[patch net-next 00/10] mlxsw: Offload multi-queue RED support

[patch net-next 00/10] mlxsw: Offload multi-queue RED support

[Jiri Pirko]

From: Jiri Pirko <jiri@mellanox.com>

Nogah says:

Support a two level hierarchy of offloaded qdiscs in mlxsw, with sch_prio
being the root qdisc and sch_red as the children.

                +----------+
                | sch_prio |
                +----+-----+
                     |
                     |
    +----------------------------------+
    |                |                 |
    |                |                 |
    |                |                 |
+---v---+       +----v---+       +-----v--+
|sch_red|       |sch_red |       |sch_red |
+-------+       +--------+       +--------+

When setting sch_prio as the root qdisc on a physical port, mlxsw will
offload it. When adding it with sch_red as a child qdisc, it will offload
it as well.
Relocating child qdisc or connecting them to more then one child will
result in unoffloading them. Relocating child qdisc more then once is
highly unrecommended and might cause a miss match between the kernel
configuration and the offloaded one. The offloaded configuration will be
aligned with the one shown in the show command.
Changing the priomap parameter of sch_prio might cause a band that its
configuration was changed and it has offloaded sch_red set on it, to lose
some stats data as if sch_red was unoffloaded and offloaded again. However,
it won't affect the data on this band that will have sch_red continuously.

Patch 1 adds support for setting RED as the child of root qdisc.
Patches 2-4 add support for RED bstasts for offloaded child qdiscs.
Patches 5-6 handle backlog related changes for offloaded child qdiscs.
Patches 7-8 update PRIO in mlxsw to be able to have RED as child on its
bands.
Patch 9 adds offload handles for PRIO graft operations. In mlxsw it will
cause the driver to stop offloading the child in question.

Nogah Frankel (10):
  mlxsw: spectrum: qdiscs: Support qdisc per tclass
  mlxsw: spectrum: Add priority counters
  mlxsw: spectrum: qdiscs: Add priority map per qdisc
  mlxsw: spectrum: qdiscs: Collect stats for sch_red based on priomap
  mlxsw: spectrum: qdiscs: Update backlog handling of a child qdiscs
  net: sch: Don't warn on missmatching qlen and backlog for offloaded
    qdiscs
  mlxsw: spectrum: Update sch_prio stats to include sch_red related
    drops
  mlxsw: spectrum: qdiscs: prio: Delete child qdiscs when removing bands
  net: sch: prio: Add offload ability for grafting a child
  mlxsw: spectrum: qdiscs: prio: Handle graft command

 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     |  10 +
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |   3 +
 .../net/ethernet/mellanox/mlxsw/spectrum_qdisc.c   | 206 ++++++++++++++++++---
 include/net/pkt_cls.h                              |   8 +
 net/sched/sch_api.c                                |   7 +-
 net/sched/sch_prio.c                               |  45 ++++-
 6 files changed, 244 insertions(+), 35 deletions(-)

-- 
2.14.3

[patch net-next 01/10] mlxsw: spectrum: qdiscs: Support qdisc per tclass

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Add the option to set a qdisc per tclass.  Match the qdisc to the tclass by
parent ID. Supported currently for sch_red only.
It allows offloading sch_prio as root qdisc and sch_red as its child.
(However, doing so might corrupt the stats for both parent and child.)

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Reviewed-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  1 +
 .../net/ethernet/mellanox/mlxsw/spectrum_qdisc.c   | 62 ++++++++++++++++++----
 2 files changed, 53 insertions(+), 10 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index 2673310f92da..cc9786ff3b05 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -247,6 +247,7 @@ struct mlxsw_sp_port {
 	struct mlxsw_sp_port_sample *sample;
 	struct list_head vlans_list;
 	struct mlxsw_sp_qdisc *root_qdisc;
+	struct mlxsw_sp_qdisc *tclass_qdiscs;
 	unsigned acl_rule_count;
 	struct mlxsw_sp_acl_block *ing_acl_block;
 	struct mlxsw_sp_acl_block *eg_acl_block;
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
index 0b7670459051..858846a019ae 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
@@ -42,6 +42,8 @@
 #include "reg.h"
 
 #define MLXSW_SP_PRIO_BAND_TO_TCLASS(band) (IEEE_8021QAZ_MAX_TCS - band - 1)
+#define MLXSW_SP_PRIO_CHILD_TO_TCLASS(child) \
+	MLXSW_SP_PRIO_BAND_TO_TCLASS((child - 1))
 
 enum mlxsw_sp_qdisc_type {
 	MLXSW_SP_QDISC_NO_QDISC,
@@ -99,6 +101,26 @@ mlxsw_sp_qdisc_compare(struct mlxsw_sp_qdisc *mlxsw_sp_qdisc, u32 handle,
 	       mlxsw_sp_qdisc->handle == handle;
 }
 
