[patch net-next 00/17] mlxsw: Introduce support for Data Center Bridging

[patch net-next 00/17] mlxsw: Introduce support for Data Center Bridging

[Jiri Pirko]

From: Jiri Pirko <jiri@mellanox.com>

Ido says:

This patchset introduces support for Quality of Service (QoS) as part of the
IEEE Data Center Bridiging (DCB) standards.

Patches 1-9 do the required device initialization. Specifically, patches 1-6
initialize the ports' headroom buffers, which are used at ingress to store
incoming packets while they go through the switch's pipeline. Patches 7-9
complete them by initializing the egress scheduling.

The pipeline mentioned above determines the packet's egress port(s) and
traffic class. Ideally, once out of the pipeline the packet moves to the
switch's shared buffer (to be introduced in Jiri's patchset, currently
default values are used) and scheduled for transmission according to its
traffic class. The egress scheduling is configured according to the 802.1Qaz
standard, which is part of the DCB infrastructure supported by Linux. This
is introduced in patches 10-12.

Even after going through the pipeline packets are not always eligible to
enter the shared buffer. This is determined by the amount of available space
and the quotas associated with the packet. However, if flow control is
enabled and the packet is associated with the lossless flow, then it will
stay in the headroom and won't be discarded. This is introduced in patches
13-17.

Please check individual commit messages for more info, as I tried to keep
them pretty detailed.

Thanks.

Ido Schimmel (17):
  mlxsw: reg: Add Port Prio To Buffer register
  mlxsw: spectrum: Map all switch priorities to priority group 0
  mlxsw: spectrum: Add bytes to cells helper
  mlxsw: spectrum: Correctly configure headroom size
  mlxsw: reg: Use correct PBMC register length
  mlxsw: spectrum: Set port's shared buffer size to 0
  mlxsw: reg: Add QoS ETS Element Configuration register
  mlxsw: reg: Add QoS Switch Traffic Class Table register
  mlxsw: spectrum: Initialize egress scheduling
  mlxsw: spectrum: Introduce support for Data Center Bridging (DCB)
  mlxsw: spectrum: Add IEEE 802.1Qaz ETS support
  mlxsw: spectrum: Allow setting maximum rate for a TC
  mlxsw: reg: Add Port Flow Control Configuration register
  mlxsw: reg: Add lossless settings for PBMC register
  mlxsw: spectrum: Add support for PAUSE frames
  mlxsw: reg: Introduce per priority counters
  mlxsw: spectrum: Add IEEE 802.1Qbb PFC support

 drivers/net/ethernet/mellanox/mlxsw/Kconfig        |   8 +
 drivers/net/ethernet/mellanox/mlxsw/Makefile       |   1 +
 drivers/net/ethernet/mellanox/mlxsw/reg.h          | 507 ++++++++++++++++++++-
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     | 263 ++++++++++-
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  61 +++
 .../net/ethernet/mellanox/mlxsw/spectrum_buffers.c | 108 +++--
 drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c | 480 +++++++++++++++++++
 drivers/net/ethernet/mellanox/mlxsw/switchx2.c     |   3 +-
 8 files changed, 1381 insertions(+), 50 deletions(-)
 create mode 100644 drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c

-- 
2.5.5

[patch net-next 01/17] mlxsw: reg: Add Port Prio To Buffer register

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

When packets ingress the switch they are assigned a switch priority
number that dictates the packet's priority group (PG) buffer in the
port's headroom buffer.

Add the Port Prio To Buffer (PPTB) register, which configures the switch
priority to PG mapping.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/reg.h | 83 +++++++++++++++++++++++++++++++
 1 file changed, 83 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index ffe4c03..0995bee 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -2340,6 +2340,87 @@ static inline void mlxsw_reg_ppcnt_pack(char *payload, u8 local_port)
 	mlxsw_reg_ppcnt_prio_tc_set(payload, 0);
 }
 
+/* PPTB - Port Prio To Buffer Register
+ * -----------------------------------
+ * Configures the switch priority to buffer table.
+ */
+#define MLXSW_REG_PPTB_ID 0x500B
+#define MLXSW_REG_PPTB_LEN 0x0C
+
+static const struct mlxsw_reg_info mlxsw_reg_pptb = {
+	.id = MLXSW_REG_PPTB_ID,
+	.len = MLXSW_REG_PPTB_LEN,
+};
+
+enum {
+	MLXSW_REG_PPTB_MM_UM,
+	MLXSW_REG_PPTB_MM_UNICAST,
+	MLXSW_REG_PPTB_MM_MULTICAST,
+};
+
+/* reg_pptb_mm
+ * Mapping mode.
+ * 0 - Map both unicast and multicast packets to the same buffer.
+ * 1 - Map only unicast packets.
+ * 2 - Map only multicast packets.
+ * Access: Index
+ *
+ * Note: SwitchX-2 only supports the first option.
+ */
+MLXSW_ITEM32(reg, pptb, mm, 0x00, 28, 2);
+
+/* reg_pptb_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pptb, local_port, 0x00, 16, 8);
+
+/* reg_pptb_um
+ * Enables the update of the untagged_buf field.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pptb, um, 0x00, 8, 1);
+
+/* reg_pptb_pm
+ * Enables the update of the prio_to_buff field.
+ * Bit <i> is a flag for updating the mapping for switch priority <i>.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pptb, pm, 0x00, 0, 8);
+
+/* reg_pptb_prio_to_buff
+ * Mapping of switch priority <i> to one of the allocated receive port
+ * buffers.
+ * Access: RW
+ */
+MLXSW_ITEM_BIT_ARRAY(reg, pptb, prio_to_buff, 0x04, 0x04, 4);
+
+/* reg_pptb_pm_msb
+ * Enables the update of the prio_to_buff field.
+ * Bit <i> is a flag for updating the mapping for switch priority <i+8>.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pptb, pm_msb, 0x08, 24, 8);
+
+/* reg_pptb_untagged_buff
+ * Mapping of untagged frames to one of the allocated receive port buffers.
+ * Access: RW
+ *
+ * Note: In SwitchX-2 this field must be mapped to buffer 8. Reserved for
+ * Spectrum, as it maps untagged packets based on the default switch priority.
+ */
+MLXSW_ITEM32(reg, pptb, untagged_buff, 0x08, 0, 4);
+
+#define MLXSW_REG_PPTB_ALL_PRIO 0xFF
+
+static inline void mlxsw_reg_pptb_pack(char *payload, u8 local_port)
+{
+	MLXSW_REG_ZERO(pptb, payload);
+	mlxsw_reg_pptb_mm_set(payload, MLXSW_REG_PPTB_MM_UM);
+	mlxsw_reg_pptb_local_port_set(payload, local_port);
+	mlxsw_reg_pptb_pm_set(payload, MLXSW_REG_PPTB_ALL_PRIO);
+}
+
 /* PBMC - Port Buffer Management Control Register
  * ----------------------------------------------
  * The PBMC register configures and retrieves the port packet buffer
@@ -3295,6 +3376,8 @@ static inline const char *mlxsw_reg_id_str(u16 reg_id)
 		return "PAOS";
 	case MLXSW_REG_PPCNT_ID:
 		return "PPCNT";
+	case MLXSW_REG_PPTB_ID:
+		return "PPTB";
 	case MLXSW_REG_PBMC_ID:
 		return "PBMC";
 	case MLXSW_REG_PSPA_ID:
-- 
2.5.5

[patch net-next 02/17] mlxsw: spectrum: Map all switch priorities to priority group 0

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

During transmission, the skb's priority is used to map the skb to a
traffic class, where the idea is to group priorities with similar
characteristics (e.g. lossy, lossless) to the same traffic class. By
default, all priorities are mapped to traffic class 0.

In the device, we model the skb's priority as the switch priority, which
is assigned to a packet according to its PCP value and ingress port
(untagged packets are assigned the port's default switch priority - 0).

At ingress, the packet is directed to a priority group (PG) buffer in
the port's headroom buffer according to the packet's switch priority and
switch priority to buffer mapping.

While it's possible to configure the egress mapping between skb's
priority (switch priority) and traffic class, there is no mechanism to
configure the ingress mapping to a PG.

In order to keep things simple and since grouping certain priorities into
a traffic class at egress also implies they should be grouped the same
at ingress, treat a PG as the ingress counterpart of an egress traffic
class.

Having established the above, during initialization map all the switch
priorities to PG0 in accordance with the Linux defaults for traffic
class mapping.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 .../net/ethernet/mellanox/mlxsw/spectrum_buffers.c | 25 +++++++++++++++++++++-
 1 file changed, 24 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
index d59195e..c3a275b 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
@@ -34,6 +34,7 @@
 
 #include <linux/kernel.h>
 #include <linux/types.h>
+#include <linux/dcbnl.h>
 
 #include "spectrum.h"
 #include "core.h"
@@ -82,6 +83,28 @@ static int mlxsw_sp_port_pb_init(struct mlxsw_sp_port *mlxsw_sp_port)
 			       MLXSW_REG(pbmc), pbmc_pl);
 }
 
+static int mlxsw_sp_port_pb_prio_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	char pptb_pl[MLXSW_REG_PPTB_LEN];
+	int i;
+
+	mlxsw_reg_pptb_pack(pptb_pl, mlxsw_sp_port->local_port);
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
+		mlxsw_reg_pptb_prio_to_buff_set(pptb_pl, i, 0);
+	return mlxsw_reg_write(mlxsw_sp_port->mlxsw_sp->core, MLXSW_REG(pptb),
+			       pptb_pl);
+}
+
+static int mlxsw_sp_port_headroom_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	int err;
+
+	err = mlxsw_sp_port_pb_init(mlxsw_sp_port);
+	if (err)
+		return err;
+	return mlxsw_sp_port_pb_prio_init(mlxsw_sp_port);
+}
+
 #define MLXSW_SP_SB_BYTES_PER_CELL 96
 
 struct mlxsw_sp_sb_pool {
@@ -410,7 +433,7 @@ int mlxsw_sp_port_buffers_init(struct mlxsw_sp_port *mlxsw_sp_port)
 {
 	int err;
 
-	err = mlxsw_sp_port_pb_init(mlxsw_sp_port);
+	err = mlxsw_sp_port_headroom_init(mlxsw_sp_port);
 	if (err)
 		return err;
 	err = mlxsw_sp_port_sb_cms_init(mlxsw_sp_port);
-- 
2.5.5

[patch net-next 03/17] mlxsw: spectrum: Add bytes to cells helper

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

Buffers in the switch store packets in units called buffer cells. Add a
helper to convert from bytes to cells, so that the actual number of
cells required (result is round up) is returned.

Also, drop the SB (shared buffer) acronym from the BYTES_PER_CELL macro,
as this unit is also used in the ports' buffers and not only the
switch's shared buffer.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  4 ++
 .../net/ethernet/mellanox/mlxsw/spectrum_buffers.c | 64 ++++++++++------------
 2 files changed, 34 insertions(+), 34 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index d58ab0c..84dddd4 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -62,6 +62,10 @@
 
 #define MLXSW_SP_PORT_BASE_SPEED 25000	/* Mb/s */
 
+#define MLXSW_SP_BYTES_PER_CELL 96
+
+#define MLXSW_SP_BYTES_TO_CELLS(b) DIV_ROUND_UP(b, MLXSW_SP_BYTES_PER_CELL)
+
 struct mlxsw_sp_port;
 
 struct mlxsw_sp_upper {
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
index c3a275b..dadb6e1 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
@@ -105,8 +105,6 @@ static int mlxsw_sp_port_headroom_init(struct mlxsw_sp_port *mlxsw_sp_port)
 	return mlxsw_sp_port_pb_prio_init(mlxsw_sp_port);
 }
 
-#define MLXSW_SP_SB_BYTES_PER_CELL 96
-
 struct mlxsw_sp_sb_pool {
 	u8 pool;
 	enum mlxsw_reg_sbpr_dir dir;
@@ -115,11 +113,9 @@ struct mlxsw_sp_sb_pool {
 };
 
 #define MLXSW_SP_SB_POOL_INGRESS_SIZE				\
-	((15000000 - (2 * 20000 * MLXSW_PORT_MAX_PORTS)) /	\
-	 MLXSW_SP_SB_BYTES_PER_CELL)
+	(15000000 - (2 * 20000 * MLXSW_PORT_MAX_PORTS))
 #define MLXSW_SP_SB_POOL_EGRESS_SIZE				\
-	((14000000 - (8 * 1500 * MLXSW_PORT_MAX_PORTS)) /	\
-	 MLXSW_SP_SB_BYTES_PER_CELL)
+	(14000000 - (8 * 1500 * MLXSW_PORT_MAX_PORTS))
 
 #define MLXSW_SP_SB_POOL(_pool, _dir, _mode, _size)		\
 	{							\
@@ -138,14 +134,14 @@ struct mlxsw_sp_sb_pool {
 			 MLXSW_REG_SBPR_MODE_DYNAMIC, _size)
 
 static const struct mlxsw_sp_sb_pool mlxsw_sp_sb_pools[] = {
-	MLXSW_SP_SB_POOL_INGRESS(0, MLXSW_SP_SB_POOL_INGRESS_SIZE),
+	MLXSW_SP_SB_POOL_INGRESS(0, MLXSW_SP_BYTES_TO_CELLS(MLXSW_SP_SB_POOL_INGRESS_SIZE)),
 	MLXSW_SP_SB_POOL_INGRESS(1, 0),
 	MLXSW_SP_SB_POOL_INGRESS(2, 0),
 	MLXSW_SP_SB_POOL_INGRESS(3, 0),
-	MLXSW_SP_SB_POOL_EGRESS(0, MLXSW_SP_SB_POOL_EGRESS_SIZE),
+	MLXSW_SP_SB_POOL_EGRESS(0, MLXSW_SP_BYTES_TO_CELLS(MLXSW_SP_SB_POOL_EGRESS_SIZE)),
 	MLXSW_SP_SB_POOL_EGRESS(1, 0),
 	MLXSW_SP_SB_POOL_EGRESS(2, 0),
-	MLXSW_SP_SB_POOL_EGRESS(2, MLXSW_SP_SB_POOL_EGRESS_SIZE),
+	MLXSW_SP_SB_POOL_EGRESS(2, MLXSW_SP_BYTES_TO_CELLS(MLXSW_SP_SB_POOL_EGRESS_SIZE)),
 };
 
