the eswitch side. Yet you use it for the configureation of the actual VF, which is clear misuse. Please don't

Stating where they are supposed to configure the rate would be helpful.

TC?

#42 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 7 lines · Jiri Pirko
Quoted 5 lines · Tue, Aug 22, 2023 at 05:12:55PM CEST,[email protected] wrote:
Quoted 3 lines · On Tue, 22 Aug 2023 08:12:28 +0200 Jiri Pirko wrote:

NACK! Port function is there to configure the VF/SF from the eswitch side. Yet you use it for the configureation of the actual VF, which is clear misuse. Please don't

Stating where they are supposed to configure the rate would be helpful.

TC?

Our implementation is an extension to this commit 42c2eb6b1f43 ice: Implement devlink-rate API).

We are setting the Tx max & share rates of individual queues in a VF using the devlink rate API.

Here we are using DEVLINK_PORT_FLAVOUR_VIRTUAL as the attribute for the port to distinguish it from being eswitch.

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

#43 lore raw

Re: [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 10 lines · Wed, Aug 23, 2023 at 09:13:34PM CEST, [email protected] wrote:
Quoted 7 lines · Jiri Pirko
Quoted 5 lines · Tue, Aug 22, 2023 at 05:12:55PM CEST,[email protected] wrote:
Quoted 3 lines · On Tue, 22 Aug 2023 08:12:28 +0200 Jiri Pirko wrote:

NACK! Port function is there to configure the VF/SF from the eswitch side. Yet you use it for the configureation of the actual VF, which is clear misuse. Please don't

Stating where they are supposed to configure the rate would be helpful.

TC?

Quoted 8 lines

Our implementation is an extension to this commit 42c2eb6b1f43 ice: Implement devlink-rate API).

We are setting the Tx max & share rates of individual queues in a VF using the devlink rate API.

Here we are using DEVLINK_PORT_FLAVOUR_VIRTUAL as the attribute for the port to distinguish it from being eswitch.

I understand, that is a wrong object. So again, you should use "function" subobject of devlink port to configure "the other side of the wire", that means the function related to a eswitch port. Here, you are doing it for the VF directly, which is wrong. If you need some rate limiting to be configured on an actual VF, use what you use for any other nic. Offload TC.

#44 lore raw

Re: [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 15 lines · Jiri Pirko
Quoted 14 lines · Wed, Aug 23, 2023 at 09:13:34PM CEST, [email protected] wrote:
Quoted 7 lines · Jiri Pirko
Quoted 5 lines · Tue, Aug 22, 2023 at 05:12:55PM CEST,[email protected] wrote:
Quoted 3 lines · On Tue, 22 Aug 2023 08:12:28 +0200 Jiri Pirko wrote:

NACK! Port function is there to configure the VF/SF from the eswitch side. Yet you use it for the configureation of the actual VF, which is clear misuse. Please don't

Stating where they are supposed to configure the rate would be helpful.

TC?

Our implementation is an extension to this commit 42c2eb6b1f43 ice: Implement devlink-rate API).

We are setting the Tx max & share rates of individual queues in a VF using the devlink rate API.

Quoted 8 lines
Quoted 2 lines

Here we are using DEVLINK_PORT_FLAVOUR_VIRTUAL as the attribute for the port to distinguish it from being eswitch.

I understand, that is a wrong object. So again, you should use "function" subobject of devlink port to configure "the other side of the wire", that means the function related to a eswitch port. Here, you are doing it for the VF directly, which is wrong. If you need some rate limiting to be configured on an actual VF, use what you use for any other nic. Offload TC.

Thanks for detailed explanation and suggestions. Sorry for late reply as it took a bit longer to understand options.

As sysfs has similar rate configuration on per queue basis with tx_maxrate, is it a viable option for our use case (i.e allow user to configure tx rate for each allocated queue in a VF).

Pls aslo see If adding tx_minrate to sysfs tx queue entry is feasible on the current framework.

#45 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Hello Jiri & Jakub,

Thanks for looking into our last patch with devlink API. Really appreciate your candid review.

Following your suggestion, we have looked into 3 tc offload options to support queue rate limiting

#1 mq + matchall + police

#2 mq + tbf

#3 htb

all 3 tc offload options require some level of tc extensions to support VF tx queue rate limiting (tx_maxrate & tx_minrate)

htb offload requires minimal tc changes or no change with similar change done @ driver (we can share patch for review).

After discussing with Maxim Mikityanskiy( https://lore.kernel.org/netdev/[email protected]/ ), looks like sysfs interface with tx_minrate extension could be the option we can take.

Look forward your opinion & guidance. Thanks for your time!

Regards,

Jun

Quoted 34 lines · Xuejun
Quoted 15 lines · Jiri Pirko
Quoted 14 lines · Wed, Aug 23, 2023 at 09:13:34PM CEST, [email protected] wrote:
Quoted 7 lines · Jiri Pirko
Quoted 5 lines · Tue, Aug 22, 2023 at 05:12:55PM CEST,[email protected]  wrote:
Quoted 3 lines · On Tue, 22 Aug 2023 08:12:28 +0200 Jiri Pirko wrote:

NACK! Port function is there to configure the VF/SF from the eswitch side. Yet you use it for the configureation of the actual VF, which is clear misuse. Please don't

Stating where they are supposed to configure the rate would be helpful.