+static struct mlxsw_sp_qdisc *
+mlxsw_sp_qdisc_find(struct mlxsw_sp_port *mlxsw_sp_port, u32 parent,
+		    bool root_only)
+{
+	int tclass, child_index;
+
+	if (parent == TC_H_ROOT)
+		return mlxsw_sp_port->root_qdisc;
+
+	if (root_only || !mlxsw_sp_port->root_qdisc ||
+	    !mlxsw_sp_port->root_qdisc->ops ||
+	    TC_H_MAJ(parent) != mlxsw_sp_port->root_qdisc->handle ||
+	    TC_H_MIN(parent) > IEEE_8021QAZ_MAX_TCS)
+		return NULL;
+
+	child_index = TC_H_MIN(parent);
+	tclass = MLXSW_SP_PRIO_CHILD_TO_TCLASS(child_index);
+	return &mlxsw_sp_port->tclass_qdiscs[tclass];
+}
+
 static int
 mlxsw_sp_qdisc_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
 		       struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
@@ -406,11 +428,10 @@ int mlxsw_sp_setup_tc_red(struct mlxsw_sp_port *mlxsw_sp_port,
 {
 	struct mlxsw_sp_qdisc *mlxsw_sp_qdisc;
 
-	if (p->parent != TC_H_ROOT)
+	mlxsw_sp_qdisc = mlxsw_sp_qdisc_find(mlxsw_sp_port, p->parent, false);
+	if (!mlxsw_sp_qdisc)
 		return -EOPNOTSUPP;
 
-	mlxsw_sp_qdisc = mlxsw_sp_port->root_qdisc;
-
 	if (p->command == TC_RED_REPLACE)
 		return mlxsw_sp_qdisc_replace(mlxsw_sp_port, p->handle,
 					      mlxsw_sp_qdisc,
@@ -441,9 +462,12 @@ mlxsw_sp_qdisc_prio_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
 {
 	int i;
 
-	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
 		mlxsw_sp_port_prio_tc_set(mlxsw_sp_port, i,
 					  MLXSW_SP_PORT_DEFAULT_TCLASS);
+		mlxsw_sp_qdisc_destroy(mlxsw_sp_port,
+				       &mlxsw_sp_port->tclass_qdiscs[i]);
+	}
 
 	return 0;
 }
@@ -569,10 +593,10 @@ int mlxsw_sp_setup_tc_prio(struct mlxsw_sp_port *mlxsw_sp_port,
 {
 	struct mlxsw_sp_qdisc *mlxsw_sp_qdisc;
 
-	if (p->parent != TC_H_ROOT)
+	mlxsw_sp_qdisc = mlxsw_sp_qdisc_find(mlxsw_sp_port, p->parent, true);
+	if (!mlxsw_sp_qdisc)
 		return -EOPNOTSUPP;
 
-	mlxsw_sp_qdisc = mlxsw_sp_port->root_qdisc;
 	if (p->command == TC_PRIO_REPLACE)
 		return mlxsw_sp_qdisc_replace(mlxsw_sp_port, p->handle,
 					      mlxsw_sp_qdisc,
@@ -596,17 +620,35 @@ int mlxsw_sp_setup_tc_prio(struct mlxsw_sp_port *mlxsw_sp_port,
 
 int mlxsw_sp_tc_qdisc_init(struct mlxsw_sp_port *mlxsw_sp_port)
 {
-	mlxsw_sp_port->root_qdisc = kzalloc(sizeof(*mlxsw_sp_port->root_qdisc),
-					    GFP_KERNEL);
-	if (!mlxsw_sp_port->root_qdisc)
-		return -ENOMEM;
+	struct mlxsw_sp_qdisc *mlxsw_sp_qdisc;
+	int i;
 
+	mlxsw_sp_qdisc = kzalloc(sizeof(*mlxsw_sp_qdisc), GFP_KERNEL);
+	if (!mlxsw_sp_qdisc)
+		goto err_root_qdisc_init;
+
+	mlxsw_sp_port->root_qdisc = mlxsw_sp_qdisc;
 	mlxsw_sp_port->root_qdisc->tclass_num = MLXSW_SP_PORT_DEFAULT_TCLASS;
 
+	mlxsw_sp_qdisc = kzalloc(sizeof(*mlxsw_sp_qdisc) * IEEE_8021QAZ_MAX_TCS,
+				 GFP_KERNEL);
+	if (!mlxsw_sp_qdisc)
+		goto err_tclass_qdiscs_init;
+
+	mlxsw_sp_port->tclass_qdiscs = mlxsw_sp_qdisc;
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
+		mlxsw_sp_port->tclass_qdiscs[i].tclass_num = i;
+
 	return 0;
+
+err_tclass_qdiscs_init:
+	kfree(mlxsw_sp_port->root_qdisc);
+err_root_qdisc_init:
+	return -ENOMEM;
 }
 
 void mlxsw_sp_tc_qdisc_fini(struct mlxsw_sp_port *mlxsw_sp_port)
 {
+	kfree(mlxsw_sp_port->tclass_qdiscs);
 	kfree(mlxsw_sp_port->root_qdisc);
 }
-- 
2.14.3

