Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next

Pull networking updates from David Miller:

 1) Significantly shrink the core networking routing structures. Result
    of http://vger.kernel.org/~davem/seoul2017_netdev_keynote.pdf

 2) Add netdevsim driver for testing various offloads, from Jakub
    Kicinski.

 3) Support cross-chip FDB operations in DSA, from Vivien Didelot.

 4) Add a 2nd listener hash table for TCP, similar to what was done for
    UDP. From Martin KaFai Lau.

 5) Add eBPF based queue selection to tun, from Jason Wang.

 6) Lockless qdisc support, from John Fastabend.

 7) SCTP stream interleave support, from Xin Long.

 8) Smoother TCP receive autotuning, from Eric Dumazet.

 9) Lots of erspan tunneling enhancements, from William Tu.

10) Add true function call support to BPF, from Alexei Starovoitov.

11) Add explicit support for GRO HW offloading, from Michael Chan.

12) Support extack generation in more netlink subsystems. From Alexander
    Aring, Quentin Monnet, and Jakub Kicinski.

13) Add 1000BaseX, flow control, and EEE support to mvneta driver. From
    Russell King.

14) Add flow table abstraction to netfilter, from Pablo Neira Ayuso.

15) Many improvements and simplifications to the NFP driver bpf JIT,
    from Jakub Kicinski.

16) Support for ipv6 non-equal cost multipath routing, from Ido
    Schimmel.

17) Add resource abstration to devlink, from Arkadi Sharshevsky.

18) Packet scheduler classifier shared filter block support, from Jiri
    Pirko.

19) Avoid locking in act_csum, from Davide Caratti.

20) devinet_ioctl() simplifications from Al viro.

21) More TCP bpf improvements from Lawrence Brakmo.

22) Add support for onlink ipv6 route flag, similar to ipv4, from David
    Ahern.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1925 commits)
  tls: Add support for encryption using async offload accelerator
  ip6mr: fix stale iterator
  net/sched: kconfig: Remove blank help texts
  openvswitch: meter: Use 64-bit arithmetic instead of 32-bit
  tcp_nv: fix potential integer overflow in tcpnv_acked
  r8169: fix RTL8168EP take too long to complete driver initialization.
  qmi_wwan: Add support for Quectel EP06
  rtnetlink: enable IFLA_IF_NETNSID for RTM_NEWLINK
  ipmr: Fix ptrdiff_t print formatting
  ibmvnic: Wait for device response when changing MAC
  qlcnic: fix deadlock bug
  tcp: release sk_frag.page in tcp_disconnect
  ipv4: Get the address of interface correctly.
  net_sched: gen_estimator: fix lockdep splat
  net: macb: Handle HRESP error
  net/mlx5e: IPoIB, Fix copy-paste bug in flow steering refactoring
  ipv6: addrconf: break critical section in addrconf_verify_rtnl()
  ipv6: change route cache aging logic
  i40e/i40evf: Update DESC_NEEDED value to reflect larger value
  bnxt_en: cleanup DIM work on device shutdown
  ...

1 files changed, 736 insertions, 355 deletions

diff --git a/drivers/net/ethernet/qlogic/qed/qed_init_fw_funcs.c b/drivers/net/ethernet/qlogic/qed/qed_init_fw_funcs.c
index b069ad0..18fb506 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_init_fw_funcs.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_init_fw_funcs.c
@@ -31,6 +31,7 @@
  */
 
 #include <linux/types.h>
+#include <linux/crc8.h>
 #include <linux/delay.h>
 #include <linux/kernel.h>
 #include <linux/slab.h>
@@ -40,102 +41,197 @@
 #include "qed_init_ops.h"
 #include "qed_reg_addr.h"
 
+#define CDU_VALIDATION_DEFAULT_CFG	61
+
+static u16 con_region_offsets[3][NUM_OF_CONNECTION_TYPES_E4] = {
+	{400, 336, 352, 304, 304, 384, 416, 352},	/* region 3 offsets */
+	{528, 496, 416, 448, 448, 512, 544, 480},	/* region 4 offsets */
+	{608, 544, 496, 512, 576, 592, 624, 560}	/* region 5 offsets */
+};
+
+static u16 task_region_offsets[1][NUM_OF_CONNECTION_TYPES_E4] = {
+	{240, 240, 112, 0, 0, 0, 0, 96}	/* region 1 offsets */
+};
+
 /* General constants */
 #define QM_PQ_MEM_4KB(pq_size)	(pq_size ? DIV_ROUND_UP((pq_size + 1) *	\
 							QM_PQ_ELEMENT_SIZE, \
 							0x1000) : 0)
 #define QM_PQ_SIZE_256B(pq_size)	(pq_size ? DIV_ROUND_UP(pq_size, \
 								0x100) - 1 : 0)
-#define QM_INVALID_PQ_ID                        0xffff
+#define QM_INVALID_PQ_ID		0xffff
+
 /* Feature enable */
-#define QM_BYPASS_EN                            1
-#define QM_BYTE_CRD_EN                          1
+#define QM_BYPASS_EN	1
+#define QM_BYTE_CRD_EN	1
+
 /* Other PQ constants */
-#define QM_OTHER_PQS_PER_PF                     4
+#define QM_OTHER_PQS_PER_PF	4
+
 /* WFQ constants */
-#define QM_WFQ_UPPER_BOUND		62500000
-#define QM_WFQ_VP_PQ_VOQ_SHIFT          0
-#define QM_WFQ_VP_PQ_PF_SHIFT           5
-#define QM_WFQ_INC_VAL(weight)          ((weight) * 0x9000)
-#define QM_WFQ_MAX_INC_VAL                      43750000
+
+/* Upper bound in MB, 10 * burst size of 1ms in 50Gbps */
+#define QM_WFQ_UPPER_BOUND	62500000
+
+/* Bit  of VOQ in WFQ VP PQ map */
+#define QM_WFQ_VP_PQ_VOQ_SHIFT	0
+
+/* Bit  of PF in WFQ VP PQ map */
+#define QM_WFQ_VP_PQ_PF_E4_SHIFT	5
+
+/* 0x9000 = 4*9*1024 */
+#define QM_WFQ_INC_VAL(weight)	((weight) * 0x9000)
+
+/* Max WFQ increment value is 0.7 * upper bound */
+#define QM_WFQ_MAX_INC_VAL	((QM_WFQ_UPPER_BOUND * 7) / 10)
 
 /* RL constants */
-#define QM_RL_UPPER_BOUND                       62500000
-#define QM_RL_PERIOD                            5               /* in us */
-#define QM_RL_PERIOD_CLK_25M            (25 * QM_RL_PERIOD)
-#define QM_RL_MAX_INC_VAL                       43750000
-#define QM_RL_INC_VAL(rate)		max_t(u32,	\
-					      (u32)(((rate ? rate : \
-						      1000000) *    \
-						     QM_RL_PERIOD * \
-						     101) / (8 * 100)), 1)
+
+/* Period in us */
+#define QM_RL_PERIOD	5
+
+/* Period in 25MHz cycles */
+#define QM_RL_PERIOD_CLK_25M	(25 * QM_RL_PERIOD)
+
+/* RL increment value - rate is specified in mbps */
+#define QM_RL_INC_VAL(rate) ({ \
+	typeof(rate) __rate = (rate); \
+	max_t(u32, \
+	      (u32)(((__rate ? __rate : 1000000) * QM_RL_PERIOD * 101) / \
+		    (8 * 100)), \
+	      1); })
+
+/* PF RL Upper bound is set to 10 * burst size of 1ms in 50Gbps */
+#define QM_PF_RL_UPPER_BOUND	62500000
+
+/* Max PF RL increment value is 0.7 * upper bound */
+#define QM_PF_RL_MAX_INC_VAL	((QM_PF_RL_UPPER_BOUND * 7) / 10)
+
+/* Vport RL Upper bound, link speed is in Mpbs */
+#define QM_VP_RL_UPPER_BOUND(speed)	((u32)max_t(u32, \
+						    QM_RL_INC_VAL(speed), \
+						    9700 + 1000))
+
+/* Max Vport RL increment value is the Vport RL upper bound */
+#define QM_VP_RL_MAX_INC_VAL(speed)	QM_VP_RL_UPPER_BOUND(speed)
+
+/* Vport RL credit threshold in case of QM bypass */
+#define QM_VP_RL_BYPASS_THRESH_SPEED	(QM_VP_RL_UPPER_BOUND(10000) - 1)
+
 /* AFullOprtnstcCrdMask constants */
-#define QM_OPPOR_LINE_VOQ_DEF           1
-#define QM_OPPOR_FW_STOP_DEF            0
-#define QM_OPPOR_PQ_EMPTY_DEF           1
+#define QM_OPPOR_LINE_VOQ_DEF	1
+#define QM_OPPOR_FW_STOP_DEF	0
+#define QM_OPPOR_PQ_EMPTY_DEF	1
+
 /* Command Queue constants */
-#define PBF_CMDQ_PURE_LB_LINES                          150
-#define PBF_CMDQ_LINES_RT_OFFSET(voq)           (		 \
-		PBF_REG_YCMD_QS_NUM_LINES_VOQ0_RT_OFFSET + voq * \
-		(PBF_REG_YCMD_QS_NUM_LINES_VOQ1_RT_OFFSET -	 \
-		 PBF_REG_YCMD_QS_NUM_LINES_VOQ0_RT_OFFSET))
-#define PBF_BTB_GUARANTEED_RT_OFFSET(voq)       (	      \
-		PBF_REG_BTB_GUARANTEED_VOQ0_RT_OFFSET + voq * \
-		(PBF_REG_BTB_GUARANTEED_VOQ1_RT_OFFSET -      \
-		 PBF_REG_BTB_GUARANTEED_VOQ0_RT_OFFSET))
-#define QM_VOQ_LINE_CRD(pbf_cmd_lines)          ((((pbf_cmd_lines) - \
-						   4) *		     \
-						  2) | QM_LINE_CRD_REG_SIGN_BIT)
+
+/* Pure LB CmdQ lines (+spare) */
+#define PBF_CMDQ_PURE_LB_LINES	150
+
+#define PBF_CMDQ_LINES_E5_RSVD_RATIO	8
+
+#define PBF_CMDQ_LINES_RT_OFFSET(ext_voq) \
+	(PBF_REG_YCMD_QS_NUM_LINES_VOQ0_RT_OFFSET + \
+	 (ext_voq) * (PBF_REG_YCMD_QS_NUM_LINES_VOQ1_RT_OFFSET - \
+		PBF_REG_YCMD_QS_NUM_LINES_VOQ0_RT_OFFSET))
+
+#define PBF_BTB_GUARANTEED_RT_OFFSET(ext_voq) \
+	(PBF_REG_BTB_GUARANTEED_VOQ0_RT_OFFSET + \
+	 (ext_voq) * (PBF_REG_BTB_GUARANTEED_VOQ1_RT_OFFSET - \
+		PBF_REG_BTB_GUARANTEED_VOQ0_RT_OFFSET))
+
+#define QM_VOQ_LINE_CRD(pbf_cmd_lines) \
+	((((pbf_cmd_lines) - 4) * 2) | QM_LINE_CRD_REG_SIGN_BIT)
+
 /* BTB: blocks constants (block size = 256B) */
-#define BTB_JUMBO_PKT_BLOCKS            38
-#define BTB_HEADROOM_BLOCKS                     BTB_JUMBO_PKT_BLOCKS
-#define BTB_PURE_LB_FACTOR                      10
-#define BTB_PURE_LB_RATIO                       7
+
+/* 256B blocks in 9700B packet */
+#define BTB_JUMBO_PKT_BLOCKS	38
+
+/* Headroom per-port */
+#define BTB_HEADROOM_BLOCKS	BTB_JUMBO_PKT_BLOCKS
+#define BTB_PURE_LB_FACTOR	10
+
+/* Factored (hence really 0.7) */
+#define BTB_PURE_LB_RATIO	7
+
 /* QM stop command constants */
-#define QM_STOP_PQ_MASK_WIDTH           32
-#define QM_STOP_CMD_ADDR                2
-#define QM_STOP_CMD_STRUCT_SIZE         2
-#define QM_STOP_CMD_PAUSE_MASK_OFFSET   0
-#define QM_STOP_CMD_PAUSE_MASK_SHIFT    0
-#define QM_STOP_CMD_PAUSE_MASK_MASK     -1
-#define QM_STOP_CMD_GROUP_ID_OFFSET     1
-#define QM_STOP_CMD_GROUP_ID_SHIFT      16
-#define QM_STOP_CMD_GROUP_ID_MASK       15
-#define QM_STOP_CMD_PQ_TYPE_OFFSET      1
-#define QM_STOP_CMD_PQ_TYPE_SHIFT       24
-#define QM_STOP_CMD_PQ_TYPE_MASK        1
-#define QM_STOP_CMD_MAX_POLL_COUNT      100
-#define QM_STOP_CMD_POLL_PERIOD_US      500
+#define QM_STOP_PQ_MASK_WIDTH		32
+#define QM_STOP_CMD_ADDR		2
+#define QM_STOP_CMD_STRUCT_SIZE		2
+#define QM_STOP_CMD_PAUSE_MASK_OFFSET	0
+#define QM_STOP_CMD_PAUSE_MASK_SHIFT	0
+#define QM_STOP_CMD_PAUSE_MASK_MASK	-1
+#define QM_STOP_CMD_GROUP_ID_OFFSET	1
+#define QM_STOP_CMD_GROUP_ID_SHIFT	16
+#define QM_STOP_CMD_GROUP_ID_MASK	15
+#define QM_STOP_CMD_PQ_TYPE_OFFSET	1
+#define QM_STOP_CMD_PQ_TYPE_SHIFT	24
+#define QM_STOP_CMD_PQ_TYPE_MASK	1
+#define QM_STOP_CMD_MAX_POLL_COUNT	100
+#define QM_STOP_CMD_POLL_PERIOD_US	500
 
 /* QM command macros */
-#define QM_CMD_STRUCT_SIZE(cmd)			cmd ## \
-	_STRUCT_SIZE
-#define QM_CMD_SET_FIELD(var, cmd, field,				  \
-			 value)        SET_FIELD(var[cmd ## _ ## field ## \
-						     _OFFSET],		  \
-						 cmd ## _ ## field,	  \
-						 value)
-/* QM: VOQ macros */
-#define PHYS_VOQ(port, tc, max_phys_tcs_per_port) ((port) *	\
-						   (max_phys_tcs_per_port) + \
-						   (tc))
-#define LB_VOQ(port)				( \
-		MAX_PHYS_VOQS + (port))
-#define VOQ(port, tc, max_phy_tcs_pr_port)	\
-	((tc) <		\
-	 LB_TC ? PHYS_VOQ(port,		\
-			  tc,			 \
-			  max_phy_tcs_pr_port) \
-		: LB_VOQ(port))
+#define QM_CMD_STRUCT_SIZE(cmd)	cmd ## _STRUCT_SIZE
+#define QM_CMD_SET_FIELD(var, cmd, field, value) \
+	SET_FIELD(var[cmd ## _ ## field ## _OFFSET], \
+		  cmd ## _ ## field, \
+		  value)
+
+#define QM_INIT_TX_PQ_MAP(p_hwfn, map, chip, pq_id, rl_valid, vp_pq_id, rl_id, \
+			  ext_voq, wrr) \
+	do { \
+		typeof(map) __map; \
+		memset(&__map, 0, sizeof(__map)); \
+		SET_FIELD(__map.reg, QM_RF_PQ_MAP_ ## chip ## _PQ_VALID, 1); \
+		SET_FIELD(__map.reg, QM_RF_PQ_MAP_ ## chip ## _RL_VALID, \
+			  rl_valid); \
+		SET_FIELD(__map.reg, QM_RF_PQ_MAP_ ## chip ## _VP_PQ_ID, \
+			  vp_pq_id); \
+		SET_FIELD(__map.reg, QM_RF_PQ_MAP_ ## chip ## _RL_ID, rl_id); \
+		SET_FIELD(__map.reg, QM_RF_PQ_MAP_ ## chip ## _VOQ, ext_voq); \
+		SET_FIELD(__map.reg, \
+			  QM_RF_PQ_MAP_ ## chip ## _WRR_WEIGHT_GROUP, wrr); \
+		STORE_RT_REG(p_hwfn, QM_REG_TXPQMAP_RT_OFFSET + (pq_id), \
+			     *((u32 *)&__map)); \
+		(map) = __map; \
+	} while (0)
+
+#define WRITE_PQ_INFO_TO_RAM	1
+#define PQ_INFO_ELEMENT(vp, pf, tc, port, rl_valid, rl) \
+	(((vp) << 0) | ((pf) << 12) | ((tc) << 16) | ((port) << 20) | \
+	((rl_valid) << 22) | ((rl) << 24))
+#define PQ_INFO_RAM_GRC_ADDRESS(pq_id) \
+	(XSEM_REG_FAST_MEMORY + SEM_FAST_REG_INT_RAM + 21776 + (pq_id) * 4)
+
 /******************** INTERNAL IMPLEMENTATION *********************/
+
+/* Returns the external VOQ number */
+static u8 qed_get_ext_voq(struct qed_hwfn *p_hwfn,
+			  u8 port_id, u8 tc, u8 max_phys_tcs_per_port)
+{
+	if (tc == PURE_LB_TC)
+		return NUM_OF_PHYS_TCS * MAX_NUM_PORTS_BB + port_id;
+	else
+		return port_id * max_phys_tcs_per_port + tc;
+}
+
 /* Prepare PF RL enable/disable runtime init values */
 static void qed_enable_pf_rl(struct qed_hwfn *p_hwfn, bool pf_rl_en)
 {
 	STORE_RT_REG(p_hwfn, QM_REG_RLPFENABLE_RT_OFFSET, pf_rl_en ? 1 : 0);
 	if (pf_rl_en) {
+		u8 num_ext_voqs = MAX_NUM_VOQS_E4;
+		u64 voq_bit_mask = ((u64)1 << num_ext_voqs) - 1;
+
 		/* Enable RLs for all VOQs */
-		STORE_RT_REG(p_hwfn, QM_REG_RLPFVOQENABLE_RT_OFFSET,
-			     (1 << MAX_NUM_VOQS) - 1);
+		STORE_RT_REG(p_hwfn,
+			     QM_REG_RLPFVOQENABLE_RT_OFFSET,
+			     (u32)voq_bit_mask);
+		if (num_ext_voqs >= 32)
+			STORE_RT_REG(p_hwfn, QM_REG_RLPFVOQENABLE_MSB_RT_OFFSET,
+				     (u32)(voq_bit_mask >> 32));
+
 		/* Write RL period */
 		STORE_RT_REG(p_hwfn,
 			     QM_REG_RLPFPERIOD_RT_OFFSET, QM_RL_PERIOD_CLK_25M);
@@ -147,7 +243,7 @@ static void qed_enable_pf_rl(struct qed_hwfn *p_hwfn, bool pf_rl_en)
 		if (QM_BYPASS_EN)
 			STORE_RT_REG(p_hwfn,
 				     QM_REG_AFULLQMBYPTHRPFRL_RT_OFFSET,
-				     QM_RL_UPPER_BOUND);
+				     QM_PF_RL_UPPER_BOUND);
 	}
 }
 
@@ -181,7 +277,7 @@ static void qed_enable_vport_rl(struct qed_hwfn *p_hwfn, bool vport_rl_en)
 		if (QM_BYPASS_EN)
 			STORE_RT_REG(p_hwfn,
 				     QM_REG_AFULLQMBYPTHRGLBLRL_RT_OFFSET,
-				     QM_RL_UPPER_BOUND);
+				     QM_VP_RL_BYPASS_THRESH_SPEED);
 	}
 }
 
@@ -202,15 +298,15 @@ static void qed_enable_vport_wfq(struct qed_hwfn *p_hwfn, bool vport_wfq_en)
  * the specified VOQ.
  */
 static void qed_cmdq_lines_voq_rt_init(struct qed_hwfn *p_hwfn,
-				       u8 voq, u16 cmdq_lines)
+				       u8 ext_voq, u16 cmdq_lines)
 {
-	u32 qm_line_crd;
+	u32 qm_line_crd = QM_VOQ_LINE_CRD(cmdq_lines);
 
-	qm_line_crd = QM_VOQ_LINE_CRD(cmdq_lines);
-	OVERWRITE_RT_REG(p_hwfn, PBF_CMDQ_LINES_RT_OFFSET(voq),
+	OVERWRITE_RT_REG(p_hwfn, PBF_CMDQ_LINES_RT_OFFSET(ext_voq),
 			 (u32)cmdq_lines);
-	STORE_RT_REG(p_hwfn, QM_REG_VOQCRDLINE_RT_OFFSET + voq, qm_line_crd);
-	STORE_RT_REG(p_hwfn, QM_REG_VOQINITCRDLINE_RT_OFFSET + voq,
+	STORE_RT_REG(p_hwfn, QM_REG_VOQCRDLINE_RT_OFFSET + ext_voq,
+		     qm_line_crd);
+	STORE_RT_REG(p_hwfn, QM_REG_VOQINITCRDLINE_RT_OFFSET + ext_voq,
 		     qm_line_crd);
 }
 
@@ -221,43 +317,52 @@ static void qed_cmdq_lines_rt_init(
 	u8 max_phys_tcs_per_port,
 	struct init_qm_port_params port_params[MAX_NUM_PORTS])
 {
-	u8 tc, voq, port_id, num_tcs_in_port;
+	u8 tc, ext_voq, port_id, num_tcs_in_port;
+	u8 num_ext_voqs = MAX_NUM_VOQS_E4;
+
+	/* Clear PBF lines of all VOQs */
+	for (ext_voq = 0; ext_voq < num_ext_voqs; ext_voq++)
+		STORE_RT_REG(p_hwfn, PBF_CMDQ_LINES_RT_OFFSET(ext_voq), 0);
 
-	/* Clear PBF lines for all VOQs */
-	for (voq = 0; voq < MAX_NUM_VOQS; voq++)
-		STORE_RT_REG(p_hwfn, PBF_CMDQ_LINES_RT_OFFSET(voq), 0);
 	for (port_id = 0; port_id < max_ports_per_engine; port_id++) {
-		if (port_params[port_id].active) {
-			u16 phys_lines, phys_lines_per_tc;
-
-			/* find #lines to divide between active phys TCs */
-			phys_lines = port_params[port_id].num_pbf_cmd_lines -
-				     PBF_CMDQ_PURE_LB_LINES;
-			/* find #lines per active physical TC */
-			num_tcs_in_port = 0;
-			for (tc = 0; tc < NUM_OF_PHYS_TCS; tc++) {
-				if (((port_params[port_id].active_phys_tcs >>
-				      tc) & 0x1) == 1)
-					num_tcs_in_port++;
-			}
+		u16 phys_lines, phys_lines_per_tc;
 
-			phys_lines_per_tc = phys_lines / num_tcs_in_port;
-			/* init registers per active TC */
-			for (tc = 0; tc < NUM_OF_PHYS_TCS; tc++) {
-				if (((port_params[port_id].active_phys_tcs >>
-				      tc) & 0x1) != 1)
-					continue;
+		if (!port_params[port_id].active)
+			continue;
 
-				voq = PHYS_VOQ(port_id, tc,
-					       max_phys_tcs_per_port);
-				qed_cmdq_lines_voq_rt_init(p_hwfn, voq,
-							   phys_lines_per_tc);
-			}
+		/* Find number of command queue lines to divide between the
+		 * active physical TCs. In E5, 1/8 of the lines are reserved.
+		 * the lines for pure LB TC are subtracted.
+		 */
+		phys_lines = port_params[port_id].num_pbf_cmd_lines;
+		phys_lines -= PBF_CMDQ_PURE_LB_LINES;
+
+		/* Find #lines per active physical TC */
+		num_tcs_in_port = 0;
+		for (tc = 0; tc < max_phys_tcs_per_port; tc++)
+			if (((port_params[port_id].active_phys_tcs >>
+			      tc) & 0x1) == 1)
+				num_tcs_in_port++;
+		phys_lines_per_tc = phys_lines / num_tcs_in_port;
 
-			/* init registers for pure LB TC */
-			qed_cmdq_lines_voq_rt_init(p_hwfn, LB_VOQ(port_id),
-						   PBF_CMDQ_PURE_LB_LINES);
+		/* Init registers per active TC */
+		for (tc = 0; tc < max_phys_tcs_per_port; tc++) {
+			ext_voq = qed_get_ext_voq(p_hwfn,
+						  port_id,
+						  tc, max_phys_tcs_per_port);
+			if (((port_params[port_id].active_phys_tcs >>
+			      tc) & 0x1) == 1)
+				qed_cmdq_lines_voq_rt_init(p_hwfn,
+							   ext_voq,
+							   phys_lines_per_tc);
 		}
+
+		/* Init registers for pure LB TC */
+		ext_voq = qed_get_ext_voq(p_hwfn,
+					  port_id,
+					  PURE_LB_TC, max_phys_tcs_per_port);
+		qed_cmdq_lines_voq_rt_init(p_hwfn,
+					   ext_voq, PBF_CMDQ_PURE_LB_LINES);
 	}
 }
 
@@ -268,11 +373,9 @@ static void qed_btb_blocks_rt_init(
 	struct init_qm_port_params port_params[MAX_NUM_PORTS])
 {
 	u32 usable_blocks, pure_lb_blocks, phys_blocks;
-	u8 tc, voq, port_id, num_tcs_in_port;
+	u8 tc, ext_voq, port_id, num_tcs_in_port;
 
 	for (port_id = 0; port_id < max_ports_per_engine; port_id++) {
-		u32 temp;
-
 		if (!port_params[port_id].active)
 			continue;
 
@@ -280,13 +383,14 @@ static void qed_btb_blocks_rt_init(
 		usable_blocks = port_params[port_id].num_btb_blocks -
 				BTB_HEADROOM_BLOCKS;
 
-		/* find blocks per physical TC */
+		/* Find blocks per physical TC. Use factor to avoid floating
+		 * arithmethic.
+		 */
 		num_tcs_in_port = 0;
-		for (tc = 0; tc < NUM_OF_PHYS_TCS; tc++) {
+		for (tc = 0; tc < NUM_OF_PHYS_TCS; tc++)
 			if (((port_params[port_id].active_phys_tcs >>
 			      tc) & 0x1) == 1)
 				num_tcs_in_port++;
-		}
 
 		pure_lb_blocks = (usable_blocks * BTB_PURE_LB_FACTOR) /
 				 (num_tcs_in_port * BTB_PURE_LB_FACTOR +
@@ -299,47 +403,55 @@ static void qed_btb_blocks_rt_init(
 		/* Init physical TCs */
 		for (tc = 0; tc < NUM_OF_PHYS_TCS; tc++) {
 			if (((port_params[port_id].active_phys_tcs >>
-			      tc) & 0x1) != 1)
-				continue;
-
-			voq = PHYS_VOQ(port_id, tc,
-				       max_phys_tcs_per_port);
-			STORE_RT_REG(p_hwfn, PBF_BTB_GUARANTEED_RT_OFFSET(voq),
-				     phys_blocks);
+			      tc) & 0x1) == 1) {
+				ext_voq =
+					qed_get_ext_voq(p_hwfn,
+							port_id,
+							tc,
+							max_phys_tcs_per_port);
+				STORE_RT_REG(p_hwfn,
+					     PBF_BTB_GUARANTEED_RT_OFFSET
+					     (ext_voq), phys_blocks);
+			}
 		}
 
 		/* Init pure LB TC */
-		temp = LB_VOQ(port_id);
-		STORE_RT_REG(p_hwfn, PBF_BTB_GUARANTEED_RT_OFFSET(temp),
+		ext_voq = qed_get_ext_voq(p_hwfn,
+					  port_id,
+					  PURE_LB_TC, max_phys_tcs_per_port);
+		STORE_RT_REG(p_hwfn, PBF_BTB_GUARANTEED_RT_OFFSET(ext_voq),
 			     pure_lb_blocks);
 	}
 }
 
 /* Prepare Tx PQ mapping runtime init values for the specified PF */
-static void qed_tx_pq_map_rt_init(
-	struct qed_hwfn *p_hwfn,
-	struct qed_ptt *p_ptt,
-	struct qed_qm_pf_rt_init_params *p_params,
-	u32 base_mem_addr_4kb)
+static void qed_tx_pq_map_rt_init(struct qed_hwfn *p_hwfn,
+				  struct qed_ptt *p_ptt,
+				  struct qed_qm_pf_rt_init_params *p_params,
+				  u32 base_mem_addr_4kb)
 {
-	struct init_qm_vport_params *vport_params = p_params->vport_params;
-	u16 num_pqs = p_params->num_pf_pqs + p_params->num_vf_pqs;
-	u16 first_pq_group = p_params->start_pq / QM_PF_QUEUE_GROUP_SIZE;
-	u16 last_pq_group = (p_params->start_pq + num_pqs - 1) /
-			    QM_PF_QUEUE_GROUP_SIZE;
-	u16 i, pq_id, pq_group;
-
-	/* A bit per Tx PQ indicating if the PQ is associated with a VF */
 	u32 tx_pq_vf_mask[MAX_QM_TX_QUEUES / QM_PF_QUEUE_GROUP_SIZE] = { 0 };
+	struct init_qm_vport_params *vport_params = p_params->vport_params;
 	u32 num_tx_pq_vf_masks = MAX_QM_TX_QUEUES / QM_PF_QUEUE_GROUP_SIZE;
-	u32 pq_mem_4kb = QM_PQ_MEM_4KB(p_params->num_pf_cids);
-	u32 vport_pq_mem_4kb = QM_PQ_MEM_4KB(p_params->num_vf_cids);
-	u32 mem_addr_4kb = base_mem_addr_4kb;
+	u16 num_pqs, first_pq_group, last_pq_group, i, j, pq_id, pq_group;
+	struct init_qm_pq_params *pq_params = p_params->pq_params;
+	u32 pq_mem_4kb, vport_pq_mem_4kb, mem_addr_4kb;
+
+	num_pqs = p_params->num_pf_pqs + p_params->num_vf_pqs;
+
+	first_pq_group = p_params->start_pq / QM_PF_QUEUE_GROUP_SIZE;
+	last_pq_group = (p_params->start_pq + num_pqs - 1) /
+			QM_PF_QUEUE_GROUP_SIZE;
+
+	pq_mem_4kb = QM_PQ_MEM_4KB(p_params->num_pf_cids);
+	vport_pq_mem_4kb = QM_PQ_MEM_4KB(p_params->num_vf_cids);
+	mem_addr_4kb = base_mem_addr_4kb;
 
 	/* Set mapping from PQ group to PF */
 	for (pq_group = first_pq_group; pq_group <= last_pq_group; pq_group++)
 		STORE_RT_REG(p_hwfn, QM_REG_PQTX2PF_0_RT_OFFSET + pq_group,
 			     (u32)(p_params->pf_id));
+
 	/* Set PQ sizes */
 	STORE_RT_REG(p_hwfn, QM_REG_MAXPQSIZE_0_RT_OFFSET,
 		     QM_PQ_SIZE_256B(p_params->num_pf_cids));
@@ -348,58 +460,82 @@ static void qed_tx_pq_map_rt_init(
 
 	/* Go over all Tx PQs */
 	for (i = 0, pq_id = p_params->start_pq; i < num_pqs; i++, pq_id++) {
-		u8 voq = VOQ(p_params->port_id, p_params->pq_params[i].tc_id,
-			     p_params->max_phys_tcs_per_port);
-		bool is_vf_pq = (i >= p_params->num_pf_pqs);
-		struct qm_rf_pq_map tx_pq_map;
-
-		bool rl_valid = p_params->pq_params[i].rl_valid &&
-				(p_params->pq_params[i].vport_id <
-				 MAX_QM_GLOBAL_RLS);
+		u8 ext_voq, vport_id_in_pf, tc_id = pq_params[i].tc_id;
+		u32 max_qm_global_rls = MAX_QM_GLOBAL_RLS;
+		struct qm_rf_pq_map_e4 tx_pq_map;
+		bool is_vf_pq, rl_valid;
+		u16 *p_first_tx_pq_id;
+
+		ext_voq = qed_get_ext_voq(p_hwfn,
+					  p_params->port_id,
+					  tc_id,
+					  p_params->max_phys_tcs_per_port);
+		is_vf_pq = (i >= p_params->num_pf_pqs);
+		rl_valid = pq_params[i].rl_valid &&
+			   pq_params[i].vport_id < max_qm_global_rls;
 
 		/* Update first Tx PQ of VPORT/TC */
-		u8 vport_id_in_pf = p_params->pq_params[i].vport_id -
-				    p_params->start_vport;
-		u16 *pq_ids = &vport_params[vport_id_in_pf].first_tx_pq_id[0];
-		u16 first_tx_pq_id = pq_ids[p_params->pq_params[i].tc_id];
+		vport_id_in_pf = pq_params[i].vport_id - p_params->start_vport;
+		p_first_tx_pq_id =
+		    &vport_params[vport_id_in_pf].first_tx_pq_id[tc_id];
+		if (*p_first_tx_pq_id == QM_INVALID_PQ_ID) {
+			u32 map_val =
+				(ext_voq << QM_WFQ_VP_PQ_VOQ_SHIFT) |
+				(p_params->pf_id << QM_WFQ_VP_PQ_PF_E4_SHIFT);
 
-		if (first_tx_pq_id == QM_INVALID_PQ_ID) {
 			/* Create new VP PQ */
-			pq_ids[p_params->pq_params[i].tc_id] = pq_id;
-			first_tx_pq_id = pq_id;
+			*p_first_tx_pq_id = pq_id;
 
 			/* Map VP PQ to VOQ and PF */
 			STORE_RT_REG(p_hwfn,
 				     QM_REG_WFQVPMAP_RT_OFFSET +
-				     first_tx_pq_id,
-				     (voq << QM_WFQ_VP_PQ_VOQ_SHIFT) |
-				     (p_params->pf_id <<
-				      QM_WFQ_VP_PQ_PF_SHIFT));
+				     *p_first_tx_pq_id,
+				     map_val);
 		}
 
-		if (p_params->pq_params[i].rl_valid && !rl_valid)
+		/* Check RL ID */
+		if (pq_params[i].rl_valid && pq_params[i].vport_id >=
+		    max_qm_global_rls)
 			DP_NOTICE(p_hwfn,
-				  "Invalid VPORT ID for rate limiter configuration");
-		/* Fill PQ map entry */
-		memset(&tx_pq_map, 0, sizeof(tx_pq_map));
-		SET_FIELD(tx_pq_map.reg, QM_RF_PQ_MAP_PQ_VALID, 1);
-		SET_FIELD(tx_pq_map.reg,
-			  QM_RF_PQ_MAP_RL_VALID, rl_valid ? 1 : 0);
-		SET_FIELD(tx_pq_map.reg, QM_RF_PQ_MAP_VP_PQ_ID, first_tx_pq_id);
-		SET_FIELD(tx_pq_map.reg, QM_RF_PQ_MAP_RL_ID,
-			  rl_valid ?
-			  p_params->pq_params[i].vport_id : 0);
-		SET_FIELD(tx_pq_map.reg, QM_RF_PQ_MAP_VOQ, voq);
-		SET_FIELD(tx_pq_map.reg, QM_RF_PQ_MAP_WRR_WEIGHT_GROUP,
-			  p_params->pq_params[i].wrr_group);
-		/* Write PQ map entry to CAM */
-		STORE_RT_REG(p_hwfn, QM_REG_TXPQMAP_RT_OFFSET + pq_id,
-			     *((u32 *)&tx_pq_map));
-		/* Set base address */
+				  "Invalid VPORT ID for rate limiter configuration\n");
+
+		/* Prepare PQ map entry */
+		QM_INIT_TX_PQ_MAP(p_hwfn,
+				  tx_pq_map,
+				  E4,
+				  pq_id,
+				  rl_valid ? 1 : 0,
+				  *p_first_tx_pq_id,
+				  rl_valid ? pq_params[i].vport_id : 0,
+				  ext_voq, pq_params[i].wrr_group);
+
+		/* Set PQ base address */
 		STORE_RT_REG(p_hwfn,
 			     QM_REG_BASEADDRTXPQ_RT_OFFSET + pq_id,
 			     mem_addr_4kb);
 
+		/* Clear PQ pointer table entry (64 bit) */
+		if (p_params->is_pf_loading)
+			for (j = 0; j < 2; j++)
+				STORE_RT_REG(p_hwfn,
+					     QM_REG_PTRTBLTX_RT_OFFSET +
+					     (pq_id * 2) + j, 0);
+
+		/* Write PQ info to RAM */
+		if (WRITE_PQ_INFO_TO_RAM != 0) {
+			u32 pq_info = 0;
+
+			pq_info = PQ_INFO_ELEMENT(*p_first_tx_pq_id,
+						  p_params->pf_id,
+						  tc_id,
+						  p_params->port_id,
+						  rl_valid ? 1 : 0,
+						  rl_valid ?
+						  pq_params[i].vport_id : 0);
+			qed_wr(p_hwfn, p_ptt, PQ_INFO_RAM_GRC_ADDRESS(pq_id),
+			       pq_info);
+		}
+
 		/* If VF PQ, add indication to PQ VF mask */
 		if (is_vf_pq) {
 			tx_pq_vf_mask[pq_id /
@@ -421,16 +557,16 @@ static void qed_tx_pq_map_rt_init(
 
 /* Prepare Other PQ mapping runtime init values for the specified PF */
 static void qed_other_pq_map_rt_init(struct qed_hwfn *p_hwfn,
-				     u8 port_id,
 				     u8 pf_id,
+				     bool is_pf_loading,
 				     u32 num_pf_cids,
 				     u32 num_tids, u32 base_mem_addr_4kb)
 {
 	u32 pq_size, pq_mem_4kb, mem_addr_4kb;
-	u16 i, pq_id, pq_group;
+	u16 i, j, pq_id, pq_group;
 
-	/* a single other PQ group is used in each PF,
-	 * where PQ group i is used in PF i.
+	/* A single other PQ group is used in each PF, where PQ group i is used
+	 * in PF i.
 	 */
 	pq_group = pf_id;
 	pq_size = num_pf_cids + num_tids;
@@ -440,16 +576,25 @@ static void qed_other_pq_map_rt_init(struct qed_hwfn *p_hwfn,
 	/* Map PQ group to PF */
 	STORE_RT_REG(p_hwfn, QM_REG_PQOTHER2PF_0_RT_OFFSET + pq_group,
 		     (u32)(pf_id));
+
 	/* Set PQ sizes */
 	STORE_RT_REG(p_hwfn, QM_REG_MAXPQSIZE_2_RT_OFFSET,
 		     QM_PQ_SIZE_256B(pq_size));
 
-	/* Set base address */
 	for (i = 0, pq_id = pf_id * QM_PF_QUEUE_GROUP_SIZE;
 	     i < QM_OTHER_PQS_PER_PF; i++, pq_id++) {
+		/* Set PQ base address */
 		STORE_RT_REG(p_hwfn,
 			     QM_REG_BASEADDROTHERPQ_RT_OFFSET + pq_id,
 			     mem_addr_4kb);
+
+		/* Clear PQ pointer table entry */
+		if (is_pf_loading)
+			for (j = 0; j < 2; j++)
+				STORE_RT_REG(p_hwfn,
+					     QM_REG_PTRTBLOTHER_RT_OFFSET +
+					     (pq_id * 2) + j, 0);
+
 		mem_addr_4kb += pq_mem_4kb;
 	}
 }
@@ -461,16 +606,11 @@ static int qed_pf_wfq_rt_init(struct qed_hwfn *p_hwfn,
 			      struct qed_qm_pf_rt_init_params *p_params)
 {
 	u16 num_tx_pqs = p_params->num_pf_pqs + p_params->num_vf_pqs;
-	u32 crd_reg_offset;
-	u32 inc_val;
+	struct init_qm_pq_params *pq_params = p_params->pq_params;
+	u32 inc_val, crd_reg_offset;
+	u8 ext_voq;
 	u16 i;
 
-	if (p_params->pf_id < MAX_NUM_PFS_BB)
-		crd_reg_offset = QM_REG_WFQPFCRD_RT_OFFSET;
-	else
-		crd_reg_offset = QM_REG_WFQPFCRD_MSB_RT_OFFSET;
-	crd_reg_offset += p_params->pf_id % MAX_NUM_PFS_BB;
-
 	inc_val = QM_WFQ_INC_VAL(p_params->pf_wfq);
 	if (!inc_val || inc_val > QM_WFQ_MAX_INC_VAL) {
 		DP_NOTICE(p_hwfn, "Invalid PF WFQ weight configuration\n");
@@ -478,19 +618,26 @@ static int qed_pf_wfq_rt_init(struct qed_hwfn *p_hwfn,
 	}
 
 	for (i = 0; i < num_tx_pqs; i++) {
-		u8 voq = VOQ(p_params->port_id, p_params->pq_params[i].tc_id,
-			     p_params->max_phys_tcs_per_port);
-
+		ext_voq = qed_get_ext_voq(p_hwfn,
+					  p_params->port_id,
+					  pq_params[i].tc_id,
+					  p_params->max_phys_tcs_per_port);
+		crd_reg_offset =
+			(p_params->pf_id < MAX_NUM_PFS_BB ?
+			 QM_REG_WFQPFCRD_RT_OFFSET :
+			 QM_REG_WFQPFCRD_MSB_RT_OFFSET) +
+			ext_voq * MAX_NUM_PFS_BB +
+			(p_params->pf_id % MAX_NUM_PFS_BB);
 		OVERWRITE_RT_REG(p_hwfn,
-				 crd_reg_offset + voq * MAX_NUM_PFS_BB,
-				 QM_WFQ_CRD_REG_SIGN_BIT);
+				 crd_reg_offset, (u32)QM_WFQ_CRD_REG_SIGN_BIT);
 	}
 
 	STORE_RT_REG(p_hwfn,
 		     QM_REG_WFQPFUPPERBOUND_RT_OFFSET + p_params->pf_id,
-		     QM_WFQ_UPPER_BOUND | QM_WFQ_CRD_REG_SIGN_BIT);
+		     QM_WFQ_UPPER_BOUND | (u32)QM_WFQ_CRD_REG_SIGN_BIT);
 	STORE_RT_REG(p_hwfn, QM_REG_WFQPFWEIGHT_RT_OFFSET + p_params->pf_id,
 		     inc_val);
+
 	return 0;
 }
 
@@ -501,15 +648,19 @@ static int qed_pf_rl_rt_init(struct qed_hwfn *p_hwfn, u8 pf_id, u32 pf_rl)
 {
 	u32 inc_val = QM_RL_INC_VAL(pf_rl);
 
-	if (inc_val > QM_RL_MAX_INC_VAL) {
+	if (inc_val > QM_PF_RL_MAX_INC_VAL) {
 		DP_NOTICE(p_hwfn, "Invalid PF rate limit configuration\n");
 		return -1;
 	}
-	STORE_RT_REG(p_hwfn, QM_REG_RLPFCRD_RT_OFFSET + pf_id,
-		     QM_RL_CRD_REG_SIGN_BIT);
-	STORE_RT_REG(p_hwfn, QM_REG_RLPFUPPERBOUND_RT_OFFSET + pf_id,
-		     QM_RL_UPPER_BOUND | QM_RL_CRD_REG_SIGN_BIT);
+
+	STORE_RT_REG(p_hwfn,
+		     QM_REG_RLPFCRD_RT_OFFSET + pf_id,
+		     (u32)QM_RL_CRD_REG_SIGN_BIT);
+	STORE_RT_REG(p_hwfn,
+		     QM_REG_RLPFUPPERBOUND_RT_OFFSET + pf_id,
+		     QM_PF_RL_UPPER_BOUND | (u32)QM_RL_CRD_REG_SIGN_BIT);
 	STORE_RT_REG(p_hwfn, QM_REG_RLPFINCVAL_RT_OFFSET + pf_id, inc_val);
+
 	return 0;
 }
 
@@ -520,12 +671,12 @@ static int qed_vp_wfq_rt_init(struct qed_hwfn *p_hwfn,
 			      u8 num_vports,
 			      struct init_qm_vport_params *vport_params)
 {
+	u16 vport_pq_id;
 	u32 inc_val;
 	u8 tc, i;
 
 	/* Go over all PF VPORTs */
 	for (i = 0; i < num_vports; i++) {
-
 		if (!vport_params[i].vport_wfq)
 			continue;
 
@@ -536,17 +687,14 @@ static int qed_vp_wfq_rt_init(struct qed_hwfn *p_hwfn,
 			return -1;
 		}
 
-		/* each VPORT can have several VPORT PQ IDs for
-		 * different TCs
-		 */
+		/* Each VPORT can have several VPORT PQ IDs for various TCs */
 		for (tc = 0; tc < NUM_OF_TCS; tc++) {
-			u16 vport_pq_id = vport_params[i].first_tx_pq_id[tc];
-
+			vport_pq_id = vport_params[i].first_tx_pq_id[tc];
 			if (vport_pq_id != QM_INVALID_PQ_ID) {
 				STORE_RT_REG(p_hwfn,
 					     QM_REG_WFQVPCRD_RT_OFFSET +
 					     vport_pq_id,
-					     QM_WFQ_CRD_REG_SIGN_BIT);
+					     (u32)QM_WFQ_CRD_REG_SIGN_BIT);
 				STORE_RT_REG(p_hwfn,
 					     QM_REG_WFQVPWEIGHT_RT_OFFSET +
 					     vport_pq_id, inc_val);
@@ -557,12 +705,17 @@ static int qed_vp_wfq_rt_init(struct qed_hwfn *p_hwfn,
 	return 0;
 }
 
+/* Prepare VPORT RL runtime init values for the specified VPORTs.
+ * Return -1 on error.
+ */
 static int qed_vport_rl_rt_init(struct qed_hwfn *p_hwfn,
 				u8 start_vport,
 				u8 num_vports,
+				u32 link_speed,
 				struct init_qm_vport_params *vport_params)
 {
 	u8 i, vport_id;
+	u32 inc_val;
 
 	if (start_vport + num_vports >= MAX_QM_GLOBAL_RLS) {
 		DP_NOTICE(p_hwfn,
@@ -572,22 +725,22 @@ static int qed_vport_rl_rt_init(struct qed_hwfn *p_hwfn,
 
 	/* Go over all PF VPORTs */
 	for (i = 0, vport_id = start_vport; i < num_vports; i++, vport_id++) {
-		u32 inc_val = QM_RL_INC_VAL(vport_params[i].vport_rl);
-
-		if (inc_val > QM_RL_MAX_INC_VAL) {
+		inc_val = QM_RL_INC_VAL(vport_params[i].vport_rl ?
+			  vport_params[i].vport_rl :
+			  link_speed);
+		if (inc_val > QM_VP_RL_MAX_INC_VAL(link_speed)) {
 			DP_NOTICE(p_hwfn,
 				  "Invalid VPORT rate-limit configuration\n");
 			return -1;
 		}
 
-		STORE_RT_REG(p_hwfn,
-			     QM_REG_RLGLBLCRD_RT_OFFSET + vport_id,
-			     QM_RL_CRD_REG_SIGN_BIT);
+		STORE_RT_REG(p_hwfn, QM_REG_RLGLBLCRD_RT_OFFSET + vport_id,
+			     (u32)QM_RL_CRD_REG_SIGN_BIT);
 		STORE_RT_REG(p_hwfn,
 			     QM_REG_RLGLBLUPPERBOUND_RT_OFFSET + vport_id,
-			     QM_RL_UPPER_BOUND | QM_RL_CRD_REG_SIGN_BIT);
-		STORE_RT_REG(p_hwfn,
-			     QM_REG_RLGLBLINCVAL_RT_OFFSET + vport_id,
+			     QM_VP_RL_UPPER_BOUND(link_speed) |
+			     (u32)QM_RL_CRD_REG_SIGN_BIT);
+		STORE_RT_REG(p_hwfn, QM_REG_RLGLBLINCVAL_RT_OFFSET + vport_id,
 			     inc_val);
 	}
 
@@ -599,7 +752,7 @@ static bool qed_poll_on_qm_cmd_ready(struct qed_hwfn *p_hwfn,
 {
 	u32 reg_val, i;
 
-	for (i = 0, reg_val = 0; i < QM_STOP_CMD_MAX_POLL_COUNT && reg_val == 0;
+	for (i = 0, reg_val = 0; i < QM_STOP_CMD_MAX_POLL_COUNT && !reg_val;
 	     i++) {
 		udelay(QM_STOP_CMD_POLL_PERIOD_US);
 		reg_val = qed_rd(p_hwfn, p_ptt, QM_REG_SDMCMDREADY);
@@ -632,8 +785,8 @@ static bool qed_send_qm_cmd(struct qed_hwfn *p_hwfn,
 }
 
 /******************** INTERFACE IMPLEMENTATION *********************/
-u32 qed_qm_pf_mem_size(u8 pf_id,
-		       u32 num_pf_cids,
+
+u32 qed_qm_pf_mem_size(u32 num_pf_cids,
 		       u32 num_vf_cids,
 		       u32 num_tids, u16 num_pf_pqs, u16 num_vf_pqs)
 {
@@ -642,11 +795,10 @@ u32 qed_qm_pf_mem_size(u8 pf_id,
 	       QM_PQ_MEM_4KB(num_pf_cids + num_tids) * QM_OTHER_PQS_PER_PF;
 }
 
-int qed_qm_common_rt_init(
-	struct qed_hwfn *p_hwfn,
-	struct qed_qm_common_rt_init_params *p_params)
+int qed_qm_common_rt_init(struct qed_hwfn *p_hwfn,
+			  struct qed_qm_common_rt_init_params *p_params)
 {
-	/* init AFullOprtnstcCrdMask */
+	/* Init AFullOprtnstcCrdMask */
 	u32 mask = (QM_OPPOR_LINE_VOQ_DEF <<
 		    QM_RF_OPPORTUNISTIC_MASK_LINEVOQ_SHIFT) |
 		   (QM_BYTE_CRD_EN << QM_RF_OPPORTUNISTIC_MASK_BYTEVOQ_SHIFT) |
@@ -664,18 +816,31 @@ int qed_qm_common_rt_init(
 		    QM_RF_OPPORTUNISTIC_MASK_QUEUEEMPTY_SHIFT);
 
 	STORE_RT_REG(p_hwfn, QM_REG_AFULLOPRTNSTCCRDMASK_RT_OFFSET, mask);
+
+	/* Enable/disable PF RL */
 	qed_enable_pf_rl(p_hwfn, p_params->pf_rl_en);
+
+	/* Enable/disable PF WFQ */
 	qed_enable_pf_wfq(p_hwfn, p_params->pf_wfq_en);
+
+	/* Enable/disable VPORT RL */
 	qed_enable_vport_rl(p_hwfn, p_params->vport_rl_en);
+
+	/* Enable/disable VPORT WFQ */
 	qed_enable_vport_wfq(p_hwfn, p_params->vport_wfq_en);
+
+	/* Init PBF CMDQ line credit */
 	qed_cmdq_lines_rt_init(p_hwfn,
 			       p_params->max_ports_per_engine,
 			       p_params->max_phys_tcs_per_port,
 			       p_params->port_params);
+
+	/* Init BTB blocks in PBF */
 	qed_btb_blocks_rt_init(p_hwfn,
 			       p_params->max_ports_per_engine,
 			       p_params->max_phys_tcs_per_port,
 			       p_params->port_params);
+
 	return 0;
 }
 
@@ -695,24 +860,31 @@ int qed_qm_pf_rt_init(struct qed_hwfn *p_hwfn,
 			vport_params[i].first_tx_pq_id[tc] = QM_INVALID_PQ_ID;
 
 	/* Map Other PQs (if any) */
-	qed_other_pq_map_rt_init(p_hwfn, p_params->port_id, p_params->pf_id,
-				 p_params->num_pf_cids, p_params->num_tids, 0);
+	qed_other_pq_map_rt_init(p_hwfn,
+				 p_params->pf_id,
+				 p_params->is_pf_loading, p_params->num_pf_cids,
+				 p_params->num_tids, 0);
 
 	/* Map Tx PQs */
 	qed_tx_pq_map_rt_init(p_hwfn, p_ptt, p_params, other_mem_size_4kb);
 
+	/* Init PF WFQ */
 	if (p_params->pf_wfq)
 		if (qed_pf_wfq_rt_init(p_hwfn, p_params))
 			return -1;
 
+	/* Init PF RL */
 	if (qed_pf_rl_rt_init(p_hwfn, p_params->pf_id, p_params->pf_rl))
 		return -1;
 
+	/* Set VPORT WFQ */
 	if (qed_vp_wfq_rt_init(p_hwfn, p_params->num_vports, vport_params))
 		return -1;
 
+	/* Set VPORT RL */
 	if (qed_vport_rl_rt_init(p_hwfn, p_params->start_vport,
-				 p_params->num_vports, vport_params))
+				 p_params->num_vports, p_params->link_speed,
+				 vport_params))
 		return -1;
 
 	return 0;
@@ -729,6 +901,7 @@ int qed_init_pf_wfq(struct qed_hwfn *p_hwfn,
 	}
 
 	qed_wr(p_hwfn, p_ptt, QM_REG_WFQPFWEIGHT + pf_id * 4, inc_val);
+
 	return 0;
 }
 
@@ -737,14 +910,13 @@ int qed_init_pf_rl(struct qed_hwfn *p_hwfn,
 {
 	u32 inc_val = QM_RL_INC_VAL(pf_rl);
 
-	if (inc_val > QM_RL_MAX_INC_VAL) {
+	if (inc_val > QM_PF_RL_MAX_INC_VAL) {
 		DP_NOTICE(p_hwfn, "Invalid PF rate limit configuration\n");
 		return -1;
 	}
 
-	qed_wr(p_hwfn, p_ptt,
-	       QM_REG_RLPFCRD + pf_id * 4,
-	       QM_RL_CRD_REG_SIGN_BIT);
+	qed_wr(p_hwfn,
+	       p_ptt, QM_REG_RLPFCRD + pf_id * 4, (u32)QM_RL_CRD_REG_SIGN_BIT);
 	qed_wr(p_hwfn, p_ptt, QM_REG_RLPFINCVAL + pf_id * 4, inc_val);
 
 	return 0;
@@ -767,33 +939,35 @@ int qed_init_vport_wfq(struct qed_hwfn *p_hwfn,
 	for (tc = 0; tc < NUM_OF_TCS; tc++) {
 		vport_pq_id = first_tx_pq_id[tc];
 		if (vport_pq_id != QM_INVALID_PQ_ID)
-			qed_wr(p_hwfn, p_ptt,
-			       QM_REG_WFQVPWEIGHT + vport_pq_id * 4,
-			       inc_val);
+			qed_wr(p_hwfn,
+			       p_ptt,
+			       QM_REG_WFQVPWEIGHT + vport_pq_id * 4, inc_val);
 	}
 
 	return 0;
 }
 
 int qed_init_vport_rl(struct qed_hwfn *p_hwfn,
-		      struct qed_ptt *p_ptt, u8 vport_id, u32 vport_rl)
+		      struct qed_ptt *p_ptt,
+		      u8 vport_id, u32 vport_rl, u32 link_speed)
 {
-	u32 inc_val = QM_RL_INC_VAL(vport_rl);
+	u32 inc_val, max_qm_global_rls = MAX_QM_GLOBAL_RLS;
 
-	if (vport_id >= MAX_QM_GLOBAL_RLS) {
+	if (vport_id >= max_qm_global_rls) {
 		DP_NOTICE(p_hwfn,
 			  "Invalid VPORT ID for rate limiter configuration\n");
 		return -1;
 	}
 
-	if (inc_val > QM_RL_MAX_INC_VAL) {
+	inc_val = QM_RL_INC_VAL(vport_rl ? vport_rl : link_speed);
+	if (inc_val > QM_VP_RL_MAX_INC_VAL(link_speed)) {
 		DP_NOTICE(p_hwfn, "Invalid VPORT rate-limit configuration\n");
 		return -1;
 	}
 
-	qed_wr(p_hwfn, p_ptt,
-	       QM_REG_RLGLBLCRD + vport_id * 4,
-	       QM_RL_CRD_REG_SIGN_BIT);
+	qed_wr(p_hwfn,
+	       p_ptt,
+	       QM_REG_RLGLBLCRD + vport_id * 4, (u32)QM_RL_CRD_REG_SIGN_BIT);
 	qed_wr(p_hwfn, p_ptt, QM_REG_RLGLBLINCVAL + vport_id * 4, inc_val);
 
 	return 0;
@@ -805,23 +979,27 @@ bool qed_send_qm_stop_cmd(struct qed_hwfn *p_hwfn,
 			  bool is_tx_pq, u16 start_pq, u16 num_pqs)
 {
 	u32 cmd_arr[QM_CMD_STRUCT_SIZE(QM_STOP_CMD)] = { 0 };
-	u32 pq_mask = 0, last_pq = start_pq + num_pqs - 1, pq_id;
+	u32 pq_mask = 0, last_pq, pq_id;
+
+	last_pq = start_pq + num_pqs - 1;
 
 	/* Set command's PQ type */
 	QM_CMD_SET_FIELD(cmd_arr, QM_STOP_CMD, PQ_TYPE, is_tx_pq ? 0 : 1);
 
+	/* Go over requested PQs */
 	for (pq_id = start_pq; pq_id <= last_pq; pq_id++) {
 		/* Set PQ bit in mask (stop command only) */
 		if (!is_release_cmd)
-			pq_mask |= (1 << (pq_id % QM_STOP_PQ_MASK_WIDTH));
+			pq_mask |= BIT((pq_id % QM_STOP_PQ_MASK_WIDTH));
 
 		/* If last PQ or end of PQ mask, write command */
 		if ((pq_id == last_pq) ||
 		    (pq_id % QM_STOP_PQ_MASK_WIDTH ==
 		     (QM_STOP_PQ_MASK_WIDTH - 1))) {
-			QM_CMD_SET_FIELD(cmd_arr, QM_STOP_CMD,
-					 PAUSE_MASK, pq_mask);
-			QM_CMD_SET_FIELD(cmd_arr, QM_STOP_CMD,
+			QM_CMD_SET_FIELD(cmd_arr,
+					 QM_STOP_CMD, PAUSE_MASK, pq_mask);
+			QM_CMD_SET_FIELD(cmd_arr,
+					 QM_STOP_CMD,
 					 GROUP_ID,
 					 pq_id / QM_STOP_PQ_MASK_WIDTH);
 			if (!qed_send_qm_cmd(p_hwfn, p_ptt, QM_STOP_CMD_ADDR,
@@ -834,87 +1012,103 @@ bool qed_send_qm_stop_cmd(struct qed_hwfn *p_hwfn,
 	return true;
 }
 
-static void
-qed_set_tunnel_type_enable_bit(unsigned long *var, int bit, bool enable)
-{
-	if (enable)
-		set_bit(bit, var);
-	else
-		clear_bit(bit, var);
-}
 
+#define SET_TUNNEL_TYPE_ENABLE_BIT(var, offset, enable) \
+	do { \
+		typeof(var) *__p_var = &(var); \
+		typeof(offset) __offset = offset; \
+		*__p_var = (*__p_var & ~BIT(__offset)) | \
+			   ((enable) ? BIT(__offset) : 0); \
+	} while (0)
 #define PRS_ETH_TUNN_FIC_FORMAT	-188897008
 
 void qed_set_vxlan_dest_port(struct qed_hwfn *p_hwfn,
 			     struct qed_ptt *p_ptt, u16 dest_port)
 {
+	/* Update PRS register */
 	qed_wr(p_hwfn, p_ptt, PRS_REG_VXLAN_PORT, dest_port);
+
+	/* Update NIG register */
 	qed_wr(p_hwfn, p_ptt, NIG_REG_VXLAN_CTRL, dest_port);
+
+	/* Update PBF register */
 	qed_wr(p_hwfn, p_ptt, PBF_REG_VXLAN_PORT, dest_port);
 }
 
 void qed_set_vxlan_enable(struct qed_hwfn *p_hwfn,
 			  struct qed_ptt *p_ptt, bool vxlan_enable)
 {
-	unsigned long reg_val = 0;
+	u32 reg_val;
 	u8 shift;
 
+	/* Update PRS register */
 	reg_val = qed_rd(p_hwfn, p_ptt, PRS_REG_ENCAPSULATION_TYPE_EN);
 	shift = PRS_REG_ENCAPSULATION_TYPE_EN_VXLAN_ENABLE_SHIFT;
-	qed_set_tunnel_type_enable_bit(&reg_val, shift, vxlan_enable);
-
+	SET_TUNNEL_TYPE_ENABLE_BIT(reg_val, shift, vxlan_enable);
 	qed_wr(p_hwfn, p_ptt, PRS_REG_ENCAPSULATION_TYPE_EN, reg_val);
-
 	if (reg_val)
-		qed_wr(p_hwfn, p_ptt, PRS_REG_OUTPUT_FORMAT_4_0,
-		       PRS_ETH_TUNN_FIC_FORMAT);
+		qed_wr(p_hwfn,
+		       p_ptt,
+		       PRS_REG_OUTPUT_FORMAT_4_0_BB_K2,
+		       (u32)PRS_ETH_TUNN_FIC_FORMAT);
 
+	/* Update NIG register */
 	reg_val = qed_rd(p_hwfn, p_ptt, NIG_REG_ENC_TYPE_ENABLE);
 	shift = NIG_REG_ENC_TYPE_ENABLE_VXLAN_ENABLE_SHIFT;
-	qed_set_tunnel_type_enable_bit(&reg_val, shift, vxlan_enable);
-
+	SET_TUNNEL_TYPE_ENABLE_BIT(reg_val, shift, vxlan_enable);
 	qed_wr(p_hwfn, p_ptt, NIG_REG_ENC_TYPE_ENABLE, reg_val);
 
-	qed_wr(p_hwfn, p_ptt, DORQ_REG_L2_EDPM_TUNNEL_VXLAN_EN,
-	       vxlan_enable ? 1 : 0);
+	/* Update DORQ register */
+	qed_wr(p_hwfn,
+	       p_ptt, DORQ_REG_L2_EDPM_TUNNEL_VXLAN_EN, vxlan_enable ? 1 : 0);
 }
 
-void qed_set_gre_enable(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt,
+void qed_set_gre_enable(struct qed_hwfn *p_hwfn,
+			struct qed_ptt *p_ptt,
 			bool eth_gre_enable, bool ip_gre_enable)
 {
-	unsigned long reg_val = 0;
+	u32 reg_val;
 	u8 shift;
 
+	/* Update PRS register */
 	reg_val = qed_rd(p_hwfn, p_ptt, PRS_REG_ENCAPSULATION_TYPE_EN);
 	shift = PRS_REG_ENCAPSULATION_TYPE_EN_ETH_OVER_GRE_ENABLE_SHIFT;
-	qed_set_tunnel_type_enable_bit(&reg_val, shift, eth_gre_enable);
-
+	SET_TUNNEL_TYPE_ENABLE_BIT(reg_val, shift, eth_gre_enable);
 	shift = PRS_REG_ENCAPSULATION_TYPE_EN_IP_OVER_GRE_ENABLE_SHIFT;
-	qed_set_tunnel_type_enable_bit(&reg_val, shift, ip_gre_enable);
+	SET_TUNNEL_TYPE_ENABLE_BIT(reg_val, shift, ip_gre_enable);
 	qed_wr(p_hwfn, p_ptt, PRS_REG_ENCAPSULATION_TYPE_EN, reg_val);
 	if (reg_val)
-		qed_wr(p_hwfn, p_ptt, PRS_REG_OUTPUT_FORMAT_4_0,
-		       PRS_ETH_TUNN_FIC_FORMAT);
+		qed_wr(p_hwfn,
+		       p_ptt,
+		       PRS_REG_OUTPUT_FORMAT_4_0_BB_K2,
+		       (u32)PRS_ETH_TUNN_FIC_FORMAT);
 
+	/* Update NIG register */
 	reg_val = qed_rd(p_hwfn, p_ptt, NIG_REG_ENC_TYPE_ENABLE);
 	shift = NIG_REG_ENC_TYPE_ENABLE_ETH_OVER_GRE_ENABLE_SHIFT;
-	qed_set_tunnel_type_enable_bit(&reg_val, shift, eth_gre_enable);
-
+	SET_TUNNEL_TYPE_ENABLE_BIT(reg_val, shift, eth_gre_enable);
 	shift = NIG_REG_ENC_TYPE_ENABLE_IP_OVER_GRE_ENABLE_SHIFT;
-	qed_set_tunnel_type_enable_bit(&reg_val, shift, ip_gre_enable);
+	SET_TUNNEL_TYPE_ENABLE_BIT(reg_val, shift, ip_gre_enable);
 	qed_wr(p_hwfn, p_ptt, NIG_REG_ENC_TYPE_ENABLE, reg_val);
 
-	qed_wr(p_hwfn, p_ptt, DORQ_REG_L2_EDPM_TUNNEL_GRE_ETH_EN,
-	       eth_gre_enable ? 1 : 0);
-	qed_wr(p_hwfn, p_ptt, DORQ_REG_L2_EDPM_TUNNEL_GRE_IP_EN,
-	       ip_gre_enable ? 1 : 0);
+	/* Update DORQ registers */
+	qed_wr(p_hwfn,
+	       p_ptt,
+	       DORQ_REG_L2_EDPM_TUNNEL_GRE_ETH_EN, eth_gre_enable ? 1 : 0);
+	qed_wr(p_hwfn,
+	       p_ptt, DORQ_REG_L2_EDPM_TUNNEL_GRE_IP_EN, ip_gre_enable ? 1 : 0);
 }
 
 void qed_set_geneve_dest_port(struct qed_hwfn *p_hwfn,
 			      struct qed_ptt *p_ptt, u16 dest_port)
 {
+	/* Update PRS register */
 	qed_wr(p_hwfn, p_ptt, PRS_REG_NGE_PORT, dest_port);
+
+	/* Update NIG register */
 	qed_wr(p_hwfn, p_ptt, NIG_REG_NGE_PORT, dest_port);
+
+	/* Update PBF register */
 	qed_wr(p_hwfn, p_ptt, PBF_REG_NGE_PORT, dest_port);
 }
 
@@ -922,32 +1116,39 @@ void qed_set_geneve_enable(struct qed_hwfn *p_hwfn,
 			   struct qed_ptt *p_ptt,
 			   bool eth_geneve_enable, bool ip_geneve_enable)
 {
-	unsigned long reg_val = 0;
+	u32 reg_val;
 	u8 shift;
 
+	/* Update PRS register */
 	reg_val = qed_rd(p_hwfn, p_ptt, PRS_REG_ENCAPSULATION_TYPE_EN);
 	shift = PRS_REG_ENCAPSULATION_TYPE_EN_ETH_OVER_GENEVE_ENABLE_SHIFT;
-	qed_set_tunnel_type_enable_bit(&reg_val, shift, eth_geneve_enable);
-
+	SET_TUNNEL_TYPE_ENABLE_BIT(reg_val, shift, eth_geneve_enable);
 	shift = PRS_REG_ENCAPSULATION_TYPE_EN_IP_OVER_GENEVE_ENABLE_SHIFT;
-	qed_set_tunnel_type_enable_bit(&reg_val, shift, ip_geneve_enable);
-
+	SET_TUNNEL_TYPE_ENABLE_BIT(reg_val, shift, ip_geneve_enable);
 	qed_wr(p_hwfn, p_ptt, PRS_REG_ENCAPSULATION_TYPE_EN, reg_val);
 	if (reg_val)
-		qed_wr(p_hwfn, p_ptt, PRS_REG_OUTPUT_FORMAT_4_0,
-		       PRS_ETH_TUNN_FIC_FORMAT);
+		qed_wr(p_hwfn,
+		       p_ptt,
+		       PRS_REG_OUTPUT_FORMAT_4_0_BB_K2,
+		       (u32)PRS_ETH_TUNN_FIC_FORMAT);
 
+	/* Update NIG register */
 	qed_wr(p_hwfn, p_ptt, NIG_REG_NGE_ETH_ENABLE,
 	       eth_geneve_enable ? 1 : 0);
 	qed_wr(p_hwfn, p_ptt, NIG_REG_NGE_IP_ENABLE, ip_geneve_enable ? 1 : 0);
 
-	/* EDPM with geneve tunnel not supported in BB_B0 */
+	/* EDPM with geneve tunnel not supported in BB */
 	if (QED_IS_BB_B0(p_hwfn->cdev))
 		return;
 
-	qed_wr(p_hwfn, p_ptt, DORQ_REG_L2_EDPM_TUNNEL_NGE_ETH_EN,
+	/* Update DORQ registers */
+	qed_wr(p_hwfn,
+	       p_ptt,
+	       DORQ_REG_L2_EDPM_TUNNEL_NGE_ETH_EN_K2_E5,
 	       eth_geneve_enable ? 1 : 0);
-	qed_wr(p_hwfn, p_ptt, DORQ_REG_L2_EDPM_TUNNEL_NGE_IP_EN,
+	qed_wr(p_hwfn,
+	       p_ptt,
+	       DORQ_REG_L2_EDPM_TUNNEL_NGE_IP_EN_K2_E5,
 	       ip_geneve_enable ? 1 : 0);
 }
 
@@ -959,117 +1160,297 @@ void qed_set_geneve_enable(struct qed_hwfn *p_hwfn,
 #define RAM_LINE_SIZE sizeof(u64)
 #define REG_SIZE sizeof(u32)
 
-void qed_set_rfs_mode_disable(struct qed_hwfn *p_hwfn,
-			      struct qed_ptt *p_ptt, u16 pf_id)
+void qed_gft_disable(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt, u16 pf_id)
 {
-	u32 hw_addr = PRS_REG_GFT_PROFILE_MASK_RAM +
-		      pf_id * RAM_LINE_SIZE;
-
-	/*stop using gft logic */
+	/* Disable gft search for PF */
 	qed_wr(p_hwfn, p_ptt, PRS_REG_SEARCH_GFT, 0);
-	qed_wr(p_hwfn, p_ptt, PRS_REG_CM_HDR_GFT, 0x0);
+
+	/* Clean ram & cam for next gft session */
+
+	/* Zero camline */
 	qed_wr(p_hwfn, p_ptt, PRS_REG_GFT_CAM + CAM_LINE_SIZE * pf_id, 0);
-	qed_wr(p_hwfn, p_ptt, hw_addr, 0);
-	qed_wr(p_hwfn, p_ptt, hw_addr + 4, 0);
+
+	/* Zero ramline */
+	qed_wr(p_hwfn,
+	       p_ptt, PRS_REG_GFT_PROFILE_MASK_RAM + RAM_LINE_SIZE * pf_id, 0);
+	qed_wr(p_hwfn,
+	       p_ptt,
+	       PRS_REG_GFT_PROFILE_MASK_RAM + RAM_LINE_SIZE * pf_id + REG_SIZE,
+	       0);
 }
 
-void qed_set_rfs_mode_enable(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt,
-			     u16 pf_id, bool tcp, bool udp,
-			     bool ipv4, bool ipv6)
+void qed_set_gft_event_id_cm_hdr(struct qed_hwfn *p_hwfn, struct qed_ptt *p_ptt)
 {
-	union gft_cam_line_union camline;
-	struct gft_ram_line ramline;
 	u32 rfs_cm_hdr_event_id;
 
+	/* Set RFS event ID to be awakened i Tstorm By Prs */
 	rfs_cm_hdr_event_id = qed_rd(p_hwfn, p_ptt, PRS_REG_CM_HDR_GFT);
+	rfs_cm_hdr_event_id |= T_ETH_PACKET_ACTION_GFT_EVENTID <<
+			       PRS_REG_CM_HDR_GFT_EVENT_ID_SHIFT;
+	rfs_cm_hdr_event_id |= PARSER_ETH_CONN_GFT_ACTION_CM_HDR <<
+			       PRS_REG_CM_HDR_GFT_CM_HDR_SHIFT;
+	qed_wr(p_hwfn, p_ptt, PRS_REG_CM_HDR_GFT, rfs_cm_hdr_event_id);
+}
+
+void qed_gft_config(struct qed_hwfn *p_hwfn,
+		    struct qed_ptt *p_ptt,
+		    u16 pf_id,
+		    bool tcp,
+		    bool udp,
+		    bool ipv4, bool ipv6, enum gft_profile_type profile_type)
+{
+	u32 reg_val, cam_line, ram_line_lo, ram_line_hi;
 
 	if (!ipv6 && !ipv4)
 		DP_NOTICE(p_hwfn,
-			  "set_rfs_mode_enable: must accept at least on of - ipv4 or ipv6");
+			  "gft_config: must accept at least on of - ipv4 or ipv6'\n");
 	if (!tcp && !udp)
 		DP_NOTICE(p_hwfn,
-			  "set_rfs_mode_enable: must accept at least on of - udp or tcp");
+			  "gft_config: must accept at least on of - udp or tcp\n");
+	if (profile_type >= MAX_GFT_PROFILE_TYPE)
+		DP_NOTICE(p_hwfn, "gft_config: unsupported gft_profile_type\n");
 
-	rfs_cm_hdr_event_id |= T_ETH_PACKET_MATCH_RFS_EVENTID <<
-					PRS_REG_CM_HDR_GFT_EVENT_ID_SHIFT;
-	rfs_cm_hdr_event_id |= PARSER_ETH_CONN_CM_HDR <<
-					PRS_REG_CM_HDR_GFT_CM_HDR_SHIFT;
-	qed_wr(p_hwfn, p_ptt, PRS_REG_CM_HDR_GFT, rfs_cm_hdr_event_id);
+	/* Set RFS event ID to be awakened i Tstorm By Prs */
+	reg_val = T_ETH_PACKET_MATCH_RFS_EVENTID <<
+		  PRS_REG_CM_HDR_GFT_EVENT_ID_SHIFT;
+	reg_val |= PARSER_ETH_CONN_CM_HDR << PRS_REG_CM_HDR_GFT_CM_HDR_SHIFT;
+	qed_wr(p_hwfn, p_ptt, PRS_REG_CM_HDR_GFT, reg_val);
 
-	/* Configure Registers for RFS mode */
-	qed_wr(p_hwfn, p_ptt, PRS_REG_SEARCH_GFT, 1);
+	/* Do not load context only cid in PRS on match. */
 	qed_wr(p_hwfn, p_ptt, PRS_REG_LOAD_L2_FILTER, 0);
-	camline.cam_line_mapped.camline = 0;
 
-	/* Cam line is now valid!! */
-	SET_FIELD(camline.cam_line_mapped.camline,
-		  GFT_CAM_LINE_MAPPED_VALID, 1);
+	/* Do not use tenant ID exist bit for gft search */
+	qed_wr(p_hwfn, p_ptt, PRS_REG_SEARCH_TENANT_ID, 0);
 
-	/* filters are per PF!! */
-	SET_FIELD(camline.cam_line_mapped.camline,
+	/* Set Cam */
+	cam_line = 0;
+	SET_FIELD(cam_line, GFT_CAM_LINE_MAPPED_VALID, 1);
+
+	/* Filters are per PF!! */
+	SET_FIELD(cam_line,
 		  GFT_CAM_LINE_MAPPED_PF_ID_MASK,
 		  GFT_CAM_LINE_MAPPED_PF_ID_MASK_MASK);
-	SET_FIELD(camline.cam_line_mapped.camline,
-		  GFT_CAM_LINE_MAPPED_PF_ID, pf_id);
+	SET_FIELD(cam_line, GFT_CAM_LINE_MAPPED_PF_ID, pf_id);
+
 	if (!(tcp && udp)) {
-		SET_FIELD(camline.cam_line_mapped.camline,
+		SET_FIELD(cam_line,
 			  GFT_CAM_LINE_MAPPED_UPPER_PROTOCOL_TYPE_MASK,
 			  GFT_CAM_LINE_MAPPED_UPPER_PROTOCOL_TYPE_MASK_MASK);
 		if (tcp)
-			SET_FIELD(camline.cam_line_mapped.camline,
+			SET_FIELD(cam_line,
 				  GFT_CAM_LINE_MAPPED_UPPER_PROTOCOL_TYPE,
 				  GFT_PROFILE_TCP_PROTOCOL);
 		else
-			SET_FIELD(camline.cam_line_mapped.camline,
+			SET_FIELD(cam_line,
 				  GFT_CAM_LINE_MAPPED_UPPER_PROTOCOL_TYPE,
 				  GFT_PROFILE_UDP_PROTOCOL);
 	}
 
 	if (!(ipv4 && ipv6)) {
-		SET_FIELD(camline.cam_line_mapped.camline,
-			  GFT_CAM_LINE_MAPPED_IP_VERSION_MASK, 1);
+		SET_FIELD(cam_line, GFT_CAM_LINE_MAPPED_IP_VERSION_MASK, 1);
 		if (ipv4)
-			SET_FIELD(camline.cam_line_mapped.camline,
+			SET_FIELD(cam_line,
 				  GFT_CAM_LINE_MAPPED_IP_VERSION,
 				  GFT_PROFILE_IPV4);
 		else
-			SET_FIELD(camline.cam_line_mapped.camline,
+			SET_FIELD(cam_line,
 				  GFT_CAM_LINE_MAPPED_IP_VERSION,
 				  GFT_PROFILE_IPV6);
 	}
 
 	/* Write characteristics to cam */
 	qed_wr(p_hwfn, p_ptt, PRS_REG_GFT_CAM + CAM_LINE_SIZE * pf_id,
-	       camline.cam_line_mapped.camline);
-	camline.cam_line_mapped.camline = qed_rd(p_hwfn, p_ptt,
-						 PRS_REG_GFT_CAM +
-						 CAM_LINE_SIZE * pf_id);
+	       cam_line);
+	cam_line =
+	    qed_rd(p_hwfn, p_ptt, PRS_REG_GFT_CAM + CAM_LINE_SIZE * pf_id);
 
 	/* Write line to RAM - compare to filter 4 tuple */
-	ramline.lo = 0;
-	ramline.hi = 0;
-	SET_FIELD(ramline.hi, GFT_RAM_LINE_DST_IP, 1);
-	SET_FIELD(ramline.hi, GFT_RAM_LINE_SRC_IP, 1);
-	SET_FIELD(ramline.hi, GFT_RAM_LINE_OVER_IP_PROTOCOL, 1);
-	SET_FIELD(ramline.lo, GFT_RAM_LINE_ETHERTYPE, 1);
-	SET_FIELD(ramline.lo, GFT_RAM_LINE_SRC_PORT, 1);
-	SET_FIELD(ramline.lo, GFT_RAM_LINE_DST_PORT, 1);
-
-	/* Each iteration write to reg */
-	qed_wr(p_hwfn, p_ptt,
+	ram_line_lo = 0;
+	ram_line_hi = 0;
+
+	if (profile_type == GFT_PROFILE_TYPE_4_TUPLE) {
+		SET_FIELD(ram_line_hi, GFT_RAM_LINE_DST_IP, 1);
+		SET_FIELD(ram_line_hi, GFT_RAM_LINE_SRC_IP, 1);
+		SET_FIELD(ram_line_hi, GFT_RAM_LINE_OVER_IP_PROTOCOL, 1);
+		SET_FIELD(ram_line_lo, GFT_RAM_LINE_ETHERTYPE, 1);
+		SET_FIELD(ram_line_lo, GFT_RAM_LINE_SRC_PORT, 1);
+		SET_FIELD(ram_line_lo, GFT_RAM_LINE_DST_PORT, 1);
+	} else if (profile_type == GFT_PROFILE_TYPE_L4_DST_PORT) {
+		SET_FIELD(ram_line_hi, GFT_RAM_LINE_OVER_IP_PROTOCOL, 1);
+		SET_FIELD(ram_line_lo, GFT_RAM_LINE_ETHERTYPE, 1);
+		SET_FIELD(ram_line_lo, GFT_RAM_LINE_DST_PORT, 1);
+	} else if (profile_type == GFT_PROFILE_TYPE_IP_DST_PORT) {
+		SET_FIELD(ram_line_hi, GFT_RAM_LINE_DST_IP, 1);
+		SET_FIELD(ram_line_lo, GFT_RAM_LINE_ETHERTYPE, 1);
+	}
+
+	qed_wr(p_hwfn,
+	       p_ptt,
 	       PRS_REG_GFT_PROFILE_MASK_RAM + RAM_LINE_SIZE * pf_id,
-	       ramline.lo);
-	qed_wr(p_hwfn, p_ptt,
-	       PRS_REG_GFT_PROFILE_MASK_RAM + RAM_LINE_SIZE * pf_id + 4,
-	       ramline.hi);
+	       ram_line_lo);
+	qed_wr(p_hwfn,
+	       p_ptt,
+	       PRS_REG_GFT_PROFILE_MASK_RAM + RAM_LINE_SIZE * pf_id + REG_SIZE,
+	       ram_line_hi);
 
 	/* Set default profile so that no filter match will happen */
-	qed_wr(p_hwfn, p_ptt,
-	       PRS_REG_GFT_PROFILE_MASK_RAM +
-	       RAM_LINE_SIZE * PRS_GFT_CAM_LINES_NO_MATCH,
-	       ramline.lo);
-	qed_wr(p_hwfn, p_ptt,
-	       PRS_REG_GFT_PROFILE_MASK_RAM +
-	       RAM_LINE_SIZE * PRS_GFT_CAM_LINES_NO_MATCH + 4,
-	       ramline.hi);
+	qed_wr(p_hwfn,
+	       p_ptt,
+	       PRS_REG_GFT_PROFILE_MASK_RAM + RAM_LINE_SIZE *
+	       PRS_GFT_CAM_LINES_NO_MATCH, 0xffffffff);
+	qed_wr(p_hwfn,
+	       p_ptt,
+	       PRS_REG_GFT_PROFILE_MASK_RAM + RAM_LINE_SIZE *
+	       PRS_GFT_CAM_LINES_NO_MATCH + REG_SIZE, 0x3ff);
+
+	/* Enable gft search */
+	qed_wr(p_hwfn, p_ptt, PRS_REG_SEARCH_GFT, 1);
+}
+
+DECLARE_CRC8_TABLE(cdu_crc8_table);
+
+/* Calculate and return CDU validation byte per connection type/region/cid */
+static u8 qed_calc_cdu_validation_byte(u8 conn_type, u8 region, u32 cid)
+{
+	const u8 validation_cfg = CDU_VALIDATION_DEFAULT_CFG;
+	u8 crc, validation_byte = 0;
+	static u8 crc8_table_valid; /* automatically initialized to 0 */
+	u32 validation_string = 0;
+	u32 data_to_crc;
+
+	if (!crc8_table_valid) {
+		crc8_populate_msb(cdu_crc8_table, 0x07);
+		crc8_table_valid = 1;
+	}
+
+	/* The CRC is calculated on the String-to-compress:
+	 * [31:8]  = {CID[31:20],CID[11:0]}
+	 * [7:4]   = Region
+	 * [3:0]   = Type
+	 */
+	if ((validation_cfg >> CDU_CONTEXT_VALIDATION_CFG_USE_CID) & 1)
+		validation_string |= (cid & 0xFFF00000) | ((cid & 0xFFF) << 8);
+
+	if ((validation_cfg >> CDU_CONTEXT_VALIDATION_CFG_USE_REGION) & 1)
+		validation_string |= ((region & 0xF) << 4);
+
+	if ((validation_cfg >> CDU_CONTEXT_VALIDATION_CFG_USE_TYPE) & 1)
+		validation_string |= (conn_type & 0xF);
+
+	/* Convert to big-endian and calculate CRC8 */
+	data_to_crc = be32_to_cpu(validation_string);
+
+	crc = crc8(cdu_crc8_table,
+		   (u8 *)&data_to_crc, sizeof(data_to_crc), CRC8_INIT_VALUE);
+
+	/* The validation byte [7:0] is composed:
+	 * for type A validation
+	 * [7]          = active configuration bit
+	 * [6:0]        = crc[6:0]
+	 *
+	 * for type B validation
+	 * [7]          = active configuration bit
+	 * [6:3]        = connection_type[3:0]
+	 * [2:0]        = crc[2:0]
+	 */
+	validation_byte |=
+	    ((validation_cfg >>
+	      CDU_CONTEXT_VALIDATION_CFG_USE_ACTIVE) & 1) << 7;
+
+	if ((validation_cfg >>
+	     CDU_CONTEXT_VALIDATION_CFG_VALIDATION_TYPE_SHIFT) & 1)
+		validation_byte |= ((conn_type & 0xF) << 3) | (crc & 0x7);
+	else
+		validation_byte |= crc & 0x7F;
+
+	return validation_byte;
+}
+
+/* Calcualte and set validation bytes for session context */
+void qed_calc_session_ctx_validation(void *p_ctx_mem,
+				     u16 ctx_size, u8 ctx_type, u32 cid)
+{
+	u8 *x_val_ptr, *t_val_ptr, *u_val_ptr, *p_ctx;
+
+	p_ctx = (u8 * const)p_ctx_mem;
+	x_val_ptr = &p_ctx[con_region_offsets[0][ctx_type]];
+	t_val_ptr = &p_ctx[con_region_offsets[1][ctx_type]];
+	u_val_ptr = &p_ctx[con_region_offsets[2][ctx_type]];
+
+	memset(p_ctx, 0, ctx_size);
+
+	*x_val_ptr = qed_calc_cdu_validation_byte(ctx_type, 3, cid);
+	*t_val_ptr = qed_calc_cdu_validation_byte(ctx_type, 4, cid);
+	*u_val_ptr = qed_calc_cdu_validation_byte(ctx_type, 5, cid);
+}
+
+/* Calcualte and set validation bytes for task context */
+void qed_calc_task_ctx_validation(void *p_ctx_mem,
+				  u16 ctx_size, u8 ctx_type, u32 tid)
+{
+	u8 *p_ctx, *region1_val_ptr;
+
+	p_ctx = (u8 * const)p_ctx_mem;
+	region1_val_ptr = &p_ctx[task_region_offsets[0][ctx_type]];
+
+	memset(p_ctx, 0, ctx_size);
+
+	*region1_val_ptr = qed_calc_cdu_validation_byte(ctx_type, 1, tid);
+}
+
+/* Memset session context to 0 while preserving validation bytes */
+void qed_memset_session_ctx(void *p_ctx_mem, u32 ctx_size, u8 ctx_type)
+{
+	u8 *x_val_ptr, *t_val_ptr, *u_val_ptr, *p_ctx;
+	u8 x_val, t_val, u_val;
+
+	p_ctx = (u8 * const)p_ctx_mem;
+	x_val_ptr = &p_ctx[con_region_offsets[0][ctx_type]];
+	t_val_ptr = &p_ctx[con_region_offsets[1][ctx_type]];
+	u_val_ptr = &p_ctx[con_region_offsets[2][ctx_type]];
+
+	x_val = *x_val_ptr;
+	t_val = *t_val_ptr;
+	u_val = *u_val_ptr;
+
+	memset(p_ctx, 0, ctx_size);
+
+	*x_val_ptr = x_val;
+	*t_val_ptr = t_val;
+	*u_val_ptr = u_val;
+}
+
+/* Memset task context to 0 while preserving validation bytes */
+void qed_memset_task_ctx(void *p_ctx_mem, u32 ctx_size, u8 ctx_type)
+{
+	u8 *p_ctx, *region1_val_ptr;
+	u8 region1_val;
+
+	p_ctx = (u8 * const)p_ctx_mem;
+	region1_val_ptr = &p_ctx[task_region_offsets[0][ctx_type]];
+
+	region1_val = *region1_val_ptr;
+
+	memset(p_ctx, 0, ctx_size);
+
+	*region1_val_ptr = region1_val;
+}
+
+/* Enable and configure context validation */
+void qed_enable_context_validation(struct qed_hwfn *p_hwfn,
+				   struct qed_ptt *p_ptt)
+{
+	u32 ctx_validation;
+
+	/* Enable validation for connection region 3: CCFC_CTX_VALID0[31:24] */
+	ctx_validation = CDU_VALIDATION_DEFAULT_CFG << 24;
+	qed_wr(p_hwfn, p_ptt, CDU_REG_CCFC_CTX_VALID0, ctx_validation);
+
+	/* Enable validation for connection region 5: CCFC_CTX_VALID1[15:8] */
+	ctx_validation = CDU_VALIDATION_DEFAULT_CFG << 8;
+	qed_wr(p_hwfn, p_ptt, CDU_REG_CCFC_CTX_VALID1, ctx_validation);
+
+	/* Enable validation for connection region 1: TCFC_CTX_VALID0[15:8] */
+	ctx_validation = CDU_VALIDATION_DEFAULT_CFG << 8;
+	qed_wr(p_hwfn, p_ptt, CDU_REG_TCFC_CTX_VALID0, ctx_validation);
 }