Commit cedf90c0 authored by David S. Miller's avatar David S. Miller
Browse files

Merge tag 'mlx5-updates-2017-04-30' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux 



mlx5-updates-2017-04-30

Or says:
================
mlx5 neigh update

This series (whose code name is 'neigh update') from Hadar, enhances the
mlx5 TC IP tunnel offloads to deal with changes to tunnel destination
neighbours used in offloaded flows which involved encapsulation.

In order to keep track on the validity state of such neighbours, we register
a netevent notifier callback and act on NEIGH_UPDATE events: if a neighbour
becomes valid, offload the related flows to HW (the other way around when
neigh becomes invalid) and similarly when a neigh mac addresses changes.

Since this traffic is offloaded from the host OS, the neighbour for the IP
tunnel destination can mistakenly become STALE and deleted by the kernel
since its 'used' value wasn't changed. To address that, we proactively
update the neighbour 'used' value every DELAY_PROBE_TIME seconds, using
time stamps generated by the existing driver code for HW flow counters.
We use the DELAY_PROBE_TIME_UPDATE event to adjust the frequency of the updates.

Prior to the core of the series, there's a patch from Saeed that introduces an
extendable vport representor implementation scheme. It provides a separation
between the eswitch to the netdev related aspects of the representors.

We would like to thank Ido Schimmel and Ilya Lesokhin for their coaching && advice
through the long design and review cycles while we struggled to understand and
(hopefully correctly) implement the locking around the different driver flows(..) .

- Or.
=================

Misc Updates:

From Tariq:
Some small performance and trivial code optimization for mlx5 netdev driver
- Optimize poll ICOSQ completion queue
- Use prefetchw when a write is to follow
- Use u8 as ownership type in mlx5e_get_cqe()

From Eran:
- Disable LRO by default on specific setups

From Eli:
- Small cleanup for E-Switch to avoid redundant allocation

Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parents 07ff2ed0 0a0ab1d2

drivers/net/ethernet/mellanox/mlx5/core/en.h

+0 −20
Original line number Diff line number Diff line
@@ -991,20 +991,6 @@ int mlx5e_create_mdev_resources(struct mlx5_core_dev *mdev); 
void mlx5e_destroy_mdev_resources(struct mlx5_core_dev *mdev);

int mlx5e_refresh_tirs(struct mlx5e_priv *priv, bool enable_uc_lb);



struct mlx5_eswitch_rep;

int mlx5e_vport_rep_load(struct mlx5_eswitch *esw,

			 struct mlx5_eswitch_rep *rep);

void mlx5e_vport_rep_unload(struct mlx5_eswitch *esw,

			    struct mlx5_eswitch_rep *rep);

int mlx5e_nic_rep_load(struct mlx5_eswitch *esw, struct mlx5_eswitch_rep *rep);

void mlx5e_nic_rep_unload(struct mlx5_eswitch *esw,

			  struct mlx5_eswitch_rep *rep);

int mlx5e_add_sqs_fwd_rules(struct mlx5e_priv *priv);

void mlx5e_remove_sqs_fwd_rules(struct mlx5e_priv *priv);

int mlx5e_attr_get(struct net_device *dev, struct switchdev_attr *attr);

void mlx5e_handle_rx_cqe_rep(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe);

void mlx5e_update_hw_rep_counters(struct mlx5e_priv *priv);



/* common netdev helpers */

int mlx5e_create_indirect_rqt(struct mlx5e_priv *priv);
@@ -1031,12 +1017,6 @@ int mlx5e_open(struct net_device *netdev); 
void mlx5e_update_stats_work(struct work_struct *work);

u32 mlx5e_choose_lro_timeout(struct mlx5_core_dev *mdev, u32 wanted_timeout);



int mlx5e_get_offload_stats(int attr_id, const struct net_device *dev,

			    void *sp);

bool mlx5e_has_offload_stats(const struct net_device *dev, int attr_id);



bool mlx5e_is_uplink_rep(struct mlx5e_priv *priv);



