Merge branch 'packet-sampling-offload'

Jiri Pirko says:

====================
Add support for offloading packet-sampling

Yotam says:

The first patch introduces the psample module, a netlink channel dedicated
to packet sampling implemented using generic netlink. This module provides
a generic way for kernel modules to sample packets, while not being tied
to any specific subsystem like NFLOG.

The second patch adds the sample tc action, which uses psample to randomly
sample packets that match a classifier. The user can configure the psample
group number, the sampling rate and the packet's truncation (to save
kernel-user traffic).

The last two patches add the support for offloading the matchall-sample
tc command in the mlxsw driver, for ingress qdiscs.

An example for psample usage can be found in the libpsample project at:
https://github.com/Mellanox/libpsample

v1->v2:
- Reword first patch's commit message
- Fix typo in comment in second patch
- Change order of tc_sample uapi enum to match convention
- Rename act_sample action callback tcf_sample -> tcf_sample_act
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

MAINTAINERS

@@ -9957,6 +9957,13 @@ L:	linuxppc-dev@lists.ozlabs.org
 S:	Maintained
 F:	drivers/block/ps3vram.c
 
+PSAMPLE PACKET SAMPLING SUPPORT:
+M:	Yotam Gigi <yotamg@mellanox.com>
+S:	Maintained
+F:	net/psample
+F:	include/net/psample.h
+F:	include/uapi/linux/psample.h
+
 PSTORE FILESYSTEM
 M:	Anton Vorontsov <anton@enomsg.org>
 M:	Colin Cross <ccross@android.com>

drivers/net/ethernet/mellanox/mlxsw/reg.h

@@ -4965,6 +4965,46 @@ static inline void mlxsw_reg_mlcr_pack(char *payload, u8 local_port,
 					   MLXSW_REG_MLCR_DURATION_MAX : 0);
 }
 
+/* MPSC - Monitoring Packet Sampling Configuration Register
+ * --------------------------------------------------------
+ * MPSC Register is used to configure the Packet Sampling mechanism.
+ */
+#define MLXSW_REG_MPSC_ID 0x9080
+#define MLXSW_REG_MPSC_LEN 0x1C
+
+MLXSW_REG_DEFINE(mpsc, MLXSW_REG_MPSC_ID, MLXSW_REG_MPSC_LEN);
+
+/* reg_mpsc_local_port
+ * Local port number
+ * Not supported for CPU port
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, mpsc, local_port, 0x00, 16, 8);
+
+/* reg_mpsc_e
+ * Enable sampling on port local_port
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpsc, e, 0x04, 30, 1);
+
+#define MLXSW_REG_MPSC_RATE_MAX 3500000000UL
+
+/* reg_mpsc_rate
+ * Sampling rate = 1 out of rate packets (with randomization around
+ * the point). Valid values are: 1 to MLXSW_REG_MPSC_RATE_MAX
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, mpsc, rate, 0x08, 0, 32);
+
+static inline void mlxsw_reg_mpsc_pack(char *payload, u8 local_port, bool e,
+				       u32 rate)
+{
+	MLXSW_REG_ZERO(mpsc, payload);
+	mlxsw_reg_mpsc_local_port_set(payload, local_port);
+	mlxsw_reg_mpsc_e_set(payload, e);
+	mlxsw_reg_mpsc_rate_set(payload, rate);
+}
+
 /* SBPR - Shared Buffer Pools Register
  * -----------------------------------
  * The SBPR configures and retrieves the shared buffer pools and configuration.
@@ -5429,6 +5469,7 @@ static const struct mlxsw_reg_info *mlxsw_reg_infos[] = {
 	MLXSW_REG(mpat),
 	MLXSW_REG(mpar),
 	MLXSW_REG(mlcr),
+	MLXSW_REG(mpsc),
 	MLXSW_REG(sbpr),
 	MLXSW_REG(sbcm),
 	MLXSW_REG(sbpm),

drivers/net/ethernet/mellanox/mlxsw/spectrum.c

@@ -57,6 +57,7 @@
 #include <net/pkt_cls.h>
 #include <net/tc_act/tc_mirred.h>
 #include <net/netevent.h>
+#include <net/tc_act/tc_sample.h>
 
 #include "spectrum.h"
 #include "pci.h"
@@ -469,6 +470,16 @@ static void mlxsw_sp_span_mirror_remove(struct mlxsw_sp_port *from,
 	mlxsw_sp_span_inspected_port_unbind(from, span_entry, type);
 }
 
+static int mlxsw_sp_port_sample_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				    bool enable, u32 rate)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char mpsc_pl[MLXSW_REG_MPSC_LEN];
+
+	mlxsw_reg_mpsc_pack(mpsc_pl, mlxsw_sp_port->local_port, enable, rate);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(mpsc), mpsc_pl);
+}
+
 static int mlxsw_sp_port_admin_status_set(struct mlxsw_sp_port *mlxsw_sp_port,
 					  bool is_up)
 {
@@ -1218,6 +1229,51 @@ mlxsw_sp_port_del_cls_matchall_mirror(struct mlxsw_sp_port *mlxsw_sp_port,
 	mlxsw_sp_span_mirror_remove(mlxsw_sp_port, to_port, span_type);
 }
 
