From: Vladimir Oltean vladimir.oltean@nxp.com

Some devices have errata due to which they cannot report ETH_ZLEN (60) in the rx-min-frag-size. This was foreseen of course, and lldpad has logic that when we request it to advertise addFragSize 0, it will round it up to the lowest value that is _actually_ supported by the hardware.

The problem is that the selftest expects lldpad to report back to us the same value as we requested.

Make the selftest smarter by figuring out on its own what is a reasonable value to expect.

Cc: Shuah Khan shuah@kernel.org Signed-off-by: Vladimir Oltean vladimir.oltean@nxp.com Tested-by: Roger Quadros rogerq@kernel.org Signed-off-by: Roger Quadros rogerq@kernel.org --- .../selftests/net/forwarding/ethtool_mm.sh | 37 ++++++++++++++++++- 1 file changed, 35 insertions(+), 2 deletions(-)

Changelog:

v11: no change v10: no change v9: no code change. Added Kselftest maintainer in Cc. v8: no change. Moved to the beginning of series. v7: initial commit

diff --git a/tools/testing/selftests/net/forwarding/ethtool_mm.sh b/tools/testing/selftests/net/forwarding/ethtool_mm.sh index 39e736f30322..6212913f4ad1 100755 --- a/tools/testing/selftests/net/forwarding/ethtool_mm.sh +++ b/tools/testing/selftests/net/forwarding/ethtool_mm.sh @@ -155,15 +155,48 @@ manual_failed_verification_h2_to_h1() manual_failed_verification $h2 $h1 }

+smallest_supported_add_frag_size() +{ + local iface=$1 + local rx_min_frag_size= + + rx_min_frag_size=$(ethtool --json --show-mm $iface | \ + jq '.[]."rx-min-frag-size"') + + if [ $rx_min_frag_size -le 60 ]; then + echo 0 + elif [ $rx_min_frag_size -le 124 ]; then + echo 1 + elif [ $rx_min_frag_size -le 188 ]; then + echo 2 + elif [ $rx_min_frag_size -le 252 ]; then + echo 3 + else + echo "$iface: RX min frag size $rx_min_frag_size cannot be advertised over LLDP" + exit 1 + fi +} + +expected_add_frag_size() +{ + local iface=$1 + local requested=$2 + local min=$(smallest_supported_add_frag_size $iface) + + [ $requested -le $min ] && echo $min || echo $requested +} + lldp_change_add_frag_size() { local add_frag_size=$1 + local pattern=

lldptool -T -i $h1 -V addEthCaps addFragSize=$add_frag_size >/dev/null # Wait for TLVs to be received sleep 2 - lldptool -i $h2 -t -n -V addEthCaps | \ - grep -q "Additional fragment size: $add_frag_size" + pattern=$(printf "Additional fragment size: %d" \ + $(expected_add_frag_size $h1 $add_frag_size)) + lldptool -i $h2 -t -n -V addEthCaps | grep -q "$pattern" }

lldp()

From: Vladimir Oltean vladimir.oltean@nxp.com

Some devices do not support individual 'pmac' and 'emac' stats. For such devices, resort to 'aggregate' stats.

Cc: Shuah Khan shuah@kernel.org Signed-off-by: Vladimir Oltean vladimir.oltean@nxp.com Tested-by: Roger Quadros rogerq@kernel.org Signed-off-by: Roger Quadros rogerq@kernel.org --- tools/testing/selftests/net/forwarding/ethtool_mm.sh | 11 +++++++++++ tools/testing/selftests/net/forwarding/lib.sh | 9 +++++++++ 2 files changed, 20 insertions(+)

Changelog:

v11: no change v10: Fix check_ethtool_pmac_std_stats_support() logic. v9: no code change. Added Kselftest maintainer in Cc. v8: initial commit

diff --git a/tools/testing/selftests/net/forwarding/ethtool_mm.sh b/tools/testing/selftests/net/forwarding/ethtool_mm.sh index 6212913f4ad1..50d5bfb17ef1 100755 --- a/tools/testing/selftests/net/forwarding/ethtool_mm.sh +++ b/tools/testing/selftests/net/forwarding/ethtool_mm.sh @@ -25,6 +25,10 @@ traffic_test() local after= local delta=

+ if [ ${has_pmac_stats[$if]} = false ]; then + src="aggregate" + fi + before=$(ethtool_std_stats_get $if "eth-mac" "FramesTransmittedOK" $src)

$MZ $if -q -c $num_pkts -p 64 -b bcast -t ip -R $PREEMPTIBLE_PRIO @@ -317,6 +321,13 @@ for netif in ${NETIFS[@]}; do echo "SKIP: $netif does not support MAC Merge" exit $ksft_skip fi + + if check_ethtool_pmac_std_stats_support $netif eth-mac; then + has_pmac_stats[$netif]=true + else + has_pmac_stats[$netif]=false + echo "$netif does not report pMAC statistics, falling back to aggregate" + fi done

trap cleanup EXIT diff --git a/tools/testing/selftests/net/forwarding/lib.sh b/tools/testing/selftests/net/forwarding/lib.sh index e3740163c384..69ef2a40df21 100755 --- a/tools/testing/selftests/net/forwarding/lib.sh +++ b/tools/testing/selftests/net/forwarding/lib.sh @@ -155,6 +155,15 @@ check_ethtool_counter_group_support() fi }

+check_ethtool_pmac_std_stats_support() +{ + local dev=$1; shift + local grp=$1; shift + + [ 0 -ne $(ethtool --json -S $dev --all-groups --src pmac 2>/dev/null \ + | jq ".[]."$grp" | length") ] +} + check_locked_port_support() { if ! bridge -d link show | grep -q " locked"; then

Build am65-cpsw-qos only if CONFIG_TI_AM65_CPSW_TAS is enabled.

