[patch net-next RFC 0/9] qdisc RED offload

[patch net-next RFC 0/9] qdisc RED offload

[Jiri Pirko]

From: Jiri Pirko <jiri@mellanox.com>

Nogah says:

Add an offload support for RED qdisc for mlxsw driver.
The first patch adds the ability to offload RED qdisc by using
ndo_setup_tc. It gives RED three commands, to offload, change or delete
the qdisc, to get the qdisc generic stats and to get it's RED xstats.
There is no enforcement on a driver to offload or not offload the qdisc and
it is up to the driver to decide.
RED qdisc is first being created and only later graft to a parent (unless
it is a root qdisc). For that reason the return value of the offload
replace command that is called in the init process doesn't reflect actual
offload state. The offload state is determined in the dump function so it
can be reflected to the user. This function is also responsible for stats
update.

The patchses 2-3 change the name of TC_SETUP_MQPRIO & TC_SETUP_CBS to match
with the new convention of QDISC prefix.
The rest of the patchset is driver support for the qdisc. Currently only
as root qdisc that is being set on the default traffic class. It supports
only the following parameters of RED: min, max, probability and ECN mode.
Limit and burst size related params are being ignored at this moment.

Nogah Frankel (8):
  net_sch: red: Add offload ability to RED qdisc
  net_sch: mqprio: Change TC_SETUP_MQPRIO to TC_SETUP_QDISC_MQPRIO
  net_sch: cbs: Change TC_SETUP_CBS to TC_SETUP_QDISC_CBS
  mlxsw: reg: Add cwtp & cwtpm registers
  mlxsw: spectrum: Support RED qdisc offload
  mlxsw: spectrum: Collect tclass related stats periodically
  mlxsw: spectrum: Support RED xstats
  mlxsw: spectrum: Support general qdisc stats

Yuval Mintz (1):
  mlxsw: reg: Add ext and tc-cong counter groups

 drivers/net/ethernet/amd/xgbe/xgbe-drv.c           |   2 +-
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c    |   2 +-
 drivers/net/ethernet/broadcom/bnxt/bnxt.c          |   2 +-
 drivers/net/ethernet/freescale/dpaa/dpaa_eth.c     |   2 +-
 .../net/ethernet/hisilicon/hns3/hns3pf/hns3_enet.c |   2 +-
 drivers/net/ethernet/intel/fm10k/fm10k_netdev.c    |   2 +-
 drivers/net/ethernet/intel/i40e/i40e_main.c        |   2 +-
 drivers/net/ethernet/intel/igb/igb_main.c          |   2 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c      |   2 +-
 drivers/net/ethernet/mellanox/mlx4/en_netdev.c     |   2 +-
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c  |   2 +-
 drivers/net/ethernet/mellanox/mlxsw/Makefile       |   3 +-
 drivers/net/ethernet/mellanox/mlxsw/reg.h          | 206 +++++++++++++++++
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     |  36 +++
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  38 +++
 .../net/ethernet/mellanox/mlxsw/spectrum_qdisc.c   | 254 +++++++++++++++++++++
 drivers/net/ethernet/sfc/falcon/tx.c               |   2 +-
 drivers/net/ethernet/sfc/tx.c                      |   2 +-
 drivers/net/ethernet/ti/netcp_core.c               |   2 +-
 include/linux/netdevice.h                          |   5 +-
 include/net/pkt_cls.h                              |  30 +++
 include/uapi/linux/pkt_sched.h                     |   1 +
 net/sched/sch_cbs.c                                |   4 +-
 net/sched/sch_mqprio.c                             |   5 +-
 net/sched/sch_red.c                                |  90 ++++++++
 25 files changed, 679 insertions(+), 21 deletions(-)
 create mode 100644 drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c

-- 
2.9.5

[patch net-next RFC 1/9] net_sch: red: Add offload ability to RED qdisc

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Add the ability to offload RED qdisc by using ndo_setup_tc.
There are four commands for RED offloading:
* TC_RED_SET: handles set and change.
* TC_RED_DESTROY: handle qdisc destroy.
* TC_RED_STATS: update the qdiscs counters (given as reference)
* TC_RED_XSTAT: returns red xstats.

Whether RED is being offloaded is being determined every time dump action
is being called because parent change of this qdisc could change its
offload state but doesn't require any RED function to be called.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 include/linux/netdevice.h      |  1 +
 include/net/pkt_cls.h          | 30 ++++++++++++++
 include/uapi/linux/pkt_sched.h |  1 +
 net/sched/sch_red.c            | 90 ++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 122 insertions(+)

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 5e02f79..1dc9833 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -777,6 +777,7 @@ enum tc_setup_type {
 	TC_SETUP_CLSBPF,
 	TC_SETUP_BLOCK,
 	TC_SETUP_CBS,
+	TC_SETUP_QDISC_RED,
 };
 
 /* These structures hold the attributes of xdp state that are being passed
diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h
index 04caa24..6108c6a 100644
--- a/include/net/pkt_cls.h
+++ b/include/net/pkt_cls.h
@@ -712,4 +712,34 @@ struct tc_cookie {
 	u8  *data;
 	u32 len;
 };
+
+enum tc_red_command {
+	TC_RED_REPLACE,
+	TC_RED_DESTROY,
+	TC_RED_STATS,
+	TC_RED_XSTATS,
+};
+
+struct tc_red_qopt_offload_params {
+	u32 min;
+	u32 max;
+	u32 probability;
+	bool is_ecn;
+};
+struct tc_red_qopt_offload_stats {
+	struct gnet_stats_basic_packed *bstats;
+	struct gnet_stats_queue *qstats;
+};
+
+struct tc_red_qopt_offload {
+	enum tc_red_command command;
+	u32 handle;
+	u32 parent;
+	union {
+		struct tc_red_qopt_offload_params set;
+		struct tc_red_qopt_offload_stats stats;
+		struct red_stats *xstats;
+	};
+};
+
 #endif
diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
index 0e88cc2..743c42a 100644
--- a/include/uapi/linux/pkt_sched.h
+++ b/include/uapi/linux/pkt_sched.h
@@ -255,6 +255,7 @@ struct tc_red_qopt {
 #define TC_RED_ECN		1
 #define TC_RED_HARDDROP		2
 #define TC_RED_ADAPTATIVE	4
+#define TC_RED_OFFLOADED	8
 };
 
 struct tc_red_xstats {
diff --git a/net/sched/sch_red.c b/net/sched/sch_red.c
index fdfdb56..836e0e4 100644
--- a/net/sched/sch_red.c
+++ b/net/sched/sch_red.c
@@ -19,6 +19,7 @@
 #include <linux/kernel.h>
 #include <linux/skbuff.h>
 #include <net/pkt_sched.h>
+#include <net/pkt_cls.h>
 #include <net/inet_ecn.h>
 #include <net/red.h>
 
@@ -148,11 +149,27 @@ static void red_reset(struct Qdisc *sch)
 	red_restart(&q->vars);
 }
 
+static void red_unoffload(struct Qdisc *sch)
+{
+	struct net_device *dev = qdisc_dev(sch);
+	struct tc_red_qopt_offload opt = {
+		.handle = sch->handle,
+		.command = TC_RED_DESTROY,
+		.parent = sch->parent,
+	};
+
+	if (!tc_can_offload(dev) || !dev->netdev_ops->ndo_setup_tc)
+		return;
+
+	dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_RED,  &opt);
+}
+
 static void red_destroy(struct Qdisc *sch)
 {
 	struct red_sched_data *q = qdisc_priv(sch);
 
 	del_timer_sync(&q->adapt_timer);
+	red_unoffload(sch);
 	qdisc_destroy(q->qdisc);
 }
 
@@ -162,6 +179,28 @@ static const struct nla_policy red_policy[TCA_RED_MAX + 1] = {
 	[TCA_RED_MAX_P] = { .type = NLA_U32 },
 };
 
+static int red_offload(struct Qdisc *sch)
+{
+	struct red_sched_data *q = qdisc_priv(sch);
+	struct net_device *dev = qdisc_dev(sch);
+	struct tc_red_qopt_offload opt = {
+		.handle = sch->handle,
+		.command = TC_RED_REPLACE,
+		.parent = sch->parent,
+		.set = {
+			.min = q->parms.qth_min >> q->parms.Wlog,
+			.max = q->parms.qth_max >> q->parms.Wlog,
+			.probability = q->parms.max_P,
+			.is_ecn = red_use_ecn(q),
+		},
+	};
+
+	if (!tc_can_offload(dev) || !dev->netdev_ops->ndo_setup_tc)
+		return -EOPNOTSUPP;
+
+	return dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_RED, &opt);
+}
+
 static int red_change(struct Qdisc *sch, struct nlattr *opt)
 {
 	struct red_sched_data *q = qdisc_priv(sch);
@@ -219,6 +258,7 @@ static int red_change(struct Qdisc *sch, struct nlattr *opt)
 		red_start_of_idle_period(&q->vars);
 
 	sch_tree_unlock(sch);
+	red_offload(sch);
 	return 0;
 }
 
@@ -244,6 +284,32 @@ static int red_init(struct Qdisc *sch, struct nlattr *opt)
 	return red_change(sch, opt);
 }
 