TC?

Our implementation is an extension to this commit 42c2eb6b1f43 ice: Implement devlink-rate API).

We are setting the Tx max & share rates of individual queues in a VF using the devlink rate API.

Quoted 8 lines
Quoted 2 lines

Here we are using DEVLINK_PORT_FLAVOUR_VIRTUAL as the attribute for the port to distinguish it from being eswitch.

I understand, that is a wrong object. So again, you should use "function" subobject of devlink port to configure "the other side of the wire", that means the function related to a eswitch port. Here, you are doing it for the VF directly, which is wrong. If you need some rate limiting to be configured on an actual VF, use what you use for any other nic. Offload TC.

Thanks for detailed explanation and suggestions. Sorry for late reply as it took a bit longer to understand options.

As sysfs has similar rate configuration on per queue basis with tx_maxrate, is it a viable option for our use case (i.e allow user to configure tx rate for each allocated queue in a VF).

Pls aslo see If adding tx_minrate to sysfs tx queue entry is feasible on the current framework.
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

#46 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 9 lines · Fri, Nov 17, 2023 at 06:52:49AM CET, [email protected] wrote:

Hello Jiri & Jakub,

Thanks for looking into our last patch with devlink API. Really appreciate your candid review.

Following your suggestion, we have looked into 3 tc offload options to support queue rate limiting

#1 mq + matchall + police

This looks most suitable. Why it would not work?

Quoted 5 lines

#2 mq + tbf

#3 htb

Quoted 8 lines

all 3 tc offload options require some level of tc extensions to support VF tx queue rate limiting (tx_maxrate & tx_minrate)

htb offload requires minimal tc changes or no change with similar change done @ driver (we can share patch for review).

After discussing with Maxim Mikityanskiy( https://lore.kernel.org/netdev/[email protected]/
), looks like sysfs interface with tx_minrate extension could be the option

I don't undestand how any sysfs know is related to any of the tree tc solutions above.

Quoted 52 lines

we can take.

Look forward your opinion & guidance. Thanks for your time!

Regards,

Jun

Quoted 43 lines · Xuejun
Quoted 22 lines · Jiri Pirko
Quoted 14 lines · Wed, Aug 23, 2023 at 09:13:34PM CEST, [email protected] wrote:
Quoted 9 lines · Jiri Pirko
Quoted 7 lines · Tue, Aug 22, 2023 at 05:12:55PM CEST,[email protected]  wrote:
Quoted 4 lines · On Tue, 22 Aug 2023 08:12:28 +0200 Jiri Pirko wrote:

NACK! Port function is there to configure the VF/SF from the eswitch side. Yet you use it for the configureation of the
actual VF, which is
clear misuse. Please don't

Stating where they are supposed to configure the rate would be helpful.

TC?

Our implementation is an extension to this commit 42c2eb6b1f43
ice: Implement
devlink-rate API).

Quoted 7 lines

We are setting the Tx max & share rates of individual queues in a VF using
the devlink rate API.

Here we are using DEVLINK_PORT_FLAVOUR_VIRTUAL as the attribute for the port
to distinguish it from being eswitch.

Quoted 6 lines

I understand, that is a wrong object. So again, you should use "function" subobject of devlink port to configure "the other side of the wire", that means the function related to a eswitch port. Here, you are doing it for the VF directly, which is wrong. If you need some rate limiting to be configured on an actual VF, use what you use for any other nic. Offload TC.

Thanks for detailed explanation and suggestions. Sorry for late reply as it took a bit longer to understand options.

As sysfs has similar rate configuration on per queue basis with tx_maxrate, is it a viable option for our use case (i.e allow user to configure tx rate for each allocated queue in a VF).

Pls aslo see If adding tx_minrate to sysfs tx queue entry is feasible on the current framework.
_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

maintainer
#47 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 6 lines · Jiri Pirko
Quoted 5 lines · Fri, Nov 17, 2023 at 06:52:49AM CET, [email protected] wrote:

Hello Jiri & Jakub,

Thanks for looking into our last patch with devlink API. Really appreciate your candid review.

Quoted 6 lines
Quoted 4 lines

Following your suggestion, we have looked into 3 tc offload options to support queue rate limiting

#1 mq + matchall + police

This looks most suitable. Why it would not work?

AFAICS, it should work, but it does not look the most suitable to me:
beyond splitting a "simple" task in separate entities, it poses a constraint on the classification performed on the egress device.

Suppose the admin wants to limit the egress bandwidth on the given tx queue _and_ do some application specific packet classification and actions. That would not be possible right?

Thanks!

Paolo

maintainer
#48 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

In reply to Zhang, Xuejun
Quoted 9 lines · Xuejun

Thanks for looking into our last patch with devlink API. Really appreciate your candid review.

Following your suggestion, we have looked into 3 tc offload options to support queue rate limiting

#1 mq + matchall + police

#2 mq + tbf

You can extend mqprio, too, if you wanted.