Signed-off-by: Roger Quadros rogerq@kernel.org Reviewed-by: Vladimir Oltean vladimir.oltean@nxp.com --- drivers/net/ethernet/ti/Makefile | 3 ++- drivers/net/ethernet/ti/am65-cpsw-qos.c | 12 ------------ drivers/net/ethernet/ti/am65-cpsw-qos.h | 26 +++++++++++++++++++++++++ 3 files changed, 28 insertions(+), 13 deletions(-)

Changelog:

v11: no change v10: no code change. Added Reviewed-by tag. v9: no change v8: no change v7: no change v6: initial commit

diff --git a/drivers/net/ethernet/ti/Makefile b/drivers/net/ethernet/ti/Makefile index 27de1d697134..9d7cd84d1e2d 100644 --- a/drivers/net/ethernet/ti/Makefile +++ b/drivers/net/ethernet/ti/Makefile @@ -26,7 +26,8 @@ keystone_netcp_ethss-y := netcp_ethss.o netcp_sgmii.o netcp_xgbepcsr.o cpsw_ale. obj-$(CONFIG_TI_K3_CPPI_DESC_POOL) += k3-cppi-desc-pool.o

obj-$(CONFIG_TI_K3_AM65_CPSW_NUSS) += ti-am65-cpsw-nuss.o -ti-am65-cpsw-nuss-y := am65-cpsw-nuss.o cpsw_sl.o am65-cpsw-ethtool.o cpsw_ale.o am65-cpsw-qos.o +ti-am65-cpsw-nuss-y := am65-cpsw-nuss.o cpsw_sl.o am65-cpsw-ethtool.o cpsw_ale.o +ti-am65-cpsw-nuss-$(CONFIG_TI_AM65_CPSW_TAS) += am65-cpsw-qos.o ti-am65-cpsw-nuss-$(CONFIG_TI_K3_AM65_CPSW_SWITCHDEV) += am65-cpsw-switchdev.o obj-$(CONFIG_TI_K3_AM65_CPTS) += am65-cpts.o

diff --git a/drivers/net/ethernet/ti/am65-cpsw-qos.c b/drivers/net/ethernet/ti/am65-cpsw-qos.c index 9ac2ff05d501..4bc611cc4aad 100644 --- a/drivers/net/ethernet/ti/am65-cpsw-qos.c +++ b/drivers/net/ethernet/ti/am65-cpsw-qos.c @@ -571,9 +571,6 @@ static int am65_cpsw_setup_taprio(struct net_device *ndev, void *type_data) taprio->cmd != TAPRIO_CMD_DESTROY) return -EOPNOTSUPP;

- if (!IS_ENABLED(CONFIG_TI_AM65_CPSW_TAS)) - return -ENODEV; - if (!netif_running(ndev)) { dev_err(&ndev->dev, "interface is down, link speed unknown\n"); return -ENETDOWN; @@ -599,9 +596,6 @@ static int am65_cpsw_tc_query_caps(struct net_device *ndev, void *type_data) case TC_SETUP_QDISC_TAPRIO: { struct tc_taprio_caps *caps = base->caps;

- if (!IS_ENABLED(CONFIG_TI_AM65_CPSW_TAS)) - return -EOPNOTSUPP; - caps->gate_mask_per_txq = true;

return 0; @@ -806,9 +800,6 @@ void am65_cpsw_qos_link_up(struct net_device *ndev, int link_speed) { struct am65_cpsw_port *port = am65_ndev_to_port(ndev);

- if (!IS_ENABLED(CONFIG_TI_AM65_CPSW_TAS)) - return; - am65_cpsw_est_link_up(ndev, link_speed); port->qos.link_down_time = 0; } @@ -817,9 +808,6 @@ void am65_cpsw_qos_link_down(struct net_device *ndev) { struct am65_cpsw_port *port = am65_ndev_to_port(ndev);

- if (!IS_ENABLED(CONFIG_TI_AM65_CPSW_TAS)) - return; - if (!port->qos.link_down_time) port->qos.link_down_time = ktime_get();

diff --git a/drivers/net/ethernet/ti/am65-cpsw-qos.h b/drivers/net/ethernet/ti/am65-cpsw-qos.h index 0cc2a3b3d7f9..898f13a4a112 100644 --- a/drivers/net/ethernet/ti/am65-cpsw-qos.h +++ b/drivers/net/ethernet/ti/am65-cpsw-qos.h @@ -31,11 +31,37 @@ struct am65_cpsw_qos { struct am65_cpsw_ale_ratelimit ale_mc_ratelimit; };

+#if IS_ENABLED(CONFIG_TI_AM65_CPSW_TAS) int am65_cpsw_qos_ndo_setup_tc(struct net_device *ndev, enum tc_setup_type type, void *type_data); void am65_cpsw_qos_link_up(struct net_device *ndev, int link_speed); void am65_cpsw_qos_link_down(struct net_device *ndev); int am65_cpsw_qos_ndo_tx_p0_set_maxrate(struct net_device *ndev, int queue, u32 rate_mbps); void am65_cpsw_qos_tx_p0_rate_init(struct am65_cpsw_common *common); +#else +static inline int am65_cpsw_qos_ndo_setup_tc(struct net_device *ndev, + enum tc_setup_type type, + void *type_data) +{ + return -EOPNOTSUPP; +} + +static inline void am65_cpsw_qos_link_up(struct net_device *ndev, + int link_speed) +{ } + +static inline void am65_cpsw_qos_link_down(struct net_device *ndev) +{ } + +static inline int am65_cpsw_qos_ndo_tx_p0_set_maxrate(struct net_device *ndev, + int queue, + u32 rate_mbps) +{ + return 0; +} + +static inline void am65_cpsw_qos_tx_p0_rate_init(struct am65_cpsw_common *common) +{ } +#endif

#endif /* AM65_CPSW_QOS_H_ */

Handle offloading commands using switch-case in am65_cpsw_setup_taprio().

Move checks to am65_cpsw_taprio_replace().