+static int
+mlxsw_sp_port_add_cls_matchall_sample(struct mlxsw_sp_port *mlxsw_sp_port,
+				      struct tc_cls_matchall_offload *cls,
+				      const struct tc_action *a,
+				      bool ingress)
+{
+	int err;
+
+	if (!mlxsw_sp_port->sample)
+		return -EOPNOTSUPP;
+	if (rtnl_dereference(mlxsw_sp_port->sample->psample_group)) {
+		netdev_err(mlxsw_sp_port->dev, "sample already active\n");
+		return -EEXIST;
+	}
+	if (tcf_sample_rate(a) > MLXSW_REG_MPSC_RATE_MAX) {
+		netdev_err(mlxsw_sp_port->dev, "sample rate not supported\n");
+		return -EOPNOTSUPP;
+	}
+
+	rcu_assign_pointer(mlxsw_sp_port->sample->psample_group,
+			   tcf_sample_psample_group(a));
+	mlxsw_sp_port->sample->truncate = tcf_sample_truncate(a);
+	mlxsw_sp_port->sample->trunc_size = tcf_sample_trunc_size(a);
+	mlxsw_sp_port->sample->rate = tcf_sample_rate(a);
+
+	err = mlxsw_sp_port_sample_set(mlxsw_sp_port, true, tcf_sample_rate(a));
+	if (err)
+		goto err_port_sample_set;
+	return 0;
+
+err_port_sample_set:
+	RCU_INIT_POINTER(mlxsw_sp_port->sample->psample_group, NULL);
+	return err;
+}
+
+static void
+mlxsw_sp_port_del_cls_matchall_sample(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	if (!mlxsw_sp_port->sample)
+		return;
+
+	mlxsw_sp_port_sample_set(mlxsw_sp_port, false, 1);
+	RCU_INIT_POINTER(mlxsw_sp_port->sample->psample_group, NULL);
+}
+
 static int mlxsw_sp_port_add_cls_matchall(struct mlxsw_sp_port *mlxsw_sp_port,
 					  __be16 protocol,
 					  struct tc_cls_matchall_offload *cls,
@@ -1248,6 +1304,10 @@ static int mlxsw_sp_port_add_cls_matchall(struct mlxsw_sp_port *mlxsw_sp_port,
 		mirror = &mall_tc_entry->mirror;
 		err = mlxsw_sp_port_add_cls_matchall_mirror(mlxsw_sp_port,
 							    mirror, a, ingress);
+	} else if (is_tcf_sample(a) && protocol == htons(ETH_P_ALL)) {
+		mall_tc_entry->type = MLXSW_SP_PORT_MALL_SAMPLE;
+		err = mlxsw_sp_port_add_cls_matchall_sample(mlxsw_sp_port, cls,
+							    a, ingress);
 	} else {
 		err = -EOPNOTSUPP;
 	}
@@ -1281,6 +1341,9 @@ static void mlxsw_sp_port_del_cls_matchall(struct mlxsw_sp_port *mlxsw_sp_port,
 		mlxsw_sp_port_del_cls_matchall_mirror(mlxsw_sp_port,
 						      &mall_tc_entry->mirror);
 		break;
+	case MLXSW_SP_PORT_MALL_SAMPLE:
+		mlxsw_sp_port_del_cls_matchall_sample(mlxsw_sp_port);
+		break;
 	default:
 		WARN_ON(1);
 	}
@@ -2259,6 +2322,13 @@ static int __mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port,
 		goto err_alloc_stats;
 	}
 
+	mlxsw_sp_port->sample = kzalloc(sizeof(*mlxsw_sp_port->sample),
+					GFP_KERNEL);
+	if (!mlxsw_sp_port->sample) {
+		err = -ENOMEM;
+		goto err_alloc_sample;
+	}
+
 	mlxsw_sp_port->hw_stats.cache =
 		kzalloc(sizeof(*mlxsw_sp_port->hw_stats.cache), GFP_KERNEL);
 
@@ -2387,6 +2457,8 @@ static int __mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port,
 err_port_swid_set:
 	kfree(mlxsw_sp_port->hw_stats.cache);
 err_alloc_hw_stats:
+	kfree(mlxsw_sp_port->sample);
+err_alloc_sample:
 	free_percpu(mlxsw_sp_port->pcpu_stats);
 err_alloc_stats:
 	kfree(mlxsw_sp_port->untagged_vlans);
@@ -2433,6 +2505,7 @@ static void __mlxsw_sp_port_remove(struct mlxsw_sp *mlxsw_sp, u8 local_port)
 	mlxsw_sp_port_swid_set(mlxsw_sp_port, MLXSW_PORT_SWID_DISABLED_PORT);
 	mlxsw_sp_port_module_unmap(mlxsw_sp, mlxsw_sp_port->local_port);
 	kfree(mlxsw_sp_port->hw_stats.cache);
+	kfree(mlxsw_sp_port->sample);
 	free_percpu(mlxsw_sp_port->pcpu_stats);
 	kfree(mlxsw_sp_port->untagged_vlans);
 	kfree(mlxsw_sp_port->active_vlans);
@@ -2734,6 +2807,41 @@ static void mlxsw_sp_rx_listener_mark_func(struct sk_buff *skb, u8 local_port,
 	return mlxsw_sp_rx_listener_no_mark_func(skb, local_port, priv);
 }
 
