Skip to contentSkip to conversation

ice: support devlink subfunction

PATCH Superseded Michal Swiatkowski 3 participants 25 messages Apr 17, 2024 – Apr 19, 2024

Subsystems (3)

v4 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

v4 cover ice: support devlink subfunction

Hi,

This is second patchset split from subfunction patchset [1]. Important changes from previous version:
* remove unnecessary checks for devlink port type
* link correct devlink port to subfunction netdev

Follow up patchset with subfunction port representor will be the last patchset for subfunction implementation in ice. It is a little unpleasant to split it like that, because devlink port should be linked with port representor netdev. In this patchset use devlink port without linking it. It will be done correctly in the follow up when subfunction port representor is available.

Currently ice driver does not allow creating more than one networking device per physical function. The only way to have more hardware backed netdev is to use SR-IOV.

Following patchset adds support for devlink port API. For each new pcisf type port, driver allocates new VSI, configures all resources needed, including dynamically MSIX vectors, program rules and registers new netdev.

This series supports only one Tx/Rx queue pair per subfunction.

Example commands:
devlink port add pci/0000:31:00.1 flavour pcisf pfnum 1 sfnum 1000 devlink port function set pci/0000:31:00.1/1 hw_addr 00:00:00:00:03:14 devlink port function set pci/0000:31:00.1/1 state active devlink port function del pci/0000:31:00.1/1

v3 --> v4: [4]
* get rid of unnecessary checks
* fix cosmetic issues (whitespaces, wording)
* move unrelated chunks to separate patch
* move implementing activation/deactivation to last patch

v2 --> v3: [3]
* fix building issue between the patches; allocating devlink for subfunction need to include base subfunction header
* fix kdoc issues

v1 --> v2: [2]
* use correct parameters in ice_devlink_alloc() thanks to Mateusz

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

Michal Swiatkowski (1):
ice: treat subfunction VSI the same as PF VSI

Piotr Raczynski (7):
ice: add new VSI type for subfunctions
ice: export ice ndo_ops functions
ice: add basic devlink subfunctions support
ice: allocate devlink for subfunction
ice: base subfunction aux driver
ice: implement netdev for subfunction
ice: allow to activate and deactivate subfunction

 drivers/net/ethernet/intel/ice/Makefile       |   1 +
 .../net/ethernet/intel/ice/devlink/devlink.c  |  50 +-
 .../net/ethernet/intel/ice/devlink/devlink.h  |   1 +
 .../ethernet/intel/ice/devlink/devlink_port.c | 503 ++++++++++++++++++
 .../ethernet/intel/ice/devlink/devlink_port.h |  38 ++
 drivers/net/ethernet/intel/ice/ice.h          |  12 +
 drivers/net/ethernet/intel/ice/ice_base.c     |   5 +-
 drivers/net/ethernet/intel/ice/ice_dcb_lib.c  |   1 +
 drivers/net/ethernet/intel/ice/ice_lib.c      |  52 +-
 drivers/net/ethernet/intel/ice/ice_lib.h      |   3 +
 drivers/net/ethernet/intel/ice/ice_main.c     |  66 ++-
 drivers/net/ethernet/intel/ice/ice_sf_eth.c   | 327 ++++++++++++
 drivers/net/ethernet/intel/ice/ice_sf_eth.h   |  33 ++
 drivers/net/ethernet/intel/ice/ice_type.h     |   1 +
 drivers/net/ethernet/intel/ice/ice_xsk.c      |   2 +-
 15 files changed, 1048 insertions(+), 47 deletions(-)
 create mode 100644 drivers/net/ethernet/intel/ice/ice_sf_eth.c
 create mode 100644 drivers/net/ethernet/intel/ice/ice_sf_eth.h
Signature · 1 line
2.42.0
#1 lore raw

PATCH v4 1/8 ice: add new VSI type for subfunctions

From: Piotr Raczynski <[email protected]>

Add required plumbing for new VSI type dedicated to devlink subfunctions. Make sure that the vsi is properly configured and destroyed. Also allow loading XDP and AF_XDP sockets.

The first implementation of devlink subfunctions supports only one Tx/Rx queue pair per given subfunction.

---
 drivers/net/ethernet/intel/ice/ice_base.c    |  5 +++-
 drivers/net/ethernet/intel/ice/ice_dcb_lib.c |  1 +
 drivers/net/ethernet/intel/ice/ice_lib.c     | 25 ++++++++++++++++++--
 drivers/net/ethernet/intel/ice/ice_main.c    |  7 ++++--
 drivers/net/ethernet/intel/ice/ice_type.h    |  1 +
 drivers/net/ethernet/intel/ice/ice_xsk.c     |  2 +-
 6 files changed, 35 insertions(+), 6 deletions(-)
drivers/net/ethernet/intel/ice/ice_base.c +4 -1
index 687f6cb2b917..bf1a085c7087 100644
@@ -330,6 +330,9 @@ ice_setup_tx_ctx(struct ice_tx_ring *ring, struct ice_tlan_ctx *tlan_ctx, u16 pf
 		tlan_ctx->vmvf_num = hw->func_caps.vf_base_id + vsi->vf->vf_id;
 		tlan_ctx->vmvf_type = ICE_TLAN_CTX_VMVF_TYPE_VF;
 		break;
+	case ICE_VSI_SF:
+		tlan_ctx->vmvf_type = ICE_TLAN_CTX_VMVF_TYPE_VMQ;
+		break;
 	default:
 		return;
 	}
@@ -526,7 +529,7 @@ static int ice_vsi_cfg_rxq(struct ice_rx_ring *ring)
 
 	ring->rx_buf_len = ring->vsi->rx_buf_len;
 
-	if (ring->vsi->type == ICE_VSI_PF) {
+	if (ring->vsi->type == ICE_VSI_PF || ring->vsi->type == ICE_VSI_SF) {
 		if (!xdp_rxq_info_is_reg(&ring->xdp_rxq)) {
 			err = __xdp_rxq_info_reg(&ring->xdp_rxq, ring->netdev,
 						 ring->q_index,
drivers/net/ethernet/intel/ice/ice_dcb_lib.c +1 -0
index a94e7072b570..a7c510832824 100644
@@ -187,6 +187,7 @@ void ice_vsi_set_dcb_tc_cfg(struct ice_vsi *vsi)
 		vsi->tc_cfg.numtc = ice_dcb_get_num_tc(cfg);
 		break;
 	case ICE_VSI_CHNL:
+	case ICE_VSI_SF:
 		vsi->tc_cfg.ena_tc = BIT(ice_get_first_droptc(vsi));
 		vsi->tc_cfg.numtc = 1;
 		break;
drivers/net/ethernet/intel/ice/ice_lib.c +23 -2
index d06e7c82c433..06909bf3b517 100644
@@ -20,6 +20,8 @@ const char *ice_vsi_type_str(enum ice_vsi_type vsi_type)
 		return "ICE_VSI_PF";
 	case ICE_VSI_VF:
 		return "ICE_VSI_VF";
+	case ICE_VSI_SF:
+		return "ICE_VSI_SF";
 	case ICE_VSI_CTRL:
 		return "ICE_VSI_CTRL";
 	case ICE_VSI_CHNL:
@@ -141,6 +143,7 @@ static void ice_vsi_set_num_desc(struct ice_vsi *vsi)
 {
 	switch (vsi->type) {
 	case ICE_VSI_PF:
+	case ICE_VSI_SF:
 	case ICE_VSI_CTRL:
 	case ICE_VSI_LB:
 		/* a user could change the values of num_[tr]x_desc using
@@ -207,6 +210,12 @@ static void ice_vsi_set_num_qs(struct ice_vsi *vsi)
 					   max_t(int, vsi->alloc_rxq,
 						 vsi->alloc_txq));
 		break;
+	case ICE_VSI_SF:
+		vsi->alloc_txq = 1;
+		vsi->alloc_rxq = 1;
+		vsi->num_q_vectors = 1;
+		vsi->irq_dyn_alloc = true;
+		break;
 	case ICE_VSI_VF:
 		if (vf->num_req_qs)
 			vf->num_vf_qs = vf->num_req_qs;
@@ -566,6 +575,7 @@ ice_vsi_alloc_def(struct ice_vsi *vsi, struct ice_channel *ch)
 
 	switch (vsi->type) {
 	case ICE_VSI_PF:
+	case ICE_VSI_SF:
 		/* Setup default MSIX irq handler for VSI */
 		vsi->irq_handler = ice_msix_clean_rings;
 		break;
@@ -894,6 +904,11 @@ static void ice_vsi_set_rss_params(struct ice_vsi *vsi)
 					      max_rss_size);
 		vsi->rss_lut_type = ICE_LUT_PF;
 		break;
+	case ICE_VSI_SF:
+		vsi->rss_table_size = ICE_LUT_VSI_SIZE;
+		vsi->rss_size = min_t(u16, num_online_cpus(), max_rss_size);
+		vsi->rss_lut_type = ICE_LUT_VSI;
+		break;
 	case ICE_VSI_VF:
 		/* VF VSI will get a small RSS table.
 		 * For VSI_LUT, LUT size should be set to 64 bytes.
@@ -1141,6 +1156,7 @@ static void ice_set_rss_vsi_ctx(struct ice_vsi_ctx *ctxt, struct ice_vsi *vsi)
 		lut_type = ICE_AQ_VSI_Q_OPT_RSS_LUT_PF;
 		break;
 	case ICE_VSI_VF:
+	case ICE_VSI_SF:
 		/* VF VSI will gets a small RSS table which is a VSI LUT type */
 		lut_type = ICE_AQ_VSI_Q_OPT_RSS_LUT_VSI;
 		break;
@@ -1219,6 +1235,7 @@ static int ice_vsi_init(struct ice_vsi *vsi, u32 vsi_flags)
 	case ICE_VSI_PF:
 		ctxt->flags = ICE_AQ_VSI_TYPE_PF;
 		break;
+	case ICE_VSI_SF:
 	case ICE_VSI_CHNL:
 		ctxt->flags = ICE_AQ_VSI_TYPE_VMDQ2;
 		break;
@@ -2100,6 +2117,7 @@ static void ice_set_agg_vsi(struct ice_vsi *vsi)
 	case ICE_VSI_CHNL:
 	case ICE_VSI_LB:
 	case ICE_VSI_PF:
+	case ICE_VSI_SF:
 		max_agg_nodes = ICE_MAX_PF_AGG_NODES;
 		agg_node_id_start = ICE_PF_AGG_NODE_ID_START;
 		agg_node_iter = &pf->pf_agg_node[0];
@@ -2271,6 +2289,7 @@ ice_vsi_cfg_def(struct ice_vsi *vsi, struct ice_vsi_cfg_params *params)
 
 	switch (vsi->type) {
 	case ICE_VSI_CTRL:
+	case ICE_VSI_SF:
 	case ICE_VSI_PF:
 		ret = ice_vsi_alloc_q_vectors(vsi);
 		if (ret)
@@ -2662,7 +2681,8 @@ int ice_ena_vsi(struct ice_vsi *vsi, bool locked)
 
 	clear_bit(ICE_VSI_NEEDS_RESTART, vsi->state);
 
-	if (vsi->netdev && vsi->type == ICE_VSI_PF) {
+	if (vsi->netdev && (vsi->type == ICE_VSI_PF ||
+			    vsi->type == ICE_VSI_SF)) {
 		if (netif_running(vsi->netdev)) {
 			if (!locked)
 				rtnl_lock();
@@ -2691,7 +2711,8 @@ void ice_dis_vsi(struct ice_vsi *vsi, bool locked)
 
 	set_bit(ICE_VSI_NEEDS_RESTART, vsi->state);
 