Use NL_SET_ERR_MSG_MOD for error messages. Change error message from "Failed to set cycle time extension" to "cycle time extension not supported"

Signed-off-by: Roger Quadros rogerq@kernel.org Reviewed-by: Vladimir Oltean vladimir.oltean@nxp.com --- drivers/net/ethernet/ti/am65-cpsw-qos.c | 151 +++++++++++------------- 1 file changed, 71 insertions(+), 80 deletions(-)

Changelog:

v11: no change v10: no code change. Added Reviewed-by tag. v9: no change v8: don't initialize ret = 0, tact = TACT_PROG v7: don't use "\n" in NL_SET_ERR_MSG_MOD() v6: initial commit

diff --git a/drivers/net/ethernet/ti/am65-cpsw-qos.c b/drivers/net/ethernet/ti/am65-cpsw-qos.c index 4bc611cc4aad..2c97fa05a852 100644 --- a/drivers/net/ethernet/ti/am65-cpsw-qos.c +++ b/drivers/net/ethernet/ti/am65-cpsw-qos.c @@ -428,7 +428,7 @@ static void am65_cpsw_stop_est(struct net_device *ndev) am65_cpsw_timer_stop(ndev); }

-static void am65_cpsw_purge_est(struct net_device *ndev) +static void am65_cpsw_taprio_destroy(struct net_device *ndev) { struct am65_cpsw_port *port = am65_ndev_to_port(ndev);

@@ -441,29 +441,66 @@ static void am65_cpsw_purge_est(struct net_device *ndev) port->qos.est_admin = NULL; }

-static int am65_cpsw_configure_taprio(struct net_device *ndev, - struct am65_cpsw_est *est_new) +static void am65_cpsw_cp_taprio(struct tc_taprio_qopt_offload *from, + struct tc_taprio_qopt_offload *to) +{ + int i; + + *to = *from; + for (i = 0; i < from->num_entries; i++) + to->entries[i] = from->entries[i]; +} + +static int am65_cpsw_taprio_replace(struct net_device *ndev, + struct tc_taprio_qopt_offload *taprio) { struct am65_cpsw_common *common = am65_ndev_to_common(ndev); + struct netlink_ext_ack *extack = taprio->mqprio.extack; + struct am65_cpsw_port *port = am65_ndev_to_port(ndev); struct am65_cpts *cpts = common->cpts; - int ret = 0, tact = TACT_PROG; + struct am65_cpsw_est *est_new; + int ret, tact;

- am65_cpsw_est_update_state(ndev); + if (!netif_running(ndev)) { + NL_SET_ERR_MSG_MOD(extack, "interface is down, link speed unknown"); + return -ENETDOWN; + }

- if (est_new->taprio.cmd == TAPRIO_CMD_DESTROY) { - am65_cpsw_stop_est(ndev); - return ret; + if (common->pf_p0_rx_ptype_rrobin) { + NL_SET_ERR_MSG_MOD(extack, + "p0-rx-ptype-rrobin flag conflicts with taprio qdisc"); + return -EINVAL; + } + + if (port->qos.link_speed == SPEED_UNKNOWN) + return -ENOLINK; + + if (taprio->cycle_time_extension) { + NL_SET_ERR_MSG_MOD(extack, + "cycle time extension not supported"); + return -EOPNOTSUPP; }

+ est_new = devm_kzalloc(&ndev->dev, + struct_size(est_new, taprio.entries, taprio->num_entries), + GFP_KERNEL); + if (!est_new) + return -ENOMEM; + + am65_cpsw_cp_taprio(taprio, &est_new->taprio); + + am65_cpsw_est_update_state(ndev); + ret = am65_cpsw_est_check_scheds(ndev, est_new); if (ret < 0) - return ret; + goto fail;

tact = am65_cpsw_timer_act(ndev, est_new); if (tact == TACT_NEED_STOP) { - dev_err(&ndev->dev, - "Can't toggle estf timer, stop taprio first"); - return -EINVAL; + NL_SET_ERR_MSG_MOD(extack, + "Can't toggle estf timer, stop taprio first"); + ret = -EINVAL; + goto fail; }

if (tact == TACT_PROG) @@ -476,62 +513,24 @@ static int am65_cpsw_configure_taprio(struct net_device *ndev, am65_cpsw_est_set_sched_list(ndev, est_new); am65_cpsw_port_est_assign_buf_num(ndev, est_new->buf);

- am65_cpsw_est_set(ndev, est_new->taprio.cmd == TAPRIO_CMD_REPLACE); + am65_cpsw_est_set(ndev, 1);

if (tact == TACT_PROG) { ret = am65_cpsw_timer_set(ndev, est_new); if (ret) { - dev_err(&ndev->dev, "Failed to set cycle time"); - return ret; + NL_SET_ERR_MSG_MOD(extack, + "Failed to set cycle time"); + goto fail; } }

- return ret; -} - -static void am65_cpsw_cp_taprio(struct tc_taprio_qopt_offload *from, - struct tc_taprio_qopt_offload *to) -{ - int i; - - *to = *from; - for (i = 0; i < from->num_entries; i++) - to->entries[i] = from->entries[i]; -} - -static int am65_cpsw_set_taprio(struct net_device *ndev, void *type_data) -{ - struct am65_cpsw_port *port = am65_ndev_to_port(ndev); - struct tc_taprio_qopt_offload *taprio = type_data; - struct am65_cpsw_est *est_new; - int ret = 0; - - if (taprio->cycle_time_extension) { - dev_err(&ndev->dev, "Failed to set cycle time extension"); - return -EOPNOTSUPP; - } - - est_new = devm_kzalloc(&ndev->dev, - struct_size(est_new, taprio.entries, taprio->num_entries), - GFP_KERNEL); - if (!est_new) - return -ENOMEM; - - am65_cpsw_cp_taprio(taprio, &est_new->taprio); - ret = am65_cpsw_configure_taprio(ndev, est_new); - if (!ret) { - if (taprio->cmd == TAPRIO_CMD_REPLACE) { - devm_kfree(&ndev->dev, port->qos.est_admin); + devm_kfree(&ndev->dev, port->qos.est_admin); + port->qos.est_admin = est_new;