[patch net-next 02/10] mlxsw: spectrum: Add priority counters

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Add TX packets and bytes counters per switch priority per port.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Reviewed-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 10 ++++++++++
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h |  2 ++
 2 files changed, 12 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 8d2d140d7910..7c6204f701ae 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -1040,6 +1040,16 @@ mlxsw_sp_port_get_hw_xstats(struct net_device *dev,
 		xstats->tail_drop[i] =
 			mlxsw_reg_ppcnt_tc_no_buffer_discard_uc_get(ppcnt_pl);
 	}
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_get_stats_raw(dev, MLXSW_REG_PPCNT_PRIO_CNT,
+						  i, ppcnt_pl);
+		if (err)
+			continue;
+
+		xstats->tx_packets[i] = mlxsw_reg_ppcnt_tx_frames_get(ppcnt_pl);
+		xstats->tx_bytes[i] = mlxsw_reg_ppcnt_tx_octets_get(ppcnt_pl);
+	}
 }
 
 static void update_stats_cache(struct work_struct *work)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index cc9786ff3b05..d5e711d8ad71 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -210,6 +210,8 @@ struct mlxsw_sp_port_xstats {
 	u64 wred_drop[TC_MAX_QUEUE];
 	u64 tail_drop[TC_MAX_QUEUE];
 	u64 backlog[TC_MAX_QUEUE];
+	u64 tx_bytes[IEEE_8021QAZ_MAX_TCS];
+	u64 tx_packets[IEEE_8021QAZ_MAX_TCS];
 };
 
 struct mlxsw_sp_port {
-- 
2.14.3

[patch net-next 03/10] mlxsw: spectrum: qdiscs: Add priority map per qdisc

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Add priority map per qdisc, to indicate which priorities are being
directed through this qdisc.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Reviewed-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
index 858846a019ae..0e0299020d82 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
@@ -78,6 +78,7 @@ struct mlxsw_sp_qdisc_ops {
 struct mlxsw_sp_qdisc {
 	u32 handle;
 	u8 tclass_num;
+	u8 prio_bitmap;
 	union {
 		struct red_stats red;
 	} xstats_base;
@@ -467,6 +468,7 @@ mlxsw_sp_qdisc_prio_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
 					  MLXSW_SP_PORT_DEFAULT_TCLASS);
 		mlxsw_sp_qdisc_destroy(mlxsw_sp_port,
 				       &mlxsw_sp_port->tclass_qdiscs[i]);
+		mlxsw_sp_port->tclass_qdiscs[i].prio_bitmap = 0;
 	}
 
 	return 0;
@@ -494,11 +496,15 @@ mlxsw_sp_qdisc_prio_replace(struct mlxsw_sp_port *mlxsw_sp_port,
 	int tclass, i;
 	int err;
 
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
+		mlxsw_sp_port->tclass_qdiscs[i].prio_bitmap = 0;
+
 	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
 		tclass = MLXSW_SP_PRIO_BAND_TO_TCLASS(p->priomap[i]);
 		err = mlxsw_sp_port_prio_tc_set(mlxsw_sp_port, i, tclass);
 		if (err)
 			return err;
+		mlxsw_sp_port->tclass_qdiscs[tclass].prio_bitmap |= BIT(i);
 	}
 
 	return 0;
@@ -628,6 +634,7 @@ int mlxsw_sp_tc_qdisc_init(struct mlxsw_sp_port *mlxsw_sp_port)
 		goto err_root_qdisc_init;
 
 	mlxsw_sp_port->root_qdisc = mlxsw_sp_qdisc;
+	mlxsw_sp_port->root_qdisc->prio_bitmap = 0xff;
 	mlxsw_sp_port->root_qdisc->tclass_num = MLXSW_SP_PORT_DEFAULT_TCLASS;
 
 	mlxsw_sp_qdisc = kzalloc(sizeof(*mlxsw_sp_qdisc) * IEEE_8021QAZ_MAX_TCS,
-- 
2.14.3

[patch net-next 04/10] mlxsw: spectrum: qdiscs: Collect stats for sch_red based on priomap

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Priority counters count packets according to their packet priority.
Collect the stats for sch_red based on these counters, so the qdisc bstats
will be the sum of counters matching the priorities marked in the qdisc
priomap.
Changing the mapping of the priorities to bands while traffic is running
can result in losing the stats of the bands qdiscs from their last dump
call to this change, as if the qdisc was unoffloaded and re-offloaded. It
will not affect the traffic behaviour according to sch_red.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Reviewed-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 .../net/ethernet/mellanox/mlxsw/spectrum_qdisc.c   | 69 ++++++++++++++++------
 1 file changed, 50 insertions(+), 19 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
index 0e0299020d82..b722af360475 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
@@ -208,6 +208,23 @@ mlxsw_sp_qdisc_get_xstats(struct mlxsw_sp_port *mlxsw_sp_port,
 	return -EOPNOTSUPP;
 }
 
+static void
+mlxsw_sp_qdisc_bstats_per_priority_get(struct mlxsw_sp_port_xstats *xstats,
+				       u8 prio_bitmap, u64 *tx_packets,
+				       u64 *tx_bytes)
+{
+	int i;
+
+	*tx_packets = 0;
+	*tx_bytes = 0;
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		if (prio_bitmap & BIT(i)) {
+			*tx_packets += xstats->tx_packets[i];
+			*tx_bytes += xstats->tx_bytes[i];
+		}
+	}
+}
+
 static int
 mlxsw_sp_tclass_congestion_enable(struct mlxsw_sp_port *mlxsw_sp_port,
 				  int tclass_num, u32 min, u32 max,
@@ -253,17 +270,16 @@ mlxsw_sp_setup_tc_qdisc_red_clean_stats(struct mlxsw_sp_port *mlxsw_sp_port,
 	u8 tclass_num = mlxsw_sp_qdisc->tclass_num;
 	struct mlxsw_sp_qdisc_stats *stats_base;
 	struct mlxsw_sp_port_xstats *xstats;
-	struct rtnl_link_stats64 *stats;
 	struct red_stats *red_base;
 
 	xstats = &mlxsw_sp_port->periodic_hw_stats.xstats;
-	stats = &mlxsw_sp_port->periodic_hw_stats.stats;
 	stats_base = &mlxsw_sp_qdisc->stats_base;
 	red_base = &mlxsw_sp_qdisc->xstats_base.red;
 