Quoted 8 lines

#3 htb

all 3 tc offload options require some level of tc extensions to support VF tx queue rate limiting (tx_maxrate & tx_minrate)

htb offload requires minimal tc changes or no change with similar change done @ driver (we can share patch for review).

Quoted 6 lines

After discussing with Maxim Mikityanskiy(
https://lore.kernel.org/netdev/[email protected]/
), looks like sysfs interface with tx_minrate extension could be the option we can take.

Look forward your opinion & guidance. Thanks for your time!

My least favorite thing to do is to configure the same piece of silicon with 4 different SW interfaces. It's okay if we have 4 different uAPIs (user level APIs) but the driver should not be exposed to all these options.

I'm saying 4 but really I can think of 6 ways of setting maxrate :(

IMHO we need to be a bit more realistic about the notion of "offloading the SW thing" for qdiscs specifically. Normally we offload SW constructs to have a fallback and have a clear definition of functionality. I bet most data-centers will use BPF+FQ these days, so the "fallback" argument does not apply. And the "clear definition" when it comes to basic rate limiting is.. moot.

Besides we already have mqprio, sysfs maxrate, sriov ndo, devlink rate, none of which have SW fallback.

So since you asked for my opinion - my opinion is that step 1 is to create a common representation of what we already have and feed it to the drivers via a single interface. I could just be taking sysfs maxrate and feeding it to the driver via the devlink rate interface. If we have the right internals I give 0 cares about what uAPI you pick.

#49 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 40 lines · Jakub Kicinski
Quoted 9 lines · Xuejun

Thanks for looking into our last patch with devlink API. Really appreciate your candid review.

Following your suggestion, we have looked into 3 tc offload options to support queue rate limiting

#1 mq + matchall + police

#2 mq + tbf

You can extend mqprio, too, if you wanted.

Quoted 8 lines

#3 htb

all 3 tc offload options require some level of tc extensions to support VF tx queue rate limiting (tx_maxrate & tx_minrate)

htb offload requires minimal tc changes or no change with similar change done @ driver (we can share patch for review).

Quoted 6 lines

After discussing with Maxim Mikityanskiy(
https://lore.kernel.org/netdev/[email protected]/
), looks like sysfs interface with tx_minrate extension could be the option we can take.

Look forward your opinion & guidance. Thanks for your time!

My least favorite thing to do is to configure the same piece of silicon with 4 different SW interfaces. It's okay if we have 4 different uAPIs (user level APIs) but the driver should not be exposed to all these options.

I'm saying 4 but really I can think of 6 ways of setting maxrate :(

IMHO we need to be a bit more realistic about the notion of "offloading the SW thing" for qdiscs specifically. Normally we offload SW constructs to have a fallback and have a clear definition of functionality. I bet most data-centers will use BPF+FQ these days, so the "fallback" argument does not apply. And the "clear definition" when it comes to basic rate limiting is.. moot.

Quoted 8 lines

Besides we already have mqprio, sysfs maxrate, sriov ndo, devlink rate, none of which have SW fallback.

So since you asked for my opinion - my opinion is that step 1 is to create a common representation of what we already have and feed it to the drivers via a single interface. I could just be taking sysfs maxrate and feeding it to the driver via the devlink rate interface. If we have the right internals I give 0 cares about what uAPI you pick.

There is an existing ndo api for setting tx_maxrate API

int (*ndo_set_tx_maxrate)(struct net_device *dev,
                          int queue_index,
              u32 maxrate);

we could expand and modify the above API with tx_minrate and burst parameters as below int (*ndo_set_tx_rate)(struct net_device *dev,
                 int queue_index,
               int maxrate , int minrate, int burst);

queue_index: tx queue index of net device
maxrate: tx maxrate in mbps
minrate: tx min rate in mbps
burst: data burst in bytes

The proposed API would incur net/core and driver changes as follows
a) existing drivers with ndo_set_tx_maxrate support upgraded to use new ndo_set_tx_rate
b) net sysfs (replacing ndo_set_maxrate with ndo_set_tx_rate with minrate and burst set to 0, -1 means ignore)
c) Keep the existing /sys/class/net/ethx/queues/tx_nn/tx_maxrate as it is currently
d) Add sysfs entry as /sys/class/net/ethx/queues/tx_nn/tx_minrate & /sys/class/net/ethx/queues/tx_nn/burst

Let us know your thoughts.

maintainer
#50 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 9 lines · Xuejun

The proposed API would incur net/core and driver changes as follows
a) existing drivers with ndo_set_tx_maxrate support upgraded to use new ndo_set_tx_rate
b) net sysfs (replacing ndo_set_maxrate with ndo_set_tx_rate with minrate and burst set to 0, -1 means ignore)
c) Keep the existing /sys/class/net/ethx/queues/tx_nn/tx_maxrate as it is currently
d) Add sysfs entry as /sys/class/net/ethx/queues/tx_nn/tx_minrate & /sys/class/net/ethx/queues/tx_nn/burst

You described extending the sysfs API (which the ndo you mention is for) and nothing about converging the other existing APIs.

