Merge branch 'support-one-ptp-device-per-hardware-clock'

Tariq Toukan says:

====================
Support one PTP device per hardware clock

This series contains two features from Jianbo, followed by simple
cleanups.

Patches 1-9 by Jianbo add support for one PTP device per hardware clock,
described below [1].

Patches 10-12 by Jianbo add support for 200Gbps per-lane link modes in
kernel and mlx5 driver.

Patches 13-15 are simple cleanups by Gal and Carolina.

[1]
PHC (PTP hardware clock) is normally shared by multiple functions
(PF/VF/SF). mlx5 driver currently creates a separate PTP device for each
network interface that shares one PHC.

PHC can be configured to work as free running mode or real time mode.
In this series, only one PTP device is created for the shared PHC when
it is running in real time mode.

To support this feature,
* Firmware needs to support clock identity. When functions share a
  PHC, the clock identities they query are same.
* Driver dynamically allocates mlx5_clock to represent a PHC.
* New devcom component is added for hardware clock. Functions are
  grouped by the identity, and one mlx5_clock is allocated and shared
  by the functions with the same identity.
* When PTP device accesses PHC by its callbacks, the first function
  in the clock devcom list is selected to send commands to firmware.
* PPS IN event is armed on one function. It should be re-armed on
  the other one when current is unloaded.
====================

Link: https://patch.msgid.link/20250203213516.227902-1-tariqt@nvidia.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

20 files changed, 701 insertions, 172 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port.c b/drivers/net/ethernet/mellanox/mlx5/core/en/port.c
index 5f6a0605e4ae..f62fbfb67a1b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/port.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port.c
@@ -296,11 +296,16 @@ enum mlx5e_fec_supported_link_mode {
 	MLX5E_FEC_SUPPORTED_LINK_MODE_200G_2X,
 	MLX5E_FEC_SUPPORTED_LINK_MODE_400G_4X,
 	MLX5E_FEC_SUPPORTED_LINK_MODE_800G_8X,
+	MLX5E_FEC_SUPPORTED_LINK_MODE_200G_1X,
+	MLX5E_FEC_SUPPORTED_LINK_MODE_400G_2X,
+	MLX5E_FEC_SUPPORTED_LINK_MODE_800G_4X,
+	MLX5E_FEC_SUPPORTED_LINK_MODE_1600G_8X,
 	MLX5E_MAX_FEC_SUPPORTED_LINK_MODE,
 };
 
 #define MLX5E_FEC_FIRST_50G_PER_LANE_MODE MLX5E_FEC_SUPPORTED_LINK_MODE_50G_1X
 #define MLX5E_FEC_FIRST_100G_PER_LANE_MODE MLX5E_FEC_SUPPORTED_LINK_MODE_100G_1X
+#define MLX5E_FEC_FIRST_200G_PER_LANE_MODE MLX5E_FEC_SUPPORTED_LINK_MODE_200G_1X
 
 #define MLX5E_FEC_OVERRIDE_ADMIN_POLICY(buf, policy, write, link)			\
 	do {										\
@@ -320,8 +325,10 @@ static bool mlx5e_is_fec_supported_link_mode(struct mlx5_core_dev *dev,
 	return link_mode < MLX5E_FEC_FIRST_50G_PER_LANE_MODE ||
 	       (link_mode < MLX5E_FEC_FIRST_100G_PER_LANE_MODE &&
 		MLX5_CAP_PCAM_FEATURE(dev, fec_50G_per_lane_in_pplm)) ||
-	       (link_mode >= MLX5E_FEC_FIRST_100G_PER_LANE_MODE &&
-		MLX5_CAP_PCAM_FEATURE(dev, fec_100G_per_lane_in_pplm));
+	       (link_mode < MLX5E_FEC_FIRST_200G_PER_LANE_MODE &&
+		MLX5_CAP_PCAM_FEATURE(dev, fec_100G_per_lane_in_pplm)) ||
+	       (link_mode >= MLX5E_FEC_FIRST_200G_PER_LANE_MODE &&
+		MLX5_CAP_PCAM_FEATURE(dev, fec_200G_per_lane_in_pplm));
 }
 
 /* get/set FEC admin field for a given speed */
@@ -368,6 +375,18 @@ static int mlx5e_fec_admin_field(u32 *pplm, u16 *fec_policy, bool write,
 	case MLX5E_FEC_SUPPORTED_LINK_MODE_800G_8X:
 		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 800g_8x);
 		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_200G_1X:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 200g_1x);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_400G_2X:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 400g_2x);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_800G_4X:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 800g_4x);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_1600G_8X:
+		MLX5E_FEC_OVERRIDE_ADMIN_POLICY(pplm, *fec_policy, write, 1600g_8x);
+		break;
 	default:
 		return -EINVAL;
 	}
@@ -421,6 +440,18 @@ static int mlx5e_get_fec_cap_field(u32 *pplm, u16 *fec_cap,
 	case MLX5E_FEC_SUPPORTED_LINK_MODE_800G_8X:
 		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 800g_8x);
 		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_200G_1X:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 200g_1x);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_400G_2X:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 400g_2x);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_800G_4X:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 800g_4x);
+		break;
+	case MLX5E_FEC_SUPPORTED_LINK_MODE_1600G_8X:
+		*fec_cap = MLX5E_GET_FEC_OVERRIDE_CAP(pplm, 1600g_8x);
+		break;
 	default:
 		return -EINVAL;
 	}
@@ -494,6 +525,26 @@ out:
 	return 0;
 }
 
+static u16 mlx5e_remap_fec_conf_mode(enum mlx5e_fec_supported_link_mode link_mode,
+				     u16 conf_fec)
+{
+	/* RS fec in ethtool is originally mapped to MLX5E_FEC_RS_528_514.
+	 * For link modes up to 25G per lane, the value is kept.
+	 * For 50G or 100G per lane, it's remapped to MLX5E_FEC_RS_544_514.
+	 * For 200G per lane, remapped to MLX5E_FEC_RS_544_514_INTERLEAVED_QUAD.
+	 */
+	if (conf_fec != BIT(MLX5E_FEC_RS_528_514))
+		return conf_fec;
+
+	if (link_mode >= MLX5E_FEC_FIRST_200G_PER_LANE_MODE)
+		return BIT(MLX5E_FEC_RS_544_514_INTERLEAVED_QUAD);
+
+	if (link_mode >= MLX5E_FEC_FIRST_50G_PER_LANE_MODE)
+		return BIT(MLX5E_FEC_RS_544_514);
+
+	return conf_fec;
+}
+
 int mlx5e_set_fec_mode(struct mlx5_core_dev *dev, u16 fec_policy)
 {
 	bool fec_50g_per_lane = MLX5_CAP_PCAM_FEATURE(dev, fec_50G_per_lane_in_pplm);
@@ -530,14 +581,7 @@ int mlx5e_set_fec_mode(struct mlx5_core_dev *dev, u16 fec_policy)
 		if (!mlx5e_is_fec_supported_link_mode(dev, i))
 			break;
 
-		/* RS fec in ethtool is mapped to MLX5E_FEC_RS_528_514
-		 * to link modes up to 25G per lane and to
-		 * MLX5E_FEC_RS_544_514 in the new link modes based on
-		 * 50G or 100G per lane
-		 */
-		if (conf_fec == (1 << MLX5E_FEC_RS_528_514) &&
-		    i >= MLX5E_FEC_FIRST_50G_PER_LANE_MODE)
-			conf_fec = (1 << MLX5E_FEC_RS_544_514);
+		conf_fec = mlx5e_remap_fec_conf_mode(i, conf_fec);
 
 		mlx5e_get_fec_cap_field(out, &fec_caps, i);
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/port.h b/drivers/net/ethernet/mellanox/mlx5/core/en/port.h
index d1da225f35da..fa2283dd383b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/port.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/port.h
@@ -61,6 +61,7 @@ enum {
 	MLX5E_FEC_NOFEC,
 	MLX5E_FEC_FIRECODE,
 	MLX5E_FEC_RS_528_514,
+	MLX5E_FEC_RS_544_514_INTERLEAVED_QUAD = 4,
 	MLX5E_FEC_RS_544_514 = 7,
 	MLX5E_FEC_LLRS_272_257_1 = 9,
 };
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c
index afd654583b6b..131ed97ca997 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c
@@ -326,7 +326,7 @@ static int mlx5e_ptp_alloc_txqsq(struct mlx5e_ptp *c, int txq_ix,
 	int node;
 
 	sq->pdev      = c->pdev;
-	sq->clock     = &mdev->clock;
+	sq->clock     = mdev->clock;
 	sq->mkey_be   = c->mkey_be;
 	sq->netdev    = c->netdev;
 	sq->priv      = c->priv;
@@ -696,7 +696,7 @@ static int mlx5e_init_ptp_rq(struct mlx5e_ptp *c, struct mlx5e_params *params,
 	rq->pdev         = c->pdev;
 	rq->netdev       = priv->netdev;
 	rq->priv         = priv;
-	rq->clock        = &mdev->clock;
+	rq->clock        = mdev->clock;
 	rq->tstamp       = &priv->tstamp;
 	rq->mdev         = mdev;
 	rq->hw_mtu       = MLX5E_SW2HW_MTU(params, params->sw_mtu);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.h b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.h
index d6c12d0ea55b..2e528b2c34d6 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc/act/act.h
@@ -73,11 +73,6 @@ struct mlx5e_tc_act {
 	bool is_terminating_action;
 };
 
-struct mlx5e_tc_flow_action {
-	unsigned int num_entries;
-	struct flow_action_entry **entries;
-};
-
 extern struct mlx5e_tc_act mlx5e_tc_act_drop;
 extern struct mlx5e_tc_act mlx5e_tc_act_trap;
 extern struct mlx5e_tc_act mlx5e_tc_act_accept;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/trap.c b/drivers/net/ethernet/mellanox/mlx5/core/en/trap.c
index 53ca16cb9c41..140606fcd23b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/trap.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/trap.c
@@ -46,7 +46,7 @@ static void mlx5e_init_trap_rq(struct mlx5e_trap *t, struct mlx5e_params *params
 	rq->pdev         = t->pdev;
 	rq->netdev       = priv->netdev;
 	rq->priv         = priv;
-	rq->clock        = &mdev->clock;
+	rq->clock        = mdev->clock;
 	rq->tstamp       = &priv->tstamp;
 	rq->mdev         = mdev;
 	rq->hw_mtu       = MLX5E_SW2HW_MTU(params, params->sw_mtu);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c b/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c
index 94b291662087..3cc4d55613bf 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.c
@@ -289,9 +289,9 @@ static u64 mlx5e_xsk_fill_timestamp(void *_priv)
 	ts = get_cqe_ts(priv->cqe);
 
 	if (mlx5_is_real_time_rq(priv->cq->mdev) || mlx5_is_real_time_sq(priv->cq->mdev))
-		return mlx5_real_time_cyc2time(&priv->cq->mdev->clock, ts);
+		return mlx5_real_time_cyc2time(priv->cq->mdev->clock, ts);
 
-	return  mlx5_timecounter_cyc2time(&priv->cq->mdev->clock, ts);
+	return  mlx5_timecounter_cyc2time(priv->cq->mdev->clock, ts);
 }
 
 static void mlx5e_xsk_request_checksum(u16 csum_start, u16 csum_offset, void *priv)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c
index 9240cfe25d10..d743e823362a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c
@@ -72,7 +72,7 @@ static int mlx5e_init_xsk_rq(struct mlx5e_channel *c,
 	rq->netdev       = c->netdev;
 	rq->priv         = c->priv;
 	rq->tstamp       = c->tstamp;
-	rq->clock        = &mdev->clock;
+	rq->clock        = mdev->clock;
 	rq->icosq        = &c->icosq;
 	rq->ix           = c->ix;
 	rq->channel      = c;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
index cae39198b4db..f9113cb13a0c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -237,6 +237,27 @@ void mlx5e_build_ptys2ethtool_map(void)
 				       ETHTOOL_LINK_MODE_800000baseDR8_2_Full_BIT,
 				       ETHTOOL_LINK_MODE_800000baseSR8_Full_BIT,
 				       ETHTOOL_LINK_MODE_800000baseVR8_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_200GAUI_1_200GBASE_CR1_KR1, ext,
+				       ETHTOOL_LINK_MODE_200000baseCR_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseKR_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseDR_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseDR_2_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseSR_Full_BIT,
+				       ETHTOOL_LINK_MODE_200000baseVR_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_400GAUI_2_400GBASE_CR2_KR2, ext,
+				       ETHTOOL_LINK_MODE_400000baseCR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_400000baseKR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_400000baseDR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_400000baseDR2_2_Full_BIT,
+				       ETHTOOL_LINK_MODE_400000baseSR2_Full_BIT,
+				       ETHTOOL_LINK_MODE_400000baseVR2_Full_BIT);
+	MLX5_BUILD_PTYS2ETHTOOL_CONFIG(MLX5E_800GAUI_4_800GBASE_CR4_KR4, ext,
+				       ETHTOOL_LINK_MODE_800000baseCR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_800000baseKR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_800000baseDR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_800000baseDR4_2_Full_BIT,
+				       ETHTOOL_LINK_MODE_800000baseSR4_Full_BIT,
+				       ETHTOOL_LINK_MODE_800000baseVR4_Full_BIT);
 }
 
 static void mlx5e_ethtool_get_speed_arr(struct mlx5_core_dev *mdev,
@@ -931,6 +952,7 @@ static const u32 pplm_fec_2_ethtool[] = {
 	[MLX5E_FEC_RS_528_514] = ETHTOOL_FEC_RS,
 	[MLX5E_FEC_RS_544_514] = ETHTOOL_FEC_RS,
 	[MLX5E_FEC_LLRS_272_257_1] = ETHTOOL_FEC_LLRS,
+	[MLX5E_FEC_RS_544_514_INTERLEAVED_QUAD] = ETHTOOL_FEC_RS,
 };
 
 static u32 pplm2ethtool_fec(u_long fec_mode, unsigned long size)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index a814b63ed97e..2fdc86432ac0 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -737,7 +737,7 @@ static int mlx5e_init_rxq_rq(struct mlx5e_channel *c, struct mlx5e_params *param
 	rq->netdev       = c->netdev;
 	rq->priv         = c->priv;
 	rq->tstamp       = c->tstamp;
-	rq->clock        = &mdev->clock;
+	rq->clock        = mdev->clock;
 	rq->icosq        = &c->icosq;
 	rq->ix           = c->ix;
 	rq->channel      = c;
@@ -1614,7 +1614,7 @@ static int mlx5e_alloc_txqsq(struct mlx5e_channel *c,
 	int err;
 
 	sq->pdev      = c->pdev;
-	sq->clock     = &mdev->clock;
+	sq->clock     = mdev->clock;
 	sq->mkey_be   = c->mkey_be;
 	sq->netdev    = c->netdev;
 	sq->mdev      = c->mdev;
@@ -3816,8 +3816,11 @@ static int mlx5e_setup_tc_mqprio(struct mlx5e_priv *priv,
 	/* MQPRIO is another toplevel qdisc that can't be attached
 	 * simultaneously with the offloaded HTB.
 	 */
-	if (WARN_ON(mlx5e_selq_is_htb_enabled(&priv->selq)))
-		return -EINVAL;
+	if (mlx5e_selq_is_htb_enabled(&priv->selq)) {
+		NL_SET_ERR_MSG_MOD(mqprio->extack,
+				   "MQPRIO cannot be configured when HTB offload is enabled.");
+		return -EOPNOTSUPP;
+	}
 
 	switch (mqprio->mode) {
 	case TC_MQPRIO_MODE_DCB:
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c b/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c
index bde79cac33a9..d832a12ffec0 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lag/port_sel.c
@@ -97,7 +97,7 @@ static int mlx5_lag_create_port_sel_table(struct mlx5_lag *ldev,
 						mlx5_del_flow_rules(lag_definer->rules[idx]);
 					}
 					j = ldev->buckets;
-				};
+				}
 				goto destroy_fg;
 			}
 		}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c
index d61a1a9297c9..65a94e46edcf 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c
@@ -43,6 +43,8 @@
 #include <linux/cpufeature.h>
 #endif /* CONFIG_X86 */
 
+#define MLX5_RT_CLOCK_IDENTITY_SIZE MLX5_FLD_SZ_BYTES(mrtcq_reg, rt_clock_identity)
+
 enum {
 	MLX5_PIN_MODE_IN		= 0x0,
 	MLX5_PIN_MODE_OUT		= 0x1,
@@ -77,6 +79,56 @@ enum {
 	MLX5_MTUTC_OPERATION_ADJUST_TIME_EXTENDED_MAX = 200000,
 };
 
+struct mlx5_clock_dev_state {
+	struct mlx5_core_dev *mdev;
+	struct mlx5_devcom_comp_dev *compdev;
+	struct mlx5_nb pps_nb;
+	struct work_struct out_work;
+};
+
+struct mlx5_clock_priv {
+	struct mlx5_clock clock;
+	struct mlx5_core_dev *mdev;
+	struct mutex lock; /* protect mdev and used in PTP callbacks */
+	struct mlx5_core_dev *event_mdev;
+};
+
+static struct mlx5_clock_priv *clock_priv(struct mlx5_clock *clock)
+{
+	return container_of(clock, struct mlx5_clock_priv, clock);
+}
+
+static void mlx5_clock_lockdep_assert(struct mlx5_clock *clock)
+{
+	if (!clock->shared)
+		return;
+
+	lockdep_assert(lockdep_is_held(&clock_priv(clock)->lock));
+}
+
+static struct mlx5_core_dev *mlx5_clock_mdev_get(struct mlx5_clock *clock)
+{
+	mlx5_clock_lockdep_assert(clock);
+
+	return clock_priv(clock)->mdev;
+}
+
+static void mlx5_clock_lock(struct mlx5_clock *clock)
+{
+	if (!clock->shared)
+		return;
+
+	mutex_lock(&clock_priv(clock)->lock);
+}
+
+static void mlx5_clock_unlock(struct mlx5_clock *clock)
+{
+	if (!clock->shared)
+		return;
+
+	mutex_unlock(&clock_priv(clock)->lock);
+}
+
 static bool mlx5_real_time_mode(struct mlx5_core_dev *mdev)
 {
 	return (mlx5_is_real_time_rq(mdev) || mlx5_is_real_time_sq(mdev));
@@ -94,6 +146,22 @@ static bool mlx5_modify_mtutc_allowed(struct mlx5_core_dev *mdev)
 	return MLX5_CAP_MCAM_FEATURE(mdev, ptpcyc2realtime_modify);
 }
 
+static int mlx5_clock_identity_get(struct mlx5_core_dev *mdev,
+				   u8 identify[MLX5_RT_CLOCK_IDENTITY_SIZE])
+{
+	u32 out[MLX5_ST_SZ_DW(mrtcq_reg)] = {};
+	u32 in[MLX5_ST_SZ_DW(mrtcq_reg)] = {};
+	int err;
+
+	err = mlx5_core_access_reg(mdev, in, sizeof(in),
+				   out, sizeof(out), MLX5_REG_MRTCQ, 0, 0);
+	if (!err)
+		memcpy(identify, MLX5_ADDR_OF(mrtcq_reg, out, rt_clock_identity),
+		       MLX5_RT_CLOCK_IDENTITY_SIZE);
+
+	return err;
+}
+
 static u32 mlx5_ptp_shift_constant(u32 dev_freq_khz)
 {
 	/* Optimal shift constant leads to corrections above just 1 scaled ppm.
@@ -119,21 +187,30 @@ static u32 mlx5_ptp_shift_constant(u32 dev_freq_khz)
 		   ilog2((U32_MAX / NSEC_PER_MSEC) * dev_freq_khz));
 }
 
+static s32 mlx5_clock_getmaxphase(struct mlx5_core_dev *mdev)
+{
+	return MLX5_CAP_MCAM_FEATURE(mdev, mtutc_time_adjustment_extended_range) ?
+		       MLX5_MTUTC_OPERATION_ADJUST_TIME_EXTENDED_MAX :
+			     MLX5_MTUTC_OPERATION_ADJUST_TIME_MAX;
+}
+
 static s32 mlx5_ptp_getmaxphase(struct ptp_clock_info *ptp)
 {
 	struct mlx5_clock *clock = container_of(ptp, struct mlx5_clock, ptp_info);
 	struct mlx5_core_dev *mdev;
+	s32 ret;
 
-	mdev = container_of(clock, struct mlx5_core_dev, clock);
+	mlx5_clock_lock(clock);
+	mdev = mlx5_clock_mdev_get(clock);
+	ret = mlx5_clock_getmaxphase(mdev);
+	mlx5_clock_unlock(clock);
 
-	return MLX5_CAP_MCAM_FEATURE(mdev, mtutc_time_adjustment_extended_range) ?
-		       MLX5_MTUTC_OPERATION_ADJUST_TIME_EXTENDED_MAX :
-			     MLX5_MTUTC_OPERATION_ADJUST_TIME_MAX;
+	return ret;
 }
 
 static bool mlx5_is_mtutc_time_adj_cap(struct mlx5_core_dev *mdev, s64 delta)
 {
-	s64 max = mlx5_ptp_getmaxphase(&mdev->clock.ptp_info);
+	s64 max = mlx5_clock_getmaxphase(mdev);
 
 	if (delta < -max || delta > max)
 		return false;
@@ -209,7 +286,7 @@ static int mlx5_mtctr_syncdevicetime(ktime_t *device_time,
 	if (real_time_mode)
 		*device_time = ns_to_ktime(REAL_TIME_TO_NS(device >> 32, device & U32_MAX));
 	else
-		*device_time = mlx5_timecounter_cyc2time(&mdev->clock, device);
+		*device_time = mlx5_timecounter_cyc2time(mdev->clock, device);
 
 	return 0;
 }
@@ -220,16 +297,23 @@ static int mlx5_ptp_getcrosststamp(struct ptp_clock_info *ptp,
 	struct mlx5_clock *clock = container_of(ptp, struct mlx5_clock, ptp_info);
 	struct system_time_snapshot history_begin = {0};
 	struct mlx5_core_dev *mdev;
+	int err;
 
-	mdev = container_of(clock, struct mlx5_core_dev, clock);
+	mlx5_clock_lock(clock);
+	mdev = mlx5_clock_mdev_get(clock);
 
-	if (!mlx5_is_ptm_source_time_available(mdev))
-		return -EBUSY;
+	if (!mlx5_is_ptm_source_time_available(mdev)) {
+		err = -EBUSY;
+		goto unlock;
+	}
 
 	ktime_get_snapshot(&history_begin);
 
-	return get_device_system_crosststamp(mlx5_mtctr_syncdevicetime, mdev,
-					     &history_begin, cts);
+	err = get_device_system_crosststamp(mlx5_mtctr_syncdevicetime, mdev,
+					    &history_begin, cts);
+unlock:
+	mlx5_clock_unlock(clock);
+	return err;
 }
 #endif /* CONFIG_X86 */
 
@@ -263,8 +347,7 @@ static u64 read_internal_timer(const struct cyclecounter *cc)
 {
 	struct mlx5_timer *timer = container_of(cc, struct mlx5_timer, cycles);
 	struct mlx5_clock *clock = container_of(timer, struct mlx5_clock, timer);
-	struct mlx5_core_dev *mdev = container_of(clock, struct mlx5_core_dev,
-						  clock);
+	struct mlx5_core_dev *mdev = mlx5_clock_mdev_get(clock);
 
 	return mlx5_read_time(mdev, NULL, false) & cc->mask;
 }
@@ -272,7 +355,7 @@ static u64 read_internal_timer(const struct cyclecounter *cc)
 static void mlx5_update_clock_info_page(struct mlx5_core_dev *mdev)
 {
 	struct mlx5_ib_clock_info *clock_info = mdev->clock_info;
-	struct mlx5_clock *clock = &mdev->clock;
+	struct mlx5_clock *clock = mdev->clock;
 	struct mlx5_timer *timer;
 	u32 sign;
 
@@ -295,12 +378,10 @@ static void mlx5_update_clock_info_page(struct mlx5_core_dev *mdev)
 
 static void mlx5_pps_out(struct work_struct *work)
 {
-	struct mlx5_pps *pps_info = container_of(work, struct mlx5_pps,
-						 out_work);
-	struct mlx5_clock *clock = container_of(pps_info, struct mlx5_clock,
-						pps_info);
-	struct mlx5_core_dev *mdev = container_of(clock, struct mlx5_core_dev,
-						  clock);
+	struct mlx5_clock_dev_state *clock_state = container_of(work, struct mlx5_clock_dev_state,
+								out_work);
+	struct mlx5_core_dev *mdev = clock_state->mdev;
+	struct mlx5_clock *clock = mdev->clock;
 	u32 in[MLX5_ST_SZ_DW(mtpps_reg)] = {0};
 	unsigned long flags;
 	int i;
@@ -330,7 +411,8 @@ static long mlx5_timestamp_overflow(struct ptp_clock_info *ptp_info)
 	unsigned long flags;
 
 	clock = container_of(ptp_info, struct mlx5_clock, ptp_info);
-	mdev = container_of(clock, struct mlx5_core_dev, clock);
+	mlx5_clock_lock(clock);
+	mdev = mlx5_clock_mdev_get(clock);
 	timer = &clock->timer;
 
 	if (mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR)
@@ -342,6 +424,7 @@ static long mlx5_timestamp_overflow(struct ptp_clock_info *ptp_info)
 	write_sequnlock_irqrestore(&clock->lock, flags);
 
 out:
+	mlx5_clock_unlock(clock);
 	return timer->overflow_period;
 }
 
@@ -361,15 +444,12 @@ static int mlx5_ptp_settime_real_time(struct mlx5_core_dev *mdev,
 	return mlx5_set_mtutc(mdev, in, sizeof(in));
 }
 
-static int mlx5_ptp_settime(struct ptp_clock_info *ptp, const struct timespec64 *ts)
+static int mlx5_clock_settime(struct mlx5_core_dev *mdev, struct mlx5_clock *clock,
+			      const struct timespec64 *ts)
 {
-	struct mlx5_clock *clock = container_of(ptp, struct mlx5_clock, ptp_info);
 	struct mlx5_timer *timer = &clock->timer;
-	struct mlx5_core_dev *mdev;
 	unsigned long flags;
 
-	mdev = container_of(clock, struct mlx5_core_dev, clock);
-
 	if (mlx5_modify_mtutc_allowed(mdev)) {
 		int err = mlx5_ptp_settime_real_time(mdev, ts);
 
@@ -385,6 +465,20 @@ static int mlx5_ptp_settime(struct ptp_clock_info *ptp, const struct timespec64
 	return 0;
 }
 
+static int mlx5_ptp_settime(struct ptp_clock_info *ptp, const struct timespec64 *ts)
+{
+	struct mlx5_clock *clock = container_of(ptp, struct mlx5_clock, ptp_info);
+	struct mlx5_core_dev *mdev;
+	int err;
+
+	mlx5_clock_lock(clock);
+	mdev = mlx5_clock_mdev_get(clock);
+	err = mlx5_clock_settime(mdev, clock, ts);
+	mlx5_clock_unlock(clock);
+
+	return err;
+}
+
 static
 struct timespec64 mlx5_ptp_gettimex_real_time(struct mlx5_core_dev *mdev,
 					      struct ptp_system_timestamp *sts)
@@ -404,7 +498,8 @@ static int mlx5_ptp_gettimex(struct ptp_clock_info *ptp, struct timespec64 *ts,
 	struct mlx5_core_dev *mdev;
 	u64 cycles, ns;
 
-	mdev = container_of(clock, struct mlx5_core_dev, clock);
+	mlx5_clock_lock(clock);
+	mdev = mlx5_clock_mdev_get(clock);
 	if (mlx5_real_time_mode(mdev)) {
 		*ts = mlx5_ptp_gettimex_real_time(mdev, sts);
 		goto out;
@@ -414,6 +509,7 @@ static int mlx5_ptp_gettimex(struct ptp_clock_info *ptp, struct timespec64 *ts,
 	ns = mlx5_timecounter_cyc2time(clock, cycles);
 	*ts = ns_to_timespec64(ns);
 out:
+	mlx5_clock_unlock(clock);
 	return 0;
 }
 
@@ -444,14 +540,16 @@ static int mlx5_ptp_adjtime(struct ptp_clock_info *ptp, s64 delta)
 	struct mlx5_timer *timer = &clock->timer;
 	struct mlx5_core_dev *mdev;
 	unsigned long flags;
+	int err = 0;
 
-	mdev = container_of(clock, struct mlx5_core_dev, clock);
+	mlx5_clock_lock(clock);
+	mdev = mlx5_clock_mdev_get(clock);
 
 	if (mlx5_modify_mtutc_allowed(mdev)) {
-		int err = mlx5_ptp_adjtime_real_time(mdev, delta);
+		err = mlx5_ptp_adjtime_real_time(mdev, delta);
 
 		if (err)
-			return err;
+			goto unlock;
 	}
 
 	write_seqlock_irqsave(&clock->lock, flags);
@@ -459,17 +557,23 @@ static int mlx5_ptp_adjtime(struct ptp_clock_info *ptp, s64 delta)
 	mlx5_update_clock_info_page(mdev);
 	write_sequnlock_irqrestore(&clock->lock, flags);
 
-	return 0;
+unlock:
+	mlx5_clock_unlock(clock);
+	return err;
 }
 
 static int mlx5_ptp_adjphase(struct ptp_clock_info *ptp, s32 delta)
 {
 	struct mlx5_clock *clock = container_of(ptp, struct mlx5_clock, ptp_info);
 	struct mlx5_core_dev *mdev;
+	int err;
 
-	mdev = container_of(clock, struct mlx5_core_dev, clock);
+	mlx5_clock_lock(clock);
+	mdev = mlx5_clock_mdev_get(clock);
+	err = mlx5_ptp_adjtime_real_time(mdev, delta);
+	mlx5_clock_unlock(clock);
 
-	return mlx5_ptp_adjtime_real_time(mdev, delta);
+	return err;
 }
 
 static int mlx5_ptp_freq_adj_real_time(struct mlx5_core_dev *mdev, long scaled_ppm)
@@ -498,15 +602,17 @@ static int mlx5_ptp_adjfine(struct ptp_clock_info *ptp, long scaled_ppm)
 	struct mlx5_timer *timer = &clock->timer;
 	struct mlx5_core_dev *mdev;
 	unsigned long flags;
+	int err = 0;
 	u32 mult;
 
-	mdev = container_of(clock, struct mlx5_core_dev, clock);
+	mlx5_clock_lock(clock);
+	mdev = mlx5_clock_mdev_get(clock);
 
 	if (mlx5_modify_mtutc_allowed(mdev)) {
-		int err = mlx5_ptp_freq_adj_real_time(mdev, scaled_ppm);
+		err = mlx5_ptp_freq_adj_real_time(mdev, scaled_ppm);
 
 		if (err)
-			return err;
+			goto unlock;
 	}
 
 	mult = (u32)adjust_by_scaled_ppm(timer->nominal_c_mult, scaled_ppm);
@@ -518,7 +624,9 @@ static int mlx5_ptp_adjfine(struct ptp_clock_info *ptp, long scaled_ppm)
 	write_sequnlock_irqrestore(&clock->lock, flags);
 	ptp_schedule_worker(clock->ptp, timer->overflow_period);
 
-	return 0;
+unlock:
+	mlx5_clock_unlock(clock);
+	return err;
 }
 
 static int mlx5_extts_configure(struct ptp_clock_info *ptp,
@@ -527,18 +635,14 @@ static int mlx5_extts_configure(struct ptp_clock_info *ptp,
 {
 	struct mlx5_clock *clock =
 			container_of(ptp, struct mlx5_clock, ptp_info);
-	struct mlx5_core_dev *mdev =
-			container_of(clock, struct mlx5_core_dev, clock);
 	u32 in[MLX5_ST_SZ_DW(mtpps_reg)] = {0};
+	struct mlx5_core_dev *mdev;
 	u32 field_select = 0;
 	u8 pin_mode = 0;
 	u8 pattern = 0;
 	int pin = -1;
 	int err = 0;
 
-	if (!MLX5_PPS_CAP(mdev))
-		return -EOPNOTSUPP;
-
 	/* Reject requests with unsupported flags */
 	if (rq->extts.flags & ~(PTP_ENABLE_FEATURE |
 				PTP_RISING_EDGE |
@@ -569,6 +673,14 @@ static int mlx5_extts_configure(struct ptp_clock_info *ptp,
 		field_select = MLX5_MTPPS_FS_ENABLE;
 	}
 
+	mlx5_clock_lock(clock);
+	mdev = mlx5_clock_mdev_get(clock);
+
+	if (!MLX5_PPS_CAP(mdev)) {
+		err = -EOPNOTSUPP;
+		goto unlock;
+	}
+
 	MLX5_SET(mtpps_reg, in, pin, pin);
 	MLX5_SET(mtpps_reg, in, pin_mode, pin_mode);
 	MLX5_SET(mtpps_reg, in, pattern, pattern);
@@ -577,15 +689,23 @@ static int mlx5_extts_configure(struct ptp_clock_info *ptp,
 
 	err = mlx5_set_mtpps(mdev, in, sizeof(in));
 	if (err)
-		return err;
+		goto unlock;
+
+	err = mlx5_set_mtppse(mdev, pin, 0, MLX5_EVENT_MODE_REPETETIVE & on);
+	if (err)
+		goto unlock;
+
+	clock->pps_info.pin_armed[pin] = on;
+	clock_priv(clock)->event_mdev = mdev;
 
-	return mlx5_set_mtppse(mdev, pin, 0,
-			       MLX5_EVENT_MODE_REPETETIVE & on);
+unlock:
+	mlx5_clock_unlock(clock);
+	return err;
 }
 
 static u64 find_target_cycles(struct mlx5_core_dev *mdev, s64 target_ns)
 {
-	struct mlx5_clock *clock = &mdev->clock;
+	struct mlx5_clock *clock = mdev->clock;
 	u64 cycles_now, cycles_delta;
 	u64 nsec_now, nsec_delta;
 	struct mlx5_timer *timer;
@@ -644,7 +764,7 @@ static int mlx5_perout_conf_out_pulse_duration(struct mlx5_core_dev *mdev,
 					       struct ptp_clock_request *rq,
 					       u32 *out_pulse_duration_ns)
 {
-	struct mlx5_pps *pps_info = &mdev->clock.pps_info;
+	struct mlx5_pps *pps_info = &mdev->clock->pps_info;
 	u32 out_pulse_duration;
 	struct timespec64 ts;
 
@@ -677,7 +797,7 @@ static int perout_conf_npps_real_time(struct mlx5_core_dev *mdev, struct ptp_clo
 				      u32 *field_select, u32 *out_pulse_duration_ns,
 				      u64 *period, u64 *time_stamp)
 {
-	struct mlx5_pps *pps_info = &mdev->clock.pps_info;
+	struct mlx5_pps *pps_info = &mdev->clock->pps_info;
 	struct ptp_clock_time *time = &rq->perout.start;
 	struct timespec64 ts;
 
@@ -712,26 +832,18 @@ static int mlx5_perout_configure(struct ptp_clock_info *ptp,
 {
 	struct mlx5_clock *clock =
 			container_of(ptp, struct mlx5_clock, ptp_info);
-	struct mlx5_core_dev *mdev =
-			container_of(clock, struct mlx5_core_dev, clock);
-	bool rt_mode = mlx5_real_time_mode(mdev);
 	u32 in[MLX5_ST_SZ_DW(mtpps_reg)] = {0};
 	u32 out_pulse_duration_ns = 0;
+	struct mlx5_core_dev *mdev;
 	u32 field_select = 0;
 	u64 npps_period = 0;
 	u64 time_stamp = 0;
 	u8 pin_mode = 0;
 	u8 pattern = 0;
+	bool rt_mode;
 	int pin = -1;
 	int err = 0;
 
-	if (!MLX5_PPS_CAP(mdev))
-		return -EOPNOTSUPP;
-
-	/* Reject requests with unsupported flags */
-	if (mlx5_perout_verify_flags(mdev, rq->perout.flags))
-		return -EOPNOTSUPP;
-
 	if (rq->perout.index >= clock->ptp_info.n_pins)
 		return -EINVAL;
 
@@ -740,14 +852,29 @@ static int mlx5_perout_configure(struct ptp_clock_info *ptp,
 	if (pin < 0)
 		return -EBUSY;
 
-	if (on) {
-		bool rt_mode = mlx5_real_time_mode(mdev);
+	mlx5_clock_lock(clock);
+	mdev = mlx5_clock_mdev_get(clock);
+	rt_mode = mlx5_real_time_mode(mdev);
+
+	if (!MLX5_PPS_CAP(mdev)) {
+		err = -EOPNOTSUPP;
+		goto unlock;
+	}
+
+	/* Reject requests with unsupported flags */
+	if (mlx5_perout_verify_flags(mdev, rq->perout.flags)) {
+		err = -EOPNOTSUPP;
+		goto unlock;
+	}
 
+	if (on) {
 		pin_mode = MLX5_PIN_MODE_OUT;
 		pattern = MLX5_OUT_PATTERN_PERIODIC;
 
-		if (rt_mode &&  rq->perout.start.sec > U32_MAX)
-			return -EINVAL;
+		if (rt_mode &&  rq->perout.start.sec > U32_MAX) {
+			err = -EINVAL;
+			goto unlock;
+		}
 
 		field_select |= MLX5_MTPPS_FS_PIN_MODE |
 				MLX5_MTPPS_FS_PATTERN |
@@ -760,7 +887,7 @@ static int mlx5_perout_configure(struct ptp_clock_info *ptp,
 		else
 			err = perout_conf_1pps(mdev, rq, &time_stamp, rt_mode);
 		if (err)
-			return err;
+			goto unlock;
 	}
 
 	MLX5_SET(mtpps_reg, in, pin, pin);
@@ -773,13 +900,16 @@ static int mlx5_perout_configure(struct ptp_clock_info *ptp,
 	MLX5_SET(mtpps_reg, in, out_pulse_duration_ns, out_pulse_duration_ns);
 	err = mlx5_set_mtpps(mdev, in, sizeof(in));
 	if (err)
-		return err;
+		goto unlock;
 
 	if (rt_mode)
-		return 0;
+		goto unlock;
+
+	err = mlx5_set_mtppse(mdev, pin, 0, MLX5_EVENT_MODE_REPETETIVE & on);
 
-	return mlx5_set_mtppse(mdev, pin, 0,
-			       MLX5_EVENT_MODE_REPETETIVE & on);
+unlock:
+	mlx5_clock_unlock(clock);
+	return err;
 }
 
 static int mlx5_pps_configure(struct ptp_clock_info *ptp,
@@ -866,10 +996,8 @@ static int mlx5_query_mtpps_pin_mode(struct mlx5_core_dev *mdev, u8 pin,
 				    mtpps_size, MLX5_REG_MTPPS, 0, 0);
 }
 
-static int mlx5_get_pps_pin_mode(struct mlx5_clock *clock, u8 pin)
+static int mlx5_get_pps_pin_mode(struct mlx5_core_dev *mdev, u8 pin)
 {
-	struct mlx5_core_dev *mdev = container_of(clock, struct mlx5_core_dev, clock);
-
 	u32 out[MLX5_ST_SZ_DW(mtpps_reg)] = {};
 	u8 mode;
 	int err;
@@ -888,8 +1016,9 @@ static int mlx5_get_pps_pin_mode(struct mlx5_clock *clock, u8 pin)
 	return PTP_PF_NONE;
 }
 
-static void mlx5_init_pin_config(struct mlx5_clock *clock)
+static void mlx5_init_pin_config(struct mlx5_core_dev *mdev)
 {
+	struct mlx5_clock *clock = mdev->clock;
 	int i;
 
 	if (!clock->ptp_info.n_pins)
@@ -910,15 +1039,15 @@ static void mlx5_init_pin_config(struct mlx5_clock *clock)
 			 sizeof(clock->ptp_info.pin_config[i].name),
 			 "mlx5_pps%d", i);
 		clock->ptp_info.pin_config[i].index = i;
-		clock->ptp_info.pin_config[i].func = mlx5_get_pps_pin_mode(clock, i);
+		clock->ptp_info.pin_config[i].func = mlx5_get_pps_pin_mode(mdev, i);
 		clock->ptp_info.pin_config[i].chan = 0;
 	}
 }
 
 static void mlx5_get_pps_caps(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_clock *clock = &mdev->clock;
 	u32 out[MLX5_ST_SZ_DW(mtpps_reg)] = {0};
+	struct mlx5_clock *clock = mdev->clock;
 
 	mlx5_query_mtpps(mdev, out, sizeof(out));
 
@@ -968,16 +1097,16 @@ static u64 perout_conf_next_event_timer(struct mlx5_core_dev *mdev,
 static int mlx5_pps_event(struct notifier_block *nb,
 			  unsigned long type, void *data)
 {
-	struct mlx5_clock *clock = mlx5_nb_cof(nb, struct mlx5_clock, pps_nb);
+	struct mlx5_clock_dev_state *clock_state = mlx5_nb_cof(nb, struct mlx5_clock_dev_state,
+							       pps_nb);
+	struct mlx5_core_dev *mdev = clock_state->mdev;
+	struct mlx5_clock *clock = mdev->clock;
 	struct ptp_clock_event ptp_event;
 	struct mlx5_eqe *eqe = data;
 	int pin = eqe->data.pps.pin;
-	struct mlx5_core_dev *mdev;
 	unsigned long flags;
 	u64 ns;
 
-	mdev = container_of(clock, struct mlx5_core_dev, clock);
-
 	switch (clock->ptp_info.pin_config[pin].func) {
 	case PTP_PF_EXTTS:
 		ptp_event.index = pin;
@@ -997,11 +1126,15 @@ static int mlx5_pps_event(struct notifier_block *nb,
 		ptp_clock_event(clock->ptp, &ptp_event);
 		break;
 	case PTP_PF_PEROUT:
+		if (clock->shared) {
+			mlx5_core_warn(mdev, " Received unexpected PPS out event\n");
+			break;
+		}
 		ns = perout_conf_next_event_timer(mdev, clock);
 		write_seqlock_irqsave(&clock->lock, flags);
 		clock->pps_info.start[pin] = ns;
 		write_sequnlock_irqrestore(&clock->lock, flags);
-		schedule_work(&clock->pps_info.out_work);
+		schedule_work(&clock_state->out_work);
 		break;
 	default:
 		mlx5_core_err(mdev, " Unhandled clock PPS event, func %d\n",
@@ -1013,7 +1146,7 @@ static int mlx5_pps_event(struct notifier_block *nb,
 
 static void mlx5_timecounter_init(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_clock *clock = &mdev->clock;
+	struct mlx5_clock *clock = mdev->clock;
 	struct mlx5_timer *timer = &clock->timer;
 	u32 dev_freq;
 
@@ -1029,10 +1162,10 @@ static void mlx5_timecounter_init(struct mlx5_core_dev *mdev)
 			 ktime_to_ns(ktime_get_real()));
 }
 
-static void mlx5_init_overflow_period(struct mlx5_clock *clock)
+static void mlx5_init_overflow_period(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_core_dev *mdev = container_of(clock, struct mlx5_core_dev, clock);
 	struct mlx5_ib_clock_info *clock_info = mdev->clock_info;
+	struct mlx5_clock *clock = mdev->clock;
 	struct mlx5_timer *timer = &clock->timer;
 	u64 overflow_cycles;
 	u64 frac = 0;
@@ -1065,7 +1198,7 @@ static void mlx5_init_overflow_period(struct mlx5_clock *clock)
 
 static void mlx5_init_clock_info(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_clock *clock = &mdev->clock;
+	struct mlx5_clock *clock = mdev->clock;
 	struct mlx5_ib_clock_info *info;
 	struct mlx5_timer *timer;
 
@@ -1088,7 +1221,7 @@ static void mlx5_init_clock_info(struct mlx5_core_dev *mdev)
 
 static void mlx5_init_timer_max_freq_adjustment(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_clock *clock = &mdev->clock;
+	struct mlx5_clock *clock = mdev->clock;
 	u32 out[MLX5_ST_SZ_DW(mtutc_reg)] = {};
 	u32 in[MLX5_ST_SZ_DW(mtutc_reg)] = {};
 	u8 log_max_freq_adjustment = 0;
@@ -1107,7 +1240,7 @@ static void mlx5_init_timer_max_freq_adjustment(struct mlx5_core_dev *mdev)
 
 static void mlx5_init_timer_clock(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_clock *clock = &mdev->clock;
+	struct mlx5_clock *clock = mdev->clock;
 
 	/* Configure the PHC */
 	clock->ptp_info = mlx5_ptp_clock_info;
@@ -1123,38 +1256,30 @@ static void mlx5_init_timer_clock(struct mlx5_core_dev *mdev)
 
 	mlx5_timecounter_init(mdev);
 	mlx5_init_clock_info(mdev);
-	mlx5_init_overflow_period(clock);
+	mlx5_init_overflow_period(mdev);
 
 	if (mlx5_real_time_mode(mdev)) {
 		struct timespec64 ts;
 
 		ktime_get_real_ts64(&ts);
-		mlx5_ptp_settime(&clock->ptp_info, &ts);
+		mlx5_clock_settime(mdev, clock, &ts);
 	}
 }
 
 static void mlx5_init_pps(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_clock *clock = &mdev->clock;
-
 	if (!MLX5_PPS_CAP(mdev))
 		return;
 
 	mlx5_get_pps_caps(mdev);
-	mlx5_init_pin_config(clock);
+	mlx5_init_pin_config(mdev);
 }
 
-void mlx5_init_clock(struct mlx5_core_dev *mdev)
+static void mlx5_init_clock_dev(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_clock *clock = &mdev->clock;
-
-	if (!MLX5_CAP_GEN(mdev, device_frequency_khz)) {
-		mlx5_core_warn(mdev, "invalid device_frequency_khz, aborting HW clock init\n");
-		return;
-	}
+	struct mlx5_clock *clock = mdev->clock;
 
 	seqlock_init(&clock->lock);
-	INIT_WORK(&clock->pps_info.out_work, mlx5_pps_out);
 
 	/* Initialize the device clock */
 	mlx5_init_timer_clock(mdev);
@@ -1163,35 +1288,27 @@ void mlx5_init_clock(struct mlx5_core_dev *mdev)
 	mlx5_init_pps(mdev);
 
 	clock->ptp = ptp_clock_register(&clock->ptp_info,
-					&mdev->pdev->dev);
+					clock->shared ? NULL : &mdev->pdev->dev);
 	if (IS_ERR(clock->ptp)) {
-		mlx5_core_warn(mdev, "ptp_clock_register failed %ld\n",
+		mlx5_core_warn(mdev, "%sptp_clock_register failed %ld\n",
+			       clock->shared ? "shared clock " : "",
 			       PTR_ERR(clock->ptp));
 		clock->ptp = NULL;
 	}
 
-	MLX5_NB_INIT(&clock->pps_nb, mlx5_pps_event, PPS_EVENT);
-	mlx5_eq_notifier_register(mdev, &clock->pps_nb);
-
 	if (clock->ptp)
 		ptp_schedule_worker(clock->ptp, 0);
 }
 
-void mlx5_cleanup_clock(struct mlx5_core_dev *mdev)
+static void mlx5_destroy_clock_dev(struct mlx5_core_dev *mdev)
 {
-	struct mlx5_clock *clock = &mdev->clock;
+	struct mlx5_clock *clock = mdev->clock;
 
-	if (!MLX5_CAP_GEN(mdev, device_frequency_khz))
-		return;
-
-	mlx5_eq_notifier_unregister(mdev, &clock->pps_nb);
 	if (clock->ptp) {
 		ptp_clock_unregister(clock->ptp);
 		clock->ptp = NULL;
 	}
 
-	cancel_work_sync(&clock->pps_info.out_work);
-
 	if (mdev->clock_info) {
 		free_page((unsigned long)mdev->clock_info);
 		mdev->clock_info = NULL;
@@ -1199,3 +1316,248 @@ void mlx5_cleanup_clock(struct mlx5_core_dev *mdev)
 
 	kfree(clock->ptp_info.pin_config);
 }
+
+static void mlx5_clock_free(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_clock_priv *cpriv = clock_priv(mdev->clock);
+
+	mlx5_destroy_clock_dev(mdev);
+	mutex_destroy(&cpriv->lock);
+	kfree(cpriv);
+	mdev->clock = NULL;
+}
+
+static int mlx5_clock_alloc(struct mlx5_core_dev *mdev, bool shared)
+{
+	struct mlx5_clock_priv *cpriv;
+	struct mlx5_clock *clock;
+
+	cpriv = kzalloc(sizeof(*cpriv), GFP_KERNEL);
+	if (!cpriv)
+		return -ENOMEM;
+
+	mutex_init(&cpriv->lock);
+	cpriv->mdev = mdev;
+	clock = &cpriv->clock;
+	clock->shared = shared;
+	mdev->clock = clock;
+	mlx5_clock_lock(clock);
+	mlx5_init_clock_dev(mdev);
+	mlx5_clock_unlock(clock);
+
+	if (!clock->shared)
+		return 0;
+
+	if (!clock->ptp) {
+		mlx5_core_warn(mdev, "failed to create ptp dev shared by multiple functions");
+		mlx5_clock_free(mdev);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static void mlx5_shared_clock_register(struct mlx5_core_dev *mdev, u64 key)
+{
+	struct mlx5_core_dev *peer_dev, *next = NULL;
+	struct mlx5_devcom_comp_dev *pos;
+
+	mdev->clock_state->compdev = mlx5_devcom_register_component(mdev->priv.devc,
+								    MLX5_DEVCOM_SHARED_CLOCK,
+								    key, NULL, mdev);
+	if (IS_ERR(mdev->clock_state->compdev))
+		return;
+
+	mlx5_devcom_comp_lock(mdev->clock_state->compdev);
+	mlx5_devcom_for_each_peer_entry(mdev->clock_state->compdev, peer_dev, pos) {
+		if (peer_dev->clock) {
+			next = peer_dev;
+			break;
+		}
+	}
+
+	if (next) {
+		mdev->clock = next->clock;
+		/* clock info is shared among all the functions using the same clock */
+		mdev->clock_info = next->clock_info;
+	} else {
+		mlx5_clock_alloc(mdev, true);
+	}
+	mlx5_devcom_comp_unlock(mdev->clock_state->compdev);
+
+	if (!mdev->clock) {
+		mlx5_devcom_unregister_component(mdev->clock_state->compdev);
+		mdev->clock_state->compdev = NULL;
+	}
+}
+
+static void mlx5_shared_clock_unregister(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_core_dev *peer_dev, *next = NULL;
+	struct mlx5_clock *clock = mdev->clock;
+	struct mlx5_devcom_comp_dev *pos;
+
+	mlx5_devcom_comp_lock(mdev->clock_state->compdev);
+	mlx5_devcom_for_each_peer_entry(mdev->clock_state->compdev, peer_dev, pos) {
+		if (peer_dev->clock && peer_dev != mdev) {
+			next = peer_dev;
+			break;
+		}
+	}
+
+	if (next) {
+		struct mlx5_clock_priv *cpriv = clock_priv(clock);
+
+		mlx5_clock_lock(clock);
+		if (mdev == cpriv->mdev)
+			cpriv->mdev = next;
+		mlx5_clock_unlock(clock);
+	} else {
+		mlx5_clock_free(mdev);
+	}
+
+	mdev->clock = NULL;
+	mdev->clock_info = NULL;
+	mlx5_devcom_comp_unlock(mdev->clock_state->compdev);
+
+	mlx5_devcom_unregister_component(mdev->clock_state->compdev);
+}
+
+static void mlx5_clock_arm_pps_in_event(struct mlx5_clock *clock,
+					struct mlx5_core_dev *new_mdev,
+					struct mlx5_core_dev *old_mdev)
+{
+	struct ptp_clock_info *ptp_info = &clock->ptp_info;
+	struct mlx5_clock_priv *cpriv = clock_priv(clock);
+	int i;
+
+	for (i = 0; i < ptp_info->n_pins; i++) {
+		if (ptp_info->pin_config[i].func != PTP_PF_EXTTS ||
+		    !clock->pps_info.pin_armed[i])
+			continue;
+
+		if (new_mdev) {
+			mlx5_set_mtppse(new_mdev, i, 0, MLX5_EVENT_MODE_REPETETIVE);
+			cpriv->event_mdev = new_mdev;
+		} else {
+			cpriv->event_mdev = NULL;
+		}
+
+		if (old_mdev)
+			mlx5_set_mtppse(old_mdev, i, 0, MLX5_EVENT_MODE_DISABLE);
+	}
+}
+
+void mlx5_clock_load(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_clock *clock = mdev->clock;
+	struct mlx5_clock_priv *cpriv;
+
+	if (!MLX5_CAP_GEN(mdev, device_frequency_khz))
+		return;
+
+	INIT_WORK(&mdev->clock_state->out_work, mlx5_pps_out);
+	MLX5_NB_INIT(&mdev->clock_state->pps_nb, mlx5_pps_event, PPS_EVENT);
+	mlx5_eq_notifier_register(mdev, &mdev->clock_state->pps_nb);
+
+	if (!clock->shared) {
+		mlx5_clock_arm_pps_in_event(clock, mdev, NULL);
+		return;
+	}
+
+	cpriv = clock_priv(clock);
+	mlx5_devcom_comp_lock(mdev->clock_state->compdev);
+	mlx5_clock_lock(clock);
+	if (mdev == cpriv->mdev && mdev != cpriv->event_mdev)
+		mlx5_clock_arm_pps_in_event(clock, mdev, cpriv->event_mdev);
+	mlx5_clock_unlock(clock);
+	mlx5_devcom_comp_unlock(mdev->clock_state->compdev);
+}
+
+void mlx5_clock_unload(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_core_dev *peer_dev, *next = NULL;
+	struct mlx5_clock *clock = mdev->clock;
+	struct mlx5_devcom_comp_dev *pos;
+
+	if (!MLX5_CAP_GEN(mdev, device_frequency_khz))
+		return;
+
+	if (!clock->shared) {
+		mlx5_clock_arm_pps_in_event(clock, NULL, mdev);
+		goto out;
+	}
+
+	mlx5_devcom_comp_lock(mdev->clock_state->compdev);
+	mlx5_devcom_for_each_peer_entry(mdev->clock_state->compdev, peer_dev, pos) {
+		if (peer_dev->clock && peer_dev != mdev) {
+			next = peer_dev;
+			break;
+		}
+	}
+
+	mlx5_clock_lock(clock);
+	if (mdev == clock_priv(clock)->event_mdev)
+		mlx5_clock_arm_pps_in_event(clock, next, mdev);
+	mlx5_clock_unlock(clock);
+	mlx5_devcom_comp_unlock(mdev->clock_state->compdev);
+
+out:
+	mlx5_eq_notifier_unregister(mdev, &mdev->clock_state->pps_nb);
+	cancel_work_sync(&mdev->clock_state->out_work);
+}
+
+static struct mlx5_clock null_clock;
+
+int mlx5_init_clock(struct mlx5_core_dev *mdev)
+{
+	u8 identity[MLX5_RT_CLOCK_IDENTITY_SIZE];
+	struct mlx5_clock_dev_state *clock_state;
+	u64 key;
+	int err;
+
+	if (!MLX5_CAP_GEN(mdev, device_frequency_khz)) {
+		mdev->clock = &null_clock;
+		mlx5_core_warn(mdev, "invalid device_frequency_khz, aborting HW clock init\n");
+		return 0;
+	}
+
+	clock_state = kzalloc(sizeof(*clock_state), GFP_KERNEL);
+	if (!clock_state)
+		return -ENOMEM;
+	clock_state->mdev = mdev;
+	mdev->clock_state = clock_state;
+
+	if (MLX5_CAP_MCAM_REG3(mdev, mrtcq) && mlx5_real_time_mode(mdev)) {
+		if (mlx5_clock_identity_get(mdev, identity)) {
+			mlx5_core_warn(mdev, "failed to get rt clock identity, create ptp dev per function\n");
+		} else {
+			memcpy(&key, &identity, sizeof(key));
+			mlx5_shared_clock_register(mdev, key);
+		}
+	}
+
+	if (!mdev->clock) {
+		err = mlx5_clock_alloc(mdev, false);
+		if (err) {
+			kfree(clock_state);
+			mdev->clock_state = NULL;
+			return err;
+		}
+	}
+
+	return 0;
+}
+
+void mlx5_cleanup_clock(struct mlx5_core_dev *mdev)
+{
+	if (!MLX5_CAP_GEN(mdev, device_frequency_khz))
+		return;
+
+	if (mdev->clock->shared)
+		mlx5_shared_clock_unregister(mdev);
+	else
+		mlx5_clock_free(mdev);
+	kfree(mdev->clock_state);
+	mdev->clock_state = NULL;
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.h
index bd95b9f8d143..c18a652c0faa 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.h
@@ -33,6 +33,35 @@
 #ifndef __LIB_CLOCK_H__
 #define __LIB_CLOCK_H__
 
+#include <linux/ptp_clock_kernel.h>
+
+#define MAX_PIN_NUM	8
+struct mlx5_pps {
+	u8                         pin_caps[MAX_PIN_NUM];
+	u64                        start[MAX_PIN_NUM];
+	u8                         enabled;
+	u64                        min_npps_period;
+	u64                        min_out_pulse_duration_ns;
+	bool                       pin_armed[MAX_PIN_NUM];
+};
+
+struct mlx5_timer {
+	struct cyclecounter        cycles;
+	struct timecounter         tc;
+	u32                        nominal_c_mult;
+	unsigned long              overflow_period;
+};
+
+struct mlx5_clock {
+	seqlock_t                  lock;
+	struct hwtstamp_config     hwtstamp_config;
+	struct ptp_clock          *ptp;
+	struct ptp_clock_info      ptp_info;
+	struct mlx5_pps            pps_info;
+	struct mlx5_timer          timer;
+	bool                       shared;
+};
+
 static inline bool mlx5_is_real_time_rq(struct mlx5_core_dev *mdev)
 {
 	u8 rq_ts_format_cap = MLX5_CAP_GEN(mdev, rq_ts_format);
@@ -54,12 +83,14 @@ static inline bool mlx5_is_real_time_sq(struct mlx5_core_dev *mdev)
 typedef ktime_t (*cqe_ts_to_ns)(struct mlx5_clock *, u64);
 
 #if IS_ENABLED(CONFIG_PTP_1588_CLOCK)
-void mlx5_init_clock(struct mlx5_core_dev *mdev);
+int mlx5_init_clock(struct mlx5_core_dev *mdev);
 void mlx5_cleanup_clock(struct mlx5_core_dev *mdev);
+void mlx5_clock_load(struct mlx5_core_dev *mdev);
+void mlx5_clock_unload(struct mlx5_core_dev *mdev);
 
 static inline int mlx5_clock_get_ptp_index(struct mlx5_core_dev *mdev)
 {
-	return mdev->clock.ptp ? ptp_clock_index(mdev->clock.ptp) : -1;
+	return mdev->clock->ptp ? ptp_clock_index(mdev->clock->ptp) : -1;
 }
 
 static inline ktime_t mlx5_timecounter_cyc2time(struct mlx5_clock *clock,
@@ -87,8 +118,10 @@ static inline ktime_t mlx5_real_time_cyc2time(struct mlx5_clock *clock,
 	return ns_to_ktime(time);
 }
 #else
-static inline void mlx5_init_clock(struct mlx5_core_dev *mdev) {}
+static inline int mlx5_init_clock(struct mlx5_core_dev *mdev) { return 0; }
 static inline void mlx5_cleanup_clock(struct mlx5_core_dev *mdev) {}
+static inline void mlx5_clock_load(struct mlx5_core_dev *mdev) {}
+static inline void mlx5_clock_unload(struct mlx5_core_dev *mdev) {}
 static inline int mlx5_clock_get_ptp_index(struct mlx5_core_dev *mdev)
 {
 	return -1;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/devcom.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/devcom.h
index d58032dd0df7..c79699b94a02 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/devcom.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/devcom.h
@@ -11,6 +11,7 @@ enum mlx5_devcom_component {
 	MLX5_DEVCOM_MPV,
 	MLX5_DEVCOM_HCA_PORTS,
 	MLX5_DEVCOM_SD_GROUP,
+	MLX5_DEVCOM_SHARED_CLOCK,
 	MLX5_DEVCOM_NUM_COMPONENTS,
 };
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c
index ec956c4bcebd..710633d5fdbe 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c
@@ -1038,7 +1038,11 @@ static int mlx5_init_once(struct mlx5_core_dev *dev)
 
 	mlx5_init_reserved_gids(dev);
 
-	mlx5_init_clock(dev);
+	err = mlx5_init_clock(dev);
+	if (err) {
+		mlx5_core_err(dev, "failed to initialize hardware clock\n");
+		goto err_tables_cleanup;
+	}
 
 	dev->vxlan = mlx5_vxlan_create(dev);
 	dev->geneve = mlx5_geneve_create(dev);
@@ -1046,7 +1050,7 @@ static int mlx5_init_once(struct mlx5_core_dev *dev)
 	err = mlx5_init_rl_table(dev);
 	if (err) {
 		mlx5_core_err(dev, "Failed to init rate limiting\n");
-		goto err_tables_cleanup;
+		goto err_clock_cleanup;
 	}
 
 	err = mlx5_mpfs_init(dev);
@@ -1123,10 +1127,11 @@ err_mpfs_cleanup:
 	mlx5_mpfs_cleanup(dev);
 err_rl_cleanup:
 	mlx5_cleanup_rl_table(dev);
-err_tables_cleanup:
+err_clock_cleanup:
 	mlx5_geneve_destroy(dev->geneve);
 	mlx5_vxlan_destroy(dev->vxlan);
 	mlx5_cleanup_clock(dev);
+err_tables_cleanup:
 	mlx5_cleanup_reserved_gids(dev);
 	mlx5_cq_debugfs_cleanup(dev);
 	mlx5_fw_reset_cleanup(dev);
@@ -1359,6 +1364,8 @@ static int mlx5_load(struct mlx5_core_dev *dev)
 		goto err_eq_table;
 	}
 
+	mlx5_clock_load(dev);
+
 	err = mlx5_fw_tracer_init(dev->tracer);
 	if (err) {
 		mlx5_core_err(dev, "Failed to init FW tracer %d\n", err);
@@ -1442,6 +1449,7 @@ err_fpga_start:
 	mlx5_hv_vhca_cleanup(dev->hv_vhca);
 	mlx5_fw_reset_events_stop(dev);
 	mlx5_fw_tracer_cleanup(dev->tracer);
+	mlx5_clock_unload(dev);
 	mlx5_eq_table_destroy(dev);
 err_eq_table:
 	mlx5_irq_table_destroy(dev);
@@ -1468,6 +1476,7 @@ static void mlx5_unload(struct mlx5_core_dev *dev)
 	mlx5_hv_vhca_cleanup(dev->hv_vhca);
 	mlx5_fw_reset_events_stop(dev);
 	mlx5_fw_tracer_cleanup(dev->tracer);
+	mlx5_clock_unload(dev);
 	mlx5_eq_table_destroy(dev);
 	mlx5_irq_table_destroy(dev);
 	mlx5_pagealloc_stop(dev);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/port.c b/drivers/net/ethernet/mellanox/mlx5/core/port.c
index 50931584132b..3995df064101 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/port.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/port.c
@@ -1105,6 +1105,9 @@ static const u32 mlx5e_ext_link_speed[MLX5E_EXT_LINK_MODES_NUMBER] = {
 	[MLX5E_200GAUI_2_200GBASE_CR2_KR2] = 200000,
 	[MLX5E_400GAUI_4_400GBASE_CR4_KR4] = 400000,
 	[MLX5E_800GAUI_8_800GBASE_CR8_KR8] = 800000,
+	[MLX5E_200GAUI_1_200GBASE_CR1_KR1] = 200000,
+	[MLX5E_400GAUI_2_400GBASE_CR2_KR2] = 400000,
+	[MLX5E_800GAUI_4_800GBASE_CR4_KR4] = 800000,
 };
 
 int mlx5_port_query_eth_proto(struct mlx5_core_dev *dev, u8 port, bool ext,
diff --git a/drivers/net/phy/phy-core.c b/drivers/net/phy/phy-core.c
index 6bf3ec985f3d..f181f05cb429 100644
--- a/drivers/net/phy/phy-core.c
+++ b/drivers/net/phy/phy-core.c
@@ -13,7 +13,7 @@
  */
 const char *phy_speed_to_str(int speed)
 {
-	BUILD_BUG_ON_MSG(__ETHTOOL_LINK_MODE_MASK_NBITS != 103,
+	BUILD_BUG_ON_MSG(__ETHTOOL_LINK_MODE_MASK_NBITS != 121,
 		"Enum ethtool_link_mode_bit_indices and phylib are out of sync. "
 		"If a speed or mode has been added please update phy_speed_to_str "
 		"and the PHY settings array.\n");
@@ -169,6 +169,12 @@ static const struct phy_setting settings[] = {
 	PHY_SETTING( 800000, FULL, 800000baseDR8_2_Full		),
 	PHY_SETTING( 800000, FULL, 800000baseSR8_Full		),
 	PHY_SETTING( 800000, FULL, 800000baseVR8_Full		),
+	PHY_SETTING( 800000, FULL, 800000baseCR4_Full		),
+	PHY_SETTING( 800000, FULL, 800000baseKR4_Full		),
+	PHY_SETTING( 800000, FULL, 800000baseDR4_Full		),
+	PHY_SETTING( 800000, FULL, 800000baseDR4_2_Full		),
+	PHY_SETTING( 800000, FULL, 800000baseSR4_Full		),
+	PHY_SETTING( 800000, FULL, 800000baseVR4_Full		),
 	/* 400G */
 	PHY_SETTING( 400000, FULL, 400000baseCR8_Full		),
 	PHY_SETTING( 400000, FULL, 400000baseKR8_Full		),
@@ -180,6 +186,12 @@ static const struct phy_setting settings[] = {
 	PHY_SETTING( 400000, FULL, 400000baseLR4_ER4_FR4_Full	),
 	PHY_SETTING( 400000, FULL, 400000baseDR4_Full		),
 	PHY_SETTING( 400000, FULL, 400000baseSR4_Full		),
+	PHY_SETTING( 400000, FULL, 400000baseCR2_Full		),
+	PHY_SETTING( 400000, FULL, 400000baseKR2_Full		),
+	PHY_SETTING( 400000, FULL, 400000baseDR2_Full		),
+	PHY_SETTING( 400000, FULL, 400000baseDR2_2_Full		),
+	PHY_SETTING( 400000, FULL, 400000baseSR2_Full		),
+	PHY_SETTING( 400000, FULL, 400000baseVR2_Full		),
 	/* 200G */
 	PHY_SETTING( 200000, FULL, 200000baseCR4_Full		),
 	PHY_SETTING( 200000, FULL, 200000baseKR4_Full		),
@@ -191,6 +203,12 @@ static const struct phy_setting settings[] = {
 	PHY_SETTING( 200000, FULL, 200000baseLR2_ER2_FR2_Full	),
 	PHY_SETTING( 200000, FULL, 200000baseDR2_Full		),
 	PHY_SETTING( 200000, FULL, 200000baseSR2_Full		),
+	PHY_SETTING( 200000, FULL, 200000baseCR_Full		),
+	PHY_SETTING( 200000, FULL, 200000baseKR_Full		),
+	PHY_SETTING( 200000, FULL, 200000baseDR_Full		),
+	PHY_SETTING( 200000, FULL, 200000baseDR_2_Full		),
+	PHY_SETTING( 200000, FULL, 200000baseSR_Full		),
+	PHY_SETTING( 200000, FULL, 200000baseVR_Full		),
 	/* 100G */
 	PHY_SETTING( 100000, FULL, 100000baseCR4_Full		),
 	PHY_SETTING( 100000, FULL, 100000baseKR4_Full		),
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index af86097641b0..46bd7550adf8 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -54,7 +54,6 @@
 #include <linux/mlx5/doorbell.h>
 #include <linux/mlx5/eq.h>
 #include <linux/timecounter.h>
-#include <linux/ptp_clock_kernel.h>
 #include <net/devlink.h>
 
 #define MLX5_ADEV_NAME "mlx5_core"
@@ -679,33 +678,8 @@ struct mlx5_rsvd_gids {
 	struct ida ida;
 };
 
-#define MAX_PIN_NUM	8
-struct mlx5_pps {
-	u8                         pin_caps[MAX_PIN_NUM];
-	struct work_struct         out_work;
-	u64                        start[MAX_PIN_NUM];
-	u8                         enabled;
-	u64                        min_npps_period;
-	u64                        min_out_pulse_duration_ns;
-};
-
-struct mlx5_timer {
-	struct cyclecounter        cycles;
-	struct timecounter         tc;
-	u32                        nominal_c_mult;
-	unsigned long              overflow_period;
-};
-
-struct mlx5_clock {
-	struct mlx5_nb             pps_nb;
-	seqlock_t                  lock;
-	struct hwtstamp_config     hwtstamp_config;
-	struct ptp_clock          *ptp;
-	struct ptp_clock_info      ptp_info;
-	struct mlx5_pps            pps_info;
-	struct mlx5_timer          timer;
-};
-
+struct mlx5_clock;
+struct mlx5_clock_dev_state;
 struct mlx5_dm;
 struct mlx5_fw_tracer;
 struct mlx5_vxlan;
@@ -789,7 +763,8 @@ struct mlx5_core_dev {
 #ifdef CONFIG_MLX5_FPGA
 	struct mlx5_fpga_device *fpga;
 #endif
-	struct mlx5_clock        clock;
+	struct mlx5_clock       *clock;
+	struct mlx5_clock_dev_state *clock_state;
 	struct mlx5_ib_clock_info  *clock_info;
 	struct mlx5_fw_tracer   *tracer;
 	struct mlx5_rsc_dump    *rsc_dump;
diff --git a/include/linux/mlx5/port.h b/include/linux/mlx5/port.h
index e68d42b8ce65..fd625e0dd869 100644
--- a/include/linux/mlx5/port.h
+++ b/include/linux/mlx5/port.h
@@ -115,9 +115,12 @@ enum mlx5e_ext_link_mode {
 	MLX5E_100GAUI_1_100GBASE_CR_KR		= 11,
 	MLX5E_200GAUI_4_200GBASE_CR4_KR4	= 12,
 	MLX5E_200GAUI_2_200GBASE_CR2_KR2	= 13,
+	MLX5E_200GAUI_1_200GBASE_CR1_KR1	= 14,
 	MLX5E_400GAUI_8_400GBASE_CR8		= 15,
 	MLX5E_400GAUI_4_400GBASE_CR4_KR4	= 16,
+	MLX5E_400GAUI_2_400GBASE_CR2_KR2	= 17,
 	MLX5E_800GAUI_8_800GBASE_CR8_KR8	= 19,
+	MLX5E_800GAUI_4_800GBASE_CR4_KR4	= 20,
 	MLX5E_EXT_LINK_MODES_NUMBER,
 };
 
diff --git a/include/uapi/linux/ethtool.h b/include/uapi/linux/ethtool.h
index d1089b88efc7..e0bd726f84c1 100644
--- a/include/uapi/linux/ethtool.h
+++ b/include/uapi/linux/ethtool.h
@@ -2057,6 +2057,24 @@ enum ethtool_link_mode_bit_indices {
 	ETHTOOL_LINK_MODE_10baseT1S_Half_BIT		 = 100,
 	ETHTOOL_LINK_MODE_10baseT1S_P2MP_Half_BIT	 = 101,
 	ETHTOOL_LINK_MODE_10baseT1BRR_Full_BIT		 = 102,
+	ETHTOOL_LINK_MODE_200000baseCR_Full_BIT		 = 103,
+	ETHTOOL_LINK_MODE_200000baseKR_Full_BIT		 = 104,
+	ETHTOOL_LINK_MODE_200000baseDR_Full_BIT		 = 105,
+	ETHTOOL_LINK_MODE_200000baseDR_2_Full_BIT	 = 106,
+	ETHTOOL_LINK_MODE_200000baseSR_Full_BIT		 = 107,
+	ETHTOOL_LINK_MODE_200000baseVR_Full_BIT		 = 108,
+	ETHTOOL_LINK_MODE_400000baseCR2_Full_BIT	 = 109,
+	ETHTOOL_LINK_MODE_400000baseKR2_Full_BIT	 = 110,
+	ETHTOOL_LINK_MODE_400000baseDR2_Full_BIT	 = 111,
+	ETHTOOL_LINK_MODE_400000baseDR2_2_Full_BIT	 = 112,
+	ETHTOOL_LINK_MODE_400000baseSR2_Full_BIT	 = 113,
+	ETHTOOL_LINK_MODE_400000baseVR2_Full_BIT	 = 114,
+	ETHTOOL_LINK_MODE_800000baseCR4_Full_BIT	 = 115,
+	ETHTOOL_LINK_MODE_800000baseKR4_Full_BIT	 = 116,
+	ETHTOOL_LINK_MODE_800000baseDR4_Full_BIT	 = 117,
+	ETHTOOL_LINK_MODE_800000baseDR4_2_Full_BIT	 = 118,
+	ETHTOOL_LINK_MODE_800000baseSR4_Full_BIT	 = 119,
+	ETHTOOL_LINK_MODE_800000baseVR4_Full_BIT	 = 120,
 
 	/* must be last entry */
 	__ETHTOOL_LINK_MODE_MASK_NBITS
diff --git a/net/ethtool/common.c b/net/ethtool/common.c
index 2bd77c94f9f1..5489d0c9d13f 100644
--- a/net/ethtool/common.c
+++ b/net/ethtool/common.c
@@ -213,6 +213,24 @@ const char link_mode_names[][ETH_GSTRING_LEN] = {
 	__DEFINE_LINK_MODE_NAME(10, T1S, Half),
 	__DEFINE_LINK_MODE_NAME(10, T1S_P2MP, Half),
 	__DEFINE_LINK_MODE_NAME(10, T1BRR, Full),
+	__DEFINE_LINK_MODE_NAME(200000, CR, Full),
+	__DEFINE_LINK_MODE_NAME(200000, KR, Full),
+	__DEFINE_LINK_MODE_NAME(200000, DR, Full),
+	__DEFINE_LINK_MODE_NAME(200000, DR_2, Full),
+	__DEFINE_LINK_MODE_NAME(200000, SR, Full),
+	__DEFINE_LINK_MODE_NAME(200000, VR, Full),
+	__DEFINE_LINK_MODE_NAME(400000, CR2, Full),
+	__DEFINE_LINK_MODE_NAME(400000, KR2, Full),
+	__DEFINE_LINK_MODE_NAME(400000, DR2, Full),
+	__DEFINE_LINK_MODE_NAME(400000, DR2_2, Full),
+	__DEFINE_LINK_MODE_NAME(400000, SR2, Full),
+	__DEFINE_LINK_MODE_NAME(400000, VR2, Full),
+	__DEFINE_LINK_MODE_NAME(800000, CR4, Full),
+	__DEFINE_LINK_MODE_NAME(800000, KR4, Full),
+	__DEFINE_LINK_MODE_NAME(800000, DR4, Full),
+	__DEFINE_LINK_MODE_NAME(800000, DR4_2, Full),
+	__DEFINE_LINK_MODE_NAME(800000, SR4, Full),
+	__DEFINE_LINK_MODE_NAME(800000, VR4, Full),
 };
 static_assert(ARRAY_SIZE(link_mode_names) == __ETHTOOL_LINK_MODE_MASK_NBITS);
 
@@ -221,8 +239,11 @@ static_assert(ARRAY_SIZE(link_mode_names) == __ETHTOOL_LINK_MODE_MASK_NBITS);
 #define __LINK_MODE_LANES_CR4		4
 #define __LINK_MODE_LANES_CR8		8
 #define __LINK_MODE_LANES_DR		1
+#define __LINK_MODE_LANES_DR_2		1
 #define __LINK_MODE_LANES_DR2		2
+#define __LINK_MODE_LANES_DR2_2		2
 #define __LINK_MODE_LANES_DR4		4
+#define __LINK_MODE_LANES_DR4_2		4
 #define __LINK_MODE_LANES_DR8		8
 #define __LINK_MODE_LANES_KR		1
 #define __LINK_MODE_LANES_KR2		2
@@ -251,6 +272,9 @@ static_assert(ARRAY_SIZE(link_mode_names) == __ETHTOOL_LINK_MODE_MASK_NBITS);
 #define __LINK_MODE_LANES_T1L		1
 #define __LINK_MODE_LANES_T1S		1
 #define __LINK_MODE_LANES_T1S_P2MP	1
+#define __LINK_MODE_LANES_VR		1
+#define __LINK_MODE_LANES_VR2		2
+#define __LINK_MODE_LANES_VR4		4
 #define __LINK_MODE_LANES_VR8		8
 #define __LINK_MODE_LANES_DR8_2		8
 #define __LINK_MODE_LANES_T1BRR		1
@@ -378,6 +402,24 @@ const struct link_mode_info link_mode_params[] = {
 	__DEFINE_LINK_MODE_PARAMS(10, T1S, Half),
 	__DEFINE_LINK_MODE_PARAMS(10, T1S_P2MP, Half),
 	__DEFINE_LINK_MODE_PARAMS(10, T1BRR, Full),
+	__DEFINE_LINK_MODE_PARAMS(200000, CR, Full),
+	__DEFINE_LINK_MODE_PARAMS(200000, KR, Full),
+	__DEFINE_LINK_MODE_PARAMS(200000, DR, Full),
+	__DEFINE_LINK_MODE_PARAMS(200000, DR_2, Full),
+	__DEFINE_LINK_MODE_PARAMS(200000, SR, Full),
+	__DEFINE_LINK_MODE_PARAMS(200000, VR, Full),
+	__DEFINE_LINK_MODE_PARAMS(400000, CR2, Full),
+	__DEFINE_LINK_MODE_PARAMS(400000, KR2, Full),
+	__DEFINE_LINK_MODE_PARAMS(400000, DR2, Full),
+	__DEFINE_LINK_MODE_PARAMS(400000, DR2_2, Full),
+	__DEFINE_LINK_MODE_PARAMS(400000, SR2, Full),
+	__DEFINE_LINK_MODE_PARAMS(400000, VR2, Full),
+	__DEFINE_LINK_MODE_PARAMS(800000, CR4, Full),
+	__DEFINE_LINK_MODE_PARAMS(800000, KR4, Full),
+	__DEFINE_LINK_MODE_PARAMS(800000, DR4, Full),
+	__DEFINE_LINK_MODE_PARAMS(800000, DR4_2, Full),
+	__DEFINE_LINK_MODE_PARAMS(800000, SR4, Full),
+	__DEFINE_LINK_MODE_PARAMS(800000, VR4, Full),
 };
 static_assert(ARRAY_SIZE(link_mode_params) == __ETHTOOL_LINK_MODE_MASK_NBITS);