-	stats_base->tx_packets = stats->tx_packets;
-	stats_base->tx_bytes = stats->tx_bytes;
-
+	mlxsw_sp_qdisc_bstats_per_priority_get(xstats,
+					       mlxsw_sp_qdisc->prio_bitmap,
+					       &stats_base->tx_packets,
+					       &stats_base->tx_bytes);
 	red_base->prob_mark = xstats->ecn;
 	red_base->prob_drop = xstats->wred_drop[tclass_num];
 	red_base->pdrop = xstats->tail_drop[tclass_num];
@@ -380,14 +396,16 @@ mlxsw_sp_qdisc_get_red_stats(struct mlxsw_sp_port *mlxsw_sp_port,
 	u8 tclass_num = mlxsw_sp_qdisc->tclass_num;
 	struct mlxsw_sp_qdisc_stats *stats_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;
 	stats_base = &mlxsw_sp_qdisc->stats_base;
 
-	tx_bytes = stats->tx_bytes - stats_base->tx_bytes;
-	tx_packets = stats->tx_packets - stats_base->tx_packets;
+	mlxsw_sp_qdisc_bstats_per_priority_get(xstats,
+					       mlxsw_sp_qdisc->prio_bitmap,
+					       &tx_packets, &tx_bytes);
+	tx_bytes = tx_bytes - stats_base->tx_bytes;
+	tx_packets = tx_packets - stats_base->tx_packets;
+
 	overlimits = xstats->wred_drop[tclass_num] + xstats->ecn -
 		     stats_base->overlimits;
 	drops = xstats->wred_drop[tclass_num] + xstats->tail_drop[tclass_num] -
@@ -493,18 +511,31 @@ mlxsw_sp_qdisc_prio_replace(struct mlxsw_sp_port *mlxsw_sp_port,
 			    void *params)
 {
 	struct tc_prio_qopt_offload_params *p = params;
-	int tclass, i;
+	struct mlxsw_sp_qdisc *child_qdisc;
+	int tclass, i, band;
+	u8 old_priomap;
 	int err;
 
-	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
-		mlxsw_sp_port->tclass_qdiscs[i].prio_bitmap = 0;
-
-	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
-		tclass = MLXSW_SP_PRIO_BAND_TO_TCLASS(p->priomap[i]);
-		err = mlxsw_sp_port_prio_tc_set(mlxsw_sp_port, i, tclass);
-		if (err)
-			return err;
-		mlxsw_sp_port->tclass_qdiscs[tclass].prio_bitmap |= BIT(i);
+	for (band = 0; band < p->bands; band++) {
+		tclass = MLXSW_SP_PRIO_BAND_TO_TCLASS(band);
+		child_qdisc = &mlxsw_sp_port->tclass_qdiscs[tclass];
+		old_priomap = child_qdisc->prio_bitmap;
+		child_qdisc->prio_bitmap = 0;
+		for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+			if (p->priomap[i] == band) {
+				child_qdisc->prio_bitmap |= BIT(i);
+				if (BIT(i) & old_priomap)
+					continue;
+				err = mlxsw_sp_port_prio_tc_set(mlxsw_sp_port,
+								i, tclass);
+				if (err)
+					return err;
+			}
+		}
+		if (old_priomap != child_qdisc->prio_bitmap &&
+		    child_qdisc->ops && child_qdisc->ops->clean_stats)
+			child_qdisc->ops->clean_stats(mlxsw_sp_port,
+						      child_qdisc);
 	}
 
 	return 0;
-- 
2.14.3

[patch net-next 05/10] mlxsw: spectrum: qdiscs: Update backlog handling of a child qdiscs

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

When removing a child qdisc its backlog will be decreased from the parent
backlog. The driver backlog count should do the same.
When the parent changes its configuration, the child might need to clean
its stats. However, the backlog can't be cleaned with the rest of the
stats, because it reflects a momentary value that needs to be synced with
the core, not the history of the qdisc.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Reviewed-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
index b722af360475..5ddaafc8aa18 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
@@ -294,6 +294,12 @@ static int
 mlxsw_sp_qdisc_red_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
 			   struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
 {
+	struct mlxsw_sp_qdisc *root_qdisc = mlxsw_sp_port->root_qdisc;
+
+	if (root_qdisc != mlxsw_sp_qdisc)
+		root_qdisc->stats_base.backlog -=
+					mlxsw_sp_qdisc->stats_base.backlog;
+
 	return mlxsw_sp_tclass_congestion_disable(mlxsw_sp_port,
 						  mlxsw_sp_qdisc->tclass_num);
 }
@@ -358,6 +364,7 @@ mlxsw_sp_qdisc_red_unoffload(struct mlxsw_sp_port *mlxsw_sp_port,
 	backlog = mlxsw_sp_cells_bytes(mlxsw_sp_port->mlxsw_sp,
 				       mlxsw_sp_qdisc->stats_base.backlog);
 	p->qstats->backlog -= backlog;
+	mlxsw_sp_qdisc->stats_base.backlog = 0;
 }
 
 static int
@@ -512,7 +519,7 @@ mlxsw_sp_qdisc_prio_replace(struct mlxsw_sp_port *mlxsw_sp_port,
 {
 	struct tc_prio_qopt_offload_params *p = params;
 	struct mlxsw_sp_qdisc *child_qdisc;
-	int tclass, i, band;
+	int tclass, i, band, backlog;
 	u8 old_priomap;
 	int err;
 