#51 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 10 lines · Jakub Kicinski
Quoted 9 lines · Xuejun

The proposed API would incur net/core and driver changes as follows
a) existing drivers with ndo_set_tx_maxrate support upgraded to use new ndo_set_tx_rate
b) net sysfs (replacing ndo_set_maxrate with ndo_set_tx_rate with minrate and burst set to 0, -1 means ignore)
c) Keep the existing /sys/class/net/ethx/queues/tx_nn/tx_maxrate as it is currently
d) Add sysfs entry as /sys/class/net/ethx/queues/tx_nn/tx_minrate & /sys/class/net/ethx/queues/tx_nn/burst

Quoted 2 lines

You described extending the sysfs API (which the ndo you mention is for) and nothing about converging the other existing APIs.

This is extension of ndo_set_tx_maxrate to include per queue parameters of tx_minrate and burst.

devlink rate api includes tx_maxrate and tx_minrate, it is intended for port rate configurations.

With regarding to tc mqprio, it is being used to configure queue group per tc.

For sriov ndo ndo_set_vf_rate, that has been used for overall VF rate configuration, not for queue based rate configuration.

It seems there are differences on intent of the aforementioned APIs.

Our use case here is to allow user (i.e @ uAPI) to configure tx rates of max rate & min rate per VF queue.Hence we are inclined to ndo_set_tx_maxrate extension.

maintainer
#52 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 9 lines · Xuejun

This is extension of ndo_set_tx_maxrate to include per queue parameters of tx_minrate and burst.

devlink rate api includes tx_maxrate and tx_minrate, it is intended for port rate configurations.

With regarding to tc mqprio, it is being used to configure queue group per tc.

Quoted 8 lines

For sriov ndo ndo_set_vf_rate, that has been used for overall VF rate configuration, not for queue based rate configuration.

It seems there are differences on intent of the aforementioned APIs.

Our use case here is to allow user (i.e @ uAPI) to configure tx rates of max rate & min rate per VF queue.Hence we are inclined to ndo_set_tx_maxrate extension.

I said:

So since you asked for my opinion - my opinion is that step 1 is to create a common representation of what we already have and feed it to the drivers via a single interface. I could just be taking sysfs maxrate and feeding it to the driver via the devlink rate interface. If we have the right internals I give 0 cares about what uAPI you pick.

https://lore.kernel.org/all/[email protected]/

Again, the first step is creating a common kernel <> driver interface which can be used to send to the driver the configuration from the existing 4 interfaces.

maintainer
#53 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 27 lines · Jakub Kicinski
Quoted 9 lines · Xuejun

This is extension of ndo_set_tx_maxrate to include per queue parameters of tx_minrate and burst.

devlink rate api includes tx_maxrate and tx_minrate, it is intended for port rate configurations.

With regarding to tc mqprio, it is being used to configure queue group per tc.

Quoted 8 lines

For sriov ndo ndo_set_vf_rate, that has been used for overall VF rate configuration, not for queue based rate configuration.

It seems there are differences on intent of the aforementioned APIs.

Our use case here is to allow user (i.e @ uAPI) to configure tx rates of max rate & min rate per VF queue.Hence we are inclined to ndo_set_tx_maxrate extension.

I said:

So since you asked for my opinion - my opinion is that step 1 is to create a common representation of what we already have and feed it to the drivers via a single interface. I could just be taking sysfs maxrate and feeding it to the driver via the devlink rate interface. If we have the right internals I give 0 cares about what uAPI you pick.

Quoted 5 lines

https://lore.kernel.org/all/[email protected]/

Again, the first step is creating a common kernel <> driver interface which can be used to send to the driver the configuration from the existing 4 interfaces.

Together with Simon, I spent some time on the above. We think the ndo_setup_tc(TC_SETUP_QDISC_TBF) hook could be used as common basis for this offloads, with some small extensions (adding a 'max_rate' param, too).

The idea would be:
- 'fixing' sch_btf so that the s/w path became a no-op when h/w offload is enabled
- extend sch_btf to support max rate
- do the relevant ice implementation
- ndo_set_tx_maxrate could be replaced with the mentioned ndo call (the latter interface is a strict super-set of former)
- ndo_set_vf_rate could also be replaced with the mentioned ndo call (with another small extension to the offload data)

I think mqprio deserves it's own separate offload interface, as it covers multiple tasks other than shaping (grouping queues and mapping priority to classes)

In the long run we could have a generic implementation of the ndo_setup_tc(TC_SETUP_QDISC_TBF) in term of devlink rate adding a generic way to fetch the devlink_port instance corresponding to the given netdev and mapping the TBF features to the devlink_rate API.

Not starting this due to what Jiri mentioned [1].

WDYT?

Thanks,

Paolo and Simon

[1] https://lore.kernel.org/netdev/ZORRzEBcUDEjMniz@nanopsycho/

maintainer
#54 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 4 lines · On Thu, 14 Dec 2023 21:29:51 +0100 Paolo Abeni wrote:

Together with Simon, I spent some time on the above. We think the ndo_setup_tc(TC_SETUP_QDISC_TBF) hook could be used as common basis for this offloads, with some small extensions (adding a 'max_rate' param, too).

uAPI aside, why would we use ndo_setup_tc(TC_SETUP_QDISC_TBF) to implement common basis?