-	if (vsi->type == ICE_VSI_PF && vsi->netdev) {
+	if (vsi->netdev && (vsi->type == ICE_VSI_PF ||
+			    vsi->type == ICE_VSI_SF)) {
 		if (netif_running(vsi->netdev)) {
 			if (!locked)
 				rtnl_lock();
drivers/net/ethernet/intel/ice/ice_main.c +5 -2
index 9d3f6683339e..8a4d9029434f 100644
@@ -2946,6 +2946,9 @@ int ice_vsi_determine_xdp_res(struct ice_vsi *vsi)
 	if (avail < cpus / 2)
 		return -ENOMEM;
 
+	if (vsi->type == ICE_VSI_SF)
+		avail = vsi->alloc_txq;
+
 	vsi->num_xdp_txq = min_t(u16, avail, cpus);
 
 	if (vsi->num_xdp_txq < cpus)
@@ -3061,8 +3064,8 @@ static int ice_xdp(struct net_device *dev, struct netdev_bpf *xdp)
 	struct ice_netdev_priv *np = netdev_priv(dev);
 	struct ice_vsi *vsi = np->vsi;
 
-	if (vsi->type != ICE_VSI_PF) {
-		NL_SET_ERR_MSG_MOD(xdp->extack, "XDP can be loaded only on PF VSI");
+	if (vsi->type != ICE_VSI_PF && vsi->type != ICE_VSI_SF) {
+		NL_SET_ERR_MSG_MOD(xdp->extack, "XDP can be loaded only on PF or SF VSI");
 		return -EINVAL;
 	}
 
drivers/net/ethernet/intel/ice/ice_type.h +1 -0
index f0796a93f428..cdb923c1664a 100644
@@ -150,6 +150,7 @@ enum ice_vsi_type {
 	ICE_VSI_CTRL = 3,	/* equates to ICE_VSI_PF with 1 queue pair */
 	ICE_VSI_CHNL = 4,
 	ICE_VSI_LB = 6,
+	ICE_VSI_SF = 9,
 };
 
 struct ice_link_status {
drivers/net/ethernet/intel/ice/ice_xsk.c +1 -1
index aa81d1162b81..64f4927efa29 100644
@@ -288,7 +288,7 @@ ice_xsk_pool_enable(struct ice_vsi *vsi, struct xsk_buff_pool *pool, u16 qid)
 {
 	int err;
 
-	if (vsi->type != ICE_VSI_PF)
+	if (vsi->type != ICE_VSI_PF && vsi->type != ICE_VSI_SF)
 		return -EINVAL;
 
 	if (qid >= vsi->netdev->real_num_rx_queues ||
Signature · 1 line
2.42.0
#2 lore raw

PATCH v4 2/8 ice: export ice ndo_ops functions

In reply to Michal Swiatkowski

From: Piotr Raczynski <[email protected]>

Make some of the netdevice_ops functions visible from outside for another VSI type created netdev.

---
 drivers/net/ethernet/intel/ice/ice.h      |  8 +++++
 drivers/net/ethernet/intel/ice/ice_lib.c  | 22 ++++++++++++++
 drivers/net/ethernet/intel/ice/ice_lib.h  |  1 +
 drivers/net/ethernet/intel/ice/ice_main.c | 37 ++++-------------------
 4 files changed, 37 insertions(+), 31 deletions(-)
drivers/net/ethernet/intel/ice/ice.h +8 -0
index 67a3236ab1fc..5f7757a44b72 100644
@@ -982,6 +982,14 @@ void ice_unload(struct ice_pf *pf);
 void ice_adv_lnk_speed_maps_init(void);
 int ice_init_dev(struct ice_pf *pf);
 void ice_deinit_dev(struct ice_pf *pf);
+int ice_change_mtu(struct net_device *netdev, int new_mtu);
+void ice_tx_timeout(struct net_device *netdev, unsigned int txqueue);
+int ice_xdp(struct net_device *dev, struct netdev_bpf *xdp);
+void ice_set_netdev_features(struct net_device *netdev);
+int ice_vlan_rx_add_vid(struct net_device *netdev, __be16 proto, u16 vid);
+int ice_vlan_rx_kill_vid(struct net_device *netdev, __be16 proto, u16 vid);
+void ice_get_stats64(struct net_device *netdev,
+		     struct rtnl_link_stats64 *stats);
 
 /**
  * ice_set_rdma_cap - enable RDMA support
drivers/net/ethernet/intel/ice/ice_lib.c +22 -0
index 06909bf3b517..cff0bb6ba428 100644
@@ -2846,6 +2846,28 @@ void ice_vsi_set_napi_queues(struct ice_vsi *vsi)
 		ice_q_vector_set_napi_queues(vsi->q_vectors[i]);
 }
 
+/**
+ * ice_napi_add - register NAPI handler for the VSI
+ * @vsi: VSI for which NAPI handler is to be registered
+ *
+ * This function is only called in the driver's load path. Registering the NAPI
+ * handler is done in ice_vsi_alloc_q_vector() for all other cases (i.e. resume,
+ * reset/rebuild, etc.)
+ */
+void ice_napi_add(struct ice_vsi *vsi)
+{
+	int v_idx;
+
+	if (!vsi->netdev)
+		return;
+
+	ice_for_each_q_vector(vsi, v_idx) {
+		netif_napi_add(vsi->netdev, &vsi->q_vectors[v_idx]->napi,
+			       ice_napi_poll);
+		__ice_q_vector_set_napi_queues(vsi->q_vectors[v_idx], false);
+	}
+}
+
 /**
  * ice_vsi_release - Delete a VSI and free its resources
  * @vsi: the VSI being removed
drivers/net/ethernet/intel/ice/ice_lib.h +1 -0
index 9cd23afe5f15..a57213062b7f 100644
@@ -90,6 +90,7 @@ void __ice_q_vector_set_napi_queues(struct ice_q_vector *q_vector, bool locked);
 void ice_q_vector_set_napi_queues(struct ice_q_vector *q_vector);
 
 void ice_vsi_set_napi_queues(struct ice_vsi *vsi);
+void ice_napi_add(struct ice_vsi *vsi);
 
 int ice_vsi_release(struct ice_vsi *vsi);
 
drivers/net/ethernet/intel/ice/ice_main.c +6 -31
index 8a4d9029434f..058d2a8836b0 100644
@@ -3059,7 +3059,7 @@ static int ice_xdp_safe_mode(struct net_device __always_unused *dev,
  * @dev: netdevice
  * @xdp: XDP command
  */
-static int ice_xdp(struct net_device *dev, struct netdev_bpf *xdp)
+int ice_xdp(struct net_device *dev, struct netdev_bpf *xdp)
 {
 	struct ice_netdev_priv *np = netdev_priv(dev);
 	struct ice_vsi *vsi = np->vsi;
@@ -3518,28 +3518,6 @@ static int ice_req_irq_msix_misc(struct ice_pf *pf)
 	return 0;
 }
 
-/**
- * ice_napi_add - register NAPI handler for the VSI
- * @vsi: VSI for which NAPI handler is to be registered
- *
- * This function is only called in the driver's load path. Registering the NAPI
- * handler is done in ice_vsi_alloc_q_vector() for all other cases (i.e. resume,
- * reset/rebuild, etc.)
- */
-static void ice_napi_add(struct ice_vsi *vsi)
-{
-	int v_idx;
-
-	if (!vsi->netdev)
-		return;
-
-	ice_for_each_q_vector(vsi, v_idx) {
-		netif_napi_add(vsi->netdev, &vsi->q_vectors[v_idx]->napi,
-			       ice_napi_poll);
-		__ice_q_vector_set_napi_queues(vsi->q_vectors[v_idx], false);
-	}
-}
-
 /**
  * ice_set_ops - set netdev and ethtools ops for the given netdev
  * @vsi: the VSI associated with the new netdev
@@ -3573,7 +3551,7 @@ static void ice_set_ops(struct ice_vsi *vsi)
  * ice_set_netdev_features - set features for the given netdev
  * @netdev: netdev instance
  */
-static void ice_set_netdev_features(struct net_device *netdev)
+void ice_set_netdev_features(struct net_device *netdev)
 {
 	struct ice_pf *pf = ice_netdev_to_pf(netdev);
 	bool is_dvm_ena = ice_is_dvm_ena(&pf->hw);
@@ -3755,8 +3733,7 @@ ice_lb_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi)
  *
  * net_device_ops implementation for adding VLAN IDs
  */
-static int
-ice_vlan_rx_add_vid(struct net_device *netdev, __be16 proto, u16 vid)
+int ice_vlan_rx_add_vid(struct net_device *netdev, __be16 proto, u16 vid)
 {
 	struct ice_netdev_priv *np = netdev_priv(netdev);
 	struct ice_vsi_vlan_ops *vlan_ops;
@@ -3818,8 +3795,7 @@ ice_vlan_rx_add_vid(struct net_device *netdev, __be16 proto, u16 vid)
  *
  * net_device_ops implementation for removing VLAN IDs
  */
-static int
-ice_vlan_rx_kill_vid(struct net_device *netdev, __be16 proto, u16 vid)
+int ice_vlan_rx_kill_vid(struct net_device *netdev, __be16 proto, u16 vid)
 {
 	struct ice_netdev_priv *np = netdev_priv(netdev);
 	struct ice_vsi_vlan_ops *vlan_ops;
@@ -7052,7 +7028,6 @@ void ice_update_pf_stats(struct ice_pf *pf)
  * @netdev: network interface device structure
  * @stats: main device statistics structure
  */
-static
 void ice_get_stats64(struct net_device *netdev, struct rtnl_link_stats64 *stats)
 {
 	struct ice_netdev_priv *np = netdev_priv(netdev);
@@ -7725,7 +7700,7 @@ static void ice_rebuild(struct ice_pf *pf, enum ice_reset_req reset_type)
  *
  * Returns 0 on success, negative on failure
  */
-static int ice_change_mtu(struct net_device *netdev, int new_mtu)
+int ice_change_mtu(struct net_device *netdev, int new_mtu)
 {
 	struct ice_netdev_priv *np = netdev_priv(netdev);
 	struct ice_vsi *vsi = np->vsi;
@@ -8149,7 +8124,7 @@ ice_bridge_setlink(struct net_device *dev, struct nlmsghdr *nlh,
  * @netdev: network interface device structure
  * @txqueue: Tx queue
  */
-static void ice_tx_timeout(struct net_device *netdev, unsigned int txqueue)
+void ice_tx_timeout(struct net_device *netdev, unsigned int txqueue)
 {
 	struct ice_netdev_priv *np = netdev_priv(netdev);
 	struct ice_tx_ring *tx_ring = NULL;
Signature · 1 line
2.42.0
#3 lore raw

PATCH v4 3/8 ice: add basic devlink subfunctions support

In reply to Michal Swiatkowski

From: Piotr Raczynski <[email protected]>

Implement devlink port handlers responsible for ethernet type devlink subfunctions. Create subfunction devlink port and setup all resources needed for a subfunction netdev to operate. Configure new VSI for each new subfunction, initialize and configure interrupts and Tx/Rx resources. Set correct MAC filters and create new netdev.

For now, subfunction is limited to only one Tx/Rx queue pair.

Only allocate new subfunction VSI with devlink port new command. This makes sure that real resources are configured only when a new subfunction gets activated. Allocate and free subfunction MSIX interrupt vectors using new API calls with pci_msix_alloc_irq_at and pci_msix_free_irq.

Support both automatic and manual subfunction numbers. If no subfunction number is provided, use xa_alloc to pick a number automatically. This will find the first free index and use that as the number. This reduces burden on users in the simple case where a specific number is not required. It may also be slightly faster to check that a number exists since xarray lookup should be faster than a linear scan of the dyn_ports xarray.