 #define MLXSW_SP_SB_POOLS_LEN ARRAY_SIZE(mlxsw_sp_sb_pools)
@@ -201,7 +197,7 @@ struct mlxsw_sp_sb_cm {
 	MLXSW_SP_SB_CM(_tc, MLXSW_REG_SBCM_DIR_EGRESS, 104, 2, 3)
 
 static const struct mlxsw_sp_sb_cm mlxsw_sp_sb_cms[] = {
-	MLXSW_SP_SB_CM_INGRESS(0, 10000 / MLXSW_SP_SB_BYTES_PER_CELL, 8),
+	MLXSW_SP_SB_CM_INGRESS(0, MLXSW_SP_BYTES_TO_CELLS(10000), 8),
 	MLXSW_SP_SB_CM_INGRESS(1, 0, 0),
 	MLXSW_SP_SB_CM_INGRESS(2, 0, 0),
 	MLXSW_SP_SB_CM_INGRESS(3, 0, 0),
@@ -209,15 +205,15 @@ static const struct mlxsw_sp_sb_cm mlxsw_sp_sb_cms[] = {
 	MLXSW_SP_SB_CM_INGRESS(5, 0, 0),
 	MLXSW_SP_SB_CM_INGRESS(6, 0, 0),
 	MLXSW_SP_SB_CM_INGRESS(7, 0, 0),
-	MLXSW_SP_SB_CM_INGRESS(9, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff),
-	MLXSW_SP_SB_CM_EGRESS(0, 1500 / MLXSW_SP_SB_BYTES_PER_CELL, 9),
-	MLXSW_SP_SB_CM_EGRESS(1, 1500 / MLXSW_SP_SB_BYTES_PER_CELL, 9),
-	MLXSW_SP_SB_CM_EGRESS(2, 1500 / MLXSW_SP_SB_BYTES_PER_CELL, 9),
-	MLXSW_SP_SB_CM_EGRESS(3, 1500 / MLXSW_SP_SB_BYTES_PER_CELL, 9),
-	MLXSW_SP_SB_CM_EGRESS(4, 1500 / MLXSW_SP_SB_BYTES_PER_CELL, 9),
-	MLXSW_SP_SB_CM_EGRESS(5, 1500 / MLXSW_SP_SB_BYTES_PER_CELL, 9),
-	MLXSW_SP_SB_CM_EGRESS(6, 1500 / MLXSW_SP_SB_BYTES_PER_CELL, 9),
-	MLXSW_SP_SB_CM_EGRESS(7, 1500 / MLXSW_SP_SB_BYTES_PER_CELL, 9),
+	MLXSW_SP_SB_CM_INGRESS(9, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff),
+	MLXSW_SP_SB_CM_EGRESS(0, MLXSW_SP_BYTES_TO_CELLS(1500), 9),
+	MLXSW_SP_SB_CM_EGRESS(1, MLXSW_SP_BYTES_TO_CELLS(1500), 9),
+	MLXSW_SP_SB_CM_EGRESS(2, MLXSW_SP_BYTES_TO_CELLS(1500), 9),
+	MLXSW_SP_SB_CM_EGRESS(3, MLXSW_SP_BYTES_TO_CELLS(1500), 9),
+	MLXSW_SP_SB_CM_EGRESS(4, MLXSW_SP_BYTES_TO_CELLS(1500), 9),
+	MLXSW_SP_SB_CM_EGRESS(5, MLXSW_SP_BYTES_TO_CELLS(1500), 9),
+	MLXSW_SP_SB_CM_EGRESS(6, MLXSW_SP_BYTES_TO_CELLS(1500), 9),
+	MLXSW_SP_SB_CM_EGRESS(7, MLXSW_SP_BYTES_TO_CELLS(1500), 9),
 	MLXSW_SP_SB_CM_EGRESS(8, 0, 0),
 	MLXSW_SP_SB_CM_EGRESS(9, 0, 0),
 	MLXSW_SP_SB_CM_EGRESS(10, 0, 0),
@@ -376,21 +372,21 @@ struct mlxsw_sp_sb_mm {
 	}
 
 static const struct mlxsw_sp_sb_mm mlxsw_sp_sb_mms[] = {
-	MLXSW_SP_SB_MM(0, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(1, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(2, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(3, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(4, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(5, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(6, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(7, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(8, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(9, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(10, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(11, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(12, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(13, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
-	MLXSW_SP_SB_MM(14, 20000 / MLXSW_SP_SB_BYTES_PER_CELL, 0xff, 0),
+	MLXSW_SP_SB_MM(0, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(1, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(2, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(3, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(4, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(5, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(6, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(7, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(8, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(9, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(10, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(11, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(12, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(13, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
+	MLXSW_SP_SB_MM(14, MLXSW_SP_BYTES_TO_CELLS(20000), 0xff, 0),
 };
 
 #define MLXSW_SP_SB_MMS_LEN ARRAY_SIZE(mlxsw_sp_sb_mms)
-- 
2.5.5

[patch net-next 04/17] mlxsw: spectrum: Correctly configure headroom size

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

When packets ingress the switch they are assigned a switch priority and
directed to the corresponding priority group (PG) buffer in the port's
headroom buffer.

Since we now map all switch priorities to priority group 0 (PG0) by
default, there is no need to allocate the other priority groups during
initialization. The only exception is PG9, which is used for control
traffic.

At minimum, the PG should be able to store the currently classified
packet (pipeline latency isn't 0) and also the packets arriving during
the classification time. However, an incoming packet will not be
buffered if there is no available MTU-sized buffer space for storing it.

The buffer needed to accommodate for pipeline latency is variable and
needs to take into account both the current link speed and current
latency of the pipeline, which is time-dependent. Testing showed that
setting the PG's size to twice the current MTU is optimal.

Since PG9 is used strictly for control packets and not subject to flow
control, we are not going to resize it according to user configuration,
so we simply set it according to worst case scenario, which is twice the
maximum MTU.

In any case, later patches in the series will allow a user to direct
lossless flows to other PGs than PG0 and set their size to accommodate
for round-trip propagation delay.

The above change also requires us to resize the PG buffer whenever the
port's MTU is changed.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     | 25 +++++++++++++++++++++-
 .../net/ethernet/mellanox/mlxsw/spectrum_buffers.c | 19 ++++++++--------
 2 files changed, 34 insertions(+), 10 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index cb5f36e..4576d59 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -449,16 +449,39 @@ static int mlxsw_sp_port_set_mac_address(struct net_device *dev, void *p)
 	return 0;
 }
 
+static int mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				      int mtu)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u16 pg_size = 2 * MLXSW_SP_BYTES_TO_CELLS(mtu);
+	char pbmc_pl[MLXSW_REG_PBMC_LEN];
+	int err;
+
+	mlxsw_reg_pbmc_pack(pbmc_pl, mlxsw_sp_port->local_port, 0, 0);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(pbmc), pbmc_pl);
+	if (err)
+		return err;
+	mlxsw_reg_pbmc_lossy_buffer_pack(pbmc_pl, 0, pg_size);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pbmc), pbmc_pl);
+}
+
 static int mlxsw_sp_port_change_mtu(struct net_device *dev, int mtu)
 {
 	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
 	int err;
 
-	err = mlxsw_sp_port_mtu_set(mlxsw_sp_port, mtu);
+	err = mlxsw_sp_port_headroom_set(mlxsw_sp_port, mtu);
 	if (err)
 		return err;
+	err = mlxsw_sp_port_mtu_set(mlxsw_sp_port, mtu);
+	if (err)
+		goto err_port_mtu_set;
 	dev->mtu = mtu;
 	return 0;
+
+err_port_mtu_set:
+	mlxsw_sp_port_headroom_set(mlxsw_sp_port, dev->mtu);
+	return err;
 }
 
 static struct rtnl_link_stats64 *
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
index dadb6e1..e7a5b73 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
@@ -35,6 +35,7 @@
 #include <linux/kernel.h>
 #include <linux/types.h>
 #include <linux/dcbnl.h>
+#include <linux/if_ether.h>
 
 #include "spectrum.h"
 #include "core.h"
@@ -53,15 +54,15 @@ struct mlxsw_sp_pb {
 	}
 
 static const struct mlxsw_sp_pb mlxsw_sp_pbs[] = {
-	MLXSW_SP_PB(0, 208),
-	MLXSW_SP_PB(1, 208),
-	MLXSW_SP_PB(2, 208),
-	MLXSW_SP_PB(3, 208),
-	MLXSW_SP_PB(4, 208),
-	MLXSW_SP_PB(5, 208),
-	MLXSW_SP_PB(6, 208),
-	MLXSW_SP_PB(7, 208),
-	MLXSW_SP_PB(9, 208),
+	MLXSW_SP_PB(0, 2 * MLXSW_SP_BYTES_TO_CELLS(ETH_FRAME_LEN)),
+	MLXSW_SP_PB(1, 0),
+	MLXSW_SP_PB(2, 0),
+	MLXSW_SP_PB(3, 0),
+	MLXSW_SP_PB(4, 0),
+	MLXSW_SP_PB(5, 0),
+	MLXSW_SP_PB(6, 0),
+	MLXSW_SP_PB(7, 0),
+	MLXSW_SP_PB(9, 2 * MLXSW_SP_BYTES_TO_CELLS(MLXSW_PORT_MAX_MTU)),
 };
 
 #define MLXSW_SP_PBS_LEN ARRAY_SIZE(mlxsw_sp_pbs)
-- 
2.5.5

[patch net-next 05/17] mlxsw: reg: Use correct PBMC register length

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

The last field of the PBMC register is at offset 0x64 and its size is
0x8, so the correct register's length is 0x6C bytes.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/reg.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index 0995bee..f08a17f 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -2427,7 +2427,7 @@ static inline void mlxsw_reg_pptb_pack(char *payload, u8 local_port)
  * allocation for different Prios, and the Pause threshold management.
  */
 #define MLXSW_REG_PBMC_ID 0x500C
-#define MLXSW_REG_PBMC_LEN 0x68
+#define MLXSW_REG_PBMC_LEN 0x6C
 
 static const struct mlxsw_reg_info mlxsw_reg_pbmc = {
 	.id = MLXSW_REG_PBMC_ID,
-- 
2.5.5

[patch net-next 06/17] mlxsw: spectrum: Set port's shared buffer size to 0

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

In addition to the priority group (PG) buffers in the headroom, the
device enables the allocation of headroom shared buffer, which can
be shared between different PGs.

However, we are not going to use the headroom shared buffer and instead
allow the user to use its size for PGs or the switch's shared buffer.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/reg.h              | 2 ++
 drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index f08a17f..370914e 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -2455,6 +2455,8 @@ MLXSW_ITEM32(reg, pbmc, xoff_timer_value, 0x04, 16, 16);
  */
 MLXSW_ITEM32(reg, pbmc, xoff_refresh, 0x04, 0, 16);
 
+#define MLXSW_REG_PBMC_PORT_SHARED_BUF_IDX 11
+
 /* reg_pbmc_buf_lossy
  * The field indicates if the buffer is lossy.
  * 0 - Lossless
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
index e7a5b73..97c8d53 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_buffers.c
@@ -80,6 +80,8 @@ static int mlxsw_sp_port_pb_init(struct mlxsw_sp_port *mlxsw_sp_port)
 		pb = &mlxsw_sp_pbs[i];
 		mlxsw_reg_pbmc_lossy_buffer_pack(pbmc_pl, pb->index, pb->size);
 	}
+	mlxsw_reg_pbmc_lossy_buffer_pack(pbmc_pl,
+					 MLXSW_REG_PBMC_PORT_SHARED_BUF_IDX, 0);
 	return mlxsw_reg_write(mlxsw_sp_port->mlxsw_sp->core,
 			       MLXSW_REG(pbmc), pbmc_pl);
 }
-- 
2.5.5

[patch net-next 07/17] mlxsw: reg: Add QoS ETS Element Configuration register

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

We are going to introduce support for DCB, so we need to be able to
configure the traffic selection algorithm (TSA) used by each traffic
class (TC), as well as the bandwidth percentage allocated to each TC in
case of ETS.