Is it not cleaner to have a separate driver API, with its ops and capabilities?

Quoted 14 lines

The idea would be:
- 'fixing' sch_btf so that the s/w path became a no-op when h/w offload is enabled
- extend sch_btf to support max rate
- do the relevant ice implementation
- ndo_set_tx_maxrate could be replaced with the mentioned ndo call (the latter interface is a strict super-set of former)
- ndo_set_vf_rate could also be replaced with the mentioned ndo call (with another small extension to the offload data)

I think mqprio deserves it's own separate offload interface, as it covers multiple tasks other than shaping (grouping queues and mapping priority to classes)

Quoted 6 lines

In the long run we could have a generic implementation of the ndo_setup_tc(TC_SETUP_QDISC_TBF) in term of devlink rate adding a generic way to fetch the devlink_port instance corresponding to the given netdev and mapping the TBF features to the devlink_rate API.

Not starting this due to what Jiri mentioned [1].

Jiri, AFAIU, is against using devlink rate *uAPI* to configure network rate limiting. That's separate from the internal representation.

maintainer
#55 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 11 lines · Jakub Kicinski
Quoted 4 lines · On Thu, 14 Dec 2023 21:29:51 +0100 Paolo Abeni wrote:

Together with Simon, I spent some time on the above. We think the ndo_setup_tc(TC_SETUP_QDISC_TBF) hook could be used as common basis for this offloads, with some small extensions (adding a 'max_rate' param, too).

uAPI aside, why would we use ndo_setup_tc(TC_SETUP_QDISC_TBF) to implement common basis?

Is it not cleaner to have a separate driver API, with its ops and capabilities?

We understand one of the end goal is consolidating the existing rate- related in kernel interfaces. Adding a new one does not feel a good starting to reach that goal, see [1] & [2] ;). ndo_setup_tc() feels like the natural choice for H/W offload and TBF is the existing interface IMHO nearest to the requirements here.

The devlink rate API could be a possible alternative...

Quoted 19 lines
Quoted 19 lines

The idea would be:
- 'fixing' sch_btf so that the s/w path became a no-op when h/w offload is enabled
- extend sch_btf to support max rate
- do the relevant ice implementation
- ndo_set_tx_maxrate could be replaced with the mentioned ndo call (the latter interface is a strict super-set of former)
- ndo_set_vf_rate could also be replaced with the mentioned ndo call (with another small extension to the offload data)

I think mqprio deserves it's own separate offload interface, as it covers multiple tasks other than shaping (grouping queues and mapping priority to classes)

In the long run we could have a generic implementation of the ndo_setup_tc(TC_SETUP_QDISC_TBF) in term of devlink rate adding a generic way to fetch the devlink_port instance corresponding to the given netdev and mapping the TBF features to the devlink_rate API.

Quoted 4 lines
Quoted 1 line

Not starting this due to what Jiri mentioned [1].

Jiri, AFAIU, is against using devlink rate *uAPI* to configure network rate limiting. That's separate from the internal representation.

... with a couples of caveats:

1) AFAICS devlink (and/or devlink_port) does not have fine grained, per queue representation and intel want to be able to configure shaping on per queue basis. I think/hope we don't want to bring the discussion to extending the devlink interface with queue support, I fear that will block us for a long time. Perhaps I’m missing or misunderstanding something here. Otherwise in retrospect this looks like a reasonable point to completely avoid devlink here.

2) My understanding of Jiri statement was more restrictive. @Jiri it would great if could share your genuine interpretation: are you ok with using the devlink_port rate API as a basis to replace ndo_set_tx_maxrate() (via dev->devlink_port->devlink->) and possibly ndo_set_vf_rate(). Note the given the previous point, this option would still feel problematic.

Cheers,

Paolo

[1] https://xkcd.com/927/
[2] https://www.youtube.com/watch?v=f8kO_L-pDwo

maintainer
#56 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 7 lines · Paolo Abeni

1) AFAICS devlink (and/or devlink_port) does not have fine grained, per queue representation and intel want to be able to configure shaping on per queue basis. I think/hope we don't want to bring the discussion to extending the devlink interface with queue support, I fear that will block us for a long time. Perhaps I’m missing or misunderstanding something here. Otherwise in retrospect this looks like a reasonable point to completely avoid devlink here.

Note to self: never send a message to the ML before my 3rd morning coffee.

This thread started with Intel trying to using devlink rate for their use-case, apparently slamming my doubt above.

My understanding is that in the patches the queue devlink <> queue relationship was kept inside the driver and not exposed to the devlink level.

If we want to use the devlink rate api to replace e.g. ndo_set_tx_maxrate, we would need a devlink queue(id) or the like, hence this point.

Cheer,

Paolo

#57 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

In reply to Paolo Abeni
Quoted 13 lines · Fri, Dec 15, 2023 at 12:06:52PM CET, [email protected] wrote:
Quoted 11 lines · Jakub Kicinski
Quoted 4 lines · On Thu, 14 Dec 2023 21:29:51 +0100 Paolo Abeni wrote:

Together with Simon, I spent some time on the above. We think the ndo_setup_tc(TC_SETUP_QDISC_TBF) hook could be used as common basis for this offloads, with some small extensions (adding a 'max_rate' param, too).

uAPI aside, why would we use ndo_setup_tc(TC_SETUP_QDISC_TBF) to implement common basis?