- port->qos.est_admin = est_new; - } else { - devm_kfree(&ndev->dev, est_new); - am65_cpsw_purge_est(ndev); - } - } else { - devm_kfree(&ndev->dev, est_new); - } + return 0;

+fail: + devm_kfree(&ndev->dev, est_new); return ret; }

@@ -558,34 +557,26 @@ static void am65_cpsw_est_link_up(struct net_device *ndev, int link_speed) return;

purge_est: - am65_cpsw_purge_est(ndev); + am65_cpsw_taprio_destroy(ndev); }

static int am65_cpsw_setup_taprio(struct net_device *ndev, void *type_data) { - struct am65_cpsw_port *port = am65_ndev_to_port(ndev); struct tc_taprio_qopt_offload *taprio = type_data; - struct am65_cpsw_common *common = port->common; - - if (taprio->cmd != TAPRIO_CMD_REPLACE && - taprio->cmd != TAPRIO_CMD_DESTROY) - return -EOPNOTSUPP; - - if (!netif_running(ndev)) { - dev_err(&ndev->dev, "interface is down, link speed unknown\n"); - return -ENETDOWN; - } - - if (common->pf_p0_rx_ptype_rrobin) { - dev_err(&ndev->dev, - "p0-rx-ptype-rrobin flag conflicts with taprio qdisc\n"); - return -EINVAL; + int err = 0; + + switch (taprio->cmd) { + case TAPRIO_CMD_REPLACE: + err = am65_cpsw_taprio_replace(ndev, taprio); + break; + case TAPRIO_CMD_DESTROY: + am65_cpsw_taprio_destroy(ndev); + break; + default: + err = -EOPNOTSUPP; }

- if (port->qos.link_speed == SPEED_UNKNOWN) - return -ENOLINK; - - return am65_cpsw_set_taprio(ndev, type_data); + return err; }

static int am65_cpsw_tc_query_caps(struct net_device *ndev, void *type_data)

Move register definitions to header file. No functional change.

Signed-off-by: Roger Quadros rogerq@kernel.org Reviewed-by: Vladimir Oltean vladimir.oltean@nxp.com --- drivers/net/ethernet/ti/am65-cpsw-qos.c | 35 ------------------------- drivers/net/ethernet/ti/am65-cpsw-qos.h | 35 +++++++++++++++++++++++++ 2 files changed, 35 insertions(+), 35 deletions(-)

Changelog:

v11: no change v10: no change v9: no code change. added Reviewed-by tag. v8: no change v7: no change v6: initial commit

diff --git a/drivers/net/ethernet/ti/am65-cpsw-qos.c b/drivers/net/ethernet/ti/am65-cpsw-qos.c index 862a98211caa..9f0a05e763d1 100644 --- a/drivers/net/ethernet/ti/am65-cpsw-qos.c +++ b/drivers/net/ethernet/ti/am65-cpsw-qos.c @@ -15,41 +15,6 @@ #include "am65-cpts.h" #include "cpsw_ale.h"

-#define AM65_CPSW_REG_CTL 0x004 -#define AM65_CPSW_PN_REG_CTL 0x004 -#define AM65_CPSW_PN_REG_FIFO_STATUS 0x050 -#define AM65_CPSW_PN_REG_EST_CTL 0x060 -#define AM65_CPSW_PN_REG_PRI_CIR(pri) (0x140 + 4 * (pri)) - -/* AM65_CPSW_REG_CTL register fields */ -#define AM65_CPSW_CTL_EST_EN BIT(18) - -/* AM65_CPSW_PN_REG_CTL register fields */ -#define AM65_CPSW_PN_CTL_EST_PORT_EN BIT(17) - -/* AM65_CPSW_PN_REG_EST_CTL register fields */ -#define AM65_CPSW_PN_EST_ONEBUF BIT(0) -#define AM65_CPSW_PN_EST_BUFSEL BIT(1) -#define AM65_CPSW_PN_EST_TS_EN BIT(2) -#define AM65_CPSW_PN_EST_TS_FIRST BIT(3) -#define AM65_CPSW_PN_EST_ONEPRI BIT(4) -#define AM65_CPSW_PN_EST_TS_PRI_MSK GENMASK(7, 5) - -/* AM65_CPSW_PN_REG_FIFO_STATUS register fields */ -#define AM65_CPSW_PN_FST_TX_PRI_ACTIVE_MSK GENMASK(7, 0) -#define AM65_CPSW_PN_FST_TX_E_MAC_ALLOW_MSK GENMASK(15, 8) -#define AM65_CPSW_PN_FST_EST_CNT_ERR BIT(16) -#define AM65_CPSW_PN_FST_EST_ADD_ERR BIT(17) -#define AM65_CPSW_PN_FST_EST_BUFACT BIT(18) - -/* EST FETCH COMMAND RAM */ -#define AM65_CPSW_FETCH_RAM_CMD_NUM 0x80 -#define AM65_CPSW_FETCH_CNT_MSK GENMASK(21, 8) -#define AM65_CPSW_FETCH_CNT_MAX (AM65_CPSW_FETCH_CNT_MSK >> 8) -#define AM65_CPSW_FETCH_CNT_OFFSET 8 -#define AM65_CPSW_FETCH_ALLOW_MSK GENMASK(7, 0) -#define AM65_CPSW_FETCH_ALLOW_MAX AM65_CPSW_FETCH_ALLOW_MSK - enum timer_act { TACT_PROG, /* need program timer */ TACT_NEED_STOP, /* need stop first */ diff --git a/drivers/net/ethernet/ti/am65-cpsw-qos.h b/drivers/net/ethernet/ti/am65-cpsw-qos.h index be4987eb8c51..e3259f27baad 100644 --- a/drivers/net/ethernet/ti/am65-cpsw-qos.h +++ b/drivers/net/ethernet/ti/am65-cpsw-qos.h @@ -31,6 +31,41 @@ struct am65_cpsw_qos { struct am65_cpsw_ale_ratelimit ale_mc_ratelimit; };

+#define AM65_CPSW_REG_CTL 0x004 +#define AM65_CPSW_PN_REG_CTL 0x004 +#define AM65_CPSW_PN_REG_FIFO_STATUS 0x050 +#define AM65_CPSW_PN_REG_EST_CTL 0x060 +#define AM65_CPSW_PN_REG_PRI_CIR(pri) (0x140 + 4 * (pri)) + +/* AM65_CPSW_REG_CTL register fields */ +#define AM65_CPSW_CTL_EST_EN BIT(18) + +/* AM65_CPSW_PN_REG_CTL register fields */ +#define AM65_CPSW_PN_CTL_EST_PORT_EN BIT(17) + +/* AM65_CPSW_PN_REG_EST_CTL register fields */ +#define AM65_CPSW_PN_EST_ONEBUF BIT(0) +#define AM65_CPSW_PN_EST_BUFSEL BIT(1) +#define AM65_CPSW_PN_EST_TS_EN BIT(2) +#define AM65_CPSW_PN_EST_TS_FIRST BIT(3) +#define AM65_CPSW_PN_EST_ONEPRI BIT(4) +#define AM65_CPSW_PN_EST_TS_PRI_MSK GENMASK(7, 5) + +/* AM65_CPSW_PN_REG_FIFO_STATUS register fields */ +#define AM65_CPSW_PN_FST_TX_PRI_ACTIVE_MSK GENMASK(7, 0) +#define AM65_CPSW_PN_FST_TX_E_MAC_ALLOW_MSK GENMASK(15, 8) +#define AM65_CPSW_PN_FST_EST_CNT_ERR BIT(16) +#define AM65_CPSW_PN_FST_EST_ADD_ERR BIT(17) +#define AM65_CPSW_PN_FST_EST_BUFACT BIT(18) + +/* EST FETCH COMMAND RAM */ +#define AM65_CPSW_FETCH_RAM_CMD_NUM 0x80 +#define AM65_CPSW_FETCH_CNT_MSK GENMASK(21, 8) +#define AM65_CPSW_FETCH_CNT_MAX (AM65_CPSW_FETCH_CNT_MSK >> 8) +#define AM65_CPSW_FETCH_CNT_OFFSET 8 +#define AM65_CPSW_FETCH_ALLOW_MSK GENMASK(7, 0) +#define AM65_CPSW_FETCH_ALLOW_MAX AM65_CPSW_FETCH_ALLOW_MSK + #if IS_ENABLED(CONFIG_TI_AM65_CPSW_QOS) int am65_cpsw_qos_ndo_setup_tc(struct net_device *ndev, enum tc_setup_type type, void *type_data);

Add driver support for viewing / changing the MAC Merge sublayer parameters and seeing the verification state machine's current state via ethtool.

As hardware does not support interrupt notification for verification events we resort to polling on link up. On link up we try a couple of times for verification success and if unsuccessful then give up.

The Frame Preemption feature is described in the Technical Reference Manual [1] in section: 12.3.1.4.6.7 Intersperced Express Traffic (IET – P802.3br/D2.0)

Due to Silicon Errata i2208 [2] we set limit min IET fragment size to 124 (excluding 4 bytes mCRC).

[1] AM62x TRM - https://www.ti.com/lit/ug/spruiv7a/spruiv7a.pdf [2] AM62x Silicon Errata - https://www.ti.com/lit/er/sprz487c/sprz487c.pdf

Signed-off-by: Roger Quadros rogerq@kernel.org Reviewed-by: Vladimir Oltean vladimir.oltean@nxp.com --- drivers/net/ethernet/ti/Kconfig | 5 +- drivers/net/ethernet/ti/am65-cpsw-ethtool.c | 167 +++++++++++++++++++ drivers/net/ethernet/ti/am65-cpsw-nuss.c | 2 + drivers/net/ethernet/ti/am65-cpsw-nuss.h | 5 + drivers/net/ethernet/ti/am65-cpsw-qos.c | 174 ++++++++++++++++++++ drivers/net/ethernet/ti/am65-cpsw-qos.h | 105 ++++++++++++ 6 files changed, 456 insertions(+), 2 deletions(-)

Changelog:

v11: - no change

v10: - no change

v9: - fix up am65_cpsw_get_eth_mac_stats() by merging patch "net: ethernet: ti: am65-cpsw: Fix get_eth_mac_stats" - remove unnecessary 'if (cfg->verify_time > 0)' check in am65_cpsw_set_mm()

v8: - fix build if CONFIG_TI_AM65_CPSW_TAS not enabled - change CONFIG_TI_AM65_CPSW_TAS to CONFIG_TI_AM65_CPSW_QOS - mention about Frame Preemption support in TI_AM65_CPSW_QOS Kconfig help - mentioned IET fragment size excludes 4 bytes mCRC in commit log

v7: - use else if - drop FIXME comment - fix lldp kselftest failure by limiting max_verify_time to spec limit of 128ms. - now passes all ethtool_mm.sh kselftests (patch 8 required)

v6: - get mutex around am65_cpsw_iet_commit_preemptible_tcs() in am65_cpsw_iet_change_preemptible_tcs() - use "preemption" instead of "pre-emption" - call am65_cpsw_setup_mqprio() from within am65_cpsw_setup_taprio() - Now works with kselftest except the last test which fails

v5: - No change

v4: - Rebase and include in the same series as mqprio support.

v3: - Rebase on top of v6.6-rc1 and mqprio support [1] - Support ethtool_ops :: get_mm_stats() - drop unused variables cmn_ctrl and verify_cnt - make am65_cpsw_iet_link_state_update() and am65_cpsw_iet_change_preemptible_tcs() static

[1] https://lore.kernel.org/all/20230918075358.5878-1-rogerq@kernel.org/

v2: - Use proper control bits for PMAC enable (AM65_CPSW_PN_CTL_IET_PORT_EN) and TX enable (AM65_CPSW_PN_IET_MAC_PENABLE) - Common IET Enable (AM65_CPSW_CTL_IET_EN) is set if any port has AM65_CPSW_PN_CTL_IET_PORT_EN set. - Fix workaround for erratum i2208. i.e. Limit rx_min_frag_size to 124 - Fix am65_cpsw_iet_get_verify_timeout_ms() to default to timeout for 1G link if link is inactive. - resize the RX FIFO based on pmac_enabled, not tx_enabled.

diff --git a/drivers/net/ethernet/ti/Kconfig b/drivers/net/ethernet/ti/Kconfig index f76532443298..be01450c20dc 100644 --- a/drivers/net/ethernet/ti/Kconfig +++ b/drivers/net/ethernet/ti/Kconfig @@ -139,8 +139,9 @@ config TI_AM65_CPSW_QOS depends on TI_K3_AM65_CPSW_NUSS && NET_SCH_TAPRIO && TI_K3_AM65_CPTS help This option enables QoS offload features in AM65 CPSW like - Time Aware Shaper (TAS) / Enhanced Scheduled Traffic (EST) - and MQPRIO qdisc offload. + Time Aware Shaper (TAS) / Enhanced Scheduled Traffic (EST), + MQPRIO qdisc offload and Frame-Preemption MAC Merge / Interspersing + Express Traffic (IET). The EST scheduler runs on CPTS and the TAS/EST schedule is updated in the Fetch RAM memory of the CPSW.

diff --git a/drivers/net/ethernet/ti/am65-cpsw-ethtool.c b/drivers/net/ethernet/ti/am65-cpsw-ethtool.c index b9e1d568604b..0a6134cdf39f 100644 --- a/drivers/net/ethernet/ti/am65-cpsw-ethtool.c +++ b/drivers/net/ethernet/ti/am65-cpsw-ethtool.c @@ -11,6 +11,7 @@ #include <linux/pm_runtime.h>

#include "am65-cpsw-nuss.h" +#include "am65-cpsw-qos.h" #include "cpsw_ale.h" #include "am65-cpts.h"

@@ -670,6 +671,9 @@ static void am65_cpsw_get_eth_mac_stats(struct net_device *ndev,

stats = port->stat_base;

+ if (s->src != ETHTOOL_MAC_STATS_SRC_AGGREGATE) + return; + s->FramesTransmittedOK = readl_relaxed(&stats->tx_good_frames); s->SingleCollisionFrames = readl_relaxed(&stats->tx_single_coll_frames); s->MultipleCollisionFrames = readl_relaxed(&stats->tx_mult_coll_frames); @@ -740,6 +744,166 @@ static int am65_cpsw_set_ethtool_priv_flags(struct net_device *ndev, u32 flags) return 0; }

+static void am65_cpsw_port_iet_rx_enable(struct am65_cpsw_port *port, bool enable) +{ + u32 val; + + val = readl(port->port_base + AM65_CPSW_PN_REG_CTL); + if (enable) + val |= AM65_CPSW_PN_CTL_IET_PORT_EN; + else + val &= ~AM65_CPSW_PN_CTL_IET_PORT_EN; + + writel(val, port->port_base + AM65_CPSW_PN_REG_CTL); + am65_cpsw_iet_common_enable(port->common); +} + +static void am65_cpsw_port_iet_tx_enable(struct am65_cpsw_port *port, bool enable) +{ + u32 val; + + val = readl(port->port_base + AM65_CPSW_PN_REG_IET_CTRL); + if (enable) + val |= AM65_CPSW_PN_IET_MAC_PENABLE; + else + val &= ~AM65_CPSW_PN_IET_MAC_PENABLE; + + writel(val, port->port_base + AM65_CPSW_PN_REG_IET_CTRL); +} + +static int am65_cpsw_get_mm(struct net_device *ndev, struct ethtool_mm_state *state) +{ + struct am65_cpsw_port *port = am65_ndev_to_port(ndev); + struct am65_cpsw_ndev_priv *priv = netdev_priv(ndev); + u32 port_ctrl, iet_ctrl, iet_status; + u32 add_frag_size; + + if (!IS_ENABLED(CONFIG_TI_AM65_CPSW_QOS)) + return -EOPNOTSUPP; + + mutex_lock(&priv->mm_lock); + + iet_ctrl = readl(port->port_base + AM65_CPSW_PN_REG_IET_CTRL); + port_ctrl = readl(port->port_base + AM65_CPSW_PN_REG_CTL); + + state->tx_enabled = !!(iet_ctrl & AM65_CPSW_PN_IET_MAC_PENABLE); + state->pmac_enabled = !!(port_ctrl & AM65_CPSW_PN_CTL_IET_PORT_EN); + + iet_status = readl(port->port_base + AM65_CPSW_PN_REG_IET_STATUS); + + if (iet_ctrl & AM65_CPSW_PN_IET_MAC_DISABLEVERIFY) + state->verify_status = ETHTOOL_MM_VERIFY_STATUS_DISABLED; + else if (iet_status & AM65_CPSW_PN_MAC_VERIFIED) + state->verify_status = ETHTOOL_MM_VERIFY_STATUS_SUCCEEDED; + else if (iet_status & AM65_CPSW_PN_MAC_VERIFY_FAIL) + state->verify_status = ETHTOOL_MM_VERIFY_STATUS_FAILED; + else + state->verify_status = ETHTOOL_MM_VERIFY_STATUS_UNKNOWN; + + add_frag_size = AM65_CPSW_PN_IET_MAC_GET_ADDFRAGSIZE(iet_ctrl); + state->tx_min_frag_size = ethtool_mm_frag_size_add_to_min(add_frag_size); + + /* Errata i2208: RX min fragment size cannot be less than 124 */ + state->rx_min_frag_size = 124; + + /* FPE active if common tx_enabled and verification success or disabled (forced) */ + state->tx_active = state->tx_enabled && + (state->verify_status == ETHTOOL_MM_VERIFY_STATUS_SUCCEEDED || + state->verify_status == ETHTOOL_MM_VERIFY_STATUS_DISABLED); + state->verify_enabled = !(iet_ctrl & AM65_CPSW_PN_IET_MAC_DISABLEVERIFY); + + state->verify_time = port->qos.iet.verify_time_ms; + + /* 802.3-2018 clause 30.14.1.6, says that the aMACMergeVerifyTime + * variable has a range between 1 and 128 ms inclusive. Limit to that. + */ + state->max_verify_time = 128; + + mutex_unlock(&priv->mm_lock); + + return 0; +} + +static int am65_cpsw_set_mm(struct net_device *ndev, struct ethtool_mm_cfg *cfg, + struct netlink_ext_ack *extack) +{ + struct am65_cpsw_port *port = am65_ndev_to_port(ndev); + struct am65_cpsw_ndev_priv *priv = netdev_priv(ndev); + struct am65_cpsw_iet *iet = &port->qos.iet; + u32 val, add_frag_size; + int err; + + if (!IS_ENABLED(CONFIG_TI_AM65_CPSW_QOS)) + return -EOPNOTSUPP; + + err = ethtool_mm_frag_size_min_to_add(cfg->tx_min_frag_size, &add_frag_size, extack); + if (err) + return err; + + mutex_lock(&priv->mm_lock); + + if (cfg->pmac_enabled) { + /* change TX & RX FIFO MAX_BLKS as per TRM recommendation */ + if (!iet->original_max_blks) + iet->original_max_blks = readl(port->port_base + AM65_CPSW_PN_REG_MAX_BLKS); + + writel(AM65_CPSW_PN_TX_RX_MAX_BLKS_IET, + port->port_base + AM65_CPSW_PN_REG_MAX_BLKS); + } else if (iet->original_max_blks) { + /* restore RX & TX FIFO MAX_BLKS */ + writel(iet->original_max_blks, + port->port_base + AM65_CPSW_PN_REG_MAX_BLKS); + } + + am65_cpsw_port_iet_rx_enable(port, cfg->pmac_enabled); + am65_cpsw_port_iet_tx_enable(port, cfg->tx_enabled); + + val = readl(port->port_base + AM65_CPSW_PN_REG_IET_CTRL); + if (cfg->verify_enabled) { + val &= ~AM65_CPSW_PN_IET_MAC_DISABLEVERIFY; + /* Reset Verify state machine. Verification won't start here. + * Verification will be done once link-up. + */ + val |= AM65_CPSW_PN_IET_MAC_LINKFAIL; + } else { + val |= AM65_CPSW_PN_IET_MAC_DISABLEVERIFY; + /* Clear LINKFAIL to allow verify/response packets */ + val &= ~AM65_CPSW_PN_IET_MAC_LINKFAIL; + } + + val &= ~AM65_CPSW_PN_IET_MAC_MAC_ADDFRAGSIZE_MASK; + val |= AM65_CPSW_PN_IET_MAC_SET_ADDFRAGSIZE(add_frag_size); + writel(val, port->port_base + AM65_CPSW_PN_REG_IET_CTRL); + + /* verify_timeout_count can only be set at valid link */ + port->qos.iet.verify_time_ms = cfg->verify_time; + + /* enable/disable preemption based on link status */ + am65_cpsw_iet_commit_preemptible_tcs(port); + + mutex_unlock(&priv->mm_lock); + + return 0; +} + +static void am65_cpsw_get_mm_stats(struct net_device *ndev, + struct ethtool_mm_stats *s) +{ + struct am65_cpsw_port *port = am65_ndev_to_port(ndev); + void __iomem *base = port->stat_base; + + s->MACMergeFrameAssOkCount = readl(base + AM65_CPSW_STATN_IET_RX_ASSEMBLY_OK); + s->MACMergeFrameAssErrorCount = readl(base + AM65_CPSW_STATN_IET_RX_ASSEMBLY_ERROR); + s->MACMergeFrameSmdErrorCount = readl(base + AM65_CPSW_STATN_IET_RX_SMD_ERROR); + /* CPSW Functional Spec states: + * "The IET stat aMACMergeFragCountRx is derived by adding the + * Receive Assembly Error count to this value. i.e. AM65_CPSW_STATN_IET_RX_FRAG" + */ + s->MACMergeFragCountRx = readl(base + AM65_CPSW_STATN_IET_RX_FRAG) + s->MACMergeFrameAssErrorCount; + s->MACMergeFragCountTx = readl(base + AM65_CPSW_STATN_IET_TX_FRAG); + s->MACMergeHoldCount = readl(base + AM65_CPSW_STATN_IET_TX_HOLD); +} + const struct ethtool_ops am65_cpsw_ethtool_ops_slave = { .begin = am65_cpsw_ethtool_op_begin, .complete = am65_cpsw_ethtool_op_complete, @@ -769,4 +933,7 @@ const struct ethtool_ops am65_cpsw_ethtool_ops_slave = { .get_eee = am65_cpsw_get_eee, .set_eee = am65_cpsw_set_eee, .nway_reset = am65_cpsw_nway_reset, + .get_mm = am65_cpsw_get_mm, + .set_mm = am65_cpsw_set_mm, + .get_mm_stats = am65_cpsw_get_mm_stats, }; diff --git a/drivers/net/ethernet/ti/am65-cpsw-nuss.c b/drivers/net/ethernet/ti/am65-cpsw-nuss.c index 2e374745f239..41e0046a52d5 100644 --- a/drivers/net/ethernet/ti/am65-cpsw-nuss.c +++ b/drivers/net/ethernet/ti/am65-cpsw-nuss.c @@ -2194,6 +2194,8 @@ am65_cpsw_nuss_init_port_ndev(struct am65_cpsw_common *common, u32 port_idx) ndev_priv = netdev_priv(port->ndev); ndev_priv->port = port; ndev_priv->msg_enable = AM65_CPSW_DEBUG; + mutex_init(&ndev_priv->mm_lock); + port->qos.link_speed = SPEED_UNKNOWN; SET_NETDEV_DEV(port->ndev, dev);