/* mlx5e generic netdev management API */

struct net_device*

mlx5e_create_netdev(struct mlx5_core_dev *mdev, const struct mlx5e_profile *profile,

drivers/net/ethernet/mellanox/mlx5/core/en_main.c

+35 −63
Original line number Diff line number Diff line
@@ -35,9 +35,10 @@ 
#include <linux/mlx5/fs.h>

#include <net/vxlan.h>

#include <linux/bpf.h>

#include "eswitch.h"

#include "en.h"

#include "en_tc.h"

#include "eswitch.h"

#include "en_rep.h"

#include "vxlan.h"



struct mlx5e_rq_param {
@@ -3784,6 +3785,12 @@ static bool cqe_compress_heuristic(u32 link_speed, u32 pci_bw) 
		(pci_bw < 40000) && (pci_bw < link_speed));

}



static bool hw_lro_heuristic(u32 link_speed, u32 pci_bw)

{

	return !(link_speed && pci_bw &&

		 (pci_bw <= 16000) && (pci_bw < link_speed));

}



void mlx5e_set_rx_cq_mode_params(struct mlx5e_params *params, u8 cq_period_mode)

{

	params->rx_cq_period_mode = cq_period_mode;
@@ -3828,6 +3835,11 @@ void mlx5e_build_nic_params(struct mlx5_core_dev *mdev, 
	params->num_channels = max_channels;

	params->num_tc       = 1;



	mlx5e_get_max_linkspeed(mdev, &link_speed);

	mlx5e_get_pci_bw(mdev, &pci_bw);

	mlx5_core_dbg(mdev, "Max link speed = %d, PCI BW = %d\n",

		      link_speed, pci_bw);



	/* SQ */

	params->log_sq_size = is_kdump_kernel() ?

		MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE :
@@ -3836,13 +3848,9 @@ void mlx5e_build_nic_params(struct mlx5_core_dev *mdev, 
	/* set CQE compression */

	params->rx_cqe_compress_def = false;

	if (MLX5_CAP_GEN(mdev, cqe_compression) &&

	     MLX5_CAP_GEN(mdev, vport_group_manager)) {

		mlx5e_get_max_linkspeed(mdev, &link_speed);

		mlx5e_get_pci_bw(mdev, &pci_bw);

		mlx5_core_dbg(mdev, "Max link speed = %d, PCI BW = %d\n",

			       link_speed, pci_bw);

	     MLX5_CAP_GEN(mdev, vport_group_manager))

		params->rx_cqe_compress_def = cqe_compress_heuristic(link_speed, pci_bw);

	}



	MLX5E_SET_PFLAG(params, MLX5E_PFLAG_RX_CQE_COMPRESS, params->rx_cqe_compress_def);



	/* RQ */
@@ -3851,7 +3859,7 @@ void mlx5e_build_nic_params(struct mlx5_core_dev *mdev, 
	/* HW LRO */

	/* TODO: && MLX5_CAP_ETH(mdev, lro_cap) */

	if (params->rq_wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ)

		params->lro_en = true;

		params->lro_en = hw_lro_heuristic(link_speed, pci_bw);

	params->lro_timeout = mlx5e_choose_lro_timeout(mdev, MLX5E_DEFAULT_LRO_TIMEOUT);



	/* CQ moderation params */
@@ -4123,48 +4131,10 @@ static int mlx5e_init_nic_tx(struct mlx5e_priv *priv) 
	return 0;

}



static void mlx5e_register_vport_rep(struct mlx5_core_dev *mdev)