Is it not cleaner to have a separate driver API, with its ops and capabilities?

Quoted 7 lines

We understand one of the end goal is consolidating the existing rate- related in kernel interfaces. Adding a new one does not feel a good starting to reach that goal, see [1] & [2] ;). ndo_setup_tc() feels like the natural choice for H/W offload and TBF is the existing interface IMHO nearest to the requirements here.

The devlink rate API could be a possible alternative...

Again, devlink rate was introduced for the rate configuration of the entity that is not present (by netdev for example) on a host. If we have netdev, let's use it.

Quoted 35 lines
Quoted 20 lines
Quoted 20 lines

The idea would be:
- 'fixing' sch_btf so that the s/w path became a no-op when h/w offload is enabled
- extend sch_btf to support max rate
- do the relevant ice implementation
- ndo_set_tx_maxrate could be replaced with the mentioned ndo call (the latter interface is a strict super-set of former)
- ndo_set_vf_rate could also be replaced with the mentioned ndo call (with another small extension to the offload data)

I think mqprio deserves it's own separate offload interface, as it covers multiple tasks other than shaping (grouping queues and mapping priority to classes)

In the long run we could have a generic implementation of the ndo_setup_tc(TC_SETUP_QDISC_TBF) in term of devlink rate adding a generic way to fetch the devlink_port instance corresponding to the given netdev and mapping the TBF features to the devlink_rate API.

Quoted 4 lines
Quoted 1 line

Not starting this due to what Jiri mentioned [1].

Jiri, AFAIU, is against using devlink rate *uAPI* to configure network rate limiting. That's separate from the internal representation.

... with a couples of caveats:

1) AFAICS devlink (and/or devlink_port) does not have fine grained, per queue representation and intel want to be able to configure shaping on per queue basis. I think/hope we don't want to bring the discussion to extending the devlink interface with queue support, I fear that will block us for a long time. Perhaps I’m missing or misunderstanding something here. Otherwise in retrospect this looks like a reasonable point to completely avoid devlink here.

Quoted 4 lines

2) My understanding of Jiri statement was more restrictive. @Jiri it would great if could share your genuine interpretation: are you ok with using the devlink_port rate API as a basis to replace ndo_set_tx_maxrate() (via dev->devlink_port->devlink->) and possibly

Does not make any sense to me.

Quoted 1 line

ndo_set_vf_rate(). Note the given the previous point, this option would

ndo_set_vf_rate() (and the rest of ndo_[gs]et_vf_*() ndo) is the legacy way. Devlink rate replaced that when switchdev eswich mode is configured by:

$ sudo devlink dev eswitch set pci/0000:08:00.1 mode switchdev

In drivers, ndo_set_vf_rate() and devlink rate are implemented in the same way. See mlx5 for example:
mlx5_esw_qos_set_vport_rate() mlx5_esw_devlink_rate_leaf_tx_share_set()

Quoted 9 lines

still feel problematic.

Cheers,

Paolo

[1] https://xkcd.com/927/
[2] https://www.youtube.com/watch?v=f8kO_L-pDwo

maintainer
#58 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

In reply to Paolo Abeni
Quoted 9 lines · On Fri, 15 Dec 2023 12:06:52 +0100 Paolo Abeni wrote:
Quoted 5 lines

uAPI aside, why would we use ndo_setup_tc(TC_SETUP_QDISC_TBF) to implement common basis?

Is it not cleaner to have a separate driver API, with its ops and capabilities?

We understand one of the end goal is consolidating the existing rate- related in kernel interfaces. Adding a new one does not feel a good starting to reach that goal, see [1] & [2] ;)

ndo_setup_tc(TC_SETUP_QDISC_TBF) is a new API, too, very much so. These attempts to build on top of existing interfaces with small tweaks is leading us to a fragmented and incompatible driver landscape.

I explained before (perhaps on the netdev call) - Qdiscs have two different offload models. "local" and "switchdev", here we want "local" AFAIU and TBF only has "switchdev" offload (take a look at the enqueue method and which drivers support it today).

"We'll extend TBF" is very much adding a new API. You'll have to add "local offload" support in TBF and no NIC driver today supports it. I'm not saying TBF is bad, but I disagree that it's any different than a new NDO for all practical purposes.

Quoted 2 lines

ndo_setup_tc() feels like the natural choice for H/W offload and TBF is the existing interface IMHO nearest to the requirements here.

I question whether something as basic as scheduling and ACLs should follow the "offload SW constructs" mantra. You are exposed to more diverse users so please don't hesitate to disagree, but AFAICT the transparent offload (user installs SW constructs and if offload is available - offload, otherwise use SW is good enough) has not played out like we have hoped.

Let's figure out what is the abstract model of scheduling / shaping within a NIC that we want to target. And then come up with a way of representing it in SW. Not which uAPI we can shoehorn into the use case.

maintainer
#59 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 4 lines · Jakub Kicinski

I explained before (perhaps on the netdev call) - Qdiscs have two different offload models. "local" and "switchdev", here we want "local" AFAIU and TBF only has "switchdev" offload (take a look at the enqueue method and which drivers support it today).