---
 .../net/ethernet/intel/ice/devlink/devlink.c  |   3 +
 .../ethernet/intel/ice/devlink/devlink_port.c | 347 ++++++++++++++++++
 .../ethernet/intel/ice/devlink/devlink_port.h |  30 ++
 drivers/net/ethernet/intel/ice/ice.h          |   4 +
 drivers/net/ethernet/intel/ice/ice_lib.c      |   5 +-
 drivers/net/ethernet/intel/ice/ice_lib.h      |   2 +
 drivers/net/ethernet/intel/ice/ice_main.c     |   9 +-
 7 files changed, 397 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/ice/devlink/devlink.c +3 -0
index b179eaccc774..661af04c8eef 100644
@@ -6,6 +6,7 @@
 #include "ice.h"
 #include "ice_lib.h"
 #include "devlink.h"
+#include "devlink_port.h"
 #include "ice_eswitch.h"
 #include "ice_fw_update.h"
 #include "ice_dcb_lib.h"
@@ -1281,6 +1282,8 @@ static const struct devlink_ops ice_devlink_ops = {
 
 	.rate_leaf_parent_set = ice_devlink_set_parent,
 	.rate_node_parent_set = ice_devlink_set_parent,
+
+	.port_new = ice_devlink_port_new,
 };
 
 static int
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +347 -0

356 lines not on this page · raw message

drivers/net/ethernet/intel/ice/devlink/devlink_port.h +30 -0
index 9223bcdb6444..196db6806019 100644
@@ -4,9 +4,39 @@
 #ifndef _DEVLINK_PORT_H_
 #define _DEVLINK_PORT_H_
 
+#include "../ice.h"
+
+/**
+ * struct ice_dynamic_port - Track dynamically added devlink port instance
+ * @hw_addr: the HW address for this port
+ * @active: true if the port has been activated
+ * @devlink_port: the associated devlink port structure
+ * @pf: pointer to the PF private structure
+ * @vsi: the VSI associated with this port
+ *
+ * An instance of a dynamically added devlink port. Each port flavour
+ */
+struct ice_dynamic_port {
+	u8 hw_addr[ETH_ALEN];
+	u8 active: 1;
+	struct devlink_port devlink_port;
+	struct ice_pf *pf;
+	struct ice_vsi *vsi;
+};
+
+void ice_dealloc_all_dynamic_ports(struct ice_pf *pf);
+
 int ice_devlink_create_pf_port(struct ice_pf *pf);
 void ice_devlink_destroy_pf_port(struct ice_pf *pf);
 int ice_devlink_create_vf_port(struct ice_vf *vf);
 void ice_devlink_destroy_vf_port(struct ice_vf *vf);
 
+#define ice_devlink_port_to_dyn(p) \
+	container_of(port, struct ice_dynamic_port, devlink_port)
+
+int
+ice_devlink_port_new(struct devlink *devlink,
+		     const struct devlink_port_new_attrs *new_attr,
+		     struct netlink_ext_ack *extack,
+		     struct devlink_port **devlink_port);
 #endif /* _DEVLINK_PORT_H_ */
drivers/net/ethernet/intel/ice/ice.h +4 -0
index 5f7757a44b72..edc93d14464c 100644
@@ -649,6 +649,9 @@ struct ice_pf {
 	struct ice_eswitch eswitch;
 	struct ice_esw_br_port *br_port;
 
+	struct xarray dyn_ports;
+	struct xarray sf_nums;
+
 #define ICE_INVALID_AGG_NODE_ID		0
 #define ICE_PF_AGG_NODE_ID_START	1
 #define ICE_MAX_PF_AGG_NODES		32
@@ -905,6 +908,7 @@ int ice_vsi_open(struct ice_vsi *vsi);
 void ice_set_ethtool_ops(struct net_device *netdev);
 void ice_set_ethtool_repr_ops(struct net_device *netdev);
 void ice_set_ethtool_safe_mode_ops(struct net_device *netdev);
+void ice_set_ethtool_sf_ops(struct net_device *netdev);
 u16 ice_get_avail_txq_count(struct ice_pf *pf);
 u16 ice_get_avail_rxq_count(struct ice_pf *pf);
 int ice_vsi_recfg_qs(struct ice_vsi *vsi, int new_rx, int new_tx, bool locked);
drivers/net/ethernet/intel/ice/ice_lib.c +3 -2
index cff0bb6ba428..3e5c0651534b 100644
@@ -7,6 +7,7 @@
 #include "ice_lib.h"
 #include "ice_fltr.h"
 #include "ice_dcb_lib.h"
+#include "ice_type.h"
 #include "ice_vsi_vlan_ops.h"
 
 /**
@@ -440,7 +441,7 @@ static int ice_vsi_alloc_ring_stats(struct ice_vsi *vsi)
  * This deallocates the VSI's queue resources, removes it from the PF's
  * VSI array if necessary, and deallocates the VSI
  */
-static void ice_vsi_free(struct ice_vsi *vsi)
+void ice_vsi_free(struct ice_vsi *vsi)
 {
 	struct ice_pf *pf = NULL;
 	struct device *dev;
@@ -612,7 +613,7 @@ ice_vsi_alloc_def(struct ice_vsi *vsi, struct ice_channel *ch)
  *
  * returns a pointer to a VSI on success, NULL on failure.
  */
-static struct ice_vsi *ice_vsi_alloc(struct ice_pf *pf)
+struct ice_vsi *ice_vsi_alloc(struct ice_pf *pf)
 {
 	struct device *dev = ice_pf_to_dev(pf);
 	struct ice_vsi *vsi = NULL;
drivers/net/ethernet/intel/ice/ice_lib.h +2 -0
index a57213062b7f..198cfd1fdca0 100644
@@ -103,6 +103,8 @@ void ice_dis_vsi(struct ice_vsi *vsi, bool locked);
 
 int ice_vsi_rebuild(struct ice_vsi *vsi, u32 vsi_flags);
 int ice_vsi_cfg(struct ice_vsi *vsi, struct ice_vsi_cfg_params *params);
+struct ice_vsi *ice_vsi_alloc(struct ice_pf *pf);
+void ice_vsi_free(struct ice_vsi *vsi);
 
 bool ice_is_reset_in_progress(unsigned long *state);
 int ice_wait_for_reset(struct ice_pf *pf, unsigned long timeout);
drivers/net/ethernet/intel/ice/ice_main.c +8 -1
index 058d2a8836b0..9bbb7b328ae9 100644
@@ -3964,6 +3964,9 @@ static void ice_deinit_pf(struct ice_pf *pf)
 
 	if (pf->ptp.clock)
 		ptp_clock_unregister(pf->ptp.clock);
+
+	xa_destroy(&pf->dyn_ports);
+	xa_destroy(&pf->sf_nums);
 }
 
 /**
@@ -4057,6 +4060,9 @@ static int ice_init_pf(struct ice_pf *pf)
 	hash_init(pf->vfs.table);
 	ice_mbx_init_snapshot(&pf->hw);
 
+	xa_init(&pf->dyn_ports);
+	xa_init(&pf->sf_nums);
+
 	return 0;
 }
 
@@ -5376,6 +5382,7 @@ static void ice_remove(struct pci_dev *pdev)
 		ice_remove_arfs(pf);
 
 	devl_lock(priv_to_devlink(pf));
+	ice_dealloc_all_dynamic_ports(pf);
 	ice_deinit_devlink(pf);
 
 	ice_unload(pf);
@@ -6670,7 +6677,7 @@ static int ice_up_complete(struct ice_vsi *vsi)
 
 	if (vsi->port_info &&
 	    (vsi->port_info->phy.link_info.link_info & ICE_AQ_LINK_UP) &&
-	    vsi->netdev && vsi->type == ICE_VSI_PF) {
+	    ((vsi->netdev && vsi->type == ICE_VSI_PF))) {
 		ice_print_link_msg(vsi, true);
 		netif_tx_start_all_queues(vsi->netdev);
 		netif_carrier_on(vsi->netdev);
Signature · 1 line
2.42.0
#4 lore raw

PATCH v4 4/8 ice: treat subfunction VSI the same as PF VSI

In reply to Michal Swiatkowski

When subfunction VSI is open the same code as for PF VSI should be executed. Also when up is complete. Reflect that in code by adding subfunction VSI to consideration.

In case of stopping, PF doesn't have additional tasks, so the same is with subfunction VSI.

---
 drivers/net/ethernet/intel/ice/ice_main.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/net/ethernet/intel/ice/ice_main.c +3 -2
index 9bbb7b328ae9..29552598ddb6 100644
@@ -6677,7 +6677,8 @@ static int ice_up_complete(struct ice_vsi *vsi)
 
 	if (vsi->port_info &&
 	    (vsi->port_info->phy.link_info.link_info & ICE_AQ_LINK_UP) &&
-	    ((vsi->netdev && vsi->type == ICE_VSI_PF))) {
+	    ((vsi->netdev && vsi->type == ICE_VSI_PF) ||
+	     (vsi->netdev && vsi->type == ICE_VSI_SF))) {
 		ice_print_link_msg(vsi, true);
 		netif_tx_start_all_queues(vsi->netdev);
 		netif_carrier_on(vsi->netdev);
@@ -7375,7 +7376,7 @@ int ice_vsi_open(struct ice_vsi *vsi)
 
 	ice_vsi_cfg_netdev_tc(vsi, vsi->tc_cfg.ena_tc);
 
-	if (vsi->type == ICE_VSI_PF) {
+	if (vsi->type == ICE_VSI_PF || vsi->type == ICE_VSI_SF) {
 		/* Notify the stack of the actual queue counts. */
 		err = netif_set_real_num_tx_queues(vsi->netdev, vsi->num_txq);
 		if (err)
Signature · 1 line
2.42.0
#5 lore raw

PATCH v4 5/8 ice: allocate devlink for subfunction

In reply to Michal Swiatkowski

From: Piotr Raczynski <[email protected]>

Make devlink allocation function generic to use it for PF and for SF.

Add function for SF devlink port creation. It will be used in next patch.

Create header file for subfunction device. Define subfunction device structure there as it is needed for devlink allocation and port creation.