{

	struct mlx5_eswitch *esw = mdev->priv.eswitch;

	int total_vfs = MLX5_TOTAL_VPORTS(mdev);

	int vport;

	u8 mac[ETH_ALEN];



	if (!MLX5_CAP_GEN(mdev, vport_group_manager))

		return;



	mlx5_query_nic_vport_mac_address(mdev, 0, mac);



	for (vport = 1; vport < total_vfs; vport++) {

		struct mlx5_eswitch_rep rep;



		rep.load = mlx5e_vport_rep_load;

		rep.unload = mlx5e_vport_rep_unload;

		rep.vport = vport;

		ether_addr_copy(rep.hw_id, mac);

		mlx5_eswitch_register_vport_rep(esw, vport, &rep);

	}

}



static void mlx5e_unregister_vport_rep(struct mlx5_core_dev *mdev)

{

	struct mlx5_eswitch *esw = mdev->priv.eswitch;

	int total_vfs = MLX5_TOTAL_VPORTS(mdev);

	int vport;



	if (!MLX5_CAP_GEN(mdev, vport_group_manager))

		return;



	for (vport = 1; vport < total_vfs; vport++)

		mlx5_eswitch_unregister_vport_rep(esw, vport);

}



static void mlx5e_nic_enable(struct mlx5e_priv *priv)

{

	struct net_device *netdev = priv->netdev;

	struct mlx5_core_dev *mdev = priv->mdev;

	struct mlx5_eswitch *esw = mdev->priv.eswitch;

	struct mlx5_eswitch_rep rep;

	u16 max_mtu;



	mlx5e_init_l2_addr(priv);
@@ -4179,16 +4149,8 @@ static void mlx5e_nic_enable(struct mlx5e_priv *priv) 


	mlx5e_enable_async_events(priv);



	if (MLX5_CAP_GEN(mdev, vport_group_manager)) {

		mlx5_query_nic_vport_mac_address(mdev, 0, rep.hw_id);

		rep.load = mlx5e_nic_rep_load;

		rep.unload = mlx5e_nic_rep_unload;

		rep.vport = FDB_UPLINK_VPORT;

		rep.netdev = netdev;

		mlx5_eswitch_register_vport_rep(esw, 0, &rep);

	}



	mlx5e_register_vport_rep(mdev);

	if (MLX5_CAP_GEN(mdev, vport_group_manager))

		mlx5e_register_vport_reps(priv);



	if (netdev->reg_state != NETREG_REGISTERED)

		return;
@@ -4212,7 +4174,6 @@ static void mlx5e_nic_enable(struct mlx5e_priv *priv) 
static void mlx5e_nic_disable(struct mlx5e_priv *priv)

{

	struct mlx5_core_dev *mdev = priv->mdev;

	struct mlx5_eswitch *esw = mdev->priv.eswitch;



	rtnl_lock();

	if (netif_running(priv->netdev))
@@ -4221,9 +4182,10 @@ static void mlx5e_nic_disable(struct mlx5e_priv *priv) 
	rtnl_unlock();



	queue_work(priv->wq, &priv->set_rx_mode_work);

	mlx5e_unregister_vport_rep(mdev);



	if (MLX5_CAP_GEN(mdev, vport_group_manager))

		mlx5_eswitch_unregister_vport_rep(esw, 0);

		mlx5e_unregister_vport_reps(priv);



	mlx5e_disable_async_events(priv);

	mlx5_lag_remove(mdev);

}
@@ -4394,7 +4356,7 @@ static void *mlx5e_add(struct mlx5_core_dev *mdev) 
{

	struct mlx5_eswitch *esw = mdev->priv.eswitch;

	int total_vfs = MLX5_TOTAL_VPORTS(mdev);

	void *ppriv = NULL;

	struct mlx5e_rep_priv *rpriv = NULL;

	void *priv;

	int vport;

	int err;
@@ -4404,10 +4366,17 @@ static void *mlx5e_add(struct mlx5_core_dev *mdev) 
	if (err)

		return NULL;



	if (MLX5_CAP_GEN(mdev, vport_group_manager))

		ppriv = &esw->offloads.vport_reps[0];

	if (MLX5_CAP_GEN(mdev, vport_group_manager)) {

		rpriv = kzalloc(sizeof(*rpriv), GFP_KERNEL);

		if (!rpriv) {

			mlx5_core_warn(mdev,

				       "Not creating net device, Failed to alloc rep priv data\n");

			return NULL;