I must admit the above is not yet clear to me.

I initially thought you meant that "local" offloads properly reconfigure the S/W datapath so that locally generated traffic would go through the expected processing (e.g. shaping) just once, while with "switchdev" offload locally generated traffic will see shaping done both by the S/W and the H/W[1].

Reading the above I now think you mean that local offloads has only effect for locally generated traffic but not on traffic forwarded via eswitch, and vice versa[2].

The drivers I looked at did not show any clue (to me).

FTR, I think that [1] is a bug worth fixing and [2] is evil ;)

Could you please clarify which is the difference exactly between them?

Quoted 15 lines

"We'll extend TBF" is very much adding a new API. You'll have to add "local offload" support in TBF and no NIC driver today supports it. I'm not saying TBF is bad, but I disagree that it's any different than a new NDO for all practical purposes.

Quoted 2 lines

ndo_setup_tc() feels like the natural choice for H/W offload and TBF is the existing interface IMHO nearest to the requirements here.

I question whether something as basic as scheduling and ACLs should follow the "offload SW constructs" mantra. You are exposed to more diverse users so please don't hesitate to disagree, but AFAICT the transparent offload (user installs SW constructs and if offload is available - offload, otherwise use SW is good enough) has not played out like we have hoped.

Quoted 4 lines

Let's figure out what is the abstract model of scheduling / shaping within a NIC that we want to target. And then come up with a way of representing it in SW. Not which uAPI we can shoehorn into the use case.

I thought the model was quite well defined since the initial submission from Intel, and is quite simple: expose TX shaping on per tx queue basis, with min rate, max rate (in bps) and burst (in bytes).

I think that making it more complex (e.g. with nesting, pkt overhead, etc) we will still not cover every possible use case and will add considerable complexity.

Quoted 1 line

Cheers,

Paolo

maintainer
#60 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

Quoted 18 lines · On Mon, 18 Dec 2023 21:12:35 +0100 Paolo Abeni wrote:
Quoted 4 lines · Jakub Kicinski

I explained before (perhaps on the netdev call) - Qdiscs have two different offload models. "local" and "switchdev", here we want "local" AFAIU and TBF only has "switchdev" offload (take a look at the enqueue method and which drivers support it today).

I must admit the above is not yet clear to me.

I initially thought you meant that "local" offloads properly reconfigure the S/W datapath so that locally generated traffic would go through the expected processing (e.g. shaping) just once, while with "switchdev" offload locally generated traffic will see shaping done both by the S/W and the H/W[1].

Reading the above I now think you mean that local offloads has only effect for locally generated traffic but not on traffic forwarded via eswitch, and vice versa[2].

Quoted 5 lines

The drivers I looked at did not show any clue (to me).

FTR, I think that [1] is a bug worth fixing and [2] is evil ;)

Could you please clarify which is the difference exactly between them?

The practical difference which you can see in the code is that "locally offloaded" qdiscs will act like a FIFO in the SW path (at least to some extent). While "switchdev" offload qdiscs act exactly the same regardless of the offload.

Neither is wrong, they are offloading different things. Qdisc offload on a representor (switchdev) offloads from the switch perspective, i.e. "ingress to host". Only fallback goes thru SW path, and should be negligible.

"Local" offload can be implemented as admission control (and is sometimes work conserving), it's on the "real" interface, it's egress, and doesn't take part in forwarding.

Quoted 7 lines
Quoted 7 lines

I question whether something as basic as scheduling and ACLs should follow the "offload SW constructs" mantra. You are exposed to more diverse users so please don't hesitate to disagree, but AFAICT the transparent offload (user installs SW constructs and if offload is available - offload, otherwise use SW is good enough) has not played out like we have hoped.

Quoted 8 lines
Quoted 4 lines

Let's figure out what is the abstract model of scheduling / shaping within a NIC that we want to target. And then come up with a way of representing it in SW. Not which uAPI we can shoehorn into the use case.

I thought the model was quite well defined since the initial submission from Intel, and is quite simple: expose TX shaping on per tx queue basis, with min rate, max rate (in bps) and burst (in bytes).

For some definition of a model, I guess. Given the confusion about switchdev vs local (ingress vs egress) - I can't agree that the model is well defined :(

What I mean is - given piece of functionality like "Tx queue shaping" you can come up with a reasonable uAPI that you can hijack and it makes sense to you. But someone else (switchdev ingress) can chose the same API to implement a different offload. Not to mention that yet another person will chose a different API to implement the same things as you :(

Off the top of my head we have at least:

- Tx DMA admission control / scheduling (which Tx DMA queue will NIC pull from)
- Rx DMA scheduling (which Rx queue will NIC push to)

- buffer/queue configuration (how to deal with buildup of packets in NIC SRAM, usually mostly for ingress)
- NIC buffer configuration (how the SRAM is allocated to queues)

- policers in the NIC forwarding logic

Let's extend this list so that it covers all reasonable NIC designs, and them work on mapping how each of them is configured?