@@ -533,9 +540,12 @@ mlxsw_sp_qdisc_prio_replace(struct mlxsw_sp_port *mlxsw_sp_port,
 			}
 		}
 		if (old_priomap != child_qdisc->prio_bitmap &&
-		    child_qdisc->ops && child_qdisc->ops->clean_stats)
+		    child_qdisc->ops && child_qdisc->ops->clean_stats) {
+			backlog = child_qdisc->stats_base.backlog;
 			child_qdisc->ops->clean_stats(mlxsw_sp_port,
 						      child_qdisc);
+			child_qdisc->stats_base.backlog = backlog;
+		}
 	}
 
 	return 0;
-- 
2.14.3

[patch net-next 06/10] net: sch: Don't warn on missmatching qlen and backlog for offloaded qdiscs

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Offloaded qdiscs are allowed to expose only parts of their statistics.
It means that if backlog is being exposed and qlen is not, it might trigger
a warning in qdisc_tree_reduce_backlog.
Do not warn in case the qdisc that was removed was an offloaded one.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Reviewed-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 net/sched/sch_api.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index 27e672c12492..68f9d942bed4 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -739,6 +739,7 @@ static u32 qdisc_alloc_handle(struct net_device *dev)
 void qdisc_tree_reduce_backlog(struct Qdisc *sch, unsigned int n,
 			       unsigned int len)
 {
+	bool qdisc_is_offloaded = sch->flags & TCQ_F_OFFLOADED;
 	const struct Qdisc_class_ops *cops;
 	unsigned long cl;
 	u32 parentid;
@@ -760,8 +761,12 @@ void qdisc_tree_reduce_backlog(struct Qdisc *sch, unsigned int n,
 		 * If child was empty even before update then backlog
 		 * counter is screwed and we skip notification because
 		 * parent class is already passive.
+		 *
+		 * If the original child was offloaded then it is allowed
+		 * to be seem as empty, so the parent is notified anyway.
 		 */
-		notify = !sch->q.qlen && !WARN_ON_ONCE(!n);
+		notify = !sch->q.qlen && !WARN_ON_ONCE(!n &&
+						       !qdisc_is_offloaded);
 		/* TODO: perform the search on a per txq basis */
 		sch = qdisc_lookup(qdisc_dev(sch), TC_H_MAJ(parentid));
 		if (sch == NULL) {
-- 
2.14.3

[patch net-next 07/10] mlxsw: spectrum: Update sch_prio stats to include sch_red related drops

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

sch_prio as root qdisc should count all the drops its children have. Since
it is possible for it to have sch_red children, it needs to count RED early
drops.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Reviewed-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
index 5ddaafc8aa18..bed7495d35d6 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
@@ -584,6 +584,7 @@ mlxsw_sp_qdisc_get_prio_stats(struct mlxsw_sp_port *mlxsw_sp_port,
 
 	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
 		drops += xstats->tail_drop[i];
+		drops += xstats->wred_drop[i];
 		backlog += xstats->backlog[i];
 	}
 	drops = drops - stats_base->drops;
@@ -619,8 +620,10 @@ mlxsw_sp_setup_tc_qdisc_prio_clean_stats(struct mlxsw_sp_port *mlxsw_sp_port,
 	stats_base->tx_bytes = stats->tx_bytes;
 
 	stats_base->drops = 0;
-	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
 		stats_base->drops += xstats->tail_drop[i];
+		stats_base->drops += xstats->wred_drop[i];
+	}
 
 	mlxsw_sp_qdisc->stats_base.backlog = 0;
 }
-- 
2.14.3

[patch net-next 08/10] mlxsw: spectrum: qdiscs: prio: Delete child qdiscs when removing bands

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

When the number the bands of sch_prio is decreased, child qdiscs on the
deleted bands would get deleted as well.
This change and deletions are being done under sch_tree_lock of the
sch_prio qdisc. Part of the destruction of qdisc is unoffloading it, if
it is offloaded. Un-offloading can't be done inside this lock.
Move the offload command to be done before reducing the number of bands,
so unoffloading of the qdiscs that are about to be deleted could be done
outside of the lock.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Reviewed-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c |  7 ++++++-
 net/sched/sch_prio.c                                 | 13 ++++++-------
 2 files changed, 12 insertions(+), 8 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
index bed7495d35d6..a2a3ac09c3bc 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
@@ -547,7 +547,12 @@ mlxsw_sp_qdisc_prio_replace(struct mlxsw_sp_port *mlxsw_sp_port,
 			child_qdisc->stats_base.backlog = backlog;
 		}
 	}
-
+	for (; band < IEEE_8021QAZ_MAX_TCS; band++) {
+		tclass = MLXSW_SP_PRIO_BAND_TO_TCLASS(band);
+		child_qdisc = &mlxsw_sp_port->tclass_qdiscs[tclass];
+		child_qdisc->prio_bitmap = 0;
+		mlxsw_sp_qdisc_destroy(mlxsw_sp_port, child_qdisc);
+	}
 	return 0;
 }
 
diff --git a/net/sched/sch_prio.c b/net/sched/sch_prio.c
index efbf51f35778..ba2d6d17d95a 100644
--- a/net/sched/sch_prio.c
+++ b/net/sched/sch_prio.c
@@ -142,9 +142,8 @@ prio_reset(struct Qdisc *sch)
 	sch->q.qlen = 0;
 }
 