+static void mlxsw_sp_rx_listener_sample_func(struct sk_buff *skb, u8 local_port,
+					     void *priv)
+{
+	struct mlxsw_sp *mlxsw_sp = priv;
+	struct mlxsw_sp_port *mlxsw_sp_port = mlxsw_sp->ports[local_port];
+	struct psample_group *psample_group;
+	u32 size;
+
+	if (unlikely(!mlxsw_sp_port)) {
+		dev_warn_ratelimited(mlxsw_sp->bus_info->dev, "Port %d: sample skb received for non-existent port\n",
+				     local_port);
+		goto out;
+	}
+	if (unlikely(!mlxsw_sp_port->sample)) {
+		dev_warn_ratelimited(mlxsw_sp->bus_info->dev, "Port %d: sample skb received on unsupported port\n",
+				     local_port);
+		goto out;
+	}
+
+	size = mlxsw_sp_port->sample->truncate ?
+		  mlxsw_sp_port->sample->trunc_size : skb->len;
+
+	rcu_read_lock();
+	psample_group = rcu_dereference(mlxsw_sp_port->sample->psample_group);
+	if (!psample_group)
+		goto out_unlock;
+	psample_sample_packet(psample_group, skb, size,
+			      mlxsw_sp_port->dev->ifindex, 0,
+			      mlxsw_sp_port->sample->rate);
+out_unlock:
+	rcu_read_unlock();
+out:
+	consume_skb(skb);
+}
+
 #define MLXSW_SP_RXL_NO_MARK(_trap_id, _action, _trap_group, _is_ctrl)	\
 	MLXSW_RXL(mlxsw_sp_rx_listener_no_mark_func, _trap_id, _action,	\
 		  _is_ctrl, SP_##_trap_group, DISCARD)
@@ -2769,6 +2877,9 @@ static const struct mlxsw_listener mlxsw_sp_listener[] = {
 	MLXSW_SP_RXL_NO_MARK(RTR_INGRESS0, TRAP_TO_CPU, REMOTE_ROUTE, false),
 	MLXSW_SP_RXL_NO_MARK(HOST_MISS_IPV4, TRAP_TO_CPU, ARP_MISS, false),
 	MLXSW_SP_RXL_NO_MARK(BGP_IPV4, TRAP_TO_CPU, BGP_IPV4, false),
+	/* PKT Sample trap */
+	MLXSW_RXL(mlxsw_sp_rx_listener_sample_func, PKT_SAMPLE, MIRROR_TO_CPU,
+		  false, SP_IP2ME, DISCARD)
 };
 
 static int mlxsw_sp_cpu_policers_set(struct mlxsw_core *mlxsw_core)

drivers/net/ethernet/mellanox/mlxsw/spectrum.h

@@ -46,6 +46,7 @@
 #include <linux/dcbnl.h>
 #include <linux/in6.h>
 #include <linux/notifier.h>
+#include <net/psample.h>
 
 #include "port.h"
 #include "core.h"
@@ -229,6 +230,7 @@ struct mlxsw_sp_span_entry {
 
 enum mlxsw_sp_port_mall_action_type {
 	MLXSW_SP_PORT_MALL_MIRROR,
+	MLXSW_SP_PORT_MALL_SAMPLE,
 };
 
 struct mlxsw_sp_port_mall_mirror_tc_entry {
@@ -315,6 +317,13 @@ struct mlxsw_sp_port_pcpu_stats {
 	u32			tx_dropped;
 };
 
+struct mlxsw_sp_port_sample {
+	struct psample_group __rcu *psample_group;
+	u32 trunc_size;
+	u32 rate;
+	bool truncate;
+};
+
 struct mlxsw_sp_port {
 	struct net_device *dev;
 	struct mlxsw_sp_port_pcpu_stats __percpu *pcpu_stats;
@@ -361,6 +370,7 @@ struct mlxsw_sp_port {
 		struct rtnl_link_stats64 *cache;
 		struct delayed_work update_dw;
 	} hw_stats;
+	struct mlxsw_sp_port_sample *sample;
 };
 
 struct mlxsw_sp_port *mlxsw_sp_port_lower_dev_hold(struct net_device *dev);

drivers/net/ethernet/mellanox/mlxsw/trap.h

@@ -54,6 +54,7 @@ enum {
 	MLXSW_TRAP_ID_IGMP_V2_REPORT = 0x32,
 	MLXSW_TRAP_ID_IGMP_V2_LEAVE = 0x33,
 	MLXSW_TRAP_ID_IGMP_V3_REPORT = 0x34,
+	MLXSW_TRAP_ID_PKT_SAMPLE = 0x38,
 	MLXSW_TRAP_ID_ARPBC = 0x50,
 	MLXSW_TRAP_ID_ARPUC = 0x51,
 	MLXSW_TRAP_ID_MTUERROR = 0x52,

include/net/psample.h

+#ifndef __NET_PSAMPLE_H
+#define __NET_PSAMPLE_H
+
+#include <uapi/linux/psample.h>
+#include <linux/module.h>
+#include <linux/list.h>
+
+struct psample_group {
+	struct list_head list;
+	struct net *net;
+	u32 group_num;
+	u32 refcount;
+	u32 seq;
+};
+
+struct psample_group *psample_group_get(struct net *net, u32 group_num);
+void psample_group_put(struct psample_group *group);
+
+#if IS_ENABLED(CONFIG_PSAMPLE)
+
+void psample_sample_packet(struct psample_group *group, struct sk_buff *skb,
+			   u32 trunc_size, int in_ifindex, int out_ifindex,
+			   u32 sample_rate);
+
+#else
+
+static inline void psample_sample_packet(struct psample_group *group,
+					 struct sk_buff *skb, u32 trunc_size,
+					 int in_ifindex, int out_ifindex,
+					 u32 sample_rate)
+{
+}
+
+#endif
+
+#endif /* __NET_PSAMPLE_H */