Add the QoS ETS Element Configuration register, which controls the
above parameters.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/reg.h | 127 ++++++++++++++++++++++++++++++
 1 file changed, 127 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index 370914e..bc08f8b 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -1805,6 +1805,131 @@ static inline void mlxsw_reg_spvmlr_pack(char *payload, u8 local_port,
 	}
 }
 
+/* QEEC - QoS ETS Element Configuration Register
+ * ---------------------------------------------
+ * Configures the ETS elements.
+ */
+#define MLXSW_REG_QEEC_ID 0x400D
+#define MLXSW_REG_QEEC_LEN 0x1C
+
+static const struct mlxsw_reg_info mlxsw_reg_qeec = {
+	.id = MLXSW_REG_QEEC_ID,
+	.len = MLXSW_REG_QEEC_LEN,
+};
+
+/* reg_qeec_local_port
+ * Local port number.
+ * Access: Index
+ *
+ * Note: CPU port is supported.
+ */
+MLXSW_ITEM32(reg, qeec, local_port, 0x00, 16, 8);
+
+enum mlxsw_reg_qeec_hr {
+	MLXSW_REG_QEEC_HIERARCY_PORT,
+	MLXSW_REG_QEEC_HIERARCY_GROUP,
+	MLXSW_REG_QEEC_HIERARCY_SUBGROUP,
+	MLXSW_REG_QEEC_HIERARCY_TC,
+};
+
+/* reg_qeec_element_hierarchy
+ * 0 - Port
+ * 1 - Group
+ * 2 - Subgroup
+ * 3 - Traffic Class
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, qeec, element_hierarchy, 0x04, 16, 4);
+
+/* reg_qeec_element_index
+ * The index of the element in the hierarchy.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, qeec, element_index, 0x04, 0, 8);
+
+/* reg_qeec_next_element_index
+ * The index of the next (lower) element in the hierarchy.
+ * Access: RW
+ *
+ * Note: Reserved for element_hierarchy 0.
+ */
+MLXSW_ITEM32(reg, qeec, next_element_index, 0x08, 0, 8);
+
+enum {
+	MLXSW_REG_QEEC_BYTES_MODE,
+	MLXSW_REG_QEEC_PACKETS_MODE,
+};
+
+/* reg_qeec_pb
+ * Packets or bytes mode.
+ * 0 - Bytes mode
+ * 1 - Packets mode
+ * Access: RW
+ *
+ * Note: Used for max shaper configuration. For Spectrum, packets mode
+ * is supported only for traffic classes of CPU port.
+ */
+MLXSW_ITEM32(reg, qeec, pb, 0x0C, 28, 1);
+
+/* reg_qeec_mase
+ * Max shaper configuration enable. Enables configuration of the max
+ * shaper on this ETS element.
+ * 0 - Disable
+ * 1 - Enable
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, mase, 0x10, 31, 1);
+
+/* A large max rate will disable the max shaper. */
+#define MLXSW_REG_QEEC_MAS_DIS	200000000	/* Kbps */
+
+/* reg_qeec_max_shaper_rate
+ * Max shaper information rate.
+ * For CPU port, can only be configured for port hierarchy.
+ * When in bytes mode, value is specified in units of 1000bps.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, max_shaper_rate, 0x10, 0, 28);
+
+/* reg_qeec_de
+ * DWRR configuration enable. Enables configuration of the dwrr and
+ * dwrr_weight.
+ * 0 - Disable
+ * 1 - Enable
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, de, 0x18, 31, 1);
+
+/* reg_qeec_dwrr
+ * Transmission selection algorithm to use on the link going down from
+ * the ETS element.
+ * 0 - Strict priority
+ * 1 - DWRR
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, dwrr, 0x18, 15, 1);
+
+/* reg_qeec_dwrr_weight
+ * DWRR weight on the link going down from the ETS element. The
+ * percentage of bandwidth guaranteed to an ETS element within
+ * its hierarchy. The sum of all weights across all ETS elements
+ * within one hierarchy should be equal to 100. Reserved when
+ * transmission selection algorithm is strict priority.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qeec, dwrr_weight, 0x18, 0, 8);
+
+static inline void mlxsw_reg_qeec_pack(char *payload, u8 local_port,
+				       enum mlxsw_reg_qeec_hr hr, u8 index,
+				       u8 next_index)
+{
+	MLXSW_REG_ZERO(qeec, payload);
+	mlxsw_reg_qeec_local_port_set(payload, local_port);
+	mlxsw_reg_qeec_element_hierarchy_set(payload, hr);
+	mlxsw_reg_qeec_element_index_set(payload, index);
+	mlxsw_reg_qeec_next_element_index_set(payload, next_index);
+}
+
 /* PMLP - Ports Module to Local Port Register
  * ------------------------------------------
  * Configures the assignment of modules to local ports.
@@ -3366,6 +3491,8 @@ static inline const char *mlxsw_reg_id_str(u16 reg_id)
 		return "SFMR";
 	case MLXSW_REG_SPVMLR_ID:
 		return "SPVMLR";
+	case MLXSW_REG_QEEC_ID:
+		return "QEEC";
 	case MLXSW_REG_PMLP_ID:
 		return "PMLP";
 	case MLXSW_REG_PMTU_ID:
-- 
2.5.5

[patch net-next 08/17] mlxsw: reg: Add QoS Switch Traffic Class Table register

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

As part of DCB ops we'll have to configure the priority to traffic class
mapping of a port.

Add the QoS Switch Traffic Class Table (QTCT) register, which configures
the mapping between the packet switch priority and traffic class on the
transmit port.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/reg.h | 55 +++++++++++++++++++++++++++++++
 1 file changed, 55 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index bc08f8b..2e58c41 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -1805,6 +1805,59 @@ static inline void mlxsw_reg_spvmlr_pack(char *payload, u8 local_port,
 	}
 }
 
+/* QTCT - QoS Switch Traffic Class Table
+ * -------------------------------------
+ * Configures the mapping between the packet switch priority and the
+ * traffic class on the transmit port.
+ */
+#define MLXSW_REG_QTCT_ID 0x400A
+#define MLXSW_REG_QTCT_LEN 0x08
+
+static const struct mlxsw_reg_info mlxsw_reg_qtct = {
+	.id = MLXSW_REG_QTCT_ID,
+	.len = MLXSW_REG_QTCT_LEN,
+};
+
+/* reg_qtct_local_port
+ * Local port number.
+ * Access: Index
+ *
+ * Note: CPU port is not supported.
+ */
+MLXSW_ITEM32(reg, qtct, local_port, 0x00, 16, 8);
+
+/* reg_qtct_sub_port
+ * Virtual port within the physical port.
+ * Should be set to 0 when virtual ports are not enabled on the port.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, qtct, sub_port, 0x00, 8, 8);
+
+/* reg_qtct_switch_prio
+ * Switch priority.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, qtct, switch_prio, 0x00, 0, 4);
+
+/* reg_qtct_tclass
+ * Traffic class.
+ * Default values:
+ * switch_prio 0 : tclass 1
+ * switch_prio 1 : tclass 0
+ * switch_prio i : tclass i, for i > 1
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, qtct, tclass, 0x04, 0, 4);
+
+static inline void mlxsw_reg_qtct_pack(char *payload, u8 local_port,
+				       u8 switch_prio, u8 tclass)
+{
+	MLXSW_REG_ZERO(qtct, payload);
+	mlxsw_reg_qtct_local_port_set(payload, local_port);
+	mlxsw_reg_qtct_switch_prio_set(payload, switch_prio);
+	mlxsw_reg_qtct_tclass_set(payload, tclass);
+}
+
 /* QEEC - QoS ETS Element Configuration Register
  * ---------------------------------------------
  * Configures the ETS elements.
@@ -3491,6 +3544,8 @@ static inline const char *mlxsw_reg_id_str(u16 reg_id)
 		return "SFMR";
 	case MLXSW_REG_SPVMLR_ID:
 		return "SPVMLR";
+	case MLXSW_REG_QTCT_ID:
+		return "QTCT";
 	case MLXSW_REG_QEEC_ID:
 		return "QEEC";
 	case MLXSW_REG_PMLP_ID:
-- 
2.5.5

[patch net-next 09/17] mlxsw: spectrum: Initialize egress scheduling

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

Before introducing support for DCB ops we should first make sure we
initialize the relevant parts in the device correctly. Specifically, the
egress scheduling.

The device supports a superset of the 802.1Qaz standard with 4 hierarchy
levels that can be linked to each other in multiple ways and with
different transmission selection algorithms (TSA) employed between them.

However, since we only intend to support the 802.1Qaz standard we
flatten the hierarchies and let the user configure via DCB ops the TSA
and max rate shaper at the subgroup hierarchy (see figure below) and the
mapping between switch priority to traffic class. By default, all switch
priorities are mapped to traffic class 0, strict priority is employed
and max shaper is disabled.

Default configuration:

         switch priority 0      ...         switch priority 7
                 +                                  +
                 |                                  |
                 +----------------------------------+
                 |
              +--v--+                          +-----+
Traffic Class |     |                          |     |
  Hierarchy   | TC0 |           ...            | TC7 |
              |     |                          |     |
              +--+--+                          +--+--+
                 |                                |
              +--v--+                          +--v--+
  Subgroup    | SG0 |                          | SG7 |
  Hierarchy   |     |                          |     |
              +-----+                          +-----+
              | TSA |                          | TSA |
              +-----+           ...            +-----+
              | MAX |                          | MAX |
              +--+--+                          +--+--+
                 |                                |
                 +---------------+----------------+
                                 |
                              +--v--+
                      Group   |     |
                    Hierarchy | GR0 |
                              |     |
                              +--+--+
                                 |
                              +--v--+
                      Port    |     |
                    Hierarchy | PR0 |
                              |     |
                              +-----+

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c | 111 +++++++++++++++++++++++++
 1 file changed, 111 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 4576d59..1243c74 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -49,6 +49,7 @@
 #include <linux/jiffies.h>
 #include <linux/bitops.h>
 #include <linux/list.h>
+#include <linux/dcbnl.h>
 #include <net/devlink.h>
 #include <net/switchdev.h>
 #include <generated/utsrelease.h>
@@ -1464,6 +1465,108 @@ mlxsw_sp_port_speed_by_width_set(struct mlxsw_sp_port *mlxsw_sp_port, u8 width)
 	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptys), ptys_pl);
 }
 
+static int mlxsw_sp_port_ets_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				 enum mlxsw_reg_qeec_hr hr, u8 index,
+				 u8 next_index, bool dwrr, u8 dwrr_weight)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qeec_pl[MLXSW_REG_QEEC_LEN];
+
+	mlxsw_reg_qeec_pack(qeec_pl, mlxsw_sp_port->local_port, hr, index,
+			    next_index);
+	mlxsw_reg_qeec_de_set(qeec_pl, true);
+	mlxsw_reg_qeec_dwrr_set(qeec_pl, dwrr);
+	mlxsw_reg_qeec_dwrr_weight_set(qeec_pl, dwrr_weight);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qeec), qeec_pl);
+}
+
+static int mlxsw_sp_port_ets_maxrate_set(struct mlxsw_sp_port *mlxsw_sp_port,
+					 enum mlxsw_reg_qeec_hr hr, u8 index,
+					 u8 next_index, u32 maxrate)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qeec_pl[MLXSW_REG_QEEC_LEN];
+
+	mlxsw_reg_qeec_pack(qeec_pl, mlxsw_sp_port->local_port, hr, index,
+			    next_index);
+	mlxsw_reg_qeec_mase_set(qeec_pl, true);
+	mlxsw_reg_qeec_max_shaper_rate_set(qeec_pl, maxrate);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qeec), qeec_pl);
+}
+
+static int mlxsw_sp_port_prio_tc_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				     u8 switch_prio, u8 tclass)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char qtct_pl[MLXSW_REG_QTCT_LEN];
+
+	mlxsw_reg_qtct_pack(qtct_pl, mlxsw_sp_port->local_port, switch_prio,
+			    tclass);
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qtct), qtct_pl);
+}
+
+static int mlxsw_sp_port_ets_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	int err, i;
+
+	/* Setup the elements hierarcy, so that each TC is linked to
+	 * one subgroup, which are all member in the same group.
+	 */
+	err = mlxsw_sp_port_ets_set(mlxsw_sp_port,
+				    MLXSW_REG_QEEC_HIERARCY_GROUP, 0, 0, false,
+				    0);
+	if (err)
+		return err;
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_ets_set(mlxsw_sp_port,
+					    MLXSW_REG_QEEC_HIERARCY_SUBGROUP, i,
+					    0, false, 0);
+		if (err)
+			return err;
+	}
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_ets_set(mlxsw_sp_port,
+					    MLXSW_REG_QEEC_HIERARCY_TC, i, i,
+					    false, 0);
+		if (err)
+			return err;
+	}
+
+	/* Make sure the max shaper is disabled in all hierarcies that
+	 * support it.
+	 */
+	err = mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port,
+					    MLXSW_REG_QEEC_HIERARCY_PORT, 0, 0,
+					    MLXSW_REG_QEEC_MAS_DIS);
+	if (err)
+		return err;
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port,
+						    MLXSW_REG_QEEC_HIERARCY_SUBGROUP,
+						    i, 0,
+						    MLXSW_REG_QEEC_MAS_DIS);
+		if (err)
+			return err;
+	}
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port,
+						    MLXSW_REG_QEEC_HIERARCY_TC,
+						    i, i,
+						    MLXSW_REG_QEEC_MAS_DIS);
+		if (err)
+			return err;
+	}
+
+	/* Map all priorities to traffic class 0. */
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_prio_tc_set(mlxsw_sp_port, i, 0);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
 static int __mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port,
 				  bool split, u8 module, u8 width)
 {
@@ -1571,6 +1674,13 @@ static int __mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port,
 		goto err_port_buffers_init;
 	}
 
+	err = mlxsw_sp_port_ets_init(mlxsw_sp_port);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to initialize ETS\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_ets_init;
+	}
+
 	mlxsw_sp_port_switchdev_init(mlxsw_sp_port);
 	err = register_netdev(dev);
 	if (err) {
@@ -1591,6 +1701,7 @@ static int __mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port,
 err_port_vlan_init:
 	unregister_netdev(dev);
 err_register_netdev:
+err_port_ets_init:
 err_port_buffers_init:
 err_port_admin_status_set:
 err_port_mtu_set:
-- 
2.5.5

[patch net-next 10/17] mlxsw: spectrum: Introduce support for Data Center Bridging (DCB)

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

Introduce basic infrastructure for DCB and add the missing ops in
following patches.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/Kconfig        |  8 +++
 drivers/net/ethernet/mellanox/mlxsw/Makefile       |  1 +
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     | 10 ++++
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     | 17 ++++++
 drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c | 65 ++++++++++++++++++++++
 5 files changed, 101 insertions(+)
 create mode 100644 drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c

diff --git a/drivers/net/ethernet/mellanox/mlxsw/Kconfig b/drivers/net/ethernet/mellanox/mlxsw/Kconfig
index 2ad7f67..5989f7c 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/Kconfig
+++ b/drivers/net/ethernet/mellanox/mlxsw/Kconfig
@@ -50,3 +50,11 @@ config MLXSW_SPECTRUM
 