---
 .../net/ethernet/intel/ice/devlink/devlink.c  | 47 ++++++++++++++---
 .../net/ethernet/intel/ice/devlink/devlink.h  |  1 +
 .../ethernet/intel/ice/devlink/devlink_port.c | 50 +++++++++++++++++++
 .../ethernet/intel/ice/devlink/devlink_port.h |  3 ++
 drivers/net/ethernet/intel/ice/ice_sf_eth.h   | 21 ++++++++
 5 files changed, 116 insertions(+), 6 deletions(-)
 create mode 100644 drivers/net/ethernet/intel/ice/ice_sf_eth.h
drivers/net/ethernet/intel/ice/devlink/devlink.c +41 -6
index 661af04c8eef..5a78bf08e731 100644
@@ -10,6 +10,7 @@
 #include "ice_eswitch.h"
 #include "ice_fw_update.h"
 #include "ice_dcb_lib.h"
+#include "ice_sf_eth.h"
 
 /* context for devlink info version reporting */
 struct ice_info_ctx {
@@ -1286,6 +1287,8 @@ static const struct devlink_ops ice_devlink_ops = {
 	.port_new = ice_devlink_port_new,
 };
 
+static const struct devlink_ops ice_sf_devlink_ops;
+
 static int
 ice_devlink_enable_roce_get(struct devlink *devlink, u32 id,
 			    struct devlink_param_gset_ctx *ctx)
@@ -1417,18 +1420,23 @@ static void ice_devlink_free(void *devlink_ptr)
 }
 
 /**
- * ice_allocate_pf - Allocate devlink and return PF structure pointer
+ * ice_devlink_alloc - Allocate devlink and return devlink priv pointer
  * @dev: the device to allocate for
+ * @priv_size: size of the priv memory
+ * @ops: pointer to devlink ops for this device
+ *
+ * Allocate a devlink instance for this device and return the private pointer
+ * The devlink memory is kept track of through devres by adding an action to
+ * remove it when unwinding.
  *
- * Allocate a devlink instance for this device and return the private area as
- * the PF structure. The devlink memory is kept track of through devres by
- * adding an action to remove it when unwinding.
+ * Return: void pointer to allocated memory
  */
-struct ice_pf *ice_allocate_pf(struct device *dev)
+static void *ice_devlink_alloc(struct device *dev, size_t priv_size,
+			       const struct devlink_ops *ops)
 {
 	struct devlink *devlink;
 
-	devlink = devlink_alloc(&ice_devlink_ops, sizeof(struct ice_pf), dev);
+	devlink = devlink_alloc(ops, priv_size, dev);
 	if (!devlink)
 		return NULL;
 
@@ -1439,6 +1447,33 @@ struct ice_pf *ice_allocate_pf(struct device *dev)
 	return devlink_priv(devlink);
 }
 
+/**
+ * ice_allocate_pf - Allocate devlink and return PF structure pointer
+ * @dev: the device to allocate for
+ *
+ * Allocate a devlink instance for PF.
+ *
+ * Return: void pointer to allocated memory
+ */
+struct ice_pf *ice_allocate_pf(struct device *dev)
+{
+	return ice_devlink_alloc(dev, sizeof(struct ice_pf), &ice_devlink_ops);
+}
+
+/**
+ * ice_allocate_sf - Allocate devlink and return SF structure pointer
+ * @dev: the device to allocate for
+ *
+ * Allocate a devlink instance for SF.
+ *
+ * Return: void pointer to allocated memory
+ */
+struct ice_sf_priv *ice_allocate_sf(struct device *dev)
+{
+	return ice_devlink_alloc(dev, sizeof(struct ice_sf_priv),
+				 &ice_sf_devlink_ops);
+}
+
 /**
  * ice_devlink_register - Register devlink interface for this PF
  * @pf: the PF to register the devlink for.
drivers/net/ethernet/intel/ice/devlink/devlink.h +1 -0
index d291c0e2e17b..1b2a5980d5e8 100644
@@ -5,6 +5,7 @@
 #define _ICE_DEVLINK_H_
 
 struct ice_pf *ice_allocate_pf(struct device *dev);
+struct ice_sf_priv *ice_allocate_sf(struct device *dev);
 
 void ice_devlink_register(struct ice_pf *pf);
 void ice_devlink_unregister(struct ice_pf *pf);
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +50 -0

56 lines not on this page · raw message

drivers/net/ethernet/intel/ice/devlink/devlink_port.h +3 -0
index 196db6806019..33f07cec6bd6 100644
@@ -5,6 +5,7 @@
 #define _DEVLINK_PORT_H_
 
 #include "../ice.h"
+#include "ice_sf_eth.h"
 
 /**
  * struct ice_dynamic_port - Track dynamically added devlink port instance
@@ -30,6 +31,8 @@ int ice_devlink_create_pf_port(struct ice_pf *pf);
 void ice_devlink_destroy_pf_port(struct ice_pf *pf);
 int ice_devlink_create_vf_port(struct ice_vf *vf);
 void ice_devlink_destroy_vf_port(struct ice_vf *vf);
+int ice_devlink_create_sf_dev_port(struct ice_sf_dev *sf_dev);
+void ice_devlink_destroy_sf_dev_port(struct ice_sf_dev *sf_dev);
 
 #define ice_devlink_port_to_dyn(p) \
 	container_of(port, struct ice_dynamic_port, devlink_port)
drivers/net/ethernet/intel/ice/ice_sf_eth.h new file +21 -0
new file mode 100644
index 000000000000..a08f8b2bceef
@@ -0,0 +1,21 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright (c) 2024, Intel Corporation. */
+
+#ifndef _ICE_SF_ETH_H_
+#define _ICE_SF_ETH_H_
+
+#include <linux/auxiliary_bus.h>
+#include "ice.h"
+
+struct ice_sf_dev {
+	struct auxiliary_device adev;
+	struct ice_dynamic_port *dyn_port;
+	struct ice_sf_priv *priv;
+};
+
+struct ice_sf_priv {
+	struct ice_sf_dev *dev;
+	struct devlink_port devlink_port;
+};
+
+#endif /* _ICE_SF_ETH_H_ */
Signature · 1 line
2.42.0
#6 lore raw

Re: [iwl-next v4 5/8] ice: allocate devlink for subfunction

Quoted 1 line · Wed, Apr 17, 2024 at 04:20:25PM CEST, [email protected] wrote:

From: Piotr Raczynski <[email protected]>

[...]

drivers/net/ethernet/intel/ice/devlink/devlink.c struct ice_pf *ice_allocate_pf(struct device *dev) · line 1471

Quoted 9 lines
drivers/net/ethernet/intel/ice/devlink/devlink.c +9 -0
+/**
+ * ice_allocate_sf - Allocate devlink and return SF structure pointer
+ * @dev: the device to allocate for
+ *
+ * Allocate a devlink instance for SF.
+ *
+ * Return: void pointer to allocated memory
+ */
+struct ice_sf_priv *ice_allocate_sf(struct device *dev)

This is devlink instance for SF auxdev. Please make sure it is properly linked with the devlink port instance using devl_port_fn_devlink_set() See mlx5 implementation for inspiration.

Quoted 5 lines
drivers/net/ethernet/intel/ice/devlink/devlink.c +5 -0
+{
+	return ice_devlink_alloc(dev, sizeof(struct ice_sf_priv),
+				 &ice_sf_devlink_ops);
+}
+

[...]

#7 lore raw

Re: [iwl-next v4 5/8] ice: allocate devlink for subfunction

Quoted 5 lines · Jiri Pirko
Quoted 1 line · Wed, Apr 17, 2024 at 04:20:25PM CEST, [email protected] wrote:

From: Piotr Raczynski <[email protected]>

[...]

Quoted 15 lines

drivers/net/ethernet/intel/ice/devlink/devlink.c struct ice_pf *ice_allocate_pf(struct device *dev) · line 1471

Quoted 9 lines
drivers/net/ethernet/intel/ice/devlink/devlink.c +9 -0
+/**
+ * ice_allocate_sf - Allocate devlink and return SF structure pointer
+ * @dev: the device to allocate for
+ *
+ * Allocate a devlink instance for SF.
+ *
+ * Return: void pointer to allocated memory
+ */
+struct ice_sf_priv *ice_allocate_sf(struct device *dev)

This is devlink instance for SF auxdev. Please make sure it is properly linked with the devlink port instance using devl_port_fn_devlink_set() See mlx5 implementation for inspiration.

I am going to do it in the last patchset. I know that it isn't the best option to split patchesets like that, but it was hard to do it differently.

Thanks,
Michal

Quoted 7 lines
Quoted 5 lines
drivers/net/ethernet/intel/ice/devlink/devlink.c +5 -0
+{
+	return ice_devlink_alloc(dev, sizeof(struct ice_sf_priv),
+				 &ice_sf_devlink_ops);
+}
+

[...]

#8 lore raw

Re: [iwl-next v4 5/8] ice: allocate devlink for subfunction

Quoted 22 lines · Thu, Apr 18, 2024 at 02:48:53PM CEST, [email protected] wrote:
Quoted 5 lines · Jiri Pirko
Quoted 1 line · Wed, Apr 17, 2024 at 04:20:25PM CEST, [email protected] wrote:

From: Piotr Raczynski <[email protected]>

[...]

Quoted 15 lines

drivers/net/ethernet/intel/ice/devlink/devlink.c struct ice_pf *ice_allocate_pf(struct device *dev) · line 1471

Quoted 9 lines
drivers/net/ethernet/intel/ice/devlink/devlink.c +9 -0
+/**
+ * ice_allocate_sf - Allocate devlink and return SF structure pointer
+ * @dev: the device to allocate for
+ *
+ * Allocate a devlink instance for SF.
+ *
+ * Return: void pointer to allocated memory
+ */
+struct ice_sf_priv *ice_allocate_sf(struct device *dev)

This is devlink instance for SF auxdev. Please make sure it is properly linked with the devlink port instance using devl_port_fn_devlink_set() See mlx5 implementation for inspiration.

Quoted 1 line

I am going to do it in the last patchset. I know that it isn't the best

Where? Either I'm blind or you don't do it.

Quoted 12 lines

option to split patchesets like that, but it was hard to do it differently.

Thanks,
Michal

Quoted 7 lines
Quoted 5 lines
drivers/net/ethernet/intel/ice/devlink/devlink.c +5 -0
+{
+	return ice_devlink_alloc(dev, sizeof(struct ice_sf_priv),
+				 &ice_sf_devlink_ops);
+}
+

[...]

#9 lore raw

Re: [iwl-next v4 5/8] ice: allocate devlink for subfunction

Quoted 23 lines · Jiri Pirko
Quoted 22 lines · Thu, Apr 18, 2024 at 02:48:53PM CEST, [email protected] wrote:
Quoted 20 lines · Jiri Pirko
Quoted 1 line · Wed, Apr 17, 2024 at 04:20:25PM CEST, [email protected] wrote:

From: Piotr Raczynski <[email protected]>

[...]

drivers/net/ethernet/intel/ice/devlink/devlink.c struct ice_pf *ice_allocate_pf(struct device *dev) · line 1471

Quoted 9 lines
drivers/net/ethernet/intel/ice/devlink/devlink.c +9 -0
+/**
+ * ice_allocate_sf - Allocate devlink and return SF structure pointer
+ * @dev: the device to allocate for
+ *
+ * Allocate a devlink instance for SF.
+ *
+ * Return: void pointer to allocated memory
+ */
+struct ice_sf_priv *ice_allocate_sf(struct device *dev)

This is devlink instance for SF auxdev. Please make sure it is properly linked with the devlink port instance using devl_port_fn_devlink_set() See mlx5 implementation for inspiration.

Quoted 5 lines
Quoted 1 line

I am going to do it in the last patchset. I know that it isn't the best

Where? Either I'm blind or you don't do it.

You told me to split few patches from first patchset [1]. We agree that there will be too many patches for one submission, so I split it into 3:
- 1/3 devlink prework (already accepted)
- 2/3 base subfunction (this patchset)
- 3/3 port representor refactor to support subfunction (I am going to include it there)

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

Thanks,
Michal

Quoted 12 lines
Quoted 12 lines

option to split patchesets like that, but it was hard to do it differently.

Thanks,
Michal

Quoted 7 lines
Quoted 5 lines
drivers/net/ethernet/intel/ice/devlink/devlink.c +5 -0
+{
+	return ice_devlink_alloc(dev, sizeof(struct ice_sf_priv),
+				 &ice_sf_devlink_ops);
+}
+

[...]

#10 lore raw

Re: [iwl-next v4 5/8] ice: allocate devlink for subfunction

Quoted 30 lines · Thu, Apr 18, 2024 at 04:46:23PM CEST, [email protected] wrote:
Quoted 23 lines · Jiri Pirko
Quoted 22 lines · Thu, Apr 18, 2024 at 02:48:53PM CEST, [email protected] wrote:
Quoted 20 lines · Jiri Pirko
Quoted 1 line · Wed, Apr 17, 2024 at 04:20:25PM CEST, [email protected] wrote:

From: Piotr Raczynski <[email protected]>

[...]

drivers/net/ethernet/intel/ice/devlink/devlink.c struct ice_pf *ice_allocate_pf(struct device *dev) · line 1471

Quoted 9 lines
drivers/net/ethernet/intel/ice/devlink/devlink.c +9 -0
+/**
+ * ice_allocate_sf - Allocate devlink and return SF structure pointer
+ * @dev: the device to allocate for
+ *
+ * Allocate a devlink instance for SF.
+ *
+ * Return: void pointer to allocated memory
+ */
+struct ice_sf_priv *ice_allocate_sf(struct device *dev)

This is devlink instance for SF auxdev. Please make sure it is properly linked with the devlink port instance using devl_port_fn_devlink_set() See mlx5 implementation for inspiration.