+static int red_dump_offload(struct Qdisc *sch, struct tc_red_qopt *opt)
+{
+	struct net_device *dev = qdisc_dev(sch);
+	struct tc_red_qopt_offload hw_stats = {
+		.handle = sch->handle,
+		.parent = sch->parent,
+		.command = TC_RED_STATS,
+		.stats.bstats = &sch->bstats,
+		.stats.qstats = &sch->qstats,
+	};
+	int err;
+
+	if (!tc_can_offload(dev) || !dev->netdev_ops->ndo_setup_tc)
+		return 0;
+
+	err = dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_RED,
+					    &hw_stats);
+	if (err == -EOPNOTSUPP)
+		return 0;
+
+	if (!err)
+		opt->flags |= TC_RED_OFFLOADED;
+
+	return err;
+}
+
 static int red_dump(struct Qdisc *sch, struct sk_buff *skb)
 {
 	struct red_sched_data *q = qdisc_priv(sch);
@@ -257,8 +323,13 @@ static int red_dump(struct Qdisc *sch, struct sk_buff *skb)
 		.Plog		= q->parms.Plog,
 		.Scell_log	= q->parms.Scell_log,
 	};
+	int err;
 
 	sch->qstats.backlog = q->qdisc->qstats.backlog;
+	err = red_dump_offload(sch, &opt);
+	if (err)
+		goto nla_put_failure;
+
 	opts = nla_nest_start(skb, TCA_OPTIONS);
 	if (opts == NULL)
 		goto nla_put_failure;
@@ -275,6 +346,7 @@ static int red_dump(struct Qdisc *sch, struct sk_buff *skb)
 static int red_dump_stats(struct Qdisc *sch, struct gnet_dump *d)
 {
 	struct red_sched_data *q = qdisc_priv(sch);
+	struct net_device *dev = qdisc_dev(sch);
 	struct tc_red_xstats st = {
 		.early	= q->stats.prob_drop + q->stats.forced_drop,
 		.pdrop	= q->stats.pdrop,
@@ -282,6 +354,24 @@ static int red_dump_stats(struct Qdisc *sch, struct gnet_dump *d)
 		.marked	= q->stats.prob_mark + q->stats.forced_mark,
 	};
 
+	if (tc_can_offload(dev) &&  dev->netdev_ops->ndo_setup_tc) {
+		struct red_stats hw_stats = {0};
+		struct tc_red_qopt_offload hw_stats_request = {
+			.handle = sch->handle,
+			.parent = sch->parent,
+			.command = TC_RED_XSTATS,
+			.xstats = &hw_stats,
+		};
+		if (!dev->netdev_ops->ndo_setup_tc(dev,
+						   TC_SETUP_QDISC_RED,
+						   &hw_stats_request)) {
+			st.early += hw_stats.prob_drop + hw_stats.forced_drop;
+			st.pdrop += hw_stats.pdrop;
+			st.other += hw_stats.other;
+			st.marked += hw_stats.prob_mark + hw_stats.forced_mark;
+		}
+	}
+
 	return gnet_stats_copy_app(d, &st, sizeof(st));
 }
 
-- 
2.9.5

[patch net-next RFC 2/9] net_sch: mqprio: Change TC_SETUP_MQPRIO to TC_SETUP_QDISC_MQPRIO

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Change TC_SETUP_MQPRIO to TC_SETUP_QDISC_MQPRIO to match the new
convention.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/amd/xgbe/xgbe-drv.c               | 2 +-
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c        | 2 +-
 drivers/net/ethernet/broadcom/bnxt/bnxt.c              | 2 +-
 drivers/net/ethernet/freescale/dpaa/dpaa_eth.c         | 2 +-
 drivers/net/ethernet/hisilicon/hns3/hns3pf/hns3_enet.c | 2 +-
 drivers/net/ethernet/intel/fm10k/fm10k_netdev.c        | 2 +-
 drivers/net/ethernet/intel/i40e/i40e_main.c            | 2 +-
 drivers/net/ethernet/intel/ixgbe/ixgbe_main.c          | 2 +-
 drivers/net/ethernet/mellanox/mlx4/en_netdev.c         | 2 +-
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c      | 2 +-
 drivers/net/ethernet/sfc/falcon/tx.c                   | 2 +-
 drivers/net/ethernet/sfc/tx.c                          | 2 +-
 drivers/net/ethernet/ti/netcp_core.c                   | 2 +-
 include/linux/netdevice.h                              | 2 +-
 net/sched/sch_mqprio.c                                 | 5 +++--
 15 files changed, 17 insertions(+), 16 deletions(-)

diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-drv.c b/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
index 3d53153..a74a8fb 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
@@ -2206,7 +2206,7 @@ static int xgbe_setup_tc(struct net_device *netdev, enum tc_setup_type type,
 	struct tc_mqprio_qopt *mqprio = type_data;
 	u8 tc;
 
-	if (type != TC_SETUP_MQPRIO)
+	if (type != TC_SETUP_QDISC_MQPRIO)
 		return -EOPNOTSUPP;
 
 	mqprio->hw = TC_MQPRIO_HW_OFFLOAD_TCS;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index 1216c1f..4c739d5 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -4289,7 +4289,7 @@ int __bnx2x_setup_tc(struct net_device *dev, enum tc_setup_type type,
 {
 	struct tc_mqprio_qopt *mqprio = type_data;
 
-	if (type != TC_SETUP_MQPRIO)
+	if (type != TC_SETUP_QDISC_MQPRIO)
 		return -EOPNOTSUPP;
 
 	mqprio->hw = TC_MQPRIO_HW_OFFLOAD_TCS;
diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
index 5ce9506..9ab2785 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
@@ -7384,7 +7384,7 @@ static int bnxt_setup_tc(struct net_device *dev, enum tc_setup_type type,
 	switch (type) {
 	case TC_SETUP_BLOCK:
 		return bnxt_setup_tc_block(dev, type_data);
-	case TC_SETUP_MQPRIO: {
+	case TC_SETUP_QDISC_MQPRIO: {
 		struct tc_mqprio_qopt *mqprio = type_data;
 
 		mqprio->hw = TC_MQPRIO_HW_OFFLOAD_TCS;
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
index a8d0be8..c9ab893 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
@@ -351,7 +351,7 @@ static int dpaa_setup_tc(struct net_device *net_dev, enum tc_setup_type type,
 	u8 num_tc;
 	int i;
 
-	if (type != TC_SETUP_MQPRIO)
+	if (type != TC_SETUP_QDISC_MQPRIO)
 		return -EOPNOTSUPP;
 
 	mqprio->hw = TC_MQPRIO_HW_OFFLOAD_TCS;
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hns3_enet.c
index c6c5b2a..39d93e6 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hns3_enet.c
@@ -1250,7 +1250,7 @@ static int hns3_setup_tc(struct net_device *netdev, void *type_data)
 static int hns3_nic_setup_tc(struct net_device *dev, enum tc_setup_type type,
 			     void *type_data)
 {
-	if (type != TC_SETUP_MQPRIO)
+	if (type != TC_SETUP_QDISC_MQPRIO)
 		return -EOPNOTSUPP;
 
 	return hns3_setup_tc(dev, type_data);
diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_netdev.c b/drivers/net/ethernet/intel/fm10k/fm10k_netdev.c
index 81e4425..adc62fb 100644
--- a/drivers/net/ethernet/intel/fm10k/fm10k_netdev.c
+++ b/drivers/net/ethernet/intel/fm10k/fm10k_netdev.c
@@ -1389,7 +1389,7 @@ static int __fm10k_setup_tc(struct net_device *dev, enum tc_setup_type type,
 {
 	struct tc_mqprio_qopt *mqprio = type_data;
 
-	if (type != TC_SETUP_MQPRIO)
+	if (type != TC_SETUP_QDISC_MQPRIO)
 		return -EOPNOTSUPP;
 
 	mqprio->hw = TC_MQPRIO_HW_OFFLOAD_TCS;
diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c b/drivers/net/ethernet/intel/i40e/i40e_main.c
index 54ff34f..97b2a82 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_main.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
@@ -6751,7 +6751,7 @@ static int i40e_setup_tc(struct net_device *netdev, void *type_data)
 static int __i40e_setup_tc(struct net_device *netdev, enum tc_setup_type type,
 			   void *type_data)
 {
-	if (type != TC_SETUP_MQPRIO)
+	if (type != TC_SETUP_QDISC_MQPRIO)
 		return -EOPNOTSUPP;
 
 	return i40e_setup_tc(netdev, type_data);
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index 7f503d3..1d176a8 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -9434,7 +9434,7 @@ static int __ixgbe_setup_tc(struct net_device *dev, enum tc_setup_type type,
 	switch (type) {
 	case TC_SETUP_BLOCK:
 		return ixgbe_setup_tc_block(dev, type_data);
-	case TC_SETUP_MQPRIO:
+	case TC_SETUP_QDISC_MQPRIO:
 		return ixgbe_setup_tc_mqprio(dev, type_data);
 	default:
 		return -EOPNOTSUPP;
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
index d611df2..9f3a9f9 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
@@ -135,7 +135,7 @@ static int __mlx4_en_setup_tc(struct net_device *dev, enum tc_setup_type type,
 {
 	struct tc_mqprio_qopt *mqprio = type_data;
 