 	  To compile this driver as a module, choose M here: the
 	  module will be called mlxsw_spectrum.
+
+config MLXSW_SPECTRUM_DCB
+	bool "Data Center Bridging (DCB) support"
+	depends on MLXSW_SPECTRUM && DCB
+	default y
+	---help---
+	  Say Y here if you want to use Data Center Bridging (DCB) in the
+	  driver.
diff --git a/drivers/net/ethernet/mellanox/mlxsw/Makefile b/drivers/net/ethernet/mellanox/mlxsw/Makefile
index 584cac4..9b5ebf8 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/Makefile
+++ b/drivers/net/ethernet/mellanox/mlxsw/Makefile
@@ -8,3 +8,4 @@ mlxsw_switchx2-objs		:= switchx2.o
 obj-$(CONFIG_MLXSW_SPECTRUM)	+= mlxsw_spectrum.o
 mlxsw_spectrum-objs		:= spectrum.o spectrum_buffers.o \
 				   spectrum_switchdev.o
+mlxsw_spectrum-$(CONFIG_MLXSW_SPECTRUM_DCB)	+= spectrum_dcb.o
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 1243c74..baaa9ea 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -1681,6 +1681,14 @@ static int __mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port,
 		goto err_port_ets_init;
 	}
 
+	/* ETS and buffers must be initialized before DCB. */
+	err = mlxsw_sp_port_dcb_init(mlxsw_sp_port);
+	if (err) {
+		dev_err(mlxsw_sp->bus_info->dev, "Port %d: Failed to initialize DCB\n",
+			mlxsw_sp_port->local_port);
+		goto err_port_dcb_init;
+	}
+
 	mlxsw_sp_port_switchdev_init(mlxsw_sp_port);
 	err = register_netdev(dev);
 	if (err) {
@@ -1701,6 +1709,7 @@ static int __mlxsw_sp_port_create(struct mlxsw_sp *mlxsw_sp, u8 local_port,
 err_port_vlan_init:
 	unregister_netdev(dev);
 err_register_netdev:
+err_port_dcb_init:
 err_port_ets_init:
 err_port_buffers_init:
 err_port_admin_status_set:
@@ -1771,6 +1780,7 @@ static void mlxsw_sp_port_remove(struct mlxsw_sp *mlxsw_sp, u8 local_port)
 	devlink_port = &mlxsw_sp_port->devlink_port;
 	devlink_port_type_clear(devlink_port);
 	unregister_netdev(mlxsw_sp_port->dev); /* This calls ndo_stop */
+	mlxsw_sp_port_dcb_fini(mlxsw_sp_port);
 	devlink_port_unregister(devlink_port);
 	mlxsw_sp_port_vports_fini(mlxsw_sp_port);
 	mlxsw_sp_port_switchdev_fini(mlxsw_sp_port);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index 84dddd4..1f50af8 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -270,4 +270,21 @@ int mlxsw_sp_vport_flood_set(struct mlxsw_sp_port *mlxsw_sp_vport, u16 vfid,
 void mlxsw_sp_port_active_vlans_del(struct mlxsw_sp_port *mlxsw_sp_port);
 int mlxsw_sp_port_pvid_set(struct mlxsw_sp_port *mlxsw_sp_port, u16 vid);
 
+#ifdef CONFIG_MLXSW_SPECTRUM_DCB
+
+int mlxsw_sp_port_dcb_init(struct mlxsw_sp_port *mlxsw_sp_port);
+void mlxsw_sp_port_dcb_fini(struct mlxsw_sp_port *mlxsw_sp_port);
+
+#else
+
+static inline int mlxsw_sp_port_dcb_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	return 0;
+}
+
+static inline void mlxsw_sp_port_dcb_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{}
+
+#endif
+
 #endif
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
new file mode 100644
index 0000000..631e980
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
@@ -0,0 +1,65 @@
+/*
+ * drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
+ * Copyright (c) 2016 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2016 Ido Schimmel <idosch@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/netdevice.h>
+#include <net/dcbnl.h>
+
+#include "spectrum.h"
+
+static u8 mlxsw_sp_dcbnl_getdcbx(struct net_device __always_unused *dev)
+{
+	return DCB_CAP_DCBX_HOST | DCB_CAP_DCBX_VER_IEEE;
+}
+
+static u8 mlxsw_sp_dcbnl_setdcbx(struct net_device __always_unused *dev,
+				 u8 mode)
+{
+	return (mode != (DCB_CAP_DCBX_HOST | DCB_CAP_DCBX_VER_IEEE)) ? 1 : 0;
+}
+
+static const struct dcbnl_rtnl_ops mlxsw_sp_dcbnl_ops = {
+	.getdcbx		= mlxsw_sp_dcbnl_getdcbx,
+	.setdcbx		= mlxsw_sp_dcbnl_setdcbx,
+};
+
+int mlxsw_sp_port_dcb_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	mlxsw_sp_port->dev->dcbnl_ops = &mlxsw_sp_dcbnl_ops;
+
+	return 0;
+}
+
+void mlxsw_sp_port_dcb_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+}
-- 
2.5.5

[patch net-next 11/17] mlxsw: spectrum: Add IEEE 802.1Qaz ETS support

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

Implement the appropriate DCB ops and allow a user to configure:
	* Priority to traffic class (TC) mapping with a total of 8
	  supported TCs
	* Transmission selection algorithm (TSA) for each TC and the
	  corresponding weights in case of weighted round robin (WRR)

As previously explained, we treat the priority group (PG) buffer in the
port's headroom as the ingress counterpart of the egress TC. Therefore,
when a certain priority to TC mapping is configured, we also configure
the port's headroom buffer.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     |  53 ++++-
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  11 +
 drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c | 229 +++++++++++++++++++++
 3 files changed, 283 insertions(+), 10 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index baaa9ea..1498e6a 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -450,22 +450,55 @@ static int mlxsw_sp_port_set_mac_address(struct net_device *dev, void *p)
 	return 0;
 }
 
-static int mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port,
-				      int mtu)
+static void mlxsw_sp_pg_buf_pack(char *pbmc_pl, int pg_index, int mtu)
 {
-	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
 	u16 pg_size = 2 * MLXSW_SP_BYTES_TO_CELLS(mtu);
+
+	mlxsw_reg_pbmc_lossy_buffer_pack(pbmc_pl, pg_index, pg_size);
+}
+
+int __mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port, int mtu,
+				 u8 *prio_tc)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
 	char pbmc_pl[MLXSW_REG_PBMC_LEN];
-	int err;
+	int i, j, err;
 
 	mlxsw_reg_pbmc_pack(pbmc_pl, mlxsw_sp_port->local_port, 0, 0);
 	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(pbmc), pbmc_pl);
 	if (err)
 		return err;
-	mlxsw_reg_pbmc_lossy_buffer_pack(pbmc_pl, 0, pg_size);
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		bool configure = false;
+
+		for (j = 0; j < IEEE_8021QAZ_MAX_TCS; j++) {
+			if (prio_tc[j] == i) {
+				configure = true;
+				break;
+			}
+		}
+
+		if (!configure)
+			continue;
+		mlxsw_sp_pg_buf_pack(pbmc_pl, i, mtu);
+	}
+
 	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pbmc), pbmc_pl);
 }
 
+static int mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				      int mtu)
+{
+	u8 def_prio_tc[IEEE_8021QAZ_MAX_TCS] = {0};
+	bool dcb_en = !!mlxsw_sp_port->dcb.ets;
+	u8 *prio_tc;
+
+	prio_tc = dcb_en ? mlxsw_sp_port->dcb.ets->prio_tc : def_prio_tc;
+
+	return __mlxsw_sp_port_headroom_set(mlxsw_sp_port, mtu, prio_tc);
+}
+
 static int mlxsw_sp_port_change_mtu(struct net_device *dev, int mtu)
 {
 	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
@@ -1465,9 +1498,9 @@ mlxsw_sp_port_speed_by_width_set(struct mlxsw_sp_port *mlxsw_sp_port, u8 width)
 	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(ptys), ptys_pl);
 }
 
-static int mlxsw_sp_port_ets_set(struct mlxsw_sp_port *mlxsw_sp_port,
-				 enum mlxsw_reg_qeec_hr hr, u8 index,
-				 u8 next_index, bool dwrr, u8 dwrr_weight)
+int mlxsw_sp_port_ets_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			  enum mlxsw_reg_qeec_hr hr, u8 index, u8 next_index,
+			  bool dwrr, u8 dwrr_weight)
 {
 	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
 	char qeec_pl[MLXSW_REG_QEEC_LEN];
@@ -1494,8 +1527,8 @@ static int mlxsw_sp_port_ets_maxrate_set(struct mlxsw_sp_port *mlxsw_sp_port,
 	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qeec), qeec_pl);
 }
 
-static int mlxsw_sp_port_prio_tc_set(struct mlxsw_sp_port *mlxsw_sp_port,
-				     u8 switch_prio, u8 tclass)
+int mlxsw_sp_port_prio_tc_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			      u8 switch_prio, u8 tclass)
 {
 	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
 	char qtct_pl[MLXSW_REG_QTCT_LEN];
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index 1f50af8..ef02081 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -42,6 +42,7 @@
 #include <linux/bitops.h>
 #include <linux/if_vlan.h>
 #include <linux/list.h>
+#include <linux/dcbnl.h>
 #include <net/switchdev.h>
 #include <net/devlink.h>
 
@@ -170,6 +171,9 @@ struct mlxsw_sp_port {
 		struct mlxsw_sp_vfid *vfid;
 		u16 vid;
 	} vport;
+	struct {
+		struct ieee_ets *ets;
+	} dcb;
 	/* 802.1Q bridge VLANs */
 	unsigned long *active_vlans;
 	unsigned long *untagged_vlans;
@@ -269,6 +273,13 @@ int mlxsw_sp_vport_flood_set(struct mlxsw_sp_port *mlxsw_sp_vport, u16 vfid,
 			     bool set, bool only_uc);
 void mlxsw_sp_port_active_vlans_del(struct mlxsw_sp_port *mlxsw_sp_port);
 int mlxsw_sp_port_pvid_set(struct mlxsw_sp_port *mlxsw_sp_port, u16 vid);
+int mlxsw_sp_port_ets_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			  enum mlxsw_reg_qeec_hr hr, u8 index, u8 next_index,
+			  bool dwrr, u8 dwrr_weight);
+int mlxsw_sp_port_prio_tc_set(struct mlxsw_sp_port *mlxsw_sp_port,
+			      u8 switch_prio, u8 tclass);
+int __mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port, int mtu,
+				 u8 *prio_tc);
 
 #ifdef CONFIG_MLXSW_SPECTRUM_DCB
 
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
index 631e980..aa5b73a 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
@@ -33,9 +33,11 @@
  */
 
 #include <linux/netdevice.h>
+#include <linux/string.h>
 #include <net/dcbnl.h>
 
 #include "spectrum.h"
+#include "reg.h"
 
 static u8 mlxsw_sp_dcbnl_getdcbx(struct net_device __always_unused *dev)
 {
@@ -48,13 +50,239 @@ static u8 mlxsw_sp_dcbnl_setdcbx(struct net_device __always_unused *dev,
 	return (mode != (DCB_CAP_DCBX_HOST | DCB_CAP_DCBX_VER_IEEE)) ? 1 : 0;
 }
 