Quoted 5 lines
Quoted 1 line

I am going to do it in the last patchset. I know that it isn't the best

Where? Either I'm blind or you don't do it.

Quoted 7 lines

You told me to split few patches from first patchset [1]. We agree that there will be too many patches for one submission, so I split it into 3:
- 1/3 devlink prework (already accepted)
- 2/3 base subfunction (this patchset)
- 3/3 port representor refactor to support subfunction (I am going to include it there)

Sorry, but how is this relevant to my suggestion to use devl_port_fn_devlink_set() which you apparently don't?

Quoted 18 lines

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

Thanks,
Michal

Quoted 12 lines
Quoted 12 lines

option to split patchesets like that, but it was hard to do it differently.

Thanks,
Michal

Quoted 7 lines
Quoted 5 lines
drivers/net/ethernet/intel/ice/devlink/devlink.c +5 -0
+{
+	return ice_devlink_alloc(dev, sizeof(struct ice_sf_priv),
+				 &ice_sf_devlink_ops);
+}
+

[...]

#11 lore raw

Re: [iwl-next v4 5/8] ice: allocate devlink for subfunction

Quoted 39 lines · Jiri Pirko
Quoted 30 lines · Thu, Apr 18, 2024 at 04:46:23PM CEST, [email protected] wrote:
Quoted 23 lines · Jiri Pirko
Quoted 22 lines · Thu, Apr 18, 2024 at 02:48:53PM CEST, [email protected] wrote:
Quoted 20 lines · Jiri Pirko
Quoted 1 line · Wed, Apr 17, 2024 at 04:20:25PM CEST, [email protected] wrote:

From: Piotr Raczynski <[email protected]>

[...]

drivers/net/ethernet/intel/ice/devlink/devlink.c struct ice_pf *ice_allocate_pf(struct device *dev) · line 1471

Quoted 9 lines
drivers/net/ethernet/intel/ice/devlink/devlink.c +9 -0
+/**
+ * ice_allocate_sf - Allocate devlink and return SF structure pointer
+ * @dev: the device to allocate for
+ *
+ * Allocate a devlink instance for SF.
+ *
+ * Return: void pointer to allocated memory
+ */
+struct ice_sf_priv *ice_allocate_sf(struct device *dev)

This is devlink instance for SF auxdev. Please make sure it is properly linked with the devlink port instance using devl_port_fn_devlink_set() See mlx5 implementation for inspiration.

Quoted 5 lines
Quoted 1 line

I am going to do it in the last patchset. I know that it isn't the best

Where? Either I'm blind or you don't do it.

Quoted 7 lines

You told me to split few patches from first patchset [1]. We agree that there will be too many patches for one submission, so I split it into 3:
- 1/3 devlink prework (already accepted)
- 2/3 base subfunction (this patchset)
- 3/3 port representor refactor to support subfunction (I am going to include it there)

Quoted 3 lines

Sorry, but how is this relevant to my suggestion to use devl_port_fn_devlink_set() which you apparently don't?

Devlink port to link with is introduced in the port representor part. Strange, but it fitted to my splitting. I can move activation/deactivation part also to this patchset (as there is no devlink port to call it on) if you want.

Quoted 19 lines
Quoted 19 lines

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

Thanks,
Michal

Quoted 12 lines
Quoted 12 lines

option to split patchesets like that, but it was hard to do it differently.

Thanks,
Michal

Quoted 7 lines
Quoted 5 lines
drivers/net/ethernet/intel/ice/devlink/devlink.c +5 -0
+{
+	return ice_devlink_alloc(dev, sizeof(struct ice_sf_priv),
+				 &ice_sf_devlink_ops);
+}
+

[...]

#12 lore raw

Re: [iwl-next v4 5/8] ice: allocate devlink for subfunction

Quoted 40 lines · Thu, Apr 18, 2024 at 06:11:38PM CEST, [email protected] wrote:
Quoted 39 lines · Jiri Pirko
Quoted 37 lines · Thu, Apr 18, 2024 at 04:46:23PM CEST, [email protected] wrote:
Quoted 23 lines · Jiri Pirko
Quoted 22 lines · Thu, Apr 18, 2024 at 02:48:53PM CEST, [email protected] wrote:
Quoted 20 lines · Jiri Pirko
Quoted 1 line · Wed, Apr 17, 2024 at 04:20:25PM CEST, [email protected] wrote:

From: Piotr Raczynski <[email protected]>

[...]

Review

Quoted 9 lines
(quoted diff) +9 -0
+/**
+ * ice_allocate_sf - Allocate devlink and return SF structure pointer
+ * @dev: the device to allocate for
+ *
+ * Allocate a devlink instance for SF.
+ *
+ * Return: void pointer to allocated memory
+ */
+struct ice_sf_priv *ice_allocate_sf(struct device *dev)

This is devlink instance for SF auxdev. Please make sure it is properly linked with the devlink port instance using devl_port_fn_devlink_set() See mlx5 implementation for inspiration.

Quoted 5 lines
Quoted 1 line

I am going to do it in the last patchset. I know that it isn't the best

Where? Either I'm blind or you don't do it.

You told me to split few patches from first patchset [1]. We agree that there will be too many patches for one submission, so I split it into 3:
- 1/3 devlink prework (already accepted)
- 2/3 base subfunction (this patchset)
- 3/3 port representor refactor to support subfunction (I am going to include it there)

Quoted 8 lines
Quoted 3 lines

Sorry, but how is this relevant to my suggestion to use devl_port_fn_devlink_set() which you apparently don't?

Devlink port to link with is introduced in the port representor part. Strange, but it fitted to my splitting. I can move activation/deactivation part also to this patchset (as there is no devlink port to call it on) if you want.

You have 7 more patches to use in this set. No problem. Please do it all at once.

Quoted 20 lines
Quoted 20 lines
Quoted 20 lines

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

Thanks,
Michal

Quoted 12 lines
Quoted 12 lines

option to split patchesets like that, but it was hard to do it differently.

Thanks,
Michal

Quoted 7 lines
Quoted 5 lines
(quoted diff) +5 -0
+{
+	return ice_devlink_alloc(dev, sizeof(struct ice_sf_priv),
+				 &ice_sf_devlink_ops);
+}
+

[...]

#13 lore raw

Re: [iwl-next v4 5/8] ice: allocate devlink for subfunction

Quoted 45 lines · Jiri Pirko
Quoted 44 lines · Thu, Apr 18, 2024 at 06:11:38PM CEST, [email protected] wrote:
Quoted 42 lines · Jiri Pirko
Quoted 30 lines · Thu, Apr 18, 2024 at 04:46:23PM CEST, [email protected] wrote:
Quoted 23 lines · Jiri Pirko
Quoted 22 lines · Thu, Apr 18, 2024 at 02:48:53PM CEST, [email protected] wrote:
Quoted 20 lines · Jiri Pirko
Quoted 1 line · Wed, Apr 17, 2024 at 04:20:25PM CEST, [email protected] wrote:

From: Piotr Raczynski <[email protected]>

[...]

Review

Quoted 9 lines
(quoted diff) +9 -0
+/**
+ * ice_allocate_sf - Allocate devlink and return SF structure pointer
+ * @dev: the device to allocate for
+ *
+ * Allocate a devlink instance for SF.
+ *
+ * Return: void pointer to allocated memory
+ */
+struct ice_sf_priv *ice_allocate_sf(struct device *dev)

This is devlink instance for SF auxdev. Please make sure it is properly linked with the devlink port instance using devl_port_fn_devlink_set() See mlx5 implementation for inspiration.

Quoted 5 lines
Quoted 1 line

I am going to do it in the last patchset. I know that it isn't the best

Where? Either I'm blind or you don't do it.

Quoted 7 lines

You told me to split few patches from first patchset [1]. We agree that there will be too many patches for one submission, so I split it into 3:
- 1/3 devlink prework (already accepted)
- 2/3 base subfunction (this patchset)
- 3/3 port representor refactor to support subfunction (I am going to include it there)

Sorry, but how is this relevant to my suggestion to use devl_port_fn_devlink_set() which you apparently don't?

Quoted 8 lines
Quoted 4 lines

Devlink port to link with is introduced in the port representor part. Strange, but it fitted to my splitting. I can move activation/deactivation part also to this patchset (as there is no devlink port to call it on) if you want.

You have 7 more patches to use in this set. No problem. Please do it all at once.

Ok, as whole will still not fit into 15 I sent preparation patchset for representor [1].

Now the patchset based on this preparation have 14 patches, so I hope it is fine (including linking that you mentioned). I will send it right after the preparation patchset is applied.

I am going on the 2 weeks vacation, so my replies will be delayed.

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

Thanks,
Michal

Quoted 21 lines
Quoted 21 lines
Quoted 21 lines
Quoted 21 lines

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

Thanks,
Michal

Quoted 12 lines
Quoted 12 lines

option to split patchesets like that, but it was hard to do it differently.

Thanks,
Michal

Quoted 7 lines
Quoted 5 lines
(quoted diff) +5 -0
+{
+	return ice_devlink_alloc(dev, sizeof(struct ice_sf_priv),
+				 &ice_sf_devlink_ops);
+}
+

[...]

#14 lore raw

PATCH v4 6/8 ice: base subfunction aux driver

In reply to Michal Swiatkowski

From: Piotr Raczynski <[email protected]>

Implement subfunction driver. It is probe when subfunction port is activated.

VSI is already created. During the probe VSI is being configured. MAC unicast and broadcast filter is added to allow traffic to pass.

---
 drivers/net/ethernet/intel/ice/Makefile     |   1 +
 drivers/net/ethernet/intel/ice/ice_main.c   |  10 ++
 drivers/net/ethernet/intel/ice/ice_sf_eth.c | 140 ++++++++++++++++++++
 drivers/net/ethernet/intel/ice/ice_sf_eth.h |   9 ++
 4 files changed, 160 insertions(+)
 create mode 100644 drivers/net/ethernet/intel/ice/ice_sf_eth.c
drivers/net/ethernet/intel/ice/Makefile +1 -0
index 03500e28ac99..4d987f5dcdc1 100644
@@ -31,6 +31,7 @@ ice-y := ice_main.o	\
 	 ice_idc.o	\
 	 devlink/devlink.o	\
 	 devlink/devlink_port.o \
+	 ice_sf_eth.o	\
 	 ice_ddp.o	\
 	 ice_fw_update.o \
 	 ice_lag.o	\
drivers/net/ethernet/intel/ice/ice_main.c +10 -0
index 29552598ddb6..f55e3340b608 100644
@@ -15,6 +15,7 @@
 #include "ice_dcb_nl.h"
 #include "devlink/devlink.h"
 #include "devlink/devlink_port.h"
+#include "ice_sf_eth.h"
 #include "ice_hwmon.h"
 /* Including ice_trace.h with CREATE_TRACE_POINTS defined will generate the
  * ice tracepoint functions. This must be done exactly once across the
@@ -5857,8 +5858,16 @@ static int __init ice_module_init(void)
 		goto err_dest_lag_wq;
 	}
 
+	status = ice_sf_driver_register();
+	if (status) {
+		pr_err("Failed to register SF driver, err %d\n", status);
+		goto err_sf_driver;
+	}
+
 	return 0;
 
+err_sf_driver:
+	pci_unregister_driver(&ice_driver);
 err_dest_lag_wq:
 	destroy_workqueue(ice_lag_wq);
 	ice_debugfs_exit();
@@ -5876,6 +5885,7 @@ module_init(ice_module_init);
  */
 static void __exit ice_module_exit(void)
 {
+	ice_sf_driver_unregister();
 	pci_unregister_driver(&ice_driver);
 	ice_debugfs_exit();
 	destroy_workqueue(ice_wq);
drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +140 -0

140 lines not on this page · raw message

drivers/net/ethernet/intel/ice/ice_sf_eth.h +9 -0
index a08f8b2bceef..e972c50f96c9 100644
@@ -18,4 +18,13 @@ struct ice_sf_priv {
 	struct devlink_port devlink_port;
 };
 
+static inline struct
+ice_sf_dev *ice_adev_to_sf_dev(struct auxiliary_device *adev)
+{
+	return container_of(adev, struct ice_sf_dev, adev);
+}
+
+int ice_sf_driver_register(void);
+void ice_sf_driver_unregister(void);
+
 #endif /* _ICE_SF_ETH_H_ */
Signature · 1 line
2.42.0
#15 lore raw

Re: [iwl-next v4 6/8] ice: base subfunction aux driver

Quoted 18 lines · Michal Swiatkowski

From: Piotr Raczynski <[email protected]>

Implement subfunction driver. It is probe when subfunction port is activated.