Quoted 3 lines

I think that making it more complex (e.g. with nesting, pkt overhead, etc) we will still not cover every possible use case and will add considerable complexity.

#61 lore raw

Re: [Intel-wired-lan] [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

In reply to Jakub Kicinski
Quoted 31 lines · Fri, Dec 15, 2023 at 02:46:04AM CET, [email protected] wrote:
Quoted 4 lines · On Thu, 14 Dec 2023 21:29:51 +0100 Paolo Abeni wrote:

Together with Simon, I spent some time on the above. We think the ndo_setup_tc(TC_SETUP_QDISC_TBF) hook could be used as common basis for this offloads, with some small extensions (adding a 'max_rate' param, too).

uAPI aside, why would we use ndo_setup_tc(TC_SETUP_QDISC_TBF) to implement common basis?

Is it not cleaner to have a separate driver API, with its ops and capabilities?

Quoted 19 lines

The idea would be:
- 'fixing' sch_btf so that the s/w path became a no-op when h/w offload is enabled
- extend sch_btf to support max rate
- do the relevant ice implementation
- ndo_set_tx_maxrate could be replaced with the mentioned ndo call (the latter interface is a strict super-set of former)
- ndo_set_vf_rate could also be replaced with the mentioned ndo call (with another small extension to the offload data)

I think mqprio deserves it's own separate offload interface, as it covers multiple tasks other than shaping (grouping queues and mapping priority to classes)

In the long run we could have a generic implementation of the ndo_setup_tc(TC_SETUP_QDISC_TBF) in term of devlink rate adding a generic way to fetch the devlink_port instance corresponding to the given netdev and mapping the TBF features to the devlink_rate API.

Quoted 4 lines
Quoted 1 line

Not starting this due to what Jiri mentioned [1].

Jiri, AFAIU, is against using devlink rate *uAPI* to configure network rate limiting. That's separate from the internal representation.

Devlink rate was introduced for configuring port functions that are connected to eswitch port. I don't see any reason to extend it for configuration of netdev on the host. We have netdev instance and other means to do it.

maintainer
#62 lore raw

Re: [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

In reply to Jiri Pirko

Hi,

please allow me to revive this old thread...

Quoted 20 lines · Jiri Pirko
Quoted 20 lines
Quoted 18 lines · Wed, Aug 23, 2023 at 09:13:34PM CEST, [email protected] wrote:
Quoted 18 lines · Jiri Pirko
Quoted 7 lines
Quoted 7 lines
Quoted 5 lines · Tue, Aug 22, 2023 at 05:12:55PM CEST,[email protected] wrote:
Quoted 5 lines · On Tue, 22 Aug 2023 08:12:28 +0200 Jiri Pirko wrote:
Quoted 3 lines
Quoted 3 lines

NACK! Port function is there to configure the VF/SF from the eswitch side. Yet you use it for the configureation of the actual VF, which is clear misuse. Please don't

Stating where they are supposed to configure the rate would be helpful.

TC?

Our implementation is an extension to this commit 42c2eb6b1f43 ice: Implement devlink-rate API).

We are setting the Tx max & share rates of individual queues in a VF using the devlink rate API.

Here we are using DEVLINK_PORT_FLAVOUR_VIRTUAL as the attribute for the port to distinguish it from being eswitch.

Quoted 6 lines
Quoted 6 lines

I understand, that is a wrong object. So again, you should use "function" subobject of devlink port to configure "the other side of the wire", that means the function related to a eswitch port. Here, you are doing it for the VF directly, which is wrong. If you need some rate limiting to be configured on an actual VF, use what you use for any other nic. Offload TC.

I have a doubt WRT the above. Don't we need something more/different here? I mean: a possible intent is limiting the amount of resources (BW in the VF -> esw direction) that the application owing the VF could use.

If that is enforced via TC on the VF side (say, a different namespace or VM), the VF user could circumvent such limit - changing the tc configuration - either by mistake or malicious action.

Looking at the thing from a different perspective, the TX B/W on the VF side is the RX B/W on the eswitch side, so the same effect could be obtained with a (new/different) API formally touching only eswitch side object. WDYT?

Thanks,

Paolo

#63 lore raw

Re: [PATCH iwl-next v4 0/5] iavf: Add devlink and devlink rate support'

In reply to Jiri Pirko
Quoted 7 lines · Jiri Pirko
Quoted 5 lines · Tue, Aug 22, 2023 at 05:12:55PM CEST, [email protected] wrote:
Quoted 3 lines · On Tue, 22 Aug 2023 08:12:28 +0200 Jiri Pirko wrote:

NACK! Port function is there to configure the VF/SF from the eswitch side. Yet you use it for the configureation of the actual VF, which is clear misuse. Please don't

Stating where they are supposed to configure the rate would be helpful.

TC?

Our implementation is an extension to this commit 42c2eb6b1f43 ice: Implement devlink-rate API).

We are setting the Tx max & share rates of individual queues in a VF using the devlink rate API.

Here we are using DEVLINK_PORT_FLAVOUR_VIRTUAL as the attribute for the port to distinguish it from being eswitch.

[resend in plain text only]

_______________________________________________
Intel-wired-lan mailing list
[email protected]
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan