Merge branch 'mlxsw-dcb'

Jiri Pirko says:

====================
mlxsw: Introduce support for Data Center Bridging

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.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

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.

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

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

@@ -1805,6 +1805,184 @@ 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.
+ */
+#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.
@@ -2141,6 +2319,145 @@ 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);
+
+#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);
+	mlxsw_reg_pfcc_local_port_set(payload, local_port);
+}
+
 /* PPCNT - Ports Performance Counters Register
  * -------------------------------------------
  * The PPCNT register retrieves per port performance counters.
@@ -2180,6 +2497,11 @@ 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
@@ -2215,6 +2537,8 @@ 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
  */
@@ -2329,15 +2653,145 @@ 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
+ * -----------------------------------
+ * 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
@@ -2346,7 +2800,7 @@ static inline void mlxsw_reg_ppcnt_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,
@@ -2374,6 +2828,8 @@ 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
@@ -2398,6 +2854,30 @@ 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)
 {
@@ -2416,6 +2896,17 @@ 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
@@ -3283,6 +3774,10 @@ 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:
 		return "PMLP";
 	case MLXSW_REG_PMTU_ID:
@@ -3293,8 +3788,12 @@ 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:
+		return "PPTB";
 	case MLXSW_REG_PBMC_ID:
 		return "PBMC";
 	case MLXSW_REG_PSPA_ID:

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>
@@ -449,16 +450,89 @@ 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,
+				 bool pause_en, bool pfc_en, u16 delay)
+{
+	u16 pg_size = 2 * MLXSW_SP_BYTES_TO_CELLS(mtu);
+
+	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_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,
+				 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;
+
+	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++) {
+		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;
+			}
+		}
+
+		if (!configure)
+			continue;
+		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);
+}
+
+static int mlxsw_sp_port_headroom_set(struct mlxsw_sp_port *mlxsw_sp_port,
+				      int mtu, bool pause_en)
+{
+	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, my_pfc);
+}
+
 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_mtu_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);
+	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, pause_en);
+	return err;
 }
 
 static struct rtnl_link_stats64 *
@@ -936,6 +1010,68 @@ 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 (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;
+	}
+
+	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);
@@ -1071,7 +1207,8 @@ 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;
@@ -1419,6 +1556,8 @@ 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,
@@ -1441,6 +1580,108 @@ 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);
 }
 
+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);
+}
+
+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);
+}
+
+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)
 {
@@ -1548,6 +1789,21 @@ 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;
+	}
+
+	/* 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) {
@@ -1568,6 +1824,8 @@ 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:
 err_port_mtu_set:
@@ -1637,6 +1895,7 @@ 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);

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>
 
@@ -62,6 +63,23 @@
 
 #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)
+
+/* Maximum delay buffer needed in case of PAUSE frames, in cells.
+ * Assumes 100m cable and maximum MTU.
+ */
+#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 {
@@ -166,6 +184,15 @@ struct mlxsw_sp_port {
 		struct mlxsw_sp_vfid *vfid;
 		u16 vid;
 	} vport;
+	struct {
+		u8 tx_pause:1,
+		   rx_pause:1;
+	} link;
+	struct {
+		struct ieee_ets *ets;
+		struct ieee_maxrate *maxrate;
+		struct ieee_pfc *pfc;
+	} dcb;
 	/* 802.1Q bridge VLANs */
 	unsigned long *active_vlans;
 	unsigned long *untagged_vlans;
@@ -174,6 +201,12 @@ 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)
 {
@@ -265,5 +298,33 @@ 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, 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);
+
+#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

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

@@ -34,6 +34,8 @@
 
 #include <linux/kernel.h>
 #include <linux/types.h>
+#include <linux/dcbnl.h>
+#include <linux/if_ether.h>
 
 #include "spectrum.h"
 #include "core.h"
@@ -52,15 +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)
@@ -78,11 +80,33 @@ 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);
 }
 
-#define MLXSW_SP_SB_BYTES_PER_CELL 96
+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);
+}
 
 struct mlxsw_sp_sb_pool {
 	u8 pool;
@@ -92,11 +116,9 @@ 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)		\
 	{							\
@@ -115,14 +137,14 @@ 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)
@@ -178,7 +200,7 @@ 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),
@@ -186,15 +208,15 @@ 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),
@@ -353,21 +375,21 @@ 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)
@@ -410,7 +432,7 @@ 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);

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

@@ -0,0 +1,480 @@
+/*
+ * 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 <linux/string.h>
+#include <linux/bitops.h>
+#include <net/dcbnl.h>
+
+#include "spectrum.h"
+#include "reg.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 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)
+{
+	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;
+
+	/* 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, pause_en,
+					   mlxsw_sp_port->dcb.pfc);
+	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 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 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,
+};
+
+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);
+}
+
+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);
+}
+
+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;
+
+	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;
+	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;
+}
+
+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);
+}

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;