		}

		rpriv->rep = &esw->offloads.vport_reps[0];

	}



	netdev = mlx5e_create_netdev(mdev, &mlx5e_nic_profile, ppriv);

	netdev = mlx5e_create_netdev(mdev, &mlx5e_nic_profile, rpriv);

	if (!netdev) {

		mlx5_core_err(mdev, "mlx5e_create_netdev failed\n");

		goto err_unregister_reps;
@@ -4439,16 +4408,19 @@ err_unregister_reps: 
	for (vport = 1; vport < total_vfs; vport++)

		mlx5_eswitch_unregister_vport_rep(esw, vport);



	kfree(rpriv);

	return NULL;

}



static void mlx5e_remove(struct mlx5_core_dev *mdev, void *vpriv)

{

	struct mlx5e_priv *priv = vpriv;

	void *ppriv = priv->ppriv;



	unregister_netdev(priv->netdev);

	mlx5e_detach(mdev, vpriv);

	mlx5e_destroy_netdev(priv);

	kfree(ppriv);

}



static void *mlx5e_get_netdev(void *vpriv)

drivers/net/ethernet/mellanox/mlx5/core/en_rep.c

+535 −39

File changed.

Preview size limit exceeded, changes collapsed.

drivers/net/ethernet/mellanox/mlx5/core/en_rep.h

0 → 100644
+145 −0
Original line number Diff line number Diff line 
/*

 * Copyright (c) 2017, Mellanox Technologies. All rights reserved.

 *

 * This software is available to you under a choice of one of two

 * licenses.  You may choose to be licensed under the terms of the GNU

 * General Public License (GPL) Version 2, available from the file

 * COPYING in the main directory of this source tree, or the

 * OpenIB.org BSD license below:

 *

 *     Redistribution and use in source and binary forms, with or

 *     without modification, are permitted provided that the following

 *     conditions are met:

 *

 *      - Redistributions of source code must retain the above

 *        copyright notice, this list of conditions and the following

 *        disclaimer.

 *

 *      - Redistributions in binary form must reproduce the above

 *        copyright notice, this list of conditions and the following

 *        disclaimer in the documentation and/or other materials

 *        provided with the distribution.

 *

 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,

 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF

 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND

 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS

 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN

 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE

 * SOFTWARE.

 */



#ifndef __MLX5E_REP_H__

#define __MLX5E_REP_H__



#include <net/ip_tunnels.h>

#include <linux/rhashtable.h>

#include "eswitch.h"

#include "en.h"



struct mlx5e_neigh_update_table {

	struct rhashtable       neigh_ht;

	/* Save the neigh hash entries in a list in addition to the hash table

	 * (neigh_ht). In order to iterate easily over the neigh entries.

	 * Used for stats query.

	 */

	struct list_head	neigh_list;

	/* protect lookup/remove operations */

	spinlock_t              encap_lock;

	struct notifier_block   netevent_nb;

	struct delayed_work     neigh_stats_work;

	unsigned long           min_interval; /* jiffies */

};



struct mlx5e_rep_priv {

	struct mlx5_eswitch_rep *rep;

	struct mlx5e_neigh_update_table neigh_update;

};



struct mlx5e_neigh {

	struct net_device *dev;

	union {

		__be32	v4;

		struct in6_addr v6;

	} dst_ip;

	int family;

};



struct mlx5e_neigh_hash_entry {

	struct rhash_head rhash_node;

	struct mlx5e_neigh m_neigh;



	/* Save the neigh hash entry in a list on the representor in

	 * addition to the hash table. In order to iterate easily over the

	 * neighbour entries. Used for stats query.

	 */

	struct list_head neigh_list;



	/* encap list sharing the same neigh */

	struct list_head encap_list;



	/* valid only when the neigh reference is taken during

	 * neigh_update_work workqueue callback.