-static int prio_offload(struct Qdisc *sch, bool enable)
+static int prio_offload(struct Qdisc *sch, struct tc_prio_qopt *qopt)
 {
-	struct prio_sched_data *q = qdisc_priv(sch);
 	struct net_device *dev = qdisc_dev(sch);
 	struct tc_prio_qopt_offload opt = {
 		.handle = sch->handle,
@@ -154,10 +153,10 @@ static int prio_offload(struct Qdisc *sch, bool enable)
 	if (!tc_can_offload(dev) || !dev->netdev_ops->ndo_setup_tc)
 		return -EOPNOTSUPP;
 
-	if (enable) {
+	if (qopt) {
 		opt.command = TC_PRIO_REPLACE;
-		opt.replace_params.bands = q->bands;
-		memcpy(&opt.replace_params.priomap, q->prio2band,
+		opt.replace_params.bands = qopt->bands;
+		memcpy(&opt.replace_params.priomap, qopt->priomap,
 		       TC_PRIO_MAX + 1);
 		opt.replace_params.qstats = &sch->qstats;
 	} else {
@@ -174,7 +173,7 @@ prio_destroy(struct Qdisc *sch)
 	struct prio_sched_data *q = qdisc_priv(sch);
 
 	tcf_block_put(q->block);
-	prio_offload(sch, false);
+	prio_offload(sch, NULL);
 	for (prio = 0; prio < q->bands; prio++)
 		qdisc_destroy(q->queues[prio]);
 }
@@ -211,6 +210,7 @@ static int prio_tune(struct Qdisc *sch, struct nlattr *opt,
 		}
 	}
 
+	prio_offload(sch, qopt);
 	sch_tree_lock(sch);
 	q->bands = qopt->bands;
 	memcpy(q->prio2band, qopt->priomap, TC_PRIO_MAX+1);
@@ -230,7 +230,6 @@ static int prio_tune(struct Qdisc *sch, struct nlattr *opt,
 	}
 
 	sch_tree_unlock(sch);
-	prio_offload(sch, true);
 	return 0;
 }
 
-- 
2.14.3

[patch net-next 09/10] net: sch: prio: Add offload ability for grafting a child

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Offload sch_prio graft command for capable drivers.
Warn in case of a failure, unless the graft was done as part of a destroy
operation (the new qdisc is a noop) or if all the qdiscs (the parent, the
old child, and the new one) are not offloaded.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Reviewed-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 include/net/pkt_cls.h |  8 ++++++++
 net/sched/sch_prio.c  | 32 ++++++++++++++++++++++++++++++++
 2 files changed, 40 insertions(+)

diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h
index 87406252f0a3..e828d31be5da 100644
--- a/include/net/pkt_cls.h
+++ b/include/net/pkt_cls.h
@@ -806,6 +806,7 @@ enum tc_prio_command {
 	TC_PRIO_REPLACE,
 	TC_PRIO_DESTROY,
 	TC_PRIO_STATS,
+	TC_PRIO_GRAFT,
 };
 
 struct tc_prio_qopt_offload_params {
@@ -818,6 +819,11 @@ struct tc_prio_qopt_offload_params {
 	struct gnet_stats_queue *qstats;
 };
 
+struct tc_prio_qopt_offload_graft_params {
+	u8 band;
+	u32 child_handle;
+};
+
 struct tc_prio_qopt_offload {
 	enum tc_prio_command command;
 	u32 handle;
@@ -825,6 +831,8 @@ struct tc_prio_qopt_offload {
 	union {
 		struct tc_prio_qopt_offload_params replace_params;
 		struct tc_qopt_offload_stats stats;
+		struct tc_prio_qopt_offload_graft_params graft_params;
 	};
 };
+
 #endif
diff --git a/net/sched/sch_prio.c b/net/sched/sch_prio.c
index ba2d6d17d95a..222e53d3d27a 100644
--- a/net/sched/sch_prio.c
+++ b/net/sched/sch_prio.c
@@ -308,12 +308,44 @@ static int prio_graft(struct Qdisc *sch, unsigned long arg, struct Qdisc *new,
 		      struct Qdisc **old, struct netlink_ext_ack *extack)
 {
 	struct prio_sched_data *q = qdisc_priv(sch);
+	struct tc_prio_qopt_offload graft_offload;
+	struct net_device *dev = qdisc_dev(sch);
 	unsigned long band = arg - 1;
+	bool any_qdisc_is_offloaded;
+	int err;
 
 	if (new == NULL)
 		new = &noop_qdisc;
 