+static int mlxsw_sp_dcbnl_ieee_getets(struct net_device *dev,
+				      struct ieee_ets *ets)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+
+	memcpy(ets, mlxsw_sp_port->dcb.ets, sizeof(*ets));
+
+	return 0;
+}
+
+static int mlxsw_sp_port_ets_validate(struct mlxsw_sp_port *mlxsw_sp_port,
+				      struct ieee_ets *ets)
+{
+	struct net_device *dev = mlxsw_sp_port->dev;
+	bool has_ets_tc = false;
+	int i, tx_bw_sum = 0;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		switch (ets->tc_tsa[i]) {
+		case IEEE_8021QAZ_TSA_STRICT:
+			break;
+		case IEEE_8021QAZ_TSA_ETS:
+			has_ets_tc = true;
+			tx_bw_sum += ets->tc_tx_bw[i];
+			break;
+		default:
+			netdev_err(dev, "Only strict priority and ETS are supported\n");
+			return -EINVAL;
+		}
+
+		if (ets->prio_tc[i] >= IEEE_8021QAZ_MAX_TCS) {
+			netdev_err(dev, "Invalid TC\n");
+			return -EINVAL;
+		}
+	}
+
+	if (has_ets_tc && tx_bw_sum != 100) {
+		netdev_err(dev, "Total ETS bandwidth should equal 100\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int mlxsw_sp_port_pg_prio_map(struct mlxsw_sp_port *mlxsw_sp_port,
+				     u8 *prio_tc)
+{
+	char pptb_pl[MLXSW_REG_PPTB_LEN];
+	int i;
+
+	mlxsw_reg_pptb_pack(pptb_pl, mlxsw_sp_port->local_port);
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
+		mlxsw_reg_pptb_prio_to_buff_set(pptb_pl, i, prio_tc[i]);
+	return mlxsw_reg_write(mlxsw_sp_port->mlxsw_sp->core, MLXSW_REG(pptb),
+			       pptb_pl);
+}
+
+static bool mlxsw_sp_ets_has_pg(u8 *prio_tc, u8 pg)
+{
+	int i;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
+		if (prio_tc[i] == pg)
+			return true;
+	return false;
+}
+
+static int mlxsw_sp_port_pg_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
+				    u8 *old_prio_tc, u8 *new_prio_tc)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	char pbmc_pl[MLXSW_REG_PBMC_LEN];
+	int err, i;
+
+	mlxsw_reg_pbmc_pack(pbmc_pl, mlxsw_sp_port->local_port, 0, 0);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(pbmc), pbmc_pl);
+	if (err)
+		return err;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		u8 pg = old_prio_tc[i];
+
+		if (!mlxsw_sp_ets_has_pg(new_prio_tc, pg))
+			mlxsw_reg_pbmc_lossy_buffer_pack(pbmc_pl, pg, 0);
+	}
+
+	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pbmc), pbmc_pl);
+}
+
+static int mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				      struct ieee_ets *ets)
+{
+	struct ieee_ets *my_ets = mlxsw_sp_port->dcb.ets;
+	struct net_device *dev = mlxsw_sp_port->dev;
+	int err;
+
+	/* Create the required PGs, but don't destroy existing ones, as
+	 * traffic is still directed to them.
+	 */
+	err = __mlxsw_sp_port_headroom_set(mlxsw_sp_port, dev->mtu,
+					   ets->prio_tc);
+	if (err) {
+		netdev_err(dev, "Failed to configure port's headroom\n");
+		return err;
+	}
+
+	err = mlxsw_sp_port_pg_prio_map(mlxsw_sp_port, ets->prio_tc);
+	if (err) {
+		netdev_err(dev, "Failed to set PG-priority mapping\n");
+		goto err_port_prio_pg_map;
+	}
+
+	err = mlxsw_sp_port_pg_destroy(mlxsw_sp_port, my_ets->prio_tc,
+				       ets->prio_tc);
+	if (err)
+		netdev_warn(dev, "Failed to remove ununsed PGs\n");
+
+	return 0;
+
+err_port_prio_pg_map:
+	mlxsw_sp_port_pg_destroy(mlxsw_sp_port, ets->prio_tc, my_ets->prio_tc);
+	return err;
+}
+
+static int __mlxsw_sp_dcbnl_ieee_setets(struct mlxsw_sp_port *mlxsw_sp_port,
+					struct ieee_ets *ets)
+{
+	struct ieee_ets *my_ets = mlxsw_sp_port->dcb.ets;
+	struct net_device *dev = mlxsw_sp_port->dev;
+	int i, err;
+
+	/* Egress configuration. */
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		bool dwrr = ets->tc_tsa[i] == IEEE_8021QAZ_TSA_ETS;
+		u8 weight = ets->tc_tx_bw[i];
+
+		err = mlxsw_sp_port_ets_set(mlxsw_sp_port,
+					    MLXSW_REG_QEEC_HIERARCY_SUBGROUP, i,
+					    0, dwrr, weight);
+		if (err) {
+			netdev_err(dev, "Failed to link subgroup ETS element %d to group\n",
+				   i);
+			goto err_port_ets_set;
+		}
+	}
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_prio_tc_set(mlxsw_sp_port, i,
+						ets->prio_tc[i]);
+		if (err) {
+			netdev_err(dev, "Failed to map prio %d to TC %d\n", i,
+				   ets->prio_tc[i]);
+			goto err_port_prio_tc_set;
+		}
+	}
+
+	/* Ingress configuration. */
+	err = mlxsw_sp_port_headroom_set(mlxsw_sp_port, ets);
+	if (err)
+		goto err_port_headroom_set;
+
+	return 0;
+
+err_port_headroom_set:
+	i = IEEE_8021QAZ_MAX_TCS;
+err_port_prio_tc_set:
+	for (i--; i >= 0; i--)
+		mlxsw_sp_port_prio_tc_set(mlxsw_sp_port, i, my_ets->prio_tc[i]);
+	i = IEEE_8021QAZ_MAX_TCS;
+err_port_ets_set:
+	for (i--; i >= 0; i--) {
+		bool dwrr = my_ets->tc_tsa[i] == IEEE_8021QAZ_TSA_ETS;
+		u8 weight = my_ets->tc_tx_bw[i];
+
+		err = mlxsw_sp_port_ets_set(mlxsw_sp_port,
+					    MLXSW_REG_QEEC_HIERARCY_SUBGROUP, i,
+					    0, dwrr, weight);
+	}
+	return err;
+}
+
+static int mlxsw_sp_dcbnl_ieee_setets(struct net_device *dev,
+				      struct ieee_ets *ets)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	int err;
+
+	err = mlxsw_sp_port_ets_validate(mlxsw_sp_port, ets);
+	if (err)
+		return err;
+
+	err = __mlxsw_sp_dcbnl_ieee_setets(mlxsw_sp_port, ets);
+	if (err)
+		return err;
+
+	memcpy(mlxsw_sp_port->dcb.ets, ets, sizeof(*ets));
+
+	return 0;
+}
+
 static const struct dcbnl_rtnl_ops mlxsw_sp_dcbnl_ops = {
+	.ieee_getets		= mlxsw_sp_dcbnl_ieee_getets,
+	.ieee_setets		= mlxsw_sp_dcbnl_ieee_setets,
+
 	.getdcbx		= mlxsw_sp_dcbnl_getdcbx,
 	.setdcbx		= mlxsw_sp_dcbnl_setdcbx,
 };
 
+static int mlxsw_sp_port_ets_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	mlxsw_sp_port->dcb.ets = kzalloc(sizeof(*mlxsw_sp_port->dcb.ets),
+					 GFP_KERNEL);
+	if (!mlxsw_sp_port->dcb.ets)
+		return -ENOMEM;
+
+	mlxsw_sp_port->dcb.ets->ets_cap = IEEE_8021QAZ_MAX_TCS;
+
+	return 0;
+}
+
+static void mlxsw_sp_port_ets_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	kfree(mlxsw_sp_port->dcb.ets);
+}
+
 int mlxsw_sp_port_dcb_init(struct mlxsw_sp_port *mlxsw_sp_port)
 {
+	int err;
+
+	err = mlxsw_sp_port_ets_init(mlxsw_sp_port);
+	if (err)
+		return err;
+
 	mlxsw_sp_port->dev->dcbnl_ops = &mlxsw_sp_dcbnl_ops;
 
 	return 0;
@@ -62,4 +290,5 @@ int mlxsw_sp_port_dcb_init(struct mlxsw_sp_port *mlxsw_sp_port)
 
 void mlxsw_sp_port_dcb_fini(struct mlxsw_sp_port *mlxsw_sp_port)
 {
+	mlxsw_sp_port_ets_fini(mlxsw_sp_port);
 }
-- 
2.5.5

[patch net-next 12/17] mlxsw: spectrum: Allow setting maximum rate for a TC

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

Allow a user to set maximum rate for a particular TC using DCB ops.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     |  6 +-
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  4 ++
 drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c | 70 ++++++++++++++++++++++
 3 files changed, 77 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 1498e6a..5f4d44e 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -1513,9 +1513,9 @@ int mlxsw_sp_port_ets_set(struct mlxsw_sp_port *mlxsw_sp_port,
 	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(qeec), qeec_pl);
 }
 
-static int mlxsw_sp_port_ets_maxrate_set(struct mlxsw_sp_port *mlxsw_sp_port,
-					 enum mlxsw_reg_qeec_hr hr, u8 index,
-					 u8 next_index, u32 maxrate)
+int mlxsw_sp_port_ets_maxrate_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				  enum mlxsw_reg_qeec_hr hr, u8 index,
+				  u8 next_index, u32 maxrate)
 {
 	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
 	char qeec_pl[MLXSW_REG_QEEC_LEN];
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index ef02081..9e1e4fe 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -173,6 +173,7 @@ struct mlxsw_sp_port {
 	} vport;
 	struct {
 		struct ieee_ets *ets;
+		struct ieee_maxrate *maxrate;
 	} dcb;
 	/* 802.1Q bridge VLANs */
 	unsigned long *active_vlans;
@@ -280,6 +281,9 @@ int mlxsw_sp_port_prio_tc_set(struct mlxsw_sp_port *mlxsw_sp_port,
 			      u8 switch_prio, u8 tclass);
 int __mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port, int mtu,
 				 u8 *prio_tc);
+int mlxsw_sp_port_ets_maxrate_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				  enum mlxsw_reg_qeec_hr hr, u8 index,
+				  u8 next_index, u32 maxrate);
 
 #ifdef CONFIG_MLXSW_SPECTRUM_DCB
 
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
index aa5b73a..257e2d4 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
@@ -250,9 +250,51 @@ static int mlxsw_sp_dcbnl_ieee_setets(struct net_device *dev,
 	return 0;
 }
 
+static int mlxsw_sp_dcbnl_ieee_getmaxrate(struct net_device *dev,
+					  struct ieee_maxrate *maxrate)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+
+	memcpy(maxrate, mlxsw_sp_port->dcb.maxrate, sizeof(*maxrate));
+
+	return 0;
+}
+
+static int mlxsw_sp_dcbnl_ieee_setmaxrate(struct net_device *dev,
+					  struct ieee_maxrate *maxrate)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	struct ieee_maxrate *my_maxrate = mlxsw_sp_port->dcb.maxrate;
+	int err, i;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port,
+						    MLXSW_REG_QEEC_HIERARCY_SUBGROUP,
+						    i, 0,
+						    maxrate->tc_maxrate[i]);
+		if (err) {
+			netdev_err(dev, "Failed to set maxrate for TC %d\n", i);
+			goto err_port_ets_maxrate_set;
+		}
+	}
+
+	memcpy(mlxsw_sp_port->dcb.maxrate, maxrate, sizeof(*maxrate));
+
+	return 0;
+
+err_port_ets_maxrate_set:
+	for (i--; i >= 0; i--)
+		mlxsw_sp_port_ets_maxrate_set(mlxsw_sp_port,
+					      MLXSW_REG_QEEC_HIERARCY_SUBGROUP,
+					      i, 0, my_maxrate->tc_maxrate[i]);
+	return err;
+}
+
 static const struct dcbnl_rtnl_ops mlxsw_sp_dcbnl_ops = {
 	.ieee_getets		= mlxsw_sp_dcbnl_ieee_getets,
 	.ieee_setets		= mlxsw_sp_dcbnl_ieee_setets,
+	.ieee_getmaxrate	= mlxsw_sp_dcbnl_ieee_getmaxrate,
+	.ieee_setmaxrate	= mlxsw_sp_dcbnl_ieee_setmaxrate,
 
 	.getdcbx		= mlxsw_sp_dcbnl_getdcbx,
 	.setdcbx		= mlxsw_sp_dcbnl_setdcbx,
@@ -275,6 +317,26 @@ static void mlxsw_sp_port_ets_fini(struct mlxsw_sp_port *mlxsw_sp_port)
 	kfree(mlxsw_sp_port->dcb.ets);
 }
 
+static int mlxsw_sp_port_maxrate_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	int i;
+
+	mlxsw_sp_port->dcb.maxrate = kmalloc(sizeof(*mlxsw_sp_port->dcb.maxrate),
+					     GFP_KERNEL);
+	if (!mlxsw_sp_port->dcb.maxrate)
+		return -ENOMEM;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++)
+		mlxsw_sp_port->dcb.maxrate->tc_maxrate[i] = MLXSW_REG_QEEC_MAS_DIS;
+
+	return 0;
+}
+
+static void mlxsw_sp_port_maxrate_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	kfree(mlxsw_sp_port->dcb.maxrate);
+}
+
 int mlxsw_sp_port_dcb_init(struct mlxsw_sp_port *mlxsw_sp_port)
 {
 	int err;
@@ -282,13 +344,21 @@ int mlxsw_sp_port_dcb_init(struct mlxsw_sp_port *mlxsw_sp_port)
 	err = mlxsw_sp_port_ets_init(mlxsw_sp_port);
 	if (err)
 		return err;
+	err = mlxsw_sp_port_maxrate_init(mlxsw_sp_port);
+	if (err)
+		goto err_port_maxrate_init;
 
 	mlxsw_sp_port->dev->dcbnl_ops = &mlxsw_sp_dcbnl_ops;
 
 	return 0;
+
+err_port_maxrate_init:
+	mlxsw_sp_port_ets_fini(mlxsw_sp_port);
+	return err;
 }
 
 void mlxsw_sp_port_dcb_fini(struct mlxsw_sp_port *mlxsw_sp_port)
 {
+	mlxsw_sp_port_maxrate_fini(mlxsw_sp_port);
 	mlxsw_sp_port_ets_fini(mlxsw_sp_port);
 }
-- 
2.5.5

[patch net-next 13/17] mlxsw: reg: Add Port Flow Control Configuration register

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

Add the Port Flow Control Configuration (PFCC) register, which
configures both flow control and Priority-based Flow Control (PFC).