include/net/tc_act/tc_sample.h

+#ifndef __NET_TC_SAMPLE_H
+#define __NET_TC_SAMPLE_H
+
+#include <net/act_api.h>
+#include <linux/tc_act/tc_sample.h>
+#include <net/psample.h>
+
+struct tcf_sample {
+	struct tc_action common;
+	u32 rate;
+	bool truncate;
+	u32 trunc_size;
+	struct psample_group __rcu *psample_group;
+	u32 psample_group_num;
+	struct list_head tcfm_list;
+	struct rcu_head rcu;
+};
+#define to_sample(a) ((struct tcf_sample *)a)
+
+static inline bool is_tcf_sample(const struct tc_action *a)
+{
+#ifdef CONFIG_NET_CLS_ACT
+	return a->ops && a->ops->type == TCA_ACT_SAMPLE;
+#else
+	return false;
+#endif
+}
+
+static inline __u32 tcf_sample_rate(const struct tc_action *a)
+{
+	return to_sample(a)->rate;
+}
+
+static inline bool tcf_sample_truncate(const struct tc_action *a)
+{
+	return to_sample(a)->truncate;
+}
+
+static inline int tcf_sample_trunc_size(const struct tc_action *a)
+{
+	return to_sample(a)->trunc_size;
+}
+
+static inline struct psample_group *
+tcf_sample_psample_group(const struct tc_action *a)
+{
+	return rcu_dereference(to_sample(a)->psample_group);
+}
+
+#endif /* __NET_TC_SAMPLE_H */

include/uapi/linux/Kbuild

@@ -305,6 +305,7 @@ header-y += netrom.h
 header-y += net_namespace.h
 header-y += net_tstamp.h
 header-y += nfc.h
+header-y += psample.h
 header-y += nfs2.h
 header-y += nfs3.h
 header-y += nfs4.h

include/uapi/linux/psample.h

+#ifndef __UAPI_PSAMPLE_H
+#define __UAPI_PSAMPLE_H
+
+enum {
+	/* sampled packet metadata */
+	PSAMPLE_ATTR_IIFINDEX,
+	PSAMPLE_ATTR_OIFINDEX,
+	PSAMPLE_ATTR_ORIGSIZE,
+	PSAMPLE_ATTR_SAMPLE_GROUP,
+	PSAMPLE_ATTR_GROUP_SEQ,
+	PSAMPLE_ATTR_SAMPLE_RATE,
+	PSAMPLE_ATTR_DATA,
+
+	/* commands attributes */
+	PSAMPLE_ATTR_GROUP_REFCOUNT,
+
+	__PSAMPLE_ATTR_MAX
+};
+
+enum psample_command {
+	PSAMPLE_CMD_SAMPLE,
+	PSAMPLE_CMD_GET_GROUP,
+	PSAMPLE_CMD_NEW_GROUP,
+	PSAMPLE_CMD_DEL_GROUP,
+};
+
+/* Can be overridden at runtime by module option */
+#define PSAMPLE_ATTR_MAX (__PSAMPLE_ATTR_MAX - 1)
+
+#define PSAMPLE_NL_MCGRP_CONFIG_NAME "config"
+#define PSAMPLE_NL_MCGRP_SAMPLE_NAME "packets"
+#define PSAMPLE_GENL_NAME "psample"
+#define PSAMPLE_GENL_VERSION 1
+
+#endif

include/uapi/linux/tc_act/Kbuild

@@ -4,6 +4,7 @@ header-y += tc_defact.h
 header-y += tc_gact.h
 header-y += tc_ipt.h
 header-y += tc_mirred.h
+header-y += tc_sample.h
 header-y += tc_nat.h
 header-y += tc_pedit.h
 header-y += tc_skbedit.h

include/uapi/linux/tc_act/tc_sample.h

+#ifndef __LINUX_TC_SAMPLE_H
+#define __LINUX_TC_SAMPLE_H
+
+#include <linux/types.h>
+#include <linux/pkt_cls.h>
+#include <linux/if_ether.h>
+
+#define TCA_ACT_SAMPLE 26
+
+struct tc_sample {
+	tc_gen;
+};
+
+enum {
+	TCA_SAMPLE_UNSPEC,
+	TCA_SAMPLE_TM,
+	TCA_SAMPLE_PARMS,
+	TCA_SAMPLE_RATE,
+	TCA_SAMPLE_TRUNC_SIZE,
+	TCA_SAMPLE_PSAMPLE_GROUP,
+	TCA_SAMPLE_PAD,
+	__TCA_SAMPLE_MAX
+};
+#define TCA_SAMPLE_MAX (__TCA_SAMPLE_MAX - 1)
+
+#endif

net/Kconfig

@@ -390,6 +390,7 @@ source "net/9p/Kconfig"
 source "net/caif/Kconfig"
 source "net/ceph/Kconfig"
 source "net/nfc/Kconfig"
+source "net/psample/Kconfig"
 
 config LWTUNNEL
 	bool "Network light weight tunnels"

net/Makefile

@@ -70,6 +70,7 @@ obj-$(CONFIG_DNS_RESOLVER)	+= dns_resolver/
 obj-$(CONFIG_CEPH_LIB)		+= ceph/
 obj-$(CONFIG_BATMAN_ADV)	+= batman-adv/
 obj-$(CONFIG_NFC)		+= nfc/
+obj-$(CONFIG_PSAMPLE)		+= psample/
 obj-$(CONFIG_OPENVSWITCH)	+= openvswitch/
 obj-$(CONFIG_VSOCKETS)	+= vmw_vsock/
 obj-$(CONFIG_MPLS)		+= mpls/

net/psample/Kconfig

+#
+# psample packet sampling configuration
+#
+
+menuconfig PSAMPLE
+	depends on NET
+	tristate "Packet-sampling netlink channel"
+	default n
+	help
+	  Say Y here to add support for packet-sampling netlink channel
+	  This netlink channel allows transferring packets alongside some
+	  metadata to userspace.
+
+	  To compile this support as a module, choose M here: the module will
+	  be called psample.