 	*old = qdisc_replace(sch, new, &q->queues[band]);
+
+	if (!tc_can_offload(dev))
+		return 0;
+
+	graft_offload.handle = sch->handle;
+	graft_offload.parent = sch->parent;
+	graft_offload.graft_params.band = band;
+	graft_offload.graft_params.child_handle = new->handle;
+	graft_offload.command = TC_PRIO_GRAFT;
+
+	err = dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_QDISC_PRIO,
+					    &graft_offload);
+
+	/* Don't report error if the graft is part of destroy operation. */
+	if (err && new != &noop_qdisc) {
+		/* Don't report error if the parent, the old child and the new
+		 * one are not offloaded.
+		 */
+		any_qdisc_is_offloaded = sch->flags & TCQ_F_OFFLOADED;
+		any_qdisc_is_offloaded |= new->flags & TCQ_F_OFFLOADED;
+		if (*old)
+			any_qdisc_is_offloaded |= (*old)->flags &
+						   TCQ_F_OFFLOADED;
+
+		if (any_qdisc_is_offloaded)
+			NL_SET_ERR_MSG(extack, "Offloading graft operation failed.");
+	}
+
 	return 0;
 }
 
-- 
2.14.3

Re: [patch net-next 09/10] net: sch: prio: Add offload ability for grafting a child

[Alexander Aring]

Hi,

On Wed, Feb 28, 2018 at 4:45 AM, Jiri Pirko <jiri@resnulli.us> wrote:
> From: Nogah Frankel <nogahf@mellanox.com>
>
> Offload sch_prio graft command for capable drivers.
> Warn in case of a failure, unless the graft was done as part of a destroy
> operation (the new qdisc is a noop) or if all the qdiscs (the parent, the
> old child, and the new one) are not offloaded.
>
> Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
> Reviewed-by: Yuval Mintz <yuvalm@mellanox.com>
> Signed-off-by: Jiri Pirko <jiri@mellanox.com>
> ---
>  include/net/pkt_cls.h |  8 ++++++++
>  net/sched/sch_prio.c  | 32 ++++++++++++++++++++++++++++++++
...
> +               if (*old)
> +                       any_qdisc_is_offloaded |= (*old)->flags &
> +                                                  TCQ_F_OFFLOADED;
> +
> +               if (any_qdisc_is_offloaded)
> +                       NL_SET_ERR_MSG(extack, "Offloading graft operation failed.");
> +       }
> +
>         return 0;
>  }
>

I guess to make extack working, you need to return an errno if failed.

- Alex

Re: [patch net-next 09/10] net: sch: prio: Add offload ability for grafting a child

[Jakub Kicinski]

On Thu, 1 Mar 2018 22:38:50 -0500, Alexander Aring wrote:
> I guess to make extack working, you need to return an errno if failed.

AFAIK extack is printed as a warning if operation did not fail.

Re: [patch net-next 09/10] net: sch: prio: Add offload ability for grafting a child

[Jiri Pirko]

Fri, Mar 02, 2018 at 04:48:40AM CET, kubakici@wp.pl wrote:
>On Thu, 1 Mar 2018 22:38:50 -0500, Alexander Aring wrote:
>> I guess to make extack working, you need to return an errno if failed.
>
>AFAIK extack is printed as a warning if operation did not fail.

Yes, I checked this and it is printed as warning.

Re: [patch net-next 09/10] net: sch: prio: Add offload ability for grafting a child

[Alexander Aring]

Hi,

On Fri, Mar 2, 2018 at 3:37 AM, Jiri Pirko <jiri@resnulli.us> wrote:
> Fri, Mar 02, 2018 at 04:48:40AM CET, kubakici@wp.pl wrote:
>>On Thu, 1 Mar 2018 22:38:50 -0500, Alexander Aring wrote:
>>> I guess to make extack working, you need to return an errno if failed.
>>
>>AFAIK extack is printed as a warning if operation did not fail.
>
> Yes, I checked this and it is printed as warning.

ouch, so far I know extack it allows only one messages delivered back
to the user space.

If we introduce a warning in the successful path here, it could be
that in the callpath (after "successful" part of this callback), that
somebody else want to add a warning and overwrites actually your
warning (even, he is not aware that this warning was set - okay I
suppose you can do another check on NULL and set a warning, that
somebody overwrites a warning :-D).

If extack messages get's append and is some kind of for_each_nested
string in netlink -> we have no problem, but I guess this not how it's
working. :-/

- Alex

Re: [patch net-next 09/10] net: sch: prio: Add offload ability for grafting a child

[Marcelo Ricardo Leitner]

On Fri, Mar 02, 2018 at 09:21:56AM -0500, Alexander Aring wrote:
> Hi,
> 
> On Fri, Mar 2, 2018 at 3:37 AM, Jiri Pirko <jiri@resnulli.us> wrote:
> > Fri, Mar 02, 2018 at 04:48:40AM CET, kubakici@wp.pl wrote:
> >>On Thu, 1 Mar 2018 22:38:50 -0500, Alexander Aring wrote:
> >>> I guess to make extack working, you need to return an errno if failed.
> >>
> >>AFAIK extack is printed as a warning if operation did not fail.
> >
> > Yes, I checked this and it is printed as warning.
> 
> ouch, so far I know extack it allows only one messages delivered back
> to the user space.
> 
> If we introduce a warning in the successful path here, it could be
> that in the callpath (after "successful" part of this callback), that
> somebody else want to add a warning and overwrites actually your
> warning (even, he is not aware that this warning was set - okay I
> suppose you can do another check on NULL and set a warning, that
> somebody overwrites a warning :-D).
> 
> If extack messages get's append and is some kind of for_each_nested
> string in netlink -> we have no problem, but I guess this not how it's
> working. :-/

IOW I guess what we are looking for here is a way to use extack to
track more than an error/warning message, but to be a bit more
complete error reporting tool.