	 */

	struct neighbour *n;

	struct work_struct neigh_update_work;



	/* neigh hash entry can be deleted only when the refcount is zero.

	 * refcount is needed to avoid neigh hash entry removal by TC, while

	 * it's used by the neigh notification call.

	 */

	refcount_t refcnt;



	/* Save the last reported time offloaded trafic pass over one of the

	 * neigh hash entry flows. Use it to periodically update the neigh

	 * 'used' value and avoid neigh deleting by the kernel.

	 */

	unsigned long reported_lastuse;

};



enum {

	/* set when the encap entry is successfully offloaded into HW */

	MLX5_ENCAP_ENTRY_VALID     = BIT(0),

};



struct mlx5e_encap_entry {

	/* neigh hash entry list of encaps sharing the same neigh */

	struct list_head encap_list;

	struct mlx5e_neigh m_neigh;

	/* a node of the eswitch encap hash table which keeping all the encap

	 * entries

	 */

	struct hlist_node encap_hlist;

	struct list_head flows;

	u32 encap_id;

	struct ip_tunnel_info tun_info;

	unsigned char h_dest[ETH_ALEN];	/* destination eth addr	*/



	struct net_device *out_dev;

	int tunnel_type;

	u8 flags;

	char *encap_header;

	int encap_size;

};



void mlx5e_register_vport_reps(struct mlx5e_priv *priv);

void mlx5e_unregister_vport_reps(struct mlx5e_priv *priv);

bool mlx5e_is_uplink_rep(struct mlx5e_priv *priv);

int mlx5e_add_sqs_fwd_rules(struct mlx5e_priv *priv);

void mlx5e_remove_sqs_fwd_rules(struct mlx5e_priv *priv);



int mlx5e_get_offload_stats(int attr_id, const struct net_device *dev, void *sp);

bool mlx5e_has_offload_stats(const struct net_device *dev, int attr_id);



int mlx5e_attr_get(struct net_device *dev, struct switchdev_attr *attr);

void mlx5e_handle_rx_cqe_rep(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe);



int mlx5e_rep_encap_entry_attach(struct mlx5e_priv *priv,

				 struct mlx5e_encap_entry *e);

void mlx5e_rep_encap_entry_detach(struct mlx5e_priv *priv,

				  struct mlx5e_encap_entry *e);



void mlx5e_rep_queue_neigh_stats_work(struct mlx5e_priv *priv);



#endif /* __MLX5E_REP_H__ */

drivers/net/ethernet/mellanox/mlx5/core/en_rx.c

+4 −2
Original line number Diff line number Diff line
@@ -39,6 +39,7 @@ 
#include "en.h"

#include "en_tc.h"

#include "eswitch.h"

#include "en_rep.h"

#include "ipoib.h"



static inline bool mlx5e_rx_hw_stamp(struct mlx5e_tstamp *tstamp)
@@ -809,7 +810,8 @@ void mlx5e_handle_rx_cqe_rep(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe) 
{

	struct net_device *netdev = rq->netdev;

	struct mlx5e_priv *priv = netdev_priv(netdev);

	struct mlx5_eswitch_rep *rep = priv->ppriv;

	struct mlx5e_rep_priv *rpriv  = priv->ppriv;

	struct mlx5_eswitch_rep *rep = rpriv->rep;

	struct mlx5e_rx_wqe *wqe;

	struct sk_buff *skb;

	__be16 wqe_counter_be;
@@ -904,7 +906,7 @@ void mlx5e_handle_rx_cqe_mpwrq(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe) 
		goto mpwrq_cqe_out;

	}



	prefetch(skb->data);

	prefetchw(skb->data);

	cqe_bcnt = mpwrq_get_cqe_byte_cnt(cqe);



	mlx5e_mpwqe_fill_rx_skb(rq, cqe, wi, cqe_bcnt, skb);

CRA Git | Maintained and supported by SUSTech CRA and CCSE