net/psample/Makefile

+#
+# Makefile for the psample netlink channel
+#
+
+obj-$(CONFIG_PSAMPLE) += psample.o

net/psample/psample.c

+/*
+ * net/psample/psample.c - Netlink channel for packet sampling
+ * Copyright (c) 2017 Yotam Gigi <yotamg@mellanox.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/skbuff.h>
+#include <linux/module.h>
+#include <net/net_namespace.h>
+#include <net/sock.h>
+#include <net/netlink.h>
+#include <net/genetlink.h>
+#include <net/psample.h>
+#include <linux/spinlock.h>
+
+#define PSAMPLE_MAX_PACKET_SIZE 0xffff
+
+static LIST_HEAD(psample_groups_list);
+static DEFINE_SPINLOCK(psample_groups_lock);
+
+/* multicast groups */
+enum psample_nl_multicast_groups {
+	PSAMPLE_NL_MCGRP_CONFIG,
+	PSAMPLE_NL_MCGRP_SAMPLE,
+};
+
+static const struct genl_multicast_group psample_nl_mcgrps[] = {
+	[PSAMPLE_NL_MCGRP_CONFIG] = { .name = PSAMPLE_NL_MCGRP_CONFIG_NAME },
+	[PSAMPLE_NL_MCGRP_SAMPLE] = { .name = PSAMPLE_NL_MCGRP_SAMPLE_NAME },
+};
+
+static struct genl_family psample_nl_family __ro_after_init;
+
+static int psample_group_nl_fill(struct sk_buff *msg,
+				 struct psample_group *group,
+				 enum psample_command cmd, u32 portid, u32 seq,
+				 int flags)
+{
+	void *hdr;
+	int ret;
+
+	hdr = genlmsg_put(msg, portid, seq, &psample_nl_family, flags, cmd);
+	if (!hdr)
+		return -EMSGSIZE;
+
+	ret = nla_put_u32(msg, PSAMPLE_ATTR_SAMPLE_GROUP, group->group_num);
+	if (ret < 0)
+		goto error;
+
+	ret = nla_put_u32(msg, PSAMPLE_ATTR_GROUP_REFCOUNT, group->refcount);
+	if (ret < 0)
+		goto error;
+
+	ret = nla_put_u32(msg, PSAMPLE_ATTR_GROUP_SEQ, group->seq);
+	if (ret < 0)
+		goto error;
+
+	genlmsg_end(msg, hdr);
+	return 0;
+
+error:
+	genlmsg_cancel(msg, hdr);
+	return -EMSGSIZE;
+}
+
+static int psample_nl_cmd_get_group_dumpit(struct sk_buff *msg,
+					   struct netlink_callback *cb)
+{
+	struct psample_group *group;
+	int start = cb->args[0];
+	int idx = 0;
+	int err;
+
+	spin_lock(&psample_groups_lock);
+	list_for_each_entry(group, &psample_groups_list, list) {
+		if (!net_eq(group->net, sock_net(msg->sk)))
+			continue;
+		if (idx < start) {
+			idx++;
+			continue;
+		}
+		err = psample_group_nl_fill(msg, group, PSAMPLE_CMD_NEW_GROUP,
+					    NETLINK_CB(cb->skb).portid,
+					    cb->nlh->nlmsg_seq, NLM_F_MULTI);
+		if (err)
+			break;
+		idx++;
+	}
+
+	spin_unlock(&psample_groups_lock);
+	cb->args[0] = idx;
+	return msg->len;
+}
+
+static const struct genl_ops psample_nl_ops[] = {
+	{
+		.cmd = PSAMPLE_CMD_GET_GROUP,
+		.dumpit = psample_nl_cmd_get_group_dumpit,
+		/* can be retrieved by unprivileged users */
+	}
+};
+
+static struct genl_family psample_nl_family __ro_after_init = {
+	.name		= PSAMPLE_GENL_NAME,
+	.version	= PSAMPLE_GENL_VERSION,
+	.maxattr	= PSAMPLE_ATTR_MAX,
+	.netnsok	= true,
+	.module		= THIS_MODULE,
+	.mcgrps		= psample_nl_mcgrps,
+	.ops		= psample_nl_ops,
+	.n_ops		= ARRAY_SIZE(psample_nl_ops),
+	.n_mcgrps	= ARRAY_SIZE(psample_nl_mcgrps),
+};
+
+static void psample_group_notify(struct psample_group *group,
+				 enum psample_command cmd)
+{
+	struct sk_buff *msg;
+	int err;
+
+	msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_ATOMIC);
+	if (!msg)
+		return;
+
+	err = psample_group_nl_fill(msg, group, cmd, 0, 0, NLM_F_MULTI);
+	if (!err)
+		genlmsg_multicast_netns(&psample_nl_family, group->net, msg, 0,
+					PSAMPLE_NL_MCGRP_CONFIG, GFP_ATOMIC);
+	else
+		nlmsg_free(msg);
+}
+
+static struct psample_group *psample_group_create(struct net *net,
+						  u32 group_num)
+{
+	struct psample_group *group;
+
+	group = kzalloc(sizeof(*group), GFP_ATOMIC);
+	if (!group)
+		return NULL;
+
+	group->net = net;
+	group->group_num = group_num;
+	list_add_tail(&group->list, &psample_groups_list);
+
+	psample_group_notify(group, PSAMPLE_CMD_NEW_GROUP);
+	return group;
+}
+
+static void psample_group_destroy(struct psample_group *group)
+{