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/reg.h | 131 ++++++++++++++++++++++++++++++
 1 file changed, 131 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index 2e58c41..b83514a 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -2319,6 +2319,135 @@ static inline void mlxsw_reg_paos_pack(char *payload, u8 local_port,
 	mlxsw_reg_paos_e_set(payload, 1);
 }
 
+/* PFCC - Ports Flow Control Configuration Register
+ * ------------------------------------------------
+ * Configures and retrieves the per port flow control configuration.
+ */
+#define MLXSW_REG_PFCC_ID 0x5007
+#define MLXSW_REG_PFCC_LEN 0x20
+
+static const struct mlxsw_reg_info mlxsw_reg_pfcc = {
+	.id = MLXSW_REG_PFCC_ID,
+	.len = MLXSW_REG_PFCC_LEN,
+};
+
+/* reg_pfcc_local_port
+ * Local port number.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pfcc, local_port, 0x00, 16, 8);
+
+/* reg_pfcc_pnat
+ * Port number access type. Determines the way local_port is interpreted:
+ * 0 - Local port number.
+ * 1 - IB / label port number.
+ * Access: Index
+ */
+MLXSW_ITEM32(reg, pfcc, pnat, 0x00, 14, 2);
+
+/* reg_pfcc_shl_cap
+ * Send to higher layers capabilities:
+ * 0 - No capability of sending Pause and PFC frames to higher layers.
+ * 1 - Device has capability of sending Pause and PFC frames to higher
+ *     layers.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pfcc, shl_cap, 0x00, 1, 1);
+
+/* reg_pfcc_shl_opr
+ * Send to higher layers operation:
+ * 0 - Pause and PFC frames are handled by the port (default).
+ * 1 - Pause and PFC frames are handled by the port and also sent to
+ *     higher layers. Only valid if shl_cap = 1.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pfcc, shl_opr, 0x00, 0, 1);
+
+/* reg_pfcc_ppan
+ * Pause policy auto negotiation.
+ * 0 - Disabled. Generate / ignore Pause frames based on pptx / pprtx.
+ * 1 - Enabled. When auto-negotiation is performed, set the Pause policy
+ *     based on the auto-negotiation resolution.
+ * Access: RW
+ *
+ * Note: The auto-negotiation advertisement is set according to pptx and
+ * pprtx. When PFC is set on Tx / Rx, ppan must be set to 0.
+ */
+MLXSW_ITEM32(reg, pfcc, ppan, 0x04, 28, 4);
+
+/* reg_pfcc_prio_mask_tx
+ * Bit per priority indicating if Tx flow control policy should be
+ * updated based on bit pfctx.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, pfcc, prio_mask_tx, 0x04, 16, 8);
+
+/* reg_pfcc_prio_mask_rx
+ * Bit per priority indicating if Rx flow control policy should be
+ * updated based on bit pfcrx.
+ * Access: WO
+ */
+MLXSW_ITEM32(reg, pfcc, prio_mask_rx, 0x04, 0, 8);
+
+/* reg_pfcc_pptx
+ * Admin Pause policy on Tx.
+ * 0 - Never generate Pause frames (default).
+ * 1 - Generate Pause frames according to Rx buffer threshold.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pfcc, pptx, 0x08, 31, 1);
+
+/* reg_pfcc_aptx
+ * Active (operational) Pause policy on Tx.
+ * 0 - Never generate Pause frames.
+ * 1 - Generate Pause frames according to Rx buffer threshold.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pfcc, aptx, 0x08, 30, 1);
+
+/* reg_pfcc_pfctx
+ * Priority based flow control policy on Tx[7:0]. Per-priority bit mask:
+ * 0 - Never generate priority Pause frames on the specified priority
+ *     (default).
+ * 1 - Generate priority Pause frames according to Rx buffer threshold on
+ *     the specified priority.
+ * Access: RW
+ *
+ * Note: pfctx and pptx must be mutually exclusive.
+ */
+MLXSW_ITEM32(reg, pfcc, pfctx, 0x08, 16, 8);
+
+/* reg_pfcc_pprx
+ * Admin Pause policy on Rx.
+ * 0 - Ignore received Pause frames (default).
+ * 1 - Respect received Pause frames.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pfcc, pprx, 0x0C, 31, 1);
+
+/* reg_pfcc_aprx
+ * Active (operational) Pause policy on Rx.
+ * 0 - Ignore received Pause frames.
+ * 1 - Respect received Pause frames.
+ * Access: RO
+ */
+MLXSW_ITEM32(reg, pfcc, aprx, 0x0C, 30, 1);
+
+/* reg_pfcc_pfcrx
+ * Priority based flow control policy on Rx[7:0]. Per-priority bit mask:
+ * 0 - Ignore incoming priority Pause frames on the specified priority
+ *     (default).
+ * 1 - Respect incoming priority Pause frames on the specified priority.
+ * Access: RW
+ */
+MLXSW_ITEM32(reg, pfcc, pfcrx, 0x0C, 16, 8);
+
+static inline void mlxsw_reg_pfcc_pack(char *payload, u8 local_port)
+{
+	MLXSW_REG_ZERO(pfcc, payload);
+	mlxsw_reg_pfcc_local_port_set(payload, local_port);
+}
+
 /* PPCNT - Ports Performance Counters Register
  * -------------------------------------------
  * The PPCNT register retrieves per port performance counters.
@@ -3558,6 +3687,8 @@ static inline const char *mlxsw_reg_id_str(u16 reg_id)
 		return "PPAD";
 	case MLXSW_REG_PAOS_ID:
 		return "PAOS";
+	case MLXSW_REG_PFCC_ID:
+		return "PFCC";
 	case MLXSW_REG_PPCNT_ID:
 		return "PPCNT";
 	case MLXSW_REG_PPTB_ID:
-- 
2.5.5

[patch net-next 14/17] mlxsw: reg: Add lossless settings for PBMC register

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

When configuring PAUSE frames and PFC we'll need to configure the
Xon/Xoff threshold for the priority group (PG) buffers.

Add the Xon/Xoff threshold fields to the PBMC register so that we can
configure these when needed.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/reg.h | 35 +++++++++++++++++++++++++++++++
 1 file changed, 35 insertions(+)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index b83514a..bcd38ff 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -2788,6 +2788,30 @@ MLXSW_ITEM32_INDEXED(reg, pbmc, buf_epsb, 0x0C, 24, 1, 0x08, 0x00, false);
  */
 MLXSW_ITEM32_INDEXED(reg, pbmc, buf_size, 0x0C, 0, 16, 0x08, 0x00, false);
 
+/* reg_pbmc_buf_xoff_threshold
+ * Once the amount of data in the buffer goes above this value, device
+ * starts sending PFC frames for all priorities associated with the
+ * buffer. Units are represented in cells. Reserved in case of lossy
+ * buffer.
+ * Access: RW
+ *
+ * Note: In Spectrum, reserved for buffer[9].
+ */
+MLXSW_ITEM32_INDEXED(reg, pbmc, buf_xoff_threshold, 0x0C, 16, 16,
+		     0x08, 0x04, false);
+
+/* reg_pbmc_buf_xon_threshold
+ * When the amount of data in the buffer goes below this value, device
+ * stops sending PFC frames for the priorities associated with the
+ * buffer. Units are represented in cells. Reserved in case of lossy
+ * buffer.
+ * Access: RW
+ *
+ * Note: In Spectrum, reserved for buffer[9].
+ */
+MLXSW_ITEM32_INDEXED(reg, pbmc, buf_xon_threshold, 0x0C, 0, 16,
+		     0x08, 0x04, false);
+
 static inline void mlxsw_reg_pbmc_pack(char *payload, u8 local_port,
 				       u16 xoff_timer_value, u16 xoff_refresh)
 {
@@ -2806,6 +2830,17 @@ static inline void mlxsw_reg_pbmc_lossy_buffer_pack(char *payload,
 	mlxsw_reg_pbmc_buf_size_set(payload, buf_index, size);
 }
 
+static inline void mlxsw_reg_pbmc_lossless_buffer_pack(char *payload,
+						       int buf_index, u16 size,
+						       u16 threshold)
+{
+	mlxsw_reg_pbmc_buf_lossy_set(payload, buf_index, 0);
+	mlxsw_reg_pbmc_buf_epsb_set(payload, buf_index, 0);
+	mlxsw_reg_pbmc_buf_size_set(payload, buf_index, size);
+	mlxsw_reg_pbmc_buf_xoff_threshold_set(payload, buf_index, threshold);
+	mlxsw_reg_pbmc_buf_xon_threshold_set(payload, buf_index, threshold);
+}
+
 /* PSPA - Port Switch Partition Allocation
  * ---------------------------------------
  * Controls the association of a port with a switch partition and enables
-- 
2.5.5

[patch net-next 15/17] mlxsw: spectrum: Add support for PAUSE frames

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

When a packet ingress the switch it's placed in its assigned priority
group (PG) buffer in the port's headroom buffer while it goes through
the switch's pipeline. After going through the pipeline - which
determines its egress port(s) and traffic class - it's moved to the
switch's shared buffer awaiting transmission.

However, some packets are not eligible to enter the shared buffer due to
exceeded quotas or insufficient space. Marking their associated PGs as
lossless will cause the packets to accumulate in the PG buffer. Another
reason for packets accumulation are complicated pipelines (e.g.
involving a lot of ACLs).

To prevent packets from being dropped a user can enable PAUSE frames on
the port. This will mark all the active PGs as lossless and set their
size according to the maximum delay, as it's not configured by user.

                         +----------------+   +
                         |                |   |
                         |                |   |
                         |                |   |
                         |                |   |
                         |                |   |
                         |                |   | Delay
                         |                |   |
                         |                |   |
                         |                |   |
                         |                |   |
                         |                |   |
    Xon/Xoff threshold   +----------------+   +
                         |                |   |
                         |                |   | 2 * MTU
                         |                |   |
                         +----------------+   +

The delay (612 [Cells]) was calculated according to worst-case scenario
involving maximum MTU and 100m cables.

After marking the PGs as lossless the device is configured to respect
incoming PAUSE frames (Rx PAUSE) and generate PAUSE frames (Tx PAUSE)
according to user's settings.

Whenever the port's headroom configuration changes we take into account
the PAUSE configuration, so that we correctly set the PG's type (lossy /
lossless), size and threshold. This can happen when:

a) The port's MTU changes, as it directly affects the PG's size.

b) A PG is created following user configuration, by binding a priority
to it.

Note that the relevant SUPPORTED flags were already mistakenly set by
the driver before this commit.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     | 85 ++++++++++++++++++++--
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     | 17 ++++-
 drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c |  3 +-
 3 files changed, 95 insertions(+), 10 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 5f4d44e..086682e 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -450,15 +450,23 @@ static int mlxsw_sp_port_set_mac_address(struct net_device *dev, void *p)
 	return 0;
 }
 
-static void mlxsw_sp_pg_buf_pack(char *pbmc_pl, int pg_index, int mtu)
+static void mlxsw_sp_pg_buf_pack(char *pbmc_pl, int pg_index, int mtu,
+				 bool pause_en)
 {
 	u16 pg_size = 2 * MLXSW_SP_BYTES_TO_CELLS(mtu);
 
-	mlxsw_reg_pbmc_lossy_buffer_pack(pbmc_pl, pg_index, pg_size);
+	if (pause_en) {
+		u16 pg_pause_size = pg_size + MLXSW_SP_PAUSE_DELAY;
+
+		mlxsw_reg_pbmc_lossless_buffer_pack(pbmc_pl, pg_index,
+						    pg_pause_size, pg_size);
+	} else {
+		mlxsw_reg_pbmc_lossy_buffer_pack(pbmc_pl, pg_index, pg_size);
+	}
 }
 
 int __mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port, int mtu,
-				 u8 *prio_tc)
+				 u8 *prio_tc, bool pause_en)
 {
 	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
 	char pbmc_pl[MLXSW_REG_PBMC_LEN];
@@ -481,14 +489,14 @@ int __mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port, int mtu,
 
 		if (!configure)
 			continue;
-		mlxsw_sp_pg_buf_pack(pbmc_pl, i, mtu);
+		mlxsw_sp_pg_buf_pack(pbmc_pl, i, mtu, pause_en);
 	}
 
 	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pbmc), pbmc_pl);
 }
 
 static int mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port,
-				      int mtu)
+				      int mtu, bool pause_en)
 {
 	u8 def_prio_tc[IEEE_8021QAZ_MAX_TCS] = {0};
 	bool dcb_en = !!mlxsw_sp_port->dcb.ets;
@@ -496,15 +504,17 @@ static int mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port,
 
 	prio_tc = dcb_en ? mlxsw_sp_port->dcb.ets->prio_tc : def_prio_tc;
 
-	return __mlxsw_sp_port_headroom_set(mlxsw_sp_port, mtu, prio_tc);
+	return __mlxsw_sp_port_headroom_set(mlxsw_sp_port, mtu, prio_tc,
+					    pause_en);
 }
 
 static int mlxsw_sp_port_change_mtu(struct net_device *dev, int mtu)
 {
 	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	bool pause_en = mlxsw_sp_port_is_pause_en(mlxsw_sp_port);
 	int err;
 
-	err = mlxsw_sp_port_headroom_set(mlxsw_sp_port, mtu);
+	err = mlxsw_sp_port_headroom_set(mlxsw_sp_port, mtu, pause_en);
 	if (err)
 		return err;
 	err = mlxsw_sp_port_mtu_set(mlxsw_sp_port, mtu);
@@ -514,7 +524,7 @@ static int mlxsw_sp_port_change_mtu(struct net_device *dev, int mtu)
 	return 0;
 
 err_port_mtu_set:
-	mlxsw_sp_port_headroom_set(mlxsw_sp_port, dev->mtu);
+	mlxsw_sp_port_headroom_set(mlxsw_sp_port, dev->mtu, pause_en);
 	return err;
 }
 
@@ -993,6 +1003,63 @@ static void mlxsw_sp_port_get_drvinfo(struct net_device *dev,
 		sizeof(drvinfo->bus_info));
 }
 