-	if (type != TC_SETUP_MQPRIO)
+	if (type != TC_SETUP_QDISC_MQPRIO)
 		return -EOPNOTSUPP;
 
 	if (mqprio->num_tc && mqprio->num_tc != MLX4_EN_NUM_UP_HIGH)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 560b208..6f47b8b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -3144,7 +3144,7 @@ int mlx5e_setup_tc(struct net_device *dev, enum tc_setup_type type,
 	case TC_SETUP_BLOCK:
 		return mlx5e_setup_tc_block(dev, type_data);
 #endif
-	case TC_SETUP_MQPRIO:
+	case TC_SETUP_QDISC_MQPRIO:
 		return mlx5e_setup_tc_mqprio(dev, type_data);
 	default:
 		return -EOPNOTSUPP;
diff --git a/drivers/net/ethernet/sfc/falcon/tx.c b/drivers/net/ethernet/sfc/falcon/tx.c
index 6a75f41..1b978d6 100644
--- a/drivers/net/ethernet/sfc/falcon/tx.c
+++ b/drivers/net/ethernet/sfc/falcon/tx.c
@@ -435,7 +435,7 @@ int ef4_setup_tc(struct net_device *net_dev, enum tc_setup_type type,
 	unsigned tc, num_tc;
 	int rc;
 
-	if (type != TC_SETUP_MQPRIO)
+	if (type != TC_SETUP_QDISC_MQPRIO)
 		return -EOPNOTSUPP;
 
 	num_tc = mqprio->num_tc;
diff --git a/drivers/net/ethernet/sfc/tx.c b/drivers/net/ethernet/sfc/tx.c
index 32bf1fe..ea27b8a 100644
--- a/drivers/net/ethernet/sfc/tx.c
+++ b/drivers/net/ethernet/sfc/tx.c
@@ -663,7 +663,7 @@ int efx_setup_tc(struct net_device *net_dev, enum tc_setup_type type,
 	unsigned tc, num_tc;
 	int rc;
 
-	if (type != TC_SETUP_MQPRIO)
+	if (type != TC_SETUP_QDISC_MQPRIO)
 		return -EOPNOTSUPP;
 
 	num_tc = mqprio->num_tc;
diff --git a/drivers/net/ethernet/ti/netcp_core.c b/drivers/net/ethernet/ti/netcp_core.c
index 437d362..15e2e30 100644
--- a/drivers/net/ethernet/ti/netcp_core.c
+++ b/drivers/net/ethernet/ti/netcp_core.c
@@ -1887,7 +1887,7 @@ static int netcp_setup_tc(struct net_device *dev, enum tc_setup_type type,
 	/* setup tc must be called under rtnl lock */
 	ASSERT_RTNL();
 
-	if (type != TC_SETUP_MQPRIO)
+	if (type != TC_SETUP_QDISC_MQPRIO)
 		return -EOPNOTSUPP;
 
 	mqprio->hw = TC_MQPRIO_HW_OFFLOAD_TCS;
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 1dc9833..732e2dc 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -770,7 +770,7 @@ typedef u16 (*select_queue_fallback_t)(struct net_device *dev,
 				       struct sk_buff *skb);
 
 enum tc_setup_type {
-	TC_SETUP_MQPRIO,
+	TC_SETUP_QDISC_MQPRIO,
 	TC_SETUP_CLSU32,
 	TC_SETUP_CLSFLOWER,
 	TC_SETUP_CLSMATCHALL,
diff --git a/net/sched/sch_mqprio.c b/net/sched/sch_mqprio.c
index 4d5ed45..b85885a9 100644
--- a/net/sched/sch_mqprio.c
+++ b/net/sched/sch_mqprio.c
@@ -50,7 +50,8 @@ static void mqprio_destroy(struct Qdisc *sch)
 		switch (priv->mode) {
 		case TC_MQPRIO_MODE_DCB:
 		case TC_MQPRIO_MODE_CHANNEL:
-			dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_MQPRIO,
+			dev->netdev_ops->ndo_setup_tc(dev,
+						      TC_SETUP_QDISC_MQPRIO,
 						      &mqprio);
 			break;
 		default:
@@ -265,7 +266,7 @@ static int mqprio_init(struct Qdisc *sch, struct nlattr *opt)
 			return -EINVAL;
 		}
 		err = dev->netdev_ops->ndo_setup_tc(dev,
-						    TC_SETUP_MQPRIO,
+						    TC_SETUP_QDISC_MQPRIO,
 						    &mqprio);
 		if (err)
 			return err;
-- 
2.9.5

[patch net-next RFC 3/9] net_sch: cbs: Change TC_SETUP_CBS to TC_SETUP_QDISC_CBS

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Change TC_SETUP_CBS to TC_SETUP_QDISC_CBS to match the new convention..

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/intel/igb/igb_main.c | 2 +-
 include/linux/netdevice.h                 | 2 +-
 net/sched/sch_cbs.c                       | 4 ++--
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/intel/igb/igb_main.c b/drivers/net/ethernet/intel/igb/igb_main.c
index b3d730f..fe0c2f4 100644
--- a/drivers/net/ethernet/intel/igb/igb_main.c
+++ b/drivers/net/ethernet/intel/igb/igb_main.c
@@ -2488,7 +2488,7 @@ static int igb_setup_tc(struct net_device *dev, enum tc_setup_type type,
 	struct igb_adapter *adapter = netdev_priv(dev);
 
 	switch (type) {
-	case TC_SETUP_CBS:
+	case TC_SETUP_QDISC_CBS:
 		return igb_offload_cbs(adapter, type_data);
 
 	default:
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 732e2dc..ec58ef8 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -776,7 +776,7 @@ enum tc_setup_type {
 	TC_SETUP_CLSMATCHALL,
 	TC_SETUP_CLSBPF,
 	TC_SETUP_BLOCK,
-	TC_SETUP_CBS,
+	TC_SETUP_QDISC_CBS,
 	TC_SETUP_QDISC_RED,
 };
 
diff --git a/net/sched/sch_cbs.c b/net/sched/sch_cbs.c
index bdb533b..7a72980 100644
--- a/net/sched/sch_cbs.c
+++ b/net/sched/sch_cbs.c
@@ -212,7 +212,7 @@ static void cbs_disable_offload(struct net_device *dev,
 	cbs.queue = q->queue;
 	cbs.enable = 0;
 
-	err = ops->ndo_setup_tc(dev, TC_SETUP_CBS, &cbs);
+	err = ops->ndo_setup_tc(dev, TC_SETUP_QDISC_CBS, &cbs);
 	if (err < 0)
 		pr_warn("Couldn't disable CBS offload for queue %d\n",
 			cbs.queue);
@@ -236,7 +236,7 @@ static int cbs_enable_offload(struct net_device *dev, struct cbs_sched_data *q,
 	cbs.idleslope = opt->idleslope;
 	cbs.sendslope = opt->sendslope;
 
-	err = ops->ndo_setup_tc(dev, TC_SETUP_CBS, &cbs);
+	err = ops->ndo_setup_tc(dev, TC_SETUP_QDISC_CBS, &cbs);
 	if (err < 0)
 		return err;
 
-- 
2.9.5

[patch net-next RFC 4/9] mlxsw: reg: Add cwtp & cwtpm registers

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

This patch adds 2 new registers:
 - Congestion WRED ECN TClass Profile Register [CWTP]
 - Congestion WRED ECN TClass and Pool Mapping Register [CWTPM]

These registers would later be needed to offload RED-related
functionality to the HW.

Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/reg.h | 187 ++++++++++++++++++++++++++++++
 1 file changed, 187 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index db6cd26..48f051f 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -1758,6 +1758,191 @@ static inline void mlxsw_reg_spvmlr_pack(char *payload, u8 local_port,
 	}
 }
 
+/* CWTP - Congetion WRED ECN TClass Profile
+ * ----------------------------------------
+ * Configures the profiles for queues of egress port and traffic class
+ */
+#define MLXSW_REG_CWTP_ID 0x2802
+#define MLXSW_REG_CWTP_BASE_LEN 0x28
+#define MLXSW_REG_CWTP_PROFILE_DATA_REC_LEN 0x08
+#define MLXSW_REG_CWTP_LEN 0x40
+
+MLXSW_REG_DEFINE(cwtp, MLXSW_REG_CWTP_ID, MLXSW_REG_CWTP_LEN);
+
+/* reg_cwtp_local_port
+ * Local port number
+ * Not supported for CPU port
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, cwtp, local_port, 0, 16, 8);
+
+/* reg_cwtp_traffic_class
+ * Traffic Class to configure
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, cwtp, traffic_class, 32, 0, 8);
+
+/* reg_cwtp_profile_min
+ * Minimum Average Queue Size of the profile in cells.
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, cwtp, profile_min, MLXSW_REG_CWTP_BASE_LEN,
+		     0, 20, MLXSW_REG_CWTP_PROFILE_DATA_REC_LEN, 0, false);
+
+/* reg_cwtp_profile_percent
+ * Percentage of WRED and ECN marking for maximum Average Queue size
+ * Range is 0 to 100, units of integer percentage
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, cwtp, profile_percent, MLXSW_REG_CWTP_BASE_LEN,
+		     24, 7, MLXSW_REG_CWTP_PROFILE_DATA_REC_LEN, 4, false);
+
+/* reg_cwtp_profile_max
+ * Maximum Average Queue size of the profile in cells
+ * Access: RW
+ */
+MLXSW_ITEM32_INDEXED(reg, cwtp, profile_max, MLXSW_REG_CWTP_BASE_LEN,
+		     0, 20, MLXSW_REG_CWTP_PROFILE_DATA_REC_LEN, 4, false);
+
+#define MLXSW_REG_CWTP_MIN_VALUE 64
+#define MLXSW_REG_CWTP_MAX_PROFILE 2
+#define MLXSW_REG_CWTP_DEFAULT_PROFILE 1
+
+static inline void mlxsw_reg_cwtp_pack(char *payload, u8 local_port,
+				       u8 traffic_class)
+{
+	int i;
+
+	MLXSW_REG_ZERO(cwtp, payload);
+	mlxsw_reg_cwtp_local_port_set(payload, local_port);
+	mlxsw_reg_cwtp_traffic_class_set(payload, traffic_class);
+
+	for (i = 0; i <= MLXSW_REG_CWTP_MAX_PROFILE; i++) {
+		mlxsw_reg_cwtp_profile_min_set(payload, i,
+					       MLXSW_REG_CWTP_MIN_VALUE);
+		mlxsw_reg_cwtp_profile_max_set(payload, i,
+					       MLXSW_REG_CWTP_MIN_VALUE);
+	}
+}
+
+#define MLXSW_REG_CWTP_PROFILE_TO_INDEX(profile) (profile - 1)
+
+static inline void
+mlxsw_reg_cwtp_profile_pack(char *payload, u8 profile, u32 min, u32 max,
+			    u32 probability)
+{
+	u8 index = MLXSW_REG_CWTP_PROFILE_TO_INDEX(profile);
+
+	mlxsw_reg_cwtp_profile_min_set(payload, index, min);
+	mlxsw_reg_cwtp_profile_max_set(payload, index, max);
+	mlxsw_reg_cwtp_profile_percent_set(payload, index, probability);
+}
+
+/* CWTPM - Congestion WRED ECN TClass and Pool Mapping
+ * ---------------------------------------------------
+ * The CWTPM register maps each egress port and traffic class to profile num.
+ */
+#define MLXSW_REG_CWTPM_ID 0x2803
+#define MLXSW_REG_CWTPM_LEN 0x44
+
+MLXSW_REG_DEFINE(cwtpm, MLXSW_REG_CWTPM_ID, MLXSW_REG_CWTPM_LEN);
+
+/* reg_cwtpm_local_port
+ * Local port number
+ * Not supported for CPU port
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, cwtpm, local_port, 0, 16, 8);
+
+/* reg_cwtpm_traffic_class
+ * Traffic Class to configure
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, cwtpm, traffic_class, 32, 0, 8);
+
+/* reg_cwtpm_ew
+ * Control enablement of WRED for traffic class:
+ * 0 - Disable
+ * 1 - Enable
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, ew, 36, 1, 1);
+
+/* reg_cwtpm_ee
+ * Control enablement of ECN for traffic class:
+ * 0 - Disable
+ * 1 - Enable
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, ee, 36, 0, 1);
+
+/* reg_cwtpm_tcp_g
+ * TCP Green Profile.
+ * Index of the profile within {port, traffic class} to use.
+ * 0 for disabling both WRED and ECN for this type of traffic.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, tcp_g, 52, 0, 2);
+
+/* reg_cwtpm_tcp_y
+ * TCP Yellow Profile.
+ * Index of the profile within {port, traffic class} to use.
+ * 0 for disabling both WRED and ECN for this type of traffic.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, tcp_y, 56, 16, 2);
+
+/* reg_cwtpm_tcp_r
+ * TCP Red Profile.
+ * Index of the profile within {port, traffic class} to use.
+ * 0 for disabling both WRED and ECN for this type of traffic.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, tcp_r, 56, 0, 2);
+
+/* reg_cwtpm_ntcp_g
+ * Non-TCP Green Profile.
+ * Index of the profile within {port, traffic class} to use.
+ * 0 for disabling both WRED and ECN for this type of traffic.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, ntcp_g, 60, 0, 2);
+
+/* reg_cwtpm_ntcp_y
+ * Non-TCP Yellow Profile.
+ * Index of the profile within {port, traffic class} to use.
+ * 0 for disabling both WRED and ECN for this type of traffic.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, ntcp_y, 64, 16, 2);
+
+/* reg_cwtpm_ntcp_r
+ * Non-TCP Red Profile.
+ * Index of the profile within {port, traffic class} to use.
+ * 0 for disabling both WRED and ECN for this type of traffic.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, cwtpm, ntcp_r, 64, 0, 2);
+
+#define MLXSW_REG_CWTPM_RESET_PROFILE 0
+
+static inline void mlxsw_reg_cwtpm_pack(char *payload, u8 local_port,
+					u8 traffic_class, u8 profile,
+					bool wred, bool ecn)
+{
+	MLXSW_REG_ZERO(cwtpm, payload);
+	mlxsw_reg_cwtpm_local_port_set(payload, local_port);
+	mlxsw_reg_cwtpm_traffic_class_set(payload, traffic_class);
+	mlxsw_reg_cwtpm_ew_set(payload, wred);
+	mlxsw_reg_cwtpm_ee_set(payload, ecn);
+	mlxsw_reg_cwtpm_tcp_g_set(payload, profile);
+	mlxsw_reg_cwtpm_tcp_y_set(payload, profile);
+	mlxsw_reg_cwtpm_tcp_r_set(payload, profile);
+	mlxsw_reg_cwtpm_ntcp_g_set(payload, profile);
+	mlxsw_reg_cwtpm_ntcp_y_set(payload, profile);
+	mlxsw_reg_cwtpm_ntcp_r_set(payload, profile);
+}
+
 /* PPBT - Policy-Engine Port Binding Table
  * ---------------------------------------
  * This register is used for configuration of the Port Binding Table.
@@ -7274,6 +7459,8 @@ static const struct mlxsw_reg_info *mlxsw_reg_infos[] = {
 	MLXSW_REG(svpe),
 	MLXSW_REG(sfmr),
 	MLXSW_REG(spvmlr),
+	MLXSW_REG(cwtp),
+	MLXSW_REG(cwtpm),
 	MLXSW_REG(ppbt),
 	MLXSW_REG(pacl),
 	MLXSW_REG(pagt),
-- 
2.9.5

[patch net-next RFC 5/9] mlxsw: spectrum: Support RED qdisc offload

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Add support for ndo_setup_tc with enum tc_setup_type value of TC_SETUP_RED.
This call sets RED qdisc on a traffic class.
This patch supports RED qdisc only as a root qdisc and set in on the
default tclass. It can be set with or without ECN.

Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/Makefile       |   3 +-
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     |   2 +
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  15 ++
 .../net/ethernet/mellanox/mlxsw/spectrum_qdisc.c   | 152 +++++++++++++++++++++
 4 files changed, 171 insertions(+), 1 deletion(-)
 create mode 100644 drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c

diff --git a/drivers/net/ethernet/mellanox/mlxsw/Makefile b/drivers/net/ethernet/mellanox/mlxsw/Makefile
index 4816504..51cd4c1 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/Makefile
+++ b/drivers/net/ethernet/mellanox/mlxsw/Makefile
@@ -18,7 +18,8 @@ mlxsw_spectrum-objs		:= spectrum.o spectrum_buffers.o \
 				   spectrum_acl.o spectrum_flower.o \
 				   spectrum_cnt.o spectrum_fid.o \
 				   spectrum_ipip.o spectrum_acl_flex_actions.o \
-				   spectrum_mr.o spectrum_mr_tcam.o
+				   spectrum_mr.o spectrum_mr_tcam.o \
+				   spectrum_qdisc.o
 mlxsw_spectrum-$(CONFIG_MLXSW_SPECTRUM_DCB)	+= spectrum_dcb.o
 mlxsw_spectrum-$(CONFIG_NET_DEVLINK) += spectrum_dpipe.o
 obj-$(CONFIG_MLXSW_MINIMAL)	+= mlxsw_minimal.o
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 0219269..fab0bb2 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -1794,6 +1794,8 @@ static int mlxsw_sp_setup_tc(struct net_device *dev, enum tc_setup_type type,
 	switch (type) {
 	case TC_SETUP_BLOCK:
 		return mlxsw_sp_setup_tc_block(mlxsw_sp_port, type_data);
+	case TC_SETUP_QDISC_RED:
+		return mlxsw_sp_setup_tc_red(mlxsw_sp_port, type_data);
 	default:
 		return -EOPNOTSUPP;
 	}
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index aa0cefb..8810c1a 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -203,6 +203,16 @@ struct mlxsw_sp_port_vlan {
 	struct list_head bridge_vlan_node;
 };
 