+	psample_group_notify(group, PSAMPLE_CMD_DEL_GROUP);
+	list_del(&group->list);
+	kfree(group);
+}
+
+static struct psample_group *
+psample_group_lookup(struct net *net, u32 group_num)
+{
+	struct psample_group *group;
+
+	list_for_each_entry(group, &psample_groups_list, list)
+		if ((group->group_num == group_num) && (group->net == net))
+			return group;
+	return NULL;
+}
+
+struct psample_group *psample_group_get(struct net *net, u32 group_num)
+{
+	struct psample_group *group;
+
+	spin_lock(&psample_groups_lock);
+
+	group = psample_group_lookup(net, group_num);
+	if (!group) {
+		group = psample_group_create(net, group_num);
+		if (!group)
+			goto out;
+	}
+	group->refcount++;
+
+out:
+	spin_unlock(&psample_groups_lock);
+	return group;
+}
+EXPORT_SYMBOL_GPL(psample_group_get);
+
+void psample_group_put(struct psample_group *group)
+{
+	spin_lock(&psample_groups_lock);
+
+	if (--group->refcount == 0)
+		psample_group_destroy(group);
+
+	spin_unlock(&psample_groups_lock);
+}
+EXPORT_SYMBOL_GPL(psample_group_put);
+
+void psample_sample_packet(struct psample_group *group, struct sk_buff *skb,
+			   u32 trunc_size, int in_ifindex, int out_ifindex,
+			   u32 sample_rate)
+{
+	struct sk_buff *nl_skb;
+	int data_len;
+	int meta_len;
+	void *data;
+	int ret;
+
+	meta_len = (in_ifindex ? nla_total_size(sizeof(u16)) : 0) +
+		   (out_ifindex ? nla_total_size(sizeof(u16)) : 0) +
+		   nla_total_size(sizeof(u32)) +	/* sample_rate */
+		   nla_total_size(sizeof(u32)) +	/* orig_size */
+		   nla_total_size(sizeof(u32)) +	/* group_num */
+		   nla_total_size(sizeof(u32));		/* seq */
+
+	data_len = min(skb->len, trunc_size);
+	if (meta_len + nla_total_size(data_len) > PSAMPLE_MAX_PACKET_SIZE)
+		data_len = PSAMPLE_MAX_PACKET_SIZE - meta_len - NLA_HDRLEN
+			    - NLA_ALIGNTO;
+
+	nl_skb = genlmsg_new(meta_len + data_len, GFP_ATOMIC);
+	if (unlikely(!nl_skb))
+		return;
+
+	data = genlmsg_put(nl_skb, 0, 0, &psample_nl_family, 0,
+			   PSAMPLE_CMD_SAMPLE);
+	if (unlikely(!data))
+		goto error;
+
+	if (in_ifindex) {
+		ret = nla_put_u16(nl_skb, PSAMPLE_ATTR_IIFINDEX, in_ifindex);
+		if (unlikely(ret < 0))
+			goto error;
+	}
+
+	if (out_ifindex) {
+		ret = nla_put_u16(nl_skb, PSAMPLE_ATTR_OIFINDEX, out_ifindex);
+		if (unlikely(ret < 0))
+			goto error;
+	}
+
+	ret = nla_put_u32(nl_skb, PSAMPLE_ATTR_SAMPLE_RATE, sample_rate);
+	if (unlikely(ret < 0))
+		goto error;
+
+	ret = nla_put_u32(nl_skb, PSAMPLE_ATTR_ORIGSIZE, skb->len);
+	if (unlikely(ret < 0))
+		goto error;
+
+	ret = nla_put_u32(nl_skb, PSAMPLE_ATTR_SAMPLE_GROUP, group->group_num);
+	if (unlikely(ret < 0))
+		goto error;
+
+	ret = nla_put_u32(nl_skb, PSAMPLE_ATTR_GROUP_SEQ, group->seq++);
+	if (unlikely(ret < 0))
+		goto error;
+
+	if (data_len) {
+		int nla_len = nla_total_size(data_len);
+		struct nlattr *nla;
+
+		nla = (struct nlattr *)skb_put(nl_skb, nla_len);
+		nla->nla_type = PSAMPLE_ATTR_DATA;
+		nla->nla_len = nla_attr_size(data_len);
+
+		if (skb_copy_bits(skb, 0, nla_data(nla), data_len))
+			goto error;
+	}
+
+	genlmsg_end(nl_skb, data);
+	genlmsg_multicast_netns(&psample_nl_family, group->net, nl_skb, 0,
+				PSAMPLE_NL_MCGRP_SAMPLE, GFP_ATOMIC);
+
+	return;
+error:
+	pr_err_ratelimited("Could not create psample log message\n");
+	nlmsg_free(nl_skb);
+}
+EXPORT_SYMBOL_GPL(psample_sample_packet);
+
+static int __init psample_module_init(void)
+{
+	return genl_register_family(&psample_nl_family);
+}
+
+static void __exit psample_module_exit(void)
+{
+	genl_unregister_family(&psample_nl_family);
+}
+
+module_init(psample_module_init);
+module_exit(psample_module_exit);
+
+MODULE_AUTHOR("Yotam Gigi <yotamg@mellanox.com>");
+MODULE_DESCRIPTION("netlink channel for packet sampling");
+MODULE_LICENSE("GPL v2");

net/sched/Kconfig

@@ -650,6 +650,18 @@ config NET_ACT_MIRRED
 	  To compile this code as a module, choose M here: the
 	  module will be called act_mirred.
 