+static void mlxsw_sp_port_get_pauseparam(struct net_device *dev,
+					 struct ethtool_pauseparam *pause)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+
+	pause->rx_pause = mlxsw_sp_port->link.rx_pause;
+	pause->tx_pause = mlxsw_sp_port->link.tx_pause;
+}
+
+static int mlxsw_sp_port_pause_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				   struct ethtool_pauseparam *pause)
+{
+	char pfcc_pl[MLXSW_REG_PFCC_LEN];
+
+	mlxsw_reg_pfcc_pack(pfcc_pl, mlxsw_sp_port->local_port);
+	mlxsw_reg_pfcc_pprx_set(pfcc_pl, pause->rx_pause);
+	mlxsw_reg_pfcc_pptx_set(pfcc_pl, pause->tx_pause);
+
+	return mlxsw_reg_write(mlxsw_sp_port->mlxsw_sp->core, MLXSW_REG(pfcc),
+			       pfcc_pl);
+}
+
+static int mlxsw_sp_port_set_pauseparam(struct net_device *dev,
+					struct ethtool_pauseparam *pause)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	bool pause_en = pause->tx_pause || pause->rx_pause;
+	int err;
+
+	if (pause->autoneg) {
+		netdev_err(dev, "PAUSE frames autonegotiation isn't supported\n");
+		return -EINVAL;
+	}
+
+	err = mlxsw_sp_port_headroom_set(mlxsw_sp_port, dev->mtu, pause_en);
+	if (err) {
+		netdev_err(dev, "Failed to configure port's headroom\n");
+		return err;
+	}
+
+	err = mlxsw_sp_port_pause_set(mlxsw_sp_port, pause);
+	if (err) {
+		netdev_err(dev, "Failed to set PAUSE parameters\n");
+		goto err_port_pause_configure;
+	}
+
+	mlxsw_sp_port->link.rx_pause = pause->rx_pause;
+	mlxsw_sp_port->link.tx_pause = pause->tx_pause;
+
+	return 0;
+
+err_port_pause_configure:
+	pause_en = mlxsw_sp_port_is_pause_en(mlxsw_sp_port);
+	mlxsw_sp_port_headroom_set(mlxsw_sp_port, dev->mtu, pause_en);
+	return err;
+}
+
 struct mlxsw_sp_port_hw_stats {
 	char str[ETH_GSTRING_LEN];
 	u64 (*getter)(char *payload);
@@ -1476,6 +1543,8 @@ static int mlxsw_sp_port_set_settings(struct net_device *dev,
 static const struct ethtool_ops mlxsw_sp_port_ethtool_ops = {
 	.get_drvinfo		= mlxsw_sp_port_get_drvinfo,
 	.get_link		= ethtool_op_get_link,
+	.get_pauseparam		= mlxsw_sp_port_get_pauseparam,
+	.set_pauseparam		= mlxsw_sp_port_set_pauseparam,
 	.get_strings		= mlxsw_sp_port_get_strings,
 	.set_phys_id		= mlxsw_sp_port_set_phys_id,
 	.get_ethtool_stats	= mlxsw_sp_port_get_stats,
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index 9e1e4fe..f4b53dd 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -67,6 +67,11 @@
 
 #define MLXSW_SP_BYTES_TO_CELLS(b) DIV_ROUND_UP(b, MLXSW_SP_BYTES_PER_CELL)
 
+/* Maximum delay buffer needed in case of PAUSE frames, in cells.
+ * Assumes 100m cable and maximum MTU.
+ */
+#define MLXSW_SP_PAUSE_DELAY 612
+
 struct mlxsw_sp_port;
 
 struct mlxsw_sp_upper {
@@ -172,6 +177,10 @@ struct mlxsw_sp_port {
 		u16 vid;
 	} vport;
 	struct {
+		u8 tx_pause:1,
+		   rx_pause:1;
+	} link;
+	struct {
 		struct ieee_ets *ets;
 		struct ieee_maxrate *maxrate;
 	} dcb;
@@ -183,6 +192,12 @@ struct mlxsw_sp_port {
 	struct devlink_port devlink_port;
 };
 
+static inline bool
+mlxsw_sp_port_is_pause_en(const struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	return mlxsw_sp_port->link.tx_pause || mlxsw_sp_port->link.rx_pause;
+}
+
 static inline struct mlxsw_sp_port *
 mlxsw_sp_port_lagged_get(struct mlxsw_sp *mlxsw_sp, u16 lag_id, u8 port_index)
 {
@@ -280,7 +295,7 @@ int mlxsw_sp_port_ets_set(struct mlxsw_sp_port *mlxsw_sp_port,
 int mlxsw_sp_port_prio_tc_set(struct mlxsw_sp_port *mlxsw_sp_port,
 			      u8 switch_prio, u8 tclass);
 int __mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port, int mtu,
-				 u8 *prio_tc);
+				 u8 *prio_tc, bool pause_en);
 int mlxsw_sp_port_ets_maxrate_set(struct mlxsw_sp_port *mlxsw_sp_port,
 				  enum mlxsw_reg_qeec_hr hr, u8 index,
 				  u8 next_index, u32 maxrate);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
index 257e2d4..8786424 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
@@ -142,6 +142,7 @@ static int mlxsw_sp_port_pg_destroy(struct mlxsw_sp_port *mlxsw_sp_port,
 static int mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port,
 				      struct ieee_ets *ets)
 {
+	bool pause_en = mlxsw_sp_port_is_pause_en(mlxsw_sp_port);
 	struct ieee_ets *my_ets = mlxsw_sp_port->dcb.ets;
 	struct net_device *dev = mlxsw_sp_port->dev;
 	int err;
@@ -150,7 +151,7 @@ static int mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port,
 	 * traffic is still directed to them.
 	 */
 	err = __mlxsw_sp_port_headroom_set(mlxsw_sp_port, dev->mtu,
-					   ets->prio_tc);
+					   ets->prio_tc, pause_en);
 	if (err) {
 		netdev_err(dev, "Failed to configure port's headroom\n");
 		return err;
-- 
2.5.5

[patch net-next 16/17] mlxsw: reg: Introduce per priority counters

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

We are going to add support for PFC as part of DCB ops, which requires us
to report the number of PFC frames sent and received per priority.

Add per priority counters in order to report number of PFC frames sent
and received per priority.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/reg.h      | 62 ++++++++++++++++++++++++--
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c |  3 +-
 drivers/net/ethernet/mellanox/mlxsw/switchx2.c |  3 +-
 3 files changed, 63 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index bcd38ff..84aacb3 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -2487,6 +2487,11 @@ MLXSW_ITEM32(reg, ppcnt, local_port, 0x00, 16, 8);
  */
 MLXSW_ITEM32(reg, ppcnt, pnat, 0x00, 14, 2);
 
+enum mlxsw_reg_ppcnt_grp {
+	MLXSW_REG_PPCNT_IEEE_8023_CNT = 0x0,
+	MLXSW_REG_PPCNT_PRIO_CNT = 0x10,
+};
+
 /* reg_ppcnt_grp
  * Performance counter group.
  * Group 63 indicates all groups. Only valid on Set() operation with
@@ -2522,6 +2527,8 @@ MLXSW_ITEM32(reg, ppcnt, clr, 0x04, 31, 1);
  */
 MLXSW_ITEM32(reg, ppcnt, prio_tc, 0x04, 0, 5);
 
+/* Ethernet IEEE 802.3 Counter Group */
+
 /* reg_ppcnt_a_frames_transmitted_ok
  * Access: RO
  */
@@ -2636,15 +2643,64 @@ MLXSW_ITEM64(reg, ppcnt, a_pause_mac_ctrl_frames_received,
 MLXSW_ITEM64(reg, ppcnt, a_pause_mac_ctrl_frames_transmitted,
 	     0x08 + 0x90, 0, 64);
 
-static inline void mlxsw_reg_ppcnt_pack(char *payload, u8 local_port)
+/* Ethernet Per Priority Group Counters */
+
+/* reg_ppcnt_rx_octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, rx_octets, 0x08 + 0x00, 0, 64);
+
+/* reg_ppcnt_rx_frames
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, rx_frames, 0x08 + 0x20, 0, 64);
+
+/* reg_ppcnt_tx_octets
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, tx_octets, 0x08 + 0x28, 0, 64);
+
+/* reg_ppcnt_tx_frames
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, tx_frames, 0x08 + 0x48, 0, 64);
+
+/* reg_ppcnt_rx_pause
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, rx_pause, 0x08 + 0x50, 0, 64);
+
+/* reg_ppcnt_rx_pause_duration
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, rx_pause_duration, 0x08 + 0x58, 0, 64);
+
+/* reg_ppcnt_tx_pause
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, tx_pause, 0x08 + 0x60, 0, 64);
+
+/* reg_ppcnt_tx_pause_duration
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, tx_pause_duration, 0x08 + 0x68, 0, 64);
+
+/* reg_ppcnt_rx_pause_transition
+ * Access: RO
+ */
+MLXSW_ITEM64(reg, ppcnt, tx_pause_transition, 0x08 + 0x70, 0, 64);
+
+static inline void mlxsw_reg_ppcnt_pack(char *payload, u8 local_port,
+					enum mlxsw_reg_ppcnt_grp grp,
+					u8 prio_tc)
 {
 	MLXSW_REG_ZERO(ppcnt, payload);
 	mlxsw_reg_ppcnt_swid_set(payload, 0);
 	mlxsw_reg_ppcnt_local_port_set(payload, local_port);
 	mlxsw_reg_ppcnt_pnat_set(payload, 0);
-	mlxsw_reg_ppcnt_grp_set(payload, 0);
+	mlxsw_reg_ppcnt_grp_set(payload, grp);
 	mlxsw_reg_ppcnt_clr_set(payload, 0);
-	mlxsw_reg_ppcnt_prio_tc_set(payload, 0);
+	mlxsw_reg_ppcnt_prio_tc_set(payload, prio_tc);
 }
 
 /* PPTB - Port Prio To Buffer Register
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 086682e..36a94a9 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -1195,7 +1195,8 @@ static void mlxsw_sp_port_get_stats(struct net_device *dev,
 	int i;
 	int err;
 
-	mlxsw_reg_ppcnt_pack(ppcnt_pl, mlxsw_sp_port->local_port);
+	mlxsw_reg_ppcnt_pack(ppcnt_pl, mlxsw_sp_port->local_port,
+			     MLXSW_REG_PPCNT_IEEE_8023_CNT, 0);
 	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ppcnt), ppcnt_pl);
 	for (i = 0; i < MLXSW_SP_PORT_HW_STATS_LEN; i++)
 		data[i] = !err ? mlxsw_sp_port_hw_stats[i].getter(ppcnt_pl) : 0;
diff --git a/drivers/net/ethernet/mellanox/mlxsw/switchx2.c b/drivers/net/ethernet/mellanox/mlxsw/switchx2.c
index 7a60a26..c49447f 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/switchx2.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/switchx2.c
@@ -518,7 +518,8 @@ static void mlxsw_sx_port_get_stats(struct net_device *dev,
 	int i;
 	int err;
 
-	mlxsw_reg_ppcnt_pack(ppcnt_pl, mlxsw_sx_port->local_port);
+	mlxsw_reg_ppcnt_pack(ppcnt_pl, mlxsw_sx_port->local_port,
+			     MLXSW_REG_PPCNT_IEEE_8023_CNT, 0);
 	err = mlxsw_reg_query(mlxsw_sx->core, MLXSW_REG(ppcnt), ppcnt_pl);
 	for (i = 0; i < MLXSW_SX_PORT_HW_STATS_LEN; i++)
 		data[i] = !err ? mlxsw_sx_port_hw_stats[i].getter(ppcnt_pl) : 0;
-- 
2.5.5

[patch net-next 17/17] mlxsw: spectrum: Add IEEE 802.1Qbb PFC support

[Jiri Pirko]

From: Ido Schimmel <idosch@mellanox.com>

Implement the appropriate DCB ops and allow a user to configure certain
traffic classes as lossless.

The operation configures PFC for both the egress (respecting PFC frames)
and ingress (sending PFC frames) parts of the port.

At egress, when a PFC frame is received for a PFC enabled priority, then
all the priorities mapped to the same TC are stopped.

At ingress, the priority group (PG) buffers to which the enabled PFC
priorities are mapped are configured to be lossless. PFC frames will be
transmitted when the Xoff threshold is crossed.

The user-supplied delay parameter is used to determine the PG's size
according to the following formula:

PG_SIZE = PG_SIZE_LOSSY + delay * CELL_FACTOR + MTU

In the worst case scenario the delay will be made up of packets that
are all of size CELL_SIZE + 1, which means each packet will require
almost twice its true size when buffered in the switch. We therefore
multiply this value by the "cell factor", which is close to 2.

Another MTU is added in case the transmitting host already started
transmitting a maximum length frame when the PFC packet was received.

As with PAUSE enabled ports, when the port's MTU is changed both the
PGs' size and threshold are adjusted accordingly.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlxsw/reg.h          |  10 ++
 drivers/net/ethernet/mellanox/mlxsw/spectrum.c     |  30 ++++--
 drivers/net/ethernet/mellanox/mlxsw/spectrum.h     |  12 ++-
 drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c | 117 ++++++++++++++++++++-
 4 files changed, 158 insertions(+), 11 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlxsw/reg.h b/drivers/net/ethernet/mellanox/mlxsw/reg.h
index 84aacb3..28f5b99 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/reg.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/reg.h
@@ -2442,6 +2442,16 @@ MLXSW_ITEM32(reg, pfcc, aprx, 0x0C, 30, 1);
  */
 MLXSW_ITEM32(reg, pfcc, pfcrx, 0x0C, 16, 8);
 
+#define MLXSW_REG_PFCC_ALL_PRIO 0xFF
+
+static inline void mlxsw_reg_pfcc_prio_pack(char *payload, u8 pfc_en)
+{
+	mlxsw_reg_pfcc_prio_mask_tx_set(payload, MLXSW_REG_PFCC_ALL_PRIO);
+	mlxsw_reg_pfcc_prio_mask_rx_set(payload, MLXSW_REG_PFCC_ALL_PRIO);
+	mlxsw_reg_pfcc_pfctx_set(payload, pfc_en);
+	mlxsw_reg_pfcc_pfcrx_set(payload, pfc_en);
+}
+
 static inline void mlxsw_reg_pfcc_pack(char *payload, u8 local_port)
 {
 	MLXSW_REG_ZERO(pfcc, payload);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
index 36a94a9..507263a 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.c
@@ -451,24 +451,27 @@ static int mlxsw_sp_port_set_mac_address(struct net_device *dev, void *p)
 }
 
 static void mlxsw_sp_pg_buf_pack(char *pbmc_pl, int pg_index, int mtu,
-				 bool pause_en)
+				 bool pause_en, bool pfc_en, u16 delay)
 {
 	u16 pg_size = 2 * MLXSW_SP_BYTES_TO_CELLS(mtu);
 
-	if (pause_en) {
-		u16 pg_pause_size = pg_size + MLXSW_SP_PAUSE_DELAY;
+	delay = pfc_en ? mlxsw_sp_pfc_delay_get(mtu, delay) :
+			 MLXSW_SP_PAUSE_DELAY;
 