VSI is already created. During the probe VSI is being configured. MAC unicast and broadcast filter is added to allow traffic to pass.

---
  drivers/net/ethernet/intel/ice/Makefile     |   1 +
  drivers/net/ethernet/intel/ice/ice_main.c   |  10 ++
  drivers/net/ethernet/intel/ice/ice_sf_eth.c | 140 ++++++++++++++++++++
  drivers/net/ethernet/intel/ice/ice_sf_eth.h |   9 ++
  4 files changed, 160 insertions(+)

create mode 100644 drivers/net/ethernet/intel/ice/ice_sf_eth.c

<...>

Quoted 5 lines
drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +5 -0
+
+/**
+ * ice_sf_driver_register - Register new auxiliary subfunction driver
+ *
+ * Return: zero on success or an error code on failure.

drivers/net/ethernet/intel/ice/ice_sf_eth.c line 135

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +10 -0
+ */
+int ice_sf_driver_register(void)
+{
+	return auxiliary_driver_register(&ice_sf_driver);
+}
+
+/**
+ * ice_sf_driver_unregister - Unregister new auxiliary subfunction driver
+ *
+ * Return: zero on success or an error code on failure.

this function doesn't return anything...

Quoted 23 lines

+ */

drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +4 -0
+void ice_sf_driver_unregister(void)
+{
+	auxiliary_driver_unregister(&ice_sf_driver);
+}
drivers/net/ethernet/intel/ice/ice_sf_eth.h +9 -0
index a08f8b2bceef..e972c50f96c9 100644
@@ -18,4 +18,13 @@ struct ice_sf_priv {
  	struct devlink_port devlink_port;
  };
 
+static inline struct
+ice_sf_dev *ice_adev_to_sf_dev(struct auxiliary_device *adev)
+{
+	return container_of(adev, struct ice_sf_dev, adev);
+}
+
+int ice_sf_driver_register(void);
+void ice_sf_driver_unregister(void);
+
  #endif /* _ICE_SF_ETH_H_ */
#16 lore raw

Re: [iwl-next v4 6/8] ice: base subfunction aux driver

Quoted 34 lines · Shay Drori
Quoted 18 lines · Michal Swiatkowski

From: Piotr Raczynski <[email protected]>

Implement subfunction driver. It is probe when subfunction port is activated.

VSI is already created. During the probe VSI is being configured. MAC unicast and broadcast filter is added to allow traffic to pass.

---
  drivers/net/ethernet/intel/ice/Makefile     |   1 +
  drivers/net/ethernet/intel/ice/ice_main.c   |  10 ++
  drivers/net/ethernet/intel/ice/ice_sf_eth.c | 140 ++++++++++++++++++++
  drivers/net/ethernet/intel/ice/ice_sf_eth.h |   9 ++
  4 files changed, 160 insertions(+)

create mode 100644 drivers/net/ethernet/intel/ice/ice_sf_eth.c

<...>

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +10 -0
+
+/**
+ * ice_sf_driver_register - Register new auxiliary subfunction driver
+ *
+ * Return: zero on success or an error code on failure.
+ */
+int ice_sf_driver_register(void)
+{
+	return auxiliary_driver_register(&ice_sf_driver);
+}
Quoted 9 lines

drivers/net/ethernet/intel/ice/ice_sf_eth.c line 135

Quoted 5 lines
drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +5 -0
+
+/**
+ * ice_sf_driver_unregister - Unregister new auxiliary subfunction driver
+ *
+ * Return: zero on success or an error code on failure.

this function doesn't return anything...

Thanks, will remove it.

Quoted 22 lines
Quoted 22 lines

+ */

drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +4 -0
+void ice_sf_driver_unregister(void)
+{
+	auxiliary_driver_unregister(&ice_sf_driver);
+}
drivers/net/ethernet/intel/ice/ice_sf_eth.h +9 -0
index a08f8b2bceef..e972c50f96c9 100644
@@ -18,4 +18,13 @@ struct ice_sf_priv {
  	struct devlink_port devlink_port;
  };
+static inline struct
+ice_sf_dev *ice_adev_to_sf_dev(struct auxiliary_device *adev)
+{
+	return container_of(adev, struct ice_sf_dev, adev);
+}
+
+int ice_sf_driver_register(void);
+void ice_sf_driver_unregister(void);
+
  #endif /* _ICE_SF_ETH_H_ */
#17 lore raw

Re: [iwl-next v4 6/8] ice: base subfunction aux driver

In reply to Michal Swiatkowski
Quoted 1 line · Wed, Apr 17, 2024 at 04:20:26PM CEST, [email protected] wrote:

From: Piotr Raczynski <[email protected]>

[...]

Quoted 18 lines
drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +18 -0
+static int ice_sf_dev_probe(struct auxiliary_device *adev,
+			    const struct auxiliary_device_id *id)
+{
+	struct ice_sf_dev *sf_dev = ice_adev_to_sf_dev(adev);
+	struct ice_dynamic_port *dyn_port = sf_dev->dyn_port;
+	struct ice_vsi_cfg_params params = {};
+	struct ice_vsi *vsi = dyn_port->vsi;
+	struct ice_pf *pf = dyn_port->pf;
+	struct device *dev = &adev->dev;
+	struct ice_sf_priv *priv;
+	struct devlink *devlink;
+	int err;
+
+	params.type = ICE_VSI_SF;
+	params.pi = pf->hw.port_info;
+	params.flags = ICE_VSI_FLAG_INIT;
+
+	priv = ice_allocate_sf(&adev->dev);

drivers/net/ethernet/intel/ice/ice_sf_eth.c line 46

Quoted 10 lines
drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +10 -0
+	if (!priv) {
+		dev_err(dev, "Subfunction devlink alloc failed");
+		return -ENOMEM;
+	}
+
+	priv->dev = sf_dev;
+	sf_dev->priv = priv;
+	devlink = priv_to_devlink(priv);
+
+	devlink_register(devlink);

Do register at the very end. Btw, currently the error path seems to be broken, leaving devlink instance allocated and registered.

Quoted 34 lines
drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +34 -0
+	devl_lock(devlink);
+
+	err = ice_vsi_cfg(vsi, &params);
+	if (err) {
+		dev_err(dev, "Subfunction vsi config failed");
+		goto err_devlink_unlock;
+	}
+
+	err = ice_devlink_create_sf_dev_port(sf_dev);
+	if (err) {
+		dev_err(dev, "Cannot add ice virtual devlink port for subfunction");
+		goto err_vsi_decfg;
+	}
+
+	err = ice_fltr_add_mac_and_broadcast(vsi, vsi->netdev->dev_addr,
+					     ICE_FWD_TO_VSI);
+	if (err) {
+		dev_err(dev, "can't add MAC filters %pM for VSI %d\n",
+			vsi->netdev->dev_addr, vsi->idx);
+		goto err_devlink_destroy;
+	}
+
+	ice_napi_add(vsi);
+	devl_unlock(devlink);
+
+	return 0;
+
+err_devlink_destroy:
+	ice_devlink_destroy_sf_dev_port(sf_dev);
+err_vsi_decfg:
+	ice_vsi_decfg(vsi);
+err_devlink_unlock:
+	devl_unlock(devlink);
+	return err;

[...]

#18 lore raw

Re: [iwl-next v4 6/8] ice: base subfunction aux driver

Quoted 29 lines · Jiri Pirko
Quoted 1 line · Wed, Apr 17, 2024 at 04:20:26PM CEST, [email protected] wrote:

From: Piotr Raczynski <[email protected]>

[...]

Quoted 23 lines
drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +23 -0
+static int ice_sf_dev_probe(struct auxiliary_device *adev,
+			    const struct auxiliary_device_id *id)
+{
+	struct ice_sf_dev *sf_dev = ice_adev_to_sf_dev(adev);
+	struct ice_dynamic_port *dyn_port = sf_dev->dyn_port;
+	struct ice_vsi_cfg_params params = {};
+	struct ice_vsi *vsi = dyn_port->vsi;
+	struct ice_pf *pf = dyn_port->pf;
+	struct device *dev = &adev->dev;
+	struct ice_sf_priv *priv;
+	struct devlink *devlink;
+	int err;
+
+	params.type = ICE_VSI_SF;
+	params.pi = pf->hw.port_info;
+	params.flags = ICE_VSI_FLAG_INIT;
+
+	priv = ice_allocate_sf(&adev->dev);
+	if (!priv) {
+		dev_err(dev, "Subfunction devlink alloc failed");
+		return -ENOMEM;
+	}
+
Quoted 9 lines

drivers/net/ethernet/intel/ice/ice_sf_eth.c line 46

Quoted 5 lines
drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +5 -0
+	priv->dev = sf_dev;
+	sf_dev->priv = priv;
+	devlink = priv_to_devlink(priv);
+
+	devlink_register(devlink);

Do register at the very end. Btw, currently the error path seems to be broken, leaving devlink instance allocated and registered.

Sure, I will fix it.

Quoted 37 lines
Quoted 35 lines
drivers/net/ethernet/intel/ice/ice_sf_eth.c new file +34 -0
+	devl_lock(devlink);
+
+	err = ice_vsi_cfg(vsi, &params);
+	if (err) {
+		dev_err(dev, "Subfunction vsi config failed");
+		goto err_devlink_unlock;
+	}
+
+	err = ice_devlink_create_sf_dev_port(sf_dev);
+	if (err) {
+		dev_err(dev, "Cannot add ice virtual devlink port for subfunction");
+		goto err_vsi_decfg;
+	}
+
+	err = ice_fltr_add_mac_and_broadcast(vsi, vsi->netdev->dev_addr,
+					     ICE_FWD_TO_VSI);
+	if (err) {
+		dev_err(dev, "can't add MAC filters %pM for VSI %d\n",
+			vsi->netdev->dev_addr, vsi->idx);
+		goto err_devlink_destroy;
+	}
+
+	ice_napi_add(vsi);
+	devl_unlock(devlink);
+
+	return 0;
+
+err_devlink_destroy:
+	ice_devlink_destroy_sf_dev_port(sf_dev);
+err_vsi_decfg:
+	ice_vsi_decfg(vsi);
+err_devlink_unlock:
+	devl_unlock(devlink);
+	return err;

[...]

#19 lore raw

PATCH v4 7/8 ice: implement netdev for subfunction

In reply to Michal Swiatkowski

From: Piotr Raczynski <[email protected]>

Configure netdevice for subfunction usecase. Mostly it is reusing ops from the PF netdevice.

SF netdev is linked to devlink port registered after SF activation.

---
 drivers/net/ethernet/intel/ice/ice_sf_eth.c | 85 ++++++++++++++++++++-
 1 file changed, 84 insertions(+), 1 deletion(-)
drivers/net/ethernet/intel/ice/ice_sf_eth.c +84 -1
index 6ecc13217544..ecef869c20a1 100644
@@ -2,11 +2,85 @@
 /* Copyright (c) 2024, Intel Corporation. */
 #include "ice.h"
 #include "ice_lib.h"
+#include "ice_txrx.h"
 #include "ice_fltr.h"
 #include "ice_sf_eth.h"
 #include "devlink/devlink_port.h"
 #include "devlink/devlink.h"
 
+static const struct net_device_ops ice_sf_netdev_ops = {
+	.ndo_open = ice_open,
+	.ndo_stop = ice_stop,
+	.ndo_start_xmit = ice_start_xmit,
+	.ndo_vlan_rx_add_vid = ice_vlan_rx_add_vid,
+	.ndo_vlan_rx_kill_vid = ice_vlan_rx_kill_vid,
+	.ndo_change_mtu = ice_change_mtu,
+	.ndo_get_stats64 = ice_get_stats64,
+	.ndo_tx_timeout = ice_tx_timeout,
+	.ndo_bpf = ice_xdp,
+	.ndo_xdp_xmit = ice_xdp_xmit,
+	.ndo_xsk_wakeup = ice_xsk_wakeup,
+};
+
+/**
+ * ice_sf_cfg_netdev - Allocate, configure and register a netdev
+ * @dyn_port: subfunction associated with configured netdev
+ * @devlink_port: subfunction devlink port to be linked with netdev
+ *
+ * Return: 0 on success, negative value on failure
+ */
+static int ice_sf_cfg_netdev(struct ice_dynamic_port *dyn_port,
+			     struct devlink_port *devlink_port)
+{
+	struct ice_vsi *vsi = dyn_port->vsi;
+	struct ice_netdev_priv *np;
+	struct net_device *netdev;
+	int err;
+
+	netdev = alloc_etherdev_mqs(sizeof(*np), vsi->alloc_txq,
+				    vsi->alloc_rxq);
+	if (!netdev)
+		return -ENOMEM;
+
+	SET_NETDEV_DEV(netdev, &vsi->back->pdev->dev);
+	set_bit(ICE_VSI_NETDEV_ALLOCD, vsi->state);
+	vsi->netdev = netdev;
+	np = netdev_priv(netdev);
+	np->vsi = vsi;
+
+	ice_set_netdev_features(netdev);
+
+	netdev->xdp_features = NETDEV_XDP_ACT_BASIC | NETDEV_XDP_ACT_REDIRECT |
+			       NETDEV_XDP_ACT_XSK_ZEROCOPY |
+			       NETDEV_XDP_ACT_RX_SG;
+
+	eth_hw_addr_set(netdev, dyn_port->hw_addr);
+	ether_addr_copy(netdev->perm_addr, dyn_port->hw_addr);
+	netdev->netdev_ops = &ice_sf_netdev_ops;
+	SET_NETDEV_DEVLINK_PORT(netdev, devlink_port);
+
+	err = register_netdev(netdev);
+	if (err) {
+		free_netdev(netdev);
+		vsi->netdev = NULL;
+		return -ENOMEM;
+	}
+	set_bit(ICE_VSI_NETDEV_REGISTERED, vsi->state);
+	netif_carrier_off(netdev);
+	netif_tx_stop_all_queues(netdev);
+
+	return 0;
+}
+
+static void ice_sf_decfg_netdev(struct ice_vsi *vsi)
+{
+	unregister_netdev(vsi->netdev);
+	clear_bit(ICE_VSI_NETDEV_REGISTERED, vsi->state);
+	free_netdev(vsi->netdev);
+	vsi->netdev = NULL;
+	clear_bit(ICE_VSI_NETDEV_ALLOCD, vsi->state);
+}
+
 /**
  * ice_sf_dev_probe - subfunction driver probe function
  * @adev: pointer to the auxiliary device
@@ -58,12 +132,18 @@ static int ice_sf_dev_probe(struct auxiliary_device *adev,
 		goto err_vsi_decfg;
 	}
 
+	err = ice_sf_cfg_netdev(dyn_port, &sf_dev->priv->devlink_port);
+	if (err) {
+		dev_err(dev, "Subfunction netdev config failed");
+		goto err_devlink_destroy;
+	}
+
 	err = ice_fltr_add_mac_and_broadcast(vsi, vsi->netdev->dev_addr,
 					     ICE_FWD_TO_VSI);
 	if (err) {
 		dev_err(dev, "can't add MAC filters %pM for VSI %d\n",
 			vsi->netdev->dev_addr, vsi->idx);
-		goto err_devlink_destroy;
+		goto err_netdev_decfg;
 	}
 
 	ice_napi_add(vsi);
@@ -71,6 +151,8 @@ static int ice_sf_dev_probe(struct auxiliary_device *adev,
 
 	return 0;
 
+err_netdev_decfg:
+	ice_sf_decfg_netdev(vsi);
 err_devlink_destroy:
 	ice_devlink_destroy_sf_dev_port(sf_dev);
 err_vsi_decfg:
@@ -98,6 +180,7 @@ static void ice_sf_dev_remove(struct auxiliary_device *adev)
 
 	ice_vsi_close(vsi);
 
+	ice_sf_decfg_netdev(vsi);
 	ice_devlink_destroy_sf_dev_port(sf_dev);
 	devl_unregister(devlink);
 	devl_unlock(devlink);
Signature · 1 line
2.42.0
#20 lore raw

PATCH v4 8/8 ice: allow to activate and deactivate subfunction

In reply to Michal Swiatkowski

From: Piotr Raczynski <[email protected]>

Use previously implemented SF aux driver. It is probe during SF activation and remove after deactivation.

---
 .../ethernet/intel/ice/devlink/devlink_port.c | 106 ++++++++++++++++++
 .../ethernet/intel/ice/devlink/devlink_port.h |   5 +
 drivers/net/ethernet/intel/ice/ice_sf_eth.c   | 104 +++++++++++++++++
 drivers/net/ethernet/intel/ice/ice_sf_eth.h   |   3 +
 4 files changed, 218 insertions(+)
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +106 -0
index 0ab4d687e311..efff286f9e3e 100644
@@ -482,6 +482,42 @@ void ice_devlink_destroy_sf_dev_port(struct ice_sf_dev *sf_dev)
 	devl_port_unregister(&sf_dev->priv->devlink_port);
 }
 
+/**
+ * ice_activate_dynamic_port - Activate a dynamic port
+ * @dyn_port: dynamic port instance to activate
+ * @extack: extack for reporting error messages
+ *
+ * Activate the dynamic port based on its flavour.
+ *
+ * Return: zero on success or an error code on failure.
+ */
+static int
+ice_activate_dynamic_port(struct ice_dynamic_port *dyn_port,
+			  struct netlink_ext_ack *extack)
+{
+	int err;
+
+	err = ice_sf_eth_activate(dyn_port, extack);
+	if (err)
+		return err;
+
+	dyn_port->active = true;
+
+	return 0;
+}
+
+/**
+ * ice_deactivate_dynamic_port - Deactivate a dynamic port
+ * @dyn_port: dynamic port instance to deactivate
+ *
+ * Undo activation of a dynamic port.
+ */
+static void ice_deactivate_dynamic_port(struct ice_dynamic_port *dyn_port)
+{
+	ice_sf_eth_deactivate(dyn_port);
+	dyn_port->active = false;
+}
+
 /**
  * ice_dealloc_dynamic_port - Deallocate and remove a dynamic port
  * @dyn_port: dynamic port instance to deallocate
@@ -494,6 +530,9 @@ static void ice_dealloc_dynamic_port(struct ice_dynamic_port *dyn_port)
 	struct devlink_port *devlink_port = &dyn_port->devlink_port;
 	struct ice_pf *pf = dyn_port->pf;
 
+	if (dyn_port->active)
+		ice_deactivate_dynamic_port(dyn_port);
+
 	xa_erase(&pf->sf_nums, devlink_port->attrs.pci_sf.sf);
 	devl_port_unregister(devlink_port);
 	ice_vsi_free(dyn_port->vsi);
@@ -638,10 +677,77 @@ ice_devlink_port_fn_hw_addr_get(struct devlink_port *port, u8 *hw_addr,
 	return 0;
 }
 
+/**
+ * ice_devlink_port_fn_state_set - devlink handler for port state set
+ * @port: pointer to devlink port
+ * @state: state to set
+ * @extack: extack for reporting error messages
+ *
+ * Activates or deactivates the port.
+ *
+ * Return: zero on success or an error code on failure.
+ */
+static int
+ice_devlink_port_fn_state_set(struct devlink_port *port,
+			      enum devlink_port_fn_state state,
+			      struct netlink_ext_ack *extack)
+{
+	struct ice_dynamic_port *dyn_port;
+
+	dyn_port = ice_devlink_port_to_dyn(port);
+
+	switch (state) {
+	case DEVLINK_PORT_FN_STATE_ACTIVE:
+		if (!dyn_port->active)
+			return ice_activate_dynamic_port(dyn_port, extack);
+		break;
+	case DEVLINK_PORT_FN_STATE_INACTIVE:
+		if (dyn_port->active)
+			ice_deactivate_dynamic_port(dyn_port);
+		break;
+	}
+
+	return 0;
+}
+
+/**
+ * ice_devlink_port_fn_state_get - devlink handler for port state get
+ * @port: pointer to devlink port
+ * @state: admin configured state of the port
+ * @opstate: current port operational state
+ * @extack: extack for reporting error messages
+ *
+ * Gets port state.
+ *
+ * Return: zero on success or an error code on failure.
+ */
+static int
+ice_devlink_port_fn_state_get(struct devlink_port *port,
+			      enum devlink_port_fn_state *state,
+			      enum devlink_port_fn_opstate *opstate,
+			      struct netlink_ext_ack *extack)
+{
+	struct ice_dynamic_port *dyn_port;
+
+	dyn_port = ice_devlink_port_to_dyn(port);
+
+	if (dyn_port->active) {
+		*state = DEVLINK_PORT_FN_STATE_ACTIVE;
+		*opstate = DEVLINK_PORT_FN_OPSTATE_ATTACHED;
+	} else {
+		*state = DEVLINK_PORT_FN_STATE_INACTIVE;
+		*opstate = DEVLINK_PORT_FN_OPSTATE_DETACHED;
+	}
+
+	return 0;
+}
+
 static const struct devlink_port_ops ice_devlink_port_sf_ops = {
 	.port_del = ice_devlink_port_del,
 	.port_fn_hw_addr_get = ice_devlink_port_fn_hw_addr_get,
 	.port_fn_hw_addr_set = ice_devlink_port_fn_hw_addr_set,
+	.port_fn_state_get = ice_devlink_port_fn_state_get,
+	.port_fn_state_set = ice_devlink_port_fn_state_set,
 };
 
 /**
drivers/net/ethernet/intel/ice/devlink/devlink_port.h +5 -0
index 33f07cec6bd6..3692b8617bd5 100644
@@ -14,6 +14,7 @@
  * @devlink_port: the associated devlink port structure
  * @pf: pointer to the PF private structure
  * @vsi: the VSI associated with this port
+ * @sf_dev: pointer to the subfunction device
  *
  * An instance of a dynamically added devlink port. Each port flavour
  */
@@ -23,6 +24,10 @@ struct ice_dynamic_port {
 	struct devlink_port devlink_port;
 	struct ice_pf *pf;
 	struct ice_vsi *vsi;
+	/* Flavour-specific implementation data */
+	union {
+		struct ice_sf_dev *sf_dev;
+	};
 };
 
 void ice_dealloc_all_dynamic_ports(struct ice_pf *pf);
drivers/net/ethernet/intel/ice/ice_sf_eth.c +104 -0

113 lines not on this page · raw message

drivers/net/ethernet/intel/ice/ice_sf_eth.h +3 -0
index e972c50f96c9..c558cad0a183 100644
@@ -27,4 +27,7 @@ ice_sf_dev *ice_adev_to_sf_dev(struct auxiliary_device *adev)
 int ice_sf_driver_register(void);
 void ice_sf_driver_unregister(void);
 
+int ice_sf_eth_activate(struct ice_dynamic_port *dyn_port,
+			struct netlink_ext_ack *extack);
+void ice_sf_eth_deactivate(struct ice_dynamic_port *dyn_port);
 #endif /* _ICE_SF_ETH_H_ */
Signature · 1 line
2.42.0
#21 lore raw

Re: [Intel-wired-lan] [iwl-next v4 8/8] ice: allow to activate and deactivate subfunction

Quoted 14 lines · Michal Swiatkowski
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +14 -0
+/**
+ * ice_devlink_port_fn_state_get - devlink handler for port state get
+ * @port: pointer to devlink port
+ * @state: admin configured state of the port
+ * @opstate: current port operational state
+ * @extack: extack for reporting error messages
+ *
+ * Gets port state.
+ *
+ * Return: zero on success or an error code on failure.
+ */
+static int
+ice_devlink_port_fn_state_get(struct devlink_port *port,
+			      enum devlink_port_fn_state *state,

drivers/net/ethernet/intel/ice/devlink/devlink_port.c ice_devlink_port_fn_hw_addr_get(struct devlink_port *port, u8 *hw_addr, · line 736

Quoted 10 lines
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +10 -0
+			      enum devlink_port_fn_opstate *opstate,
+			      struct netlink_ext_ack *extack)
+{
+	struct ice_dynamic_port *dyn_port;
+
+	dyn_port = ice_devlink_port_to_dyn(port);
+
+	if (dyn_port->active) {
+		*state = DEVLINK_PORT_FN_STATE_ACTIVE;
+		*opstate = DEVLINK_PORT_FN_OPSTATE_ATTACHED;

DEVLINK_PORT_FN_OPSTATE_ATTACHED means the SF is up/bind[1]. ice is using auxiliary bus for SFs, which means user can unbind it via the auxiliary sysfs (/sys/bus/auxiliary/drivers/ice_sf/unbind). In this case[2], you need to return:
*state = DEVLINK_PORT_FN_STATE_ACTIVE;
*opstate = DEVLINK_PORT_FN_OPSTATE_DETACHED;

[1]
Documentation from include/uapi/linux/devlink.h:

* @DEVLINK_PORT_FN_OPSTATE_ATTACHED: Driver is attached to the function. <...>
* @DEVLINK_PORT_FN_OPSTATE_DETACHED: Driver is detached from the function.

Quoted 8 lines
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +8 -0
+	} else {
+		*state = DEVLINK_PORT_FN_STATE_INACTIVE;
+		*opstate = DEVLINK_PORT_FN_OPSTATE_DETACHED;
+	}
+
+	return 0;
+}
+
#22 lore raw

Re: [iwl-next v4 8/8] ice: allow to activate and deactivate subfunction

resend as plain test

Quoted 49 lines · Shay Drori
Quoted 14 lines · Michal Swiatkowski
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +14 -0
+/**
+ * ice_devlink_port_fn_state_get - devlink handler for port state get
+ * @port: pointer to devlink port
+ * @state: admin configured state of the port
+ * @opstate: current port operational state
+ * @extack: extack for reporting error messages
+ *
+ * Gets port state.
+ *
+ * Return: zero on success or an error code on failure.
+ */
+static int
+ice_devlink_port_fn_state_get(struct devlink_port *port,
+			      enum devlink_port_fn_state *state,

drivers/net/ethernet/intel/ice/devlink/devlink_port.c ice_devlink_port_fn_hw_addr_get(struct devlink_port *port, u8 *hw_addr, · line 736

Quoted 10 lines
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +10 -0
+			      enum devlink_port_fn_opstate *opstate,
+			      struct netlink_ext_ack *extack)
+{
+	struct ice_dynamic_port *dyn_port;
+
+	dyn_port = ice_devlink_port_to_dyn(port);
+
+	if (dyn_port->active) {
+		*state = DEVLINK_PORT_FN_STATE_ACTIVE;
+		*opstate = DEVLINK_PORT_FN_OPSTATE_ATTACHED;

DEVLINK_PORT_FN_OPSTATE_ATTACHED means the SF is up/bind[1]. ice is using auxiliary bus for SFs, which means user can unbind it via the auxiliary sysfs (/sys/bus/auxiliary/drivers/ice_sf/unbind). In this case[2], you need to return:
*state = DEVLINK_PORT_FN_STATE_ACTIVE;
*opstate = DEVLINK_PORT_FN_OPSTATE_DETACHED;

[1]
Documentation from include/uapi/linux/devlink.h:

* @DEVLINK_PORT_FN_OPSTATE_ATTACHED: Driver is attached to the function. <...>
* @DEVLINK_PORT_FN_OPSTATE_DETACHED: Driver is detached from the function.

Quoted 8 lines
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +8 -0
+	} else {
+		*state = DEVLINK_PORT_FN_STATE_INACTIVE;
+		*opstate = DEVLINK_PORT_FN_OPSTATE_DETACHED;
+	}
+
+	return 0;
+}
+
#23 lore raw

Re: [iwl-next v4 8/8] ice: allow to activate and deactivate subfunction

Quoted 37 lines · Shay Drori

resend as plain test

Quoted 34 lines · Shay Drori
Quoted 14 lines · Michal Swiatkowski
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +14 -0
+/**
+ * ice_devlink_port_fn_state_get - devlink handler for port state get
+ * @port: pointer to devlink port
+ * @state: admin configured state of the port
+ * @opstate: current port operational state
+ * @extack: extack for reporting error messages
+ *
+ * Gets port state.
+ *
+ * Return: zero on success or an error code on failure.
+ */
+static int
+ice_devlink_port_fn_state_get(struct devlink_port *port,
+			      enum devlink_port_fn_state *state,

drivers/net/ethernet/intel/ice/devlink/devlink_port.c ice_devlink_port_fn_hw_addr_get(struct devlink_port *port, u8 *hw_addr, · line 736

Quoted 10 lines
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +10 -0
+			      enum devlink_port_fn_opstate *opstate,
+			      struct netlink_ext_ack *extack)
+{
+	struct ice_dynamic_port *dyn_port;
+
+	dyn_port = ice_devlink_port_to_dyn(port);
+
+	if (dyn_port->active) {
+		*state = DEVLINK_PORT_FN_STATE_ACTIVE;
+		*opstate = DEVLINK_PORT_FN_OPSTATE_ATTACHED;

DEVLINK_PORT_FN_OPSTATE_ATTACHED means the SF is up/bind[1]. ice is using auxiliary bus for SFs, which means user can unbind it via the auxiliary sysfs (/sys/bus/auxiliary/drivers/ice_sf/unbind). In this case[2], you need to return:
*state = DEVLINK_PORT_FN_STATE_ACTIVE;
*opstate = DEVLINK_PORT_FN_OPSTATE_DETACHED;

Thanks, I didn't think about unbinding/binding the aux driver via sysfs.

To be sure:
- user create the subfunction:
INACTIVE, DETACHED
- user activate it:
ACTIVE, ATTACHED
- user unbind driver:
ACTIVE, DETACHED
- user can bind it again as long as subfunction port is ACTIVE is it right?

I will fix the comment from previous patch and add state tracking for ATTACHED/DETACHED.

Thanks,
Michal

Quoted 16 lines
Quoted 16 lines

[1]
Documentation from include/uapi/linux/devlink.h:

* @DEVLINK_PORT_FN_OPSTATE_ATTACHED: Driver is attached to the function. <...>
* @DEVLINK_PORT_FN_OPSTATE_DETACHED: Driver is detached from the function.

Quoted 8 lines
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +8 -0
+	} else {
+		*state = DEVLINK_PORT_FN_STATE_INACTIVE;
+		*opstate = DEVLINK_PORT_FN_OPSTATE_DETACHED;
+	}
+
+	return 0;
+}
+
#24 lore raw

Re: [iwl-next v4 8/8] ice: allow to activate and deactivate subfunction

Quoted 42 lines · Michal Swiatkowski

External email: Use caution opening links or attachments

Quoted 37 lines · Shay Drori

resend as plain test

Quoted 34 lines · Shay Drori
Quoted 14 lines · Michal Swiatkowski
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +14 -0
+/**
+ * ice_devlink_port_fn_state_get - devlink handler for port state get
+ * @port: pointer to devlink port
+ * @state: admin configured state of the port
+ * @opstate: current port operational state
+ * @extack: extack for reporting error messages
+ *
+ * Gets port state.
+ *
+ * Return: zero on success or an error code on failure.
+ */
+static int
+ice_devlink_port_fn_state_get(struct devlink_port *port,
+                       enum devlink_port_fn_state *state,

drivers/net/ethernet/intel/ice/devlink/devlink_port.c ice_devlink_port_fn_hw_addr_get(struct devlink_port *port, u8 *hw_addr, · line 736

Quoted 10 lines
drivers/net/ethernet/intel/ice/devlink/devlink_port.c +10 -0
+                       enum devlink_port_fn_opstate *opstate,
+                       struct netlink_ext_ack *extack)
+{
+ struct ice_dynamic_port *dyn_port;
+
+ dyn_port = ice_devlink_port_to_dyn(port);
+
+ if (dyn_port->active) {
+         *state = DEVLINK_PORT_FN_STATE_ACTIVE;
+         *opstate = DEVLINK_PORT_FN_OPSTATE_ATTACHED;

DEVLINK_PORT_FN_OPSTATE_ATTACHED means the SF is up/bind[1]. ice is using auxiliary bus for SFs, which means user can unbind it via the auxiliary sysfs (/sys/bus/auxiliary/drivers/ice_sf/unbind). In this case[2], you need to return:
*state = DEVLINK_PORT_FN_STATE_ACTIVE;
*opstate = DEVLINK_PORT_FN_OPSTATE_DETACHED;

Quoted 10 lines

Thanks, I didn't think about unbinding/binding the aux driver via sysfs. > To be sure:
- user create the subfunction:
INACTIVE, DETACHED
- user activate it:
ACTIVE, ATTACHED
- user unbind driver:
ACTIVE, DETACHED
- user can bind it again as long as subfunction port is ACTIVE is it right?

yes.

Quoted 23 lines

I will fix the comment from previous patch and add state tracking for ATTACHED/DETACHED.

Thanks,
Michal

Quoted 16 lines
Quoted 16 lines

[1]
Documentation from include/uapi/linux/devlink.h:

* @DEVLINK_PORT_FN_OPSTATE_ATTACHED: Driver is attached to the function. <...>
* @DEVLINK_PORT_FN_OPSTATE_DETACHED: Driver is detached from the function.

Quoted 8 lines

+ } else {

drivers/net/ethernet/intel/ice/devlink/devlink_port.c +7 -0
+         *state = DEVLINK_PORT_FN_STATE_INACTIVE;
+         *opstate = DEVLINK_PORT_FN_OPSTATE_DETACHED;
+ }
+
+ return 0;
+}
+