+config NET_ACT_SAMPLE
+        tristate "Traffic Sampling"
+        depends on NET_CLS_ACT
+        select PSAMPLE
+        ---help---
+	  Say Y here to allow packet sampling tc action. The packet sample
+	  action consists of statistically choosing packets and sampling
+	  them using the psample module.
+
+	  To compile this code as a module, choose M here: the
+	  module will be called act_sample.
+
 config NET_ACT_IPT
         tristate "IPtables targets"
         depends on NET_CLS_ACT && NETFILTER && IP_NF_IPTABLES

net/sched/Makefile

@@ -10,6 +10,7 @@ obj-$(CONFIG_NET_CLS_ACT)	+= act_api.o
 obj-$(CONFIG_NET_ACT_POLICE)	+= act_police.o
 obj-$(CONFIG_NET_ACT_GACT)	+= act_gact.o
 obj-$(CONFIG_NET_ACT_MIRRED)	+= act_mirred.o
+obj-$(CONFIG_NET_ACT_SAMPLE)	+= act_sample.o
 obj-$(CONFIG_NET_ACT_IPT)	+= act_ipt.o
 obj-$(CONFIG_NET_ACT_NAT)	+= act_nat.o
 obj-$(CONFIG_NET_ACT_PEDIT)	+= act_pedit.o