+	if (pause_en || pfc_en)
 		mlxsw_reg_pbmc_lossless_buffer_pack(pbmc_pl, pg_index,
-						    pg_pause_size, pg_size);
-	} else {
+						    pg_size + delay, pg_size);
+	else
 		mlxsw_reg_pbmc_lossy_buffer_pack(pbmc_pl, pg_index, pg_size);
-	}
 }
 
 int __mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port, int mtu,
-				 u8 *prio_tc, bool pause_en)
+				 u8 *prio_tc, bool pause_en,
+				 struct ieee_pfc *my_pfc)
 {
 	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	u8 pfc_en = !!my_pfc ? my_pfc->pfc_en : 0;
+	u16 delay = !!my_pfc ? my_pfc->delay : 0;
 	char pbmc_pl[MLXSW_REG_PBMC_LEN];
 	int i, j, err;
 
@@ -479,9 +482,11 @@ int __mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port, int mtu,
 
 	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
 		bool configure = false;
+		bool pfc = false;
 
 		for (j = 0; j < IEEE_8021QAZ_MAX_TCS; j++) {
 			if (prio_tc[j] == i) {
+				pfc = pfc_en & BIT(j);
 				configure = true;
 				break;
 			}
@@ -489,7 +494,7 @@ int __mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port, int mtu,
 
 		if (!configure)
 			continue;
-		mlxsw_sp_pg_buf_pack(pbmc_pl, i, mtu, pause_en);
+		mlxsw_sp_pg_buf_pack(pbmc_pl, i, mtu, pause_en, pfc, delay);
 	}
 
 	return mlxsw_reg_write(mlxsw_sp->core, MLXSW_REG(pbmc), pbmc_pl);
@@ -500,12 +505,14 @@ static int mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port,
 {
 	u8 def_prio_tc[IEEE_8021QAZ_MAX_TCS] = {0};
 	bool dcb_en = !!mlxsw_sp_port->dcb.ets;
+	struct ieee_pfc *my_pfc;
 	u8 *prio_tc;
 
 	prio_tc = dcb_en ? mlxsw_sp_port->dcb.ets->prio_tc : def_prio_tc;
+	my_pfc = dcb_en ? mlxsw_sp_port->dcb.pfc : NULL;
 
 	return __mlxsw_sp_port_headroom_set(mlxsw_sp_port, mtu, prio_tc,
-					    pause_en);
+					    pause_en, my_pfc);
 }
 
 static int mlxsw_sp_port_change_mtu(struct net_device *dev, int mtu)
@@ -1032,6 +1039,11 @@ static int mlxsw_sp_port_set_pauseparam(struct net_device *dev,
 	bool pause_en = pause->tx_pause || pause->rx_pause;
 	int err;
 
+	if (mlxsw_sp_port->dcb.pfc && mlxsw_sp_port->dcb.pfc->pfc_en) {
+		netdev_err(dev, "PFC already enabled on port\n");
+		return -EINVAL;
+	}
+
 	if (pause->autoneg) {
 		netdev_err(dev, "PAUSE frames autonegotiation isn't supported\n");
 		return -EINVAL;
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
index f4b53dd..47610a5 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum.h
@@ -72,6 +72,14 @@
  */
 #define MLXSW_SP_PAUSE_DELAY 612
 
+#define MLXSW_SP_CELL_FACTOR 2	/* 2 * cell_size / (IPG + cell_size + 1) */
+
+static inline u16 mlxsw_sp_pfc_delay_get(int mtu, u16 delay)
+{
+	delay = MLXSW_SP_BYTES_TO_CELLS(DIV_ROUND_UP(delay, BITS_PER_BYTE));
+	return MLXSW_SP_CELL_FACTOR * delay + MLXSW_SP_BYTES_TO_CELLS(mtu);
+}
+
 struct mlxsw_sp_port;
 
 struct mlxsw_sp_upper {
@@ -183,6 +191,7 @@ struct mlxsw_sp_port {
 	struct {
 		struct ieee_ets *ets;
 		struct ieee_maxrate *maxrate;
+		struct ieee_pfc *pfc;
 	} dcb;
 	/* 802.1Q bridge VLANs */
 	unsigned long *active_vlans;
@@ -295,7 +304,8 @@ int mlxsw_sp_port_ets_set(struct mlxsw_sp_port *mlxsw_sp_port,
 int mlxsw_sp_port_prio_tc_set(struct mlxsw_sp_port *mlxsw_sp_port,
 			      u8 switch_prio, u8 tclass);
 int __mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port, int mtu,
-				 u8 *prio_tc, bool pause_en);
+				 u8 *prio_tc, bool pause_en,
+				 struct ieee_pfc *my_pfc);
 int mlxsw_sp_port_ets_maxrate_set(struct mlxsw_sp_port *mlxsw_sp_port,
 				  enum mlxsw_reg_qeec_hr hr, u8 index,
 				  u8 next_index, u32 maxrate);
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
index 8786424..0b32366 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_dcb.c
@@ -34,6 +34,7 @@
 
 #include <linux/netdevice.h>
 #include <linux/string.h>
+#include <linux/bitops.h>
 #include <net/dcbnl.h>
 
 #include "spectrum.h"
@@ -151,7 +152,8 @@ static int mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port,
 	 * traffic is still directed to them.
 	 */
 	err = __mlxsw_sp_port_headroom_set(mlxsw_sp_port, dev->mtu,
-					   ets->prio_tc, pause_en);
+					   ets->prio_tc, pause_en,
+					   mlxsw_sp_port->dcb.pfc);
 	if (err) {
 		netdev_err(dev, "Failed to configure port's headroom\n");
 		return err;
@@ -291,11 +293,101 @@ err_port_ets_maxrate_set:
 	return err;
 }
 
+static int mlxsw_sp_port_pfc_cnt_get(struct mlxsw_sp_port *mlxsw_sp_port,
+				     u8 prio)
+{
+	struct mlxsw_sp *mlxsw_sp = mlxsw_sp_port->mlxsw_sp;
+	struct ieee_pfc *my_pfc = mlxsw_sp_port->dcb.pfc;
+	char ppcnt_pl[MLXSW_REG_PPCNT_LEN];
+	int err;
+
+	mlxsw_reg_ppcnt_pack(ppcnt_pl, mlxsw_sp_port->local_port,
+			     MLXSW_REG_PPCNT_PRIO_CNT, prio);
+	err = mlxsw_reg_query(mlxsw_sp->core, MLXSW_REG(ppcnt), ppcnt_pl);
+	if (err)
+		return err;
+
+	my_pfc->requests[prio] = mlxsw_reg_ppcnt_tx_pause_get(ppcnt_pl);
+	my_pfc->indications[prio] = mlxsw_reg_ppcnt_rx_pause_get(ppcnt_pl);
+
+	return 0;
+}
+
+static int mlxsw_sp_dcbnl_ieee_getpfc(struct net_device *dev,
+				      struct ieee_pfc *pfc)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	int err, i;
+
+	for (i = 0; i < IEEE_8021QAZ_MAX_TCS; i++) {
+		err = mlxsw_sp_port_pfc_cnt_get(mlxsw_sp_port, i);
+		if (err) {
+			netdev_err(dev, "Failed to get PFC count for priority %d\n",
+				   i);
+			return err;
+		}
+	}
+
+	memcpy(pfc, mlxsw_sp_port->dcb.pfc, sizeof(*pfc));
+
+	return 0;
+}
+
+static int mlxsw_sp_port_pfc_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				 struct ieee_pfc *pfc)
+{
+	char pfcc_pl[MLXSW_REG_PFCC_LEN];
+
+	mlxsw_reg_pfcc_pack(pfcc_pl, mlxsw_sp_port->local_port);
+	mlxsw_reg_pfcc_prio_pack(pfcc_pl, pfc->pfc_en);
+
+	return mlxsw_reg_write(mlxsw_sp_port->mlxsw_sp->core, MLXSW_REG(pfcc),
+			       pfcc_pl);
+}
+
+static int mlxsw_sp_dcbnl_ieee_setpfc(struct net_device *dev,
+				      struct ieee_pfc *pfc)
+{
+	struct mlxsw_sp_port *mlxsw_sp_port = netdev_priv(dev);
+	int err;
+
+	if (mlxsw_sp_port->link.tx_pause || mlxsw_sp_port->link.rx_pause) {
+		netdev_err(dev, "PAUSE frames already enabled on port\n");
+		return -EINVAL;
+	}
+
+	err = __mlxsw_sp_port_headroom_set(mlxsw_sp_port, dev->mtu,
+					   mlxsw_sp_port->dcb.ets->prio_tc,
+					   false, pfc);
+	if (err) {
+		netdev_err(dev, "Failed to configure port's headroom for PFC\n");
+		return err;
+	}
+
+	err = mlxsw_sp_port_pfc_set(mlxsw_sp_port, pfc);
+	if (err) {
+		netdev_err(dev, "Failed to configure PFC\n");
+		goto err_port_pfc_set;
+	}
+
+	memcpy(mlxsw_sp_port->dcb.pfc, pfc, sizeof(*pfc));
+
+	return 0;
+
+err_port_pfc_set:
+	__mlxsw_sp_port_headroom_set(mlxsw_sp_port, dev->mtu,
+				     mlxsw_sp_port->dcb.ets->prio_tc, false,
+				     mlxsw_sp_port->dcb.pfc);
+	return err;
+}
+
 static const struct dcbnl_rtnl_ops mlxsw_sp_dcbnl_ops = {
 	.ieee_getets		= mlxsw_sp_dcbnl_ieee_getets,
 	.ieee_setets		= mlxsw_sp_dcbnl_ieee_setets,
 	.ieee_getmaxrate	= mlxsw_sp_dcbnl_ieee_getmaxrate,
 	.ieee_setmaxrate	= mlxsw_sp_dcbnl_ieee_setmaxrate,
+	.ieee_getpfc		= mlxsw_sp_dcbnl_ieee_getpfc,
+	.ieee_setpfc		= mlxsw_sp_dcbnl_ieee_setpfc,
 
 	.getdcbx		= mlxsw_sp_dcbnl_getdcbx,
 	.setdcbx		= mlxsw_sp_dcbnl_setdcbx,
@@ -338,6 +430,23 @@ static void mlxsw_sp_port_maxrate_fini(struct mlxsw_sp_port *mlxsw_sp_port)
 	kfree(mlxsw_sp_port->dcb.maxrate);
 }
 
+static int mlxsw_sp_port_pfc_init(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	mlxsw_sp_port->dcb.pfc = kzalloc(sizeof(*mlxsw_sp_port->dcb.pfc),
+					 GFP_KERNEL);
+	if (!mlxsw_sp_port->dcb.pfc)
+		return -ENOMEM;
+
+	mlxsw_sp_port->dcb.pfc->pfc_cap = IEEE_8021QAZ_MAX_TCS;
+
+	return 0;
+}
+
+static void mlxsw_sp_port_pfc_fini(struct mlxsw_sp_port *mlxsw_sp_port)
+{
+	kfree(mlxsw_sp_port->dcb.pfc);
+}
+
 int mlxsw_sp_port_dcb_init(struct mlxsw_sp_port *mlxsw_sp_port)
 {
 	int err;
@@ -348,11 +457,16 @@ int mlxsw_sp_port_dcb_init(struct mlxsw_sp_port *mlxsw_sp_port)
 	err = mlxsw_sp_port_maxrate_init(mlxsw_sp_port);
 	if (err)
 		goto err_port_maxrate_init;
+	err = mlxsw_sp_port_pfc_init(mlxsw_sp_port);
+	if (err)
+		goto err_port_pfc_init;
 
 	mlxsw_sp_port->dev->dcbnl_ops = &mlxsw_sp_dcbnl_ops;
 
 	return 0;
 
+err_port_pfc_init:
+	mlxsw_sp_port_maxrate_fini(mlxsw_sp_port);
 err_port_maxrate_init:
 	mlxsw_sp_port_ets_fini(mlxsw_sp_port);
 	return err;
@@ -360,6 +474,7 @@ err_port_maxrate_init:
 
 void mlxsw_sp_port_dcb_fini(struct mlxsw_sp_port *mlxsw_sp_port)
 {
+	mlxsw_sp_port_pfc_fini(mlxsw_sp_port);
 	mlxsw_sp_port_maxrate_fini(mlxsw_sp_port);
 	mlxsw_sp_port_ets_fini(mlxsw_sp_port);
 }
-- 
2.5.5

Re: [patch net-next 00/17] mlxsw: Introduce support for Data Center Bridging

[David Miller]

From: Jiri Pirko <jiri@resnulli.us>
Date: Wed,  6 Apr 2016 17:09:59 +0200

> From: Jiri Pirko <jiri@mellanox.com>
> 
> Ido says:
> 
> This patchset introduces support for Quality of Service (QoS) as part of the
> IEEE Data Center Bridiging (DCB) standards.
> 
> Patches 1-9 do the required device initialization. Specifically, patches 1-6
> initialize the ports' headroom buffers, which are used at ingress to store
> incoming packets while they go through the switch's pipeline. Patches 7-9
> complete them by initializing the egress scheduling.
> 
> The pipeline mentioned above determines the packet's egress port(s) and
> traffic class. Ideally, once out of the pipeline the packet moves to the
> switch's shared buffer (to be introduced in Jiri's patchset, currently
> default values are used) and scheduled for transmission according to its
> traffic class. The egress scheduling is configured according to the 802.1Qaz
> standard, which is part of the DCB infrastructure supported by Linux. This
> is introduced in patches 10-12.
> 
> Even after going through the pipeline packets are not always eligible to
> enter the shared buffer. This is determined by the amount of available space
> and the quotas associated with the packet. However, if flow control is
> enabled and the packet is associated with the lossless flow, then it will
> stay in the headroom and won't be discarded. This is introduced in patches
> 13-17.
> 
> Please check individual commit messages for more info, as I tried to keep
> them pretty detailed.

Series applied, thanks guys.