+enum mlxsw_sp_qdisc_type {
+	MLXSW_SP_QDISC_NO_QDISC,
+	MLXSW_SP_QDISC_RED,
+};
+
+struct mlxsw_sp_qdisc {
+	u32 handle;
+	enum mlxsw_sp_qdisc_type type;
+};
+
 struct mlxsw_sp_port {
 	struct net_device *dev;
 	struct mlxsw_sp_port_pcpu_stats __percpu *pcpu_stats;
@@ -236,6 +246,7 @@ struct mlxsw_sp_port {
 	} periodic_hw_stats;
 	struct mlxsw_sp_port_sample *sample;
 	struct list_head vlans_list;
+	struct mlxsw_sp_qdisc root_qdisc;
 };
 
 static inline bool
@@ -542,6 +553,10 @@ void mlxsw_sp_flower_destroy(struct mlxsw_sp_port *mlxsw_sp_port, bool ingress,
 int mlxsw_sp_flower_stats(struct mlxsw_sp_port *mlxsw_sp_port, bool ingress,
 			  struct tc_cls_flower_offload *f);
 
+/* spectrum_qdisc.c */
+int mlxsw_sp_setup_tc_red(struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct tc_red_qopt_offload *p);
+
 /* spectrum_fid.c */
 int mlxsw_sp_fid_flood_set(struct mlxsw_sp_fid *fid,
 			   enum mlxsw_sp_flood_type packet_type, u8 local_port,
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
new file mode 100644
index 0000000..e9d289e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
@@ -0,0 +1,152 @@
+/*
+ * drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
+ * Copyright (c) 2017 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2017 Nogah Frankel <nogahf@mellanox.com>
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. 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.
+ * 3. Neither the names of the copyright holders nor the names of its
+ *    contributors may be used to endorse or promote products derived from
+ *    this software without specific prior written permission.
+ *
+ * Alternatively, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") version 2 as published by the Free
+ * Software Foundation.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/netdevice.h>
+#include <net/pkt_cls.h>
+
+#include "spectrum.h"
+#include "reg.h"
+
+static int
+mlxsw_sp_tclass_congestion_enable(struct mlxsw_sp_port *mlxsw_sp_port,
+				  int tclass_num, u32 min, u32 max,
+				  u32 probability, bool is_ecn)
+{
+	char cwtp_cmd[max_t(u8, MLXSW_REG_CWTP_LEN, MLXSW_REG_CWTPM_LEN)];
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	int err;
+
+	mlxsw_reg_cwtp_pack(cwtp_cmd, mlxsw_sp_port->local_port, tclass_num);
+	mlxsw_reg_cwtp_profile_pack(cwtp_cmd, MLXSW_REG_CWTP_DEFAULT_PROFILE,
+				    roundup(min, MLXSW_REG_CWTP_MIN_VALUE),
+				    roundup(max, MLXSW_REG_CWTP_MIN_VALUE),
+				    probability);
+
+	err = mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(cwtp), cwtp_cmd);
+	if (err)
+		return err;
+
+	mlxsw_reg_cwtpm_pack(cwtp_cmd, mlxsw_sp_port->local_port, tclass_num,
+			     MLXSW_REG_CWTP_DEFAULT_PROFILE, true, is_ecn);
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(cwtpm), cwtp_cmd);
+}
+
+static int
+mlxsw_sp_tclass_congestion_disable(struct mlxsw_sp_port *mlxsw_sp_port,
+				   int tclass_num)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char cwtpm_cmd[MLXSW_REG_CWTPM_LEN];
+
+	mlxsw_reg_cwtpm_pack(cwtpm_cmd, mlxsw_sp_port->local_port, tclass_num,
+			     MLXSW_REG_CWTPM_RESET_PROFILE, false, false);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(cwtpm), cwtpm_cmd);
+}
+
+static int
+mlxsw_sp_qdisc_red_replace(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
+			   struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			   int tclass_num,
+			   struct tc_red_qopt_offload_params *p)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u32 min, max;
+	u64 prob;
+	int err = 0;
+
+	/* calculate probability in percentage */
+	prob = p->probability;
+	prob *= 100;
+	prob = DIV_ROUND_UP(prob, 1 << 16);
+	prob = DIV_ROUND_UP(prob, 1 << 16);
+	min = mlxsw_sp_bytes_cells(mlxsw_sp, p->min);
+	max = mlxsw_sp_bytes_cells(mlxsw_sp, p->max);
+	err = mlxsw_sp_tclass_congestion_enable(mlxsw_sp_port, tclass_num, min,
+						max, prob, p->is_ecn);
+	if (err)
+		goto out;
+
+	mlxsw_sp_qdisc->type = MLXSW_SP_QDISC_RED;
+	mlxsw_sp_qdisc->handle = handle;
+
+out:
+	return err;
+}
+
+static int
+mlxsw_sp_qdisc_red_destroy(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
+			   struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			   int tclass_num)
+{
+	int err;
+
+	if (mlxsw_sp_qdisc->handle != handle)
+		return 0;
+
+	err = mlxsw_sp_tclass_congestion_disable(mlxsw_sp_port, tclass_num);
+	mlxsw_sp_qdisc->handle = TC_H_UNSPEC;
+	mlxsw_sp_qdisc->type = MLXSW_SP_QDISC_NO_QDISC;
+
+	return err;
+}
+
+#define MLXSW_SP_PORT_DEFAULT_TCLASS 0
+
+int mlxsw_sp_setup_tc_red(struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct tc_red_qopt_offload *p)
+{
+	struct mlxsw_sp_qdisc *mlxsw_sp_qdisc;
+	int tclass_num;
+
+	if (p->parent != TC_H_ROOT)
+		return -EOPNOTSUPP;
+
+	mlxsw_sp_qdisc = &mlxsw_sp_port->root_qdisc;
+	tclass_num = MLXSW_SP_PORT_DEFAULT_TCLASS;
+
+	switch (p->command) {
+	case TC_RED_REPLACE:
+		return mlxsw_sp_qdisc_red_replace(mlxsw_sp_port, p->handle,
+						  mlxsw_sp_qdisc, tclass_num,
+						  &p->set);
+	case TC_RED_DESTROY:
+		return mlxsw_sp_qdisc_red_destroy(mlxsw_sp_port, p->handle,
+						  mlxsw_sp_qdisc, tclass_num);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
-- 
2.9.5

[patch net-next RFC 6/9] mlxsw: reg: Add ext and tc-cong counter groups

[Jiri Pirko]

From: Yuval Mintz <yuvalm@mellanox.com>

This adds the counter group definitions for 2 new counter groups
which are necessary for gaining ECN & wred counters.

Signed-off-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/reg.h | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index 48f051f..d1632c8 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -3341,8 +3341,10 @@ MLXSW_ITEM32(reg, ppcnt, pnat, 0x00, 14, 2);
 
 enum mlxsw_reg_ppcnt_grp {
 	MLXSW_REG_PPCNT_IEEE_8023_CNT = 0x0,
+	MLXSW_REG_PPCNT_EXT_CNT = 0x5,
 	MLXSW_REG_PPCNT_PRIO_CNT = 0x10,
 	MLXSW_REG_PPCNT_TC_CNT = 0x11,
+	MLXSW_REG_PPCNT_TC_CONG_TC = 0x13,
 };
 
 /* reg_ppcnt_grp
@@ -3358,6 +3360,7 @@ enum mlxsw_reg_ppcnt_grp {
  * 0x10: Per Priority Counters
  * 0x11: Per Traffic Class Counters
  * 0x12: Physical Layer Counters
+ * 0x13: Per Traffic Class Congestion Counters
  * Access: Index
  */
 MLXSW_ITEM32(reg, ppcnt, grp, 0x00, 0, 6);
@@ -3496,6 +3499,14 @@ MLXSW_ITEM64(reg, ppcnt, a_pause_mac_ctrl_frames_received,
 MLXSW_ITEM64(reg, ppcnt, a_pause_mac_ctrl_frames_transmitted,
 	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x90, 0, 64);
 
+/* Ethernet Extended Counter Group Counters */
+
+/* reg_ppcnt_ecn_marked
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, ecn_marked,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x08, 0, 64);
+
 /* Ethernet Per Priority Group Counters */
 
 /* reg_ppcnt_rx_octets
@@ -3571,6 +3582,14 @@ MLXSW_ITEM64(reg, ppcnt, tc_transmit_queue,
 MLXSW_ITEM64(reg, ppcnt, tc_no_buffer_discard_uc,
 	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x08, 0, 64);
 
+/* Ethernet Per Traffic Class Congestion Group Counters */
+
+/* reg_ppcnt_wred_discard
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, wred_discard,
+	     MLXSW_REG_PPCNT_COUNTERS_OFFSET + 0x00, 0, 64);
+
 static inline void mlxsw_reg_ppcnt_pack(char *payload, u8 local_port,
 					enum mlxsw_reg_ppcnt_grp grp,
 					u8 prio_tc)
-- 
2.9.5

[patch net-next RFC 7/9] mlxsw: spectrum: Collect tclass related stats periodically

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Add more statistics to be collected from the HW periodically. These stats
are tclass based (beside ECN marked packet, that exist only port based).
They are needed to expose RED qdisc stats and xstats correctly.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 34 ++++++++++++++++++++++++++
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h |  9 +++++++
 2 files changed, 43 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index fab0bb2..584216a 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -1324,6 +1324,38 @@ static int mlxsw_sp_port_get_hw_stats(struct net_device *dev,
 	return err;
 }
 
+static void
+mlxsw_sp_port_get_hw_xstats(struct net_device *dev,
+			    struct mlxsw_sp_port_xstats *xstats)
+{
+	char ppcnt_pl[MLXSW_REG_PPCNT_LEN];
+	int err, i;
+
+	err = mlxsw_sp_port_get_stats_raw(dev, MLXSW_REG_PPCNT_EXT_CNT, 0,
+					  ppcnt_pl);
+	if (!err)
+		xstats->ecn = mlxsw_reg_ppcnt_ecn_marked_get(ppcnt_pl);
+
+	for (i = 0; i < TC_MAX_QUEUE; i++) {
+		err = mlxsw_sp_port_get_stats_raw(dev,
+						  MLXSW_REG_PPCNT_TC_CONG_TC,
+						  i, ppcnt_pl);
+		if (!err)
+			xstats->wred_drop[i] =
+				mlxsw_reg_ppcnt_wred_discard_get(ppcnt_pl);
+
+		err = mlxsw_sp_port_get_stats_raw(dev, MLXSW_REG_PPCNT_TC_CNT,
+						  i, ppcnt_pl);
+		if (err)
+			continue;
+
+		xstats->backlog[i] =
+			mlxsw_reg_ppcnt_tc_transmit_queue_get(ppcnt_pl);
+		xstats->tail_drop[i] =
+			mlxsw_reg_ppcnt_tc_no_buffer_discard_uc_get(ppcnt_pl);
+	}
+}
+
 static void update_stats_cache(struct work_struct *work)
 {
 	struct mlxsw_sp_port *mlxsw_sp_port =
@@ -1335,6 +1367,8 @@ static void update_stats_cache(struct work_struct *work)
 
 	mlxsw_sp_port_get_hw_stats(mlxsw_sp_port->dev,
 				   &mlxsw_sp_port->periodic_hw_stats.stats);
+	mlxsw_sp_port_get_hw_xstats(mlxsw_sp_port->dev,
+				    &mlxsw_sp_port->periodic_hw_stats.xstats);
 
 out:
 	mlxsw_core_schedule_dw(&mlxsw_sp_port->periodic_hw_stats.update_dw,
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index 8810c1a..de81e9c 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -213,6 +213,14 @@ struct mlxsw_sp_qdisc {
 	enum mlxsw_sp_qdisc_type type;
 };
 
+/* No need an internal lock; At worse - miss a single periodic iteration */
+struct mlxsw_sp_port_xstats {
+	u64 ecn;
+	u64 wred_drop[TC_MAX_QUEUE];
+	u64 tail_drop[TC_MAX_QUEUE];
+	u64 backlog[TC_MAX_QUEUE];
+};
+
 struct mlxsw_sp_port {
 	struct net_device *dev;
 	struct mlxsw_sp_port_pcpu_stats __percpu *pcpu_stats;
@@ -242,6 +250,7 @@ struct mlxsw_sp_port {
 	struct {
 		#define MLXSW_HW_STATS_UPDATE_TIME HZ
 		struct rtnl_link_stats64 stats;
+		struct mlxsw_sp_port_xstats xstats;
 		struct delayed_work update_dw;
 	} periodic_hw_stats;
 	struct mlxsw_sp_port_sample *sample;
-- 
2.9.5

[patch net-next RFC 8/9] mlxsw: spectrum: Support RED xstats

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Add support for ndo_setup_tc with enum tc_setup_type value of
TC_SETUP_RED_XSTATS. This call returns the RED qdisc xstats from the cache
if the handle ID that is asked for matching the root qdisc ID and fails
otherwise.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  9 ++++
 .../net/ethernet/mellanox/mlxsw/spectrum_qdisc.c   | 51 ++++++++++++++++++++++
 2 files changed, 60 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index de81e9c..f3cf1d1 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -48,6 +48,7 @@
 #include <linux/notifier.h>
 #include <net/psample.h>
 #include <net/pkt_cls.h>
+#include <net/red.h>
 
 #include "port.h"
 #include "core.h"
@@ -211,6 +212,14 @@ enum mlxsw_sp_qdisc_type {
 struct mlxsw_sp_qdisc {
 	u32 handle;
 	enum mlxsw_sp_qdisc_type type;
+	struct red_stats xstats_base;
+	union {
+		struct {
+			u64 tail_drop_base;
+			u64 ecn_base;
+			u64 wred_drop_base;
+		} red;
+	} xstats;
 };
 
 /* No need an internal lock; At worse - miss a single periodic iteration */
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
index e9d289e..4c941f7 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
@@ -36,6 +36,7 @@
 #include <linux/errno.h>
 #include <linux/netdevice.h>
 #include <net/pkt_cls.h>
+#include <net/red.h>
 
 #include "spectrum.h"
 #include "reg.h"
@@ -77,6 +78,27 @@ mlxsw_sp_tclass_congestion_disable(struct mlxsw_sp_port *mlxsw_sp_port,
 	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(cwtpm), cwtpm_cmd);
 }
 
+static void
+mlxsw_sp_setup_tc_qdisc_clean_stats(struct mlxsw_sp_port *mlxsw_sp_port,
+				    struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+				    int tclass_num)
+{
+	struct red_stats *xstats_base = &mlxsw_sp_qdisc->xstats_base;
+	struct mlxsw_sp_port_xstats *xstats;
+
+	xstats = &mlxsw_sp_port->periodic_hw_stats.xstats;
+
+	switch (mlxsw_sp_qdisc->type) {
+	case MLXSW_SP_QDISC_RED:
+		xstats_base->prob_mark = xstats->ecn;
+		xstats_base->prob_drop = xstats->wred_drop[tclass_num];
+		xstats_base->pdrop = xstats->tail_drop[tclass_num];
+		break;
+	default:
+		break;
+	}
+}
+
 static int
 mlxsw_sp_qdisc_red_replace(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
 			   struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
@@ -101,6 +123,11 @@ mlxsw_sp_qdisc_red_replace(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
 		goto out;
 
 	mlxsw_sp_qdisc->type = MLXSW_SP_QDISC_RED;
+	if (mlxsw_sp_qdisc->handle != handle)
+		mlxsw_sp_setup_tc_qdisc_clean_stats(mlxsw_sp_port,
+						    mlxsw_sp_qdisc,
+						    tclass_num);
+
 	mlxsw_sp_qdisc->handle = handle;
 
 out:
@@ -124,6 +151,26 @@ mlxsw_sp_qdisc_red_destroy(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
 	return err;
 }
 
+static int
+mlxsw_sp_qdisc_get_red_xstats(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
+			      struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			      int tclass_num, struct red_stats *res)
+{
+	struct red_stats *xstats_base = &mlxsw_sp_qdisc->xstats_base;
+	struct mlxsw_sp_port_xstats *xstats;
+
+	if (mlxsw_sp_qdisc->handle != handle ||
+	    mlxsw_sp_qdisc->type != MLXSW_SP_QDISC_RED)
+		return -EOPNOTSUPP;
+
+	xstats = &mlxsw_sp_port->periodic_hw_stats.xstats;
+
+	res->prob_drop = xstats->wred_drop[tclass_num] - xstats_base->prob_drop;
+	res->prob_mark = xstats->ecn - xstats_base->prob_mark;
+	res->pdrop = xstats->tail_drop[tclass_num] - xstats_base->pdrop;
+	return 0;
+}
+
 #define MLXSW_SP_PORT_DEFAULT_TCLASS 0
 
 int mlxsw_sp_setup_tc_red(struct mlxsw_sp_port *mlxsw_sp_port,
@@ -146,6 +193,10 @@ int mlxsw_sp_setup_tc_red(struct mlxsw_sp_port *mlxsw_sp_port,
 	case TC_RED_DESTROY:
 		return mlxsw_sp_qdisc_red_destroy(mlxsw_sp_port, p->handle,
 						  mlxsw_sp_qdisc, tclass_num);
+	case TC_RED_XSTATS:
+		return mlxsw_sp_qdisc_get_red_xstats(mlxsw_sp_port, p->handle,
+						     mlxsw_sp_qdisc, tclass_num,
+						     p->xstats);
 	default:
 		return -EOPNOTSUPP;
 	}
-- 
2.9.5

[patch net-next RFC 9/9] mlxsw: spectrum: Support general qdisc stats

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Add support for ndo_setup_tc with enum tc_setup_type value of
TC_SETUP_QDISC_STATS. This call updates the generic qdisc stats from the
cache if the handle ID that is asked for matching the root qdisc ID and
fails otherwise.
Currently doesn't support qlen and rqueues.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  5 +++
 .../net/ethernet/mellanox/mlxsw/spectrum_qdisc.c   | 51 ++++++++++++++++++++++
 2 files changed, 56 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index f3cf1d1..0b0cae1 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -220,6 +220,11 @@ struct mlxsw_sp_qdisc {
 			u64 wred_drop_base;
 		} red;
 	} xstats;
+
+	u64 tx_bytes;
+	u64 tx_packets;
+	u64 drops;
+	u64 overlimits;
 };
 
 /* No need an internal lock; At worse - miss a single periodic iteration */
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
index 4c941f7..0988ba5 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
@@ -85,14 +85,24 @@ mlxsw_sp_setup_tc_qdisc_clean_stats(struct mlxsw_sp_port *mlxsw_sp_port,
 {
 	struct red_stats *xstats_base = &mlxsw_sp_qdisc->xstats_base;
 	struct mlxsw_sp_port_xstats *xstats;
+	struct rtnl_link_stats64 *stats;
 
 	xstats = &mlxsw_sp_port->periodic_hw_stats.xstats;
+	stats = &mlxsw_sp_port->periodic_hw_stats.stats;
+
+	mlxsw_sp_qdisc->tx_packets = stats->tx_packets;
+	mlxsw_sp_qdisc->tx_bytes = stats->tx_bytes;
 
 	switch (mlxsw_sp_qdisc->type) {
 	case MLXSW_SP_QDISC_RED:
 		xstats_base->prob_mark = xstats->ecn;
 		xstats_base->prob_drop = xstats->wred_drop[tclass_num];
 		xstats_base->pdrop = xstats->tail_drop[tclass_num];
+
+		mlxsw_sp_qdisc->overlimits = xstats_base->prob_drop +
+					     xstats_base->prob_mark;
+		mlxsw_sp_qdisc->drops = xstats_base->prob_drop +
+					xstats_base->pdrop;
 		break;
 	default:
 		break;
@@ -171,6 +181,43 @@ mlxsw_sp_qdisc_get_red_xstats(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
 	return 0;
 }
 
+static int
+mlxsw_sp_qdisc_get_red_stats(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle,
+			     struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			     int tclass_num,
+			     struct tc_red_qopt_offload_stats *res)
+{
+	u64 tx_bytes, tx_packets, overlimits, drops;
+	struct mlxsw_sp_port_xstats *xstats;
+	struct rtnl_link_stats64 *stats;
+
+	if (mlxsw_sp_qdisc->handle != handle ||
+	    mlxsw_sp_qdisc->type != MLXSW_SP_QDISC_RED)
+		return -EOPNOTSUPP;
+
+	xstats = &mlxsw_sp_port->periodic_hw_stats.xstats;
+	stats = &mlxsw_sp_port->periodic_hw_stats.stats;
+
+	tx_bytes = stats->tx_bytes - mlxsw_sp_qdisc->tx_bytes;
+	tx_packets = stats->tx_packets - mlxsw_sp_qdisc->tx_packets;
+	overlimits = xstats->wred_drop[tclass_num] + xstats->ecn -
+		     mlxsw_sp_qdisc->overlimits;
+	drops = xstats->wred_drop[tclass_num] + xstats->tail_drop[tclass_num] -
+		mlxsw_sp_qdisc->drops;
+
+	_bstats_update(res->bstats, tx_bytes, tx_packets);
+	res->qstats->overlimits += overlimits;
+	res->qstats->drops += drops;
+	res->qstats->backlog += mlxsw_sp_cells_bytes(mlxsw_sp_port->mlxsw_sp,
+						xstats->backlog[tclass_num]);
+
+	mlxsw_sp_qdisc->drops +=  drops;
+	mlxsw_sp_qdisc->overlimits += overlimits;
+	mlxsw_sp_qdisc->tx_bytes += tx_bytes;
+	mlxsw_sp_qdisc->tx_packets += tx_packets;
+	return 0;
+}
+
 #define MLXSW_SP_PORT_DEFAULT_TCLASS 0
 
 int mlxsw_sp_setup_tc_red(struct mlxsw_sp_port *mlxsw_sp_port,
@@ -197,6 +244,10 @@ int mlxsw_sp_setup_tc_red(struct mlxsw_sp_port *mlxsw_sp_port,
 		return mlxsw_sp_qdisc_get_red_xstats(mlxsw_sp_port, p->handle,
 						     mlxsw_sp_qdisc, tclass_num,
 						     p->xstats);
+	case TC_RED_STATS:
+		return mlxsw_sp_qdisc_get_red_stats(mlxsw_sp_port, p->handle,
+						    mlxsw_sp_qdisc, tclass_num,
+						    &p->stats);
 	default:
 		return -EOPNOTSUPP;
 	}
-- 
2.9.5

Re: [patch net-next RFC 1/9] net_sch: red: Add offload ability to RED qdisc

[David Miller]

From: Jiri Pirko <jiri@resnulli.us>
Date: Mon, 30 Oct 2017 09:56:05 +0100

> diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
> index 0e88cc2..743c42a 100644
> --- a/include/uapi/linux/pkt_sched.h
> +++ b/include/uapi/linux/pkt_sched.h
> @@ -255,6 +255,7 @@ struct tc_red_qopt {
>  #define TC_RED_ECN		1
>  #define TC_RED_HARDDROP		2
>  #define TC_RED_ADAPTATIVE	4
> +#define TC_RED_OFFLOADED	8
>  };
>  
>  struct tc_red_xstats {

What keeps a user from setting this flag in the tc_red_qopt it
passes into the a change operation?

Re: [patch net-next RFC 1/9] net_sch: red: Add offload ability to RED qdisc

[Roman Mashak]

Jiri Pirko <jiri@resnulli.us> writes:


[...]

> +static void red_unoffload(struct Qdisc *sch)
> +{
> +	struct net_device *dev = qdisc_dev(sch);
> +	struct tc_red_qopt_offload opt = {
> +		.handle = sch->handle,
> +		.command = TC_RED_DESTROY,
> +		.parent = sch->parent,
> +	};
> +
> +	if (!tc_can_offload(dev) || !dev->netdev_ops->ndo_setup_tc)
> +		return;
> +
> +	dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_RED,  &opt);
> +}
> +

> @@ -162,6 +179,28 @@ static const struct nla_policy red_policy[TCA_RED_MAX + 1] = {
>  	[TCA_RED_MAX_P] = { .type = NLA_U32 },
>  };
>  
> +static int red_offload(struct Qdisc *sch)
> +{
> +	struct red_sched_data *q = qdisc_priv(sch);
> +	struct net_device *dev = qdisc_dev(sch);
> +	struct tc_red_qopt_offload opt = {
> +		.handle = sch->handle,
> +		.command = TC_RED_REPLACE,
> +		.parent = sch->parent,
> +		.set = {
> +			.min = q->parms.qth_min >> q->parms.Wlog,
> +			.max = q->parms.qth_max >> q->parms.Wlog,
> +			.probability = q->parms.max_P,
> +			.is_ecn = red_use_ecn(q),
> +		},
> +	};
> +
> +	if (!tc_can_offload(dev) || !dev->netdev_ops->ndo_setup_tc)
> +		return -EOPNOTSUPP;
> +
> +	return dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_RED, &opt);
> +}
> +

[...]

Can't red_unoffload() and red_offload() be unified in a single API? For
example, red_offload(struct Qdisc *sch, bool enable) ?

RE: [patch net-next RFC 1/9] net_sch: red: Add offload ability to RED qdisc

[Nogah Frankel]

> -----Original Message-----
> From: David Miller [mailto:davem@davemloft.net]
> Sent: Monday, October 30, 2017 2:20 PM
> To: jiri@resnulli.us
> Cc: netdev@vger.kernel.org; Nogah Frankel <nogahf@mellanox.com>; jhs@mojatatu.com;
> xiyou.wangcong@gmail.com; mlxsw <mlxsw@mellanox.com>; andrew@lunn.ch;
> vivien.didelot@savoirfairelinux.com; f.fainelli@gmail.com; michael.chan@broadcom.com;
> ganeshgr@chelsio.com; Saeed Mahameed <saeedm@mellanox.com>; Matan Barak
> <matanb@mellanox.com>; Leon Romanovsky <leonro@mellanox.com>; Ido Schimmel
> <idosch@mellanox.com>; jakub.kicinski@netronome.com; simon.horman@netronome.com;
> pieter.jansenvanvuuren@netronome.com; john.hurley@netronome.com;
> alexander.h.duyck@intel.com; Or Gerlitz <ogerlitz@mellanox.com>;
> john.fastabend@gmail.com
> Subject: Re: [patch net-next RFC 1/9] net_sch: red: Add offload ability to RED qdisc
> 
> From: Jiri Pirko <jiri@resnulli.us>
> Date: Mon, 30 Oct 2017 09:56:05 +0100
> 
> > diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
> > index 0e88cc2..743c42a 100644
> > --- a/include/uapi/linux/pkt_sched.h
> > +++ b/include/uapi/linux/pkt_sched.h
> > @@ -255,6 +255,7 @@ struct tc_red_qopt {
> >  #define TC_RED_ECN		1
> >  #define TC_RED_HARDDROP		2
> >  #define TC_RED_ADAPTATIVE	4
> > +#define TC_RED_OFFLOADED	8
> >  };
> >
> >  struct tc_red_xstats {
> 
> What keeps a user from setting this flag in the tc_red_qopt it
> passes into the a change operation?

Nothing keeps the user from doing it, but it has no effect.
The decision to offload is the driver's only.
It is basically a read-only flag.

RE: [patch net-next RFC 1/9] net_sch: red: Add offload ability to RED qdisc

[Nogah Frankel]

> -----Original Message-----
> From: Roman Mashak [mailto:mrv@mojatatu.com]
> Sent: Monday, October 30, 2017 2:47 PM
> To: Jiri Pirko <jiri@resnulli.us>
> Cc: netdev@vger.kernel.org; davem@davemloft.net; Nogah Frankel <nogahf@mellanox.com>;
> jhs@mojatatu.com; xiyou.wangcong@gmail.com; mlxsw <mlxsw@mellanox.com>;
> andrew@lunn.ch; vivien.didelot@savoirfairelinux.com; f.fainelli@gmail.com;
> michael.chan@broadcom.com; ganeshgr@chelsio.com; Saeed Mahameed
> <saeedm@mellanox.com>; Matan Barak <matanb@mellanox.com>; Leon Romanovsky
> <leonro@mellanox.com>; Ido Schimmel <idosch@mellanox.com>;
> jakub.kicinski@netronome.com; simon.horman@netronome.com;
> pieter.jansenvanvuuren@netronome.com; john.hurley@netronome.com;
> alexander.h.duyck@intel.com; Or Gerlitz <ogerlitz@mellanox.com>;
> john.fastabend@gmail.com
> Subject: Re: [patch net-next RFC 1/9] net_sch: red: Add offload ability to RED qdisc
> 
> Jiri Pirko <jiri@resnulli.us> writes:
> 
> 
> [...]
> 
> > +static void red_unoffload(struct Qdisc *sch)
> > +{
> > +	struct net_device *dev = qdisc_dev(sch);
> > +	struct tc_red_qopt_offload opt = {
> > +		.handle = sch->handle,
> > +		.command = TC_RED_DESTROY,
> > +		.parent = sch->parent,
> > +	};
> > +
> > +	if (!tc_can_offload(dev) || !dev->netdev_ops->ndo_setup_tc)
> > +		return;
> > +
> > +	dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_RED,  &opt);
> > +}
> > +
> 
> > @@ -162,6 +179,28 @@ static const struct nla_policy red_policy[TCA_RED_MAX + 1] = {
> >  	[TCA_RED_MAX_P] = { .type = NLA_U32 },
> >  };
> >
> > +static int red_offload(struct Qdisc *sch)
> > +{
> > +	struct red_sched_data *q = qdisc_priv(sch);
> > +	struct net_device *dev = qdisc_dev(sch);
> > +	struct tc_red_qopt_offload opt = {
> > +		.handle = sch->handle,
> > +		.command = TC_RED_REPLACE,
> > +		.parent = sch->parent,
> > +		.set = {
> > +			.min = q->parms.qth_min >> q->parms.Wlog,
> > +			.max = q->parms.qth_max >> q->parms.Wlog,
> > +			.probability = q->parms.max_P,
> > +			.is_ecn = red_use_ecn(q),
> > +		},
> > +	};
> > +
> > +	if (!tc_can_offload(dev) || !dev->netdev_ops->ndo_setup_tc)
> > +		return -EOPNOTSUPP;
> > +
> > +	return dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_RED, &opt);
> > +}
> > +
> 
> [...]
> 
> Can't red_unoffload() and red_offload() be unified in a single API? For
> example, red_offload(struct Qdisc *sch, bool enable) ?

I will unify them.
Thanks.

Re: [patch net-next RFC 1/9] net_sch: red: Add offload ability to RED qdisc

[David Ahern]

On 10/30/17 7:54 AM, Nogah Frankel wrote:
>>> diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
>>> index 0e88cc2..743c42a 100644
>>> --- a/include/uapi/linux/pkt_sched.h
>>> +++ b/include/uapi/linux/pkt_sched.h
>>> @@ -255,6 +255,7 @@ struct tc_red_qopt {
>>>  #define TC_RED_ECN		1
>>>  #define TC_RED_HARDDROP		2
>>>  #define TC_RED_ADAPTATIVE	4
>>> +#define TC_RED_OFFLOADED	8
>>>  };
>>>
>>>  struct tc_red_xstats {
>>
>> What keeps a user from setting this flag in the tc_red_qopt it
>> passes into the a change operation?
> 
> Nothing keeps the user from doing it, but it has no effect.
> The decision to offload is the driver's only.
> It is basically a read-only flag.
> 

If it is read-only, then attempts to set it from userspace should cause
the command to fail.

Re: [patch net-next RFC 1/9] net_sch: red: Add offload ability to RED qdisc

[Jiri Pirko]

Mon, Oct 30, 2017 at 08:24:20PM CET, dsahern@gmail.com wrote:
>On 10/30/17 7:54 AM, Nogah Frankel wrote:
>>>> diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
>>>> index 0e88cc2..743c42a 100644
>>>> --- a/include/uapi/linux/pkt_sched.h
>>>> +++ b/include/uapi/linux/pkt_sched.h
>>>> @@ -255,6 +255,7 @@ struct tc_red_qopt {
>>>>  #define TC_RED_ECN		1
>>>>  #define TC_RED_HARDDROP		2
>>>>  #define TC_RED_ADAPTATIVE	4
>>>> +#define TC_RED_OFFLOADED	8
>>>>  };
>>>>
>>>>  struct tc_red_xstats {
>>>
>>> What keeps a user from setting this flag in the tc_red_qopt it
>>> passes into the a change operation?
>> 
>> Nothing keeps the user from doing it, but it has no effect.
>> The decision to offload is the driver's only.
>> It is basically a read-only flag.
>> 
>
>If it is read-only, then attempts to set it from userspace should cause
>the command to fail.

That breaks uapi. Consider existing userspace app sending garbage in
that bit and new kernel which randomly fails upon the garbage.

Re: [patch net-next RFC 1/9] net_sch: red: Add offload ability to RED qdisc

[David Ahern]

On 10/31/17 1:41 AM, Jiri Pirko wrote:
>>
>> If it is read-only, then attempts to set it from userspace should cause
>> the command to fail.
> 
> That breaks uapi. Consider existing userspace app sending garbage in
> that bit and new kernel which randomly fails upon the garbage.
> 

It is called syzkaller. The kernel should always validate userspace
config including checking that unused bits are 0. So you are saying tc
has never done that checking and just ignored bits that are not valid?

Re: [patch net-next RFC 1/9] net_sch: red: Add offload ability to RED qdisc

[Jiri Pirko]

Tue, Oct 31, 2017 at 03:59:08PM CET, dsahern@gmail.com wrote:
>On 10/31/17 1:41 AM, Jiri Pirko wrote:
>>>
>>> If it is read-only, then attempts to set it from userspace should cause
>>> the command to fail.
>> 
>> That breaks uapi. Consider existing userspace app sending garbage in
>> that bit and new kernel which randomly fails upon the garbage.
>> 
>
>It is called syzkaller. The kernel should always validate userspace
>config including checking that unused bits are 0. So you are saying tc
>has never done that checking and just ignored bits that are not valid?

Not sure if in 100% of cases but in this case, yes.

Re: [patch net-next RFC 2/9] net_sch: mqprio: Change TC_SETUP_MQPRIO to TC_SETUP_QDISC_MQPRIO

[Simon Horman]

On Mon, Oct 30, 2017 at 09:56:06AM +0100, Jiri Pirko wrote:
> From: Nogah Frankel <nogahf@mellanox.com>
> 
> Change TC_SETUP_MQPRIO to TC_SETUP_QDISC_MQPRIO to match the new
> convention.

Why is there a new convention? This churn in this and the following patch
warrant some justification IMHO.

Re: [patch net-next RFC 2/9] net_sch: mqprio: Change TC_SETUP_MQPRIO to TC_SETUP_QDISC_MQPRIO

[Jiri Pirko]

Wed, Nov 01, 2017 at 04:31:34PM CET, simon.horman@netronome.com wrote:
>On Mon, Oct 30, 2017 at 09:56:06AM +0100, Jiri Pirko wrote:
>> From: Nogah Frankel <nogahf@mellanox.com>
>> 
>> Change TC_SETUP_MQPRIO to TC_SETUP_QDISC_MQPRIO to match the new
>> convention.
>
>Why is there a new convention? This churn in this and the following patch
>warrant some justification IMHO.

In the enum, there is CLS*, there should be also similar prefix for
QDICS group. That's it.