net/sched/act_sample.c

+/*
+ * net/sched/act_sample.c - Packet sampling tc action
+ * Copyright (c) 2017 Yotam Gigi <yotamg@mellanox.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/errno.h>
+#include <linux/skbuff.h>
+#include <linux/rtnetlink.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/gfp.h>
+#include <net/net_namespace.h>
+#include <net/netlink.h>
+#include <net/pkt_sched.h>
+#include <linux/tc_act/tc_sample.h>
+#include <net/tc_act/tc_sample.h>
+#include <net/psample.h>
+
+#include <linux/if_arp.h>
+
+#define SAMPLE_TAB_MASK     7
+static unsigned int sample_net_id;
+static struct tc_action_ops act_sample_ops;
+
+static const struct nla_policy sample_policy[TCA_SAMPLE_MAX + 1] = {
+	[TCA_SAMPLE_PARMS]		= { .len = sizeof(struct tc_sample) },
+	[TCA_SAMPLE_RATE]		= { .type = NLA_U32 },
+	[TCA_SAMPLE_TRUNC_SIZE]		= { .type = NLA_U32 },
+	[TCA_SAMPLE_PSAMPLE_GROUP]	= { .type = NLA_U32 },
+};
+
+static int tcf_sample_init(struct net *net, struct nlattr *nla,
+			   struct nlattr *est, struct tc_action **a, int ovr,
+			   int bind)
+{
+	struct tc_action_net *tn = net_generic(net, sample_net_id);
+	struct nlattr *tb[TCA_SAMPLE_MAX + 1];
+	struct psample_group *psample_group;
+	struct tc_sample *parm;
+	struct tcf_sample *s;
+	bool exists = false;
+	int ret;
+
+	if (!nla)
+		return -EINVAL;
+	ret = nla_parse_nested(tb, TCA_SAMPLE_MAX, nla, sample_policy);
+	if (ret < 0)
+		return ret;
+	if (!tb[TCA_SAMPLE_PARMS] || !tb[TCA_SAMPLE_RATE] ||
+	    !tb[TCA_SAMPLE_PSAMPLE_GROUP])
+		return -EINVAL;
+
+	parm = nla_data(tb[TCA_SAMPLE_PARMS]);
+
+	exists = tcf_hash_check(tn, parm->index, a, bind);
+	if (exists && bind)
+		return 0;
+
+	if (!exists) {
+		ret = tcf_hash_create(tn, parm->index, est, a,
+				      &act_sample_ops, bind, false);
+		if (ret)
+			return ret;
+		ret = ACT_P_CREATED;
+	} else {
+		tcf_hash_release(*a, bind);
+		if (!ovr)
+			return -EEXIST;
+	}
+	s = to_sample(*a);
+
+	ASSERT_RTNL();
+	s->tcf_action = parm->action;
+	s->rate = nla_get_u32(tb[TCA_SAMPLE_RATE]);
+	s->psample_group_num = nla_get_u32(tb[TCA_SAMPLE_PSAMPLE_GROUP]);
+	psample_group = psample_group_get(net, s->psample_group_num);
+	if (!psample_group)
+		return -ENOMEM;
+	RCU_INIT_POINTER(s->psample_group, psample_group);
+
+	if (tb[TCA_SAMPLE_TRUNC_SIZE]) {
+		s->truncate = true;
+		s->trunc_size = nla_get_u32(tb[TCA_SAMPLE_TRUNC_SIZE]);
+	}
+
+	if (ret == ACT_P_CREATED)
+		tcf_hash_insert(tn, *a);
+	return ret;
+}
+
+static void tcf_sample_cleanup_rcu(struct rcu_head *rcu)
+{
+	struct tcf_sample *s = container_of(rcu, struct tcf_sample, rcu);
+	struct psample_group *psample_group;
+
+	psample_group = rcu_dereference_protected(s->psample_group, 1);
+	RCU_INIT_POINTER(s->psample_group, NULL);
+	psample_group_put(psample_group);
+}
+
+static void tcf_sample_cleanup(struct tc_action *a, int bind)
+{
+	struct tcf_sample *s = to_sample(a);
+
+	call_rcu(&s->rcu, tcf_sample_cleanup_rcu);
+}
+
+static bool tcf_sample_dev_ok_push(struct net_device *dev)
+{
+	switch (dev->type) {
+	case ARPHRD_TUNNEL:
+	case ARPHRD_TUNNEL6:
+	case ARPHRD_SIT:
+	case ARPHRD_IPGRE:
+	case ARPHRD_VOID:
+	case ARPHRD_NONE:
+		return false;
+	default:
+		return true;
+	}
+}
+
+static int tcf_sample_act(struct sk_buff *skb, const struct tc_action *a,
+			  struct tcf_result *res)
+{
+	struct tcf_sample *s = to_sample(a);
+	struct psample_group *psample_group;
+	int retval;
+	int size;
+	int iif;
+	int oif;
+
+	tcf_lastuse_update(&s->tcf_tm);
+	bstats_cpu_update(this_cpu_ptr(s->common.cpu_bstats), skb);
+	retval = READ_ONCE(s->tcf_action);
+
+	rcu_read_lock();
+	psample_group = rcu_dereference(s->psample_group);
+
+	/* randomly sample packets according to rate */
+	if (psample_group && (prandom_u32() % s->rate == 0)) {
+		if (!skb_at_tc_ingress(skb)) {
+			iif = skb->skb_iif;
+			oif = skb->dev->ifindex;
+		} else {
+			iif = skb->dev->ifindex;
+			oif = 0;
+		}
+
+		/* on ingress, the mac header gets popped, so push it back */
+		if (skb_at_tc_ingress(skb) && tcf_sample_dev_ok_push(skb->dev))
+			skb_push(skb, skb->mac_len);
+
+		size = s->truncate ? s->trunc_size : skb->len;
+		psample_sample_packet(psample_group, skb, size, iif, oif,
+				      s->rate);
+
+		if (skb_at_tc_ingress(skb) && tcf_sample_dev_ok_push(skb->dev))
+			skb_pull(skb, skb->mac_len);
+	}
+
+	rcu_read_unlock();
+	return retval;
+}
+
+static int tcf_sample_dump(struct sk_buff *skb, struct tc_action *a,
+			   int bind, int ref)
+{
+	unsigned char *b = skb_tail_pointer(skb);
+	struct tcf_sample *s = to_sample(a);
+	struct tc_sample opt = {
+		.index      = s->tcf_index,
+		.action     = s->tcf_action,
+		.refcnt     = s->tcf_refcnt - ref,
+		.bindcnt    = s->tcf_bindcnt - bind,
+	};
+	struct tcf_t t;
+
+	if (nla_put(skb, TCA_SAMPLE_PARMS, sizeof(opt), &opt))
+		goto nla_put_failure;
+
+	tcf_tm_dump(&t, &s->tcf_tm);
+	if (nla_put_64bit(skb, TCA_SAMPLE_TM, sizeof(t), &t, TCA_SAMPLE_PAD))
+		goto nla_put_failure;
+
+	if (nla_put_u32(skb, TCA_SAMPLE_RATE, s->rate))
+		goto nla_put_failure;
+
+	if (s->truncate)
+		if (nla_put_u32(skb, TCA_SAMPLE_TRUNC_SIZE, s->trunc_size))
+			goto nla_put_failure;
+
+	if (nla_put_u32(skb, TCA_SAMPLE_PSAMPLE_GROUP, s->psample_group_num))
+		goto nla_put_failure;
+	return skb->len;
+
+nla_put_failure:
+	nlmsg_trim(skb, b);
+	return -1;
+}
+
+static int tcf_sample_walker(struct net *net, struct sk_buff *skb,
+			     struct netlink_callback *cb, int type,
+			     const struct tc_action_ops *ops)
+{
+	struct tc_action_net *tn = net_generic(net, sample_net_id);
+
+	return tcf_generic_walker(tn, skb, cb, type, ops);
+}
+
+static int tcf_sample_search(struct net *net, struct tc_action **a, u32 index)
+{
+	struct tc_action_net *tn = net_generic(net, sample_net_id);
+
+	return tcf_hash_search(tn, a, index);
+}
+
+static struct tc_action_ops act_sample_ops = {
+	.kind	  = "sample",
+	.type	  = TCA_ACT_SAMPLE,
+	.owner	  = THIS_MODULE,
+	.act	  = tcf_sample_act,
+	.dump	  = tcf_sample_dump,
+	.init	  = tcf_sample_init,
+	.cleanup  = tcf_sample_cleanup,
+	.walk	  = tcf_sample_walker,
+	.lookup	  = tcf_sample_search,
+	.size	  = sizeof(struct tcf_sample),
+};
+
+static __net_init int sample_init_net(struct net *net)
+{
+	struct tc_action_net *tn = net_generic(net, sample_net_id);
+
+	return tc_action_net_init(tn, &act_sample_ops, SAMPLE_TAB_MASK);
+}
+
+static void __net_exit sample_exit_net(struct net *net)
+{
+	struct tc_action_net *tn = net_generic(net, sample_net_id);
+
+	tc_action_net_exit(tn);
+}
+
+static struct pernet_operations sample_net_ops = {
+	.init = sample_init_net,
+	.exit = sample_exit_net,
+	.id   = &sample_net_id,
+	.size = sizeof(struct tc_action_net),
+};
+
+static int __init sample_init_module(void)
+{
+	return tcf_register_action(&act_sample_ops, &sample_net_ops);
+}
+
+static void __exit sample_cleanup_module(void)
+{
+	tcf_unregister_action(&act_sample_ops, &sample_net_ops);
+}
+
+module_init(sample_init_module);
+module_exit(sample_cleanup_module);
+
+MODULE_AUTHOR("Yotam Gigi <yotamg@mellanox.com>");
+MODULE_DESCRIPTION("Packet sampling action");
+MODULE_LICENSE("GPL v2");