The case here is pretty much like the case with tc flower offloading,
to issue a message when it couldn't offload while it wasn't fatal for
the rule (in case SKIP_SW wasn't specified). The reason for why the
offloading couldn't happen could have been a temporary one and such
log of a warning is important for troubleshooting.

  Marcelo

[patch net-next 10/10] mlxsw: spectrum: qdiscs: prio: Handle graft command

[Jiri Pirko]

From: Nogah Frankel <nogahf@mellanox.com>

Handle graft command for an offloaded sch_prio.
Grafting a qdisc to any place other than under its original parent is not
supported by mlxsw and will cause the grafted qdisc to stop being
offloaded.

Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Reviewed-by: Yuval Mintz <yuvalm@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 .../net/ethernet/mellanox/mlxsw/spectrum_qdisc.c   | 54 ++++++++++++++++++++++
 1 file changed, 54 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
index a2a3ac09c3bc..91262b0573e3 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_qdisc.c
@@ -122,6 +122,24 @@ mlxsw_sp_qdisc_find(struct mlxsw_sp_port *mlxsw_sp_port, u32 parent,
 	return &mlxsw_sp_port->tclass_qdiscs[tclass];
 }
 
+static struct mlxsw_sp_qdisc *
+mlxsw_sp_qdisc_find_by_handle(struct mlxsw_sp_port *mlxsw_sp_port, u32 handle)
+{
+	int i;
+
+	if (mlxsw_sp_port->root_qdisc->handle == handle)
+		return mlxsw_sp_port->root_qdisc;
+
+	if (mlxsw_sp_port->root_qdisc->handle == TC_H_UNSPEC)
+		return NULL;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
+		if (mlxsw_sp_port->tclass_qdiscs[i].handle == handle)
+			return &mlxsw_sp_port->tclass_qdiscs[i];
+
+	return NULL;
+}
+
 static int
 mlxsw_sp_qdisc_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
 		       struct mlxsw_sp_qdisc *mlxsw_sp_qdisc)
@@ -643,6 +661,39 @@ static struct mlxsw_sp_qdisc_ops mlxsw_sp_qdisc_ops_prio = {
 	.clean_stats = mlxsw_sp_setup_tc_qdisc_prio_clean_stats,
 };
 
+/* Grafting is not supported in mlxsw. It will result in un-offloading of the
+ * grafted qdisc as well as the qdisc in the qdisc new location.
+ * (However, if the graft is to the location where the qdisc is already at, it
+ * will be ignored completely and won't cause un-offloading).
+ */
+static int
+mlxsw_sp_qdisc_prio_graft(struct mlxsw_sp_port *mlxsw_sp_port,
+			  struct mlxsw_sp_qdisc *mlxsw_sp_qdisc,
+			  struct tc_prio_qopt_offload_graft_params *p)
+{
+	int tclass_num = MLXSW_SP_PRIO_BAND_TO_TCLASS(p->band);
+	struct mlxsw_sp_qdisc *old_qdisc;
+
+	/* Check if the grafted qdisc is already in its "new" location. If so -
+	 * nothing needs to be done.
+	 */
+	if (p->band < IEEE_8021QAZ_MAX_TCS &&
+	    mlxsw_sp_port->tclass_qdiscs[tclass_num].handle == p->child_handle)
+		return 0;
+
+	/* See if the grafted qdisc is already offloaded on any tclass. If so,
+	 * unoffload it.
+	 */
+	old_qdisc = mlxsw_sp_qdisc_find_by_handle(mlxsw_sp_port,
+						  p->child_handle);
+	if (old_qdisc)
+		mlxsw_sp_qdisc_destroy(mlxsw_sp_port, old_qdisc);
+
+	mlxsw_sp_qdisc_destroy(mlxsw_sp_port,
+			       &mlxsw_sp_port->tclass_qdiscs[tclass_num]);
+	return -EOPNOTSUPP;
+}
+
 int mlxsw_sp_setup_tc_prio(struct mlxsw_sp_port *mlxsw_sp_port,
 			   struct tc_prio_qopt_offload *p)
 {
@@ -668,6 +719,9 @@ int mlxsw_sp_setup_tc_prio(struct mlxsw_sp_port *mlxsw_sp_port,
 	case TC_PRIO_STATS:
 		return mlxsw_sp_qdisc_get_stats(mlxsw_sp_port, mlxsw_sp_qdisc,
 						&p->stats);
+	case TC_PRIO_GRAFT:
+		return mlxsw_sp_qdisc_prio_graft(mlxsw_sp_port, mlxsw_sp_qdisc,
+						 &p->graft_params);
 	default:
 		return -EOPNOTSUPP;
 	}
-- 
2.14.3

Re: [patch net-next 00/10] mlxsw: Offload multi-queue RED support

[David Miller]

From: Jiri Pirko <jiri@resnulli.us>
Date: Wed, 28 Feb 2018 10:44:57 +0100

> Support a two level hierarchy of offloaded qdiscs in mlxsw, with sch_prio
> being the root qdisc and sch_red as the children.

Ok I applied this series.

However, we really need to have a real discussion about those stats
issues.

I understand the kinds of constraints under which you operate when
offloading, but what, if anything, can we do to close the gap
of inconsistent stats exports in the long term?

Thanks.