eth_hw_addr_set(port->ndev, port->slave.mac_addr); diff --git a/drivers/net/ethernet/ti/am65-cpsw-nuss.h b/drivers/net/ethernet/ti/am65-cpsw-nuss.h index f3dad2ab9828..1e4a045057fc 100644 --- a/drivers/net/ethernet/ti/am65-cpsw-nuss.h +++ b/drivers/net/ethernet/ti/am65-cpsw-nuss.h @@ -145,6 +145,7 @@ struct am65_cpsw_common { bool pf_p0_rx_ptype_rrobin; struct am65_cpts *cpts; int est_enabled; + bool iet_enabled;

bool is_emac_mode; u16 br_members; @@ -170,6 +171,10 @@ struct am65_cpsw_ndev_priv { struct am65_cpsw_port *port; struct am65_cpsw_ndev_stats __percpu *stats; bool offload_fwd_mark; + /* Serialize access to MAC Merge state between ethtool requests + * and link state updates + */ + struct mutex mm_lock; };

#define am65_ndev_to_priv(ndev) \ diff --git a/drivers/net/ethernet/ti/am65-cpsw-qos.c b/drivers/net/ethernet/ti/am65-cpsw-qos.c index 7790d739e74e..816e73a3d6e4 100644 --- a/drivers/net/ethernet/ti/am65-cpsw-qos.c +++ b/drivers/net/ethernet/ti/am65-cpsw-qos.c @@ -4,6 +4,7 @@ * * quality of service module includes: * Enhanced Scheduler Traffic (EST - P802.1Qbv/D2.2) + * Interspersed Express Traffic (IET - P802.3br/D2.0) */

#include <linux/pm_runtime.h> @@ -25,6 +26,8 @@ enum timer_act { TACT_SKIP_PROG, /* just buffer can be updated */ };

+static void am65_cpsw_iet_change_preemptible_tcs(struct am65_cpsw_port *port, u8 preemptible_tcs); + static u32 am65_cpsw_qos_tx_rate_calc(u32 rate_mbps, unsigned long bus_freq) { @@ -192,6 +195,8 @@ static void am65_cpsw_reset_tc_mqprio(struct net_device *ndev)

/* Reset all Queue priorities to 0 */ writel(0, port->port_base + AM65_CPSW_PN_REG_TX_PRI_MAP); + + am65_cpsw_iet_change_preemptible_tcs(port, 0); }

static int am65_cpsw_setup_mqprio(struct net_device *ndev, void *type_data) @@ -254,6 +259,7 @@ static int am65_cpsw_setup_mqprio(struct net_device *ndev, void *type_data) writel(tx_prio_map, port->port_base + AM65_CPSW_PN_REG_TX_PRI_MAP);

am65_cpsw_tx_pn_shaper_apply(port); + am65_cpsw_iet_change_preemptible_tcs(port, mqprio->preemptible_tcs);

exit_put: pm_runtime_put(common->dev); @@ -261,6 +267,171 @@ static int am65_cpsw_setup_mqprio(struct net_device *ndev, void *type_data) return ret; }

+static int am65_cpsw_iet_set_verify_timeout_count(struct am65_cpsw_port *port) +{ + int verify_time_ms = port->qos.iet.verify_time_ms; + u32 val; + + /* The number of wireside clocks contained in the verify + * timeout counter. The default is 0x1312d0 + * (10ms at 125Mhz in 1G mode). + */ + val = 125 * HZ_PER_MHZ; /* assuming 125MHz wireside clock */ + + val /= MILLIHZ_PER_HZ; /* count per ms timeout */ + val *= verify_time_ms; /* count for timeout ms */ + + if (val > AM65_CPSW_PN_MAC_VERIFY_CNT_MASK) + return -EINVAL; + + writel(val, port->port_base + AM65_CPSW_PN_REG_IET_VERIFY); + + return 0; +} + +static int am65_cpsw_iet_verify_wait(struct am65_cpsw_port *port) +{ + u32 ctrl, status; + int try; + + try = 20; + do { + /* Reset the verify state machine by writing 1 + * to LINKFAIL + */ + ctrl = readl(port->port_base + AM65_CPSW_PN_REG_IET_CTRL); + ctrl |= AM65_CPSW_PN_IET_MAC_LINKFAIL; + writel(ctrl, port->port_base + AM65_CPSW_PN_REG_IET_CTRL); + + /* Clear MAC_LINKFAIL bit to start Verify. */ + ctrl = readl(port->port_base + AM65_CPSW_PN_REG_IET_CTRL); + ctrl &= ~AM65_CPSW_PN_IET_MAC_LINKFAIL; + writel(ctrl, port->port_base + AM65_CPSW_PN_REG_IET_CTRL); + + msleep(port->qos.iet.verify_time_ms); + + status = readl(port->port_base + AM65_CPSW_PN_REG_IET_STATUS); + if (status & AM65_CPSW_PN_MAC_VERIFIED) + return 0; + + if (status & AM65_CPSW_PN_MAC_VERIFY_FAIL) { + netdev_dbg(port->ndev, + "MAC Merge verify failed, trying again\n"); + continue; + } + + if (status & AM65_CPSW_PN_MAC_RESPOND_ERR) { + netdev_dbg(port->ndev, "MAC Merge respond error\n"); + return -ENODEV; + } + + if (status & AM65_CPSW_PN_MAC_VERIFY_ERR) { + netdev_dbg(port->ndev, "MAC Merge verify error\n"); + return -ENODEV; + } + } while (try-- > 0); + + netdev_dbg(port->ndev, "MAC Merge verify timeout\n"); + return -ETIMEDOUT; +} + +static void am65_cpsw_iet_set_preempt_mask(struct am65_cpsw_port *port, u8 preemptible_tcs) +{ + u32 val; + + val = readl(port->port_base + AM65_CPSW_PN_REG_IET_CTRL); + val &= ~AM65_CPSW_PN_IET_MAC_PREMPT_MASK; + val |= AM65_CPSW_PN_IET_MAC_SET_PREEMPT(preemptible_tcs); + writel(val, port->port_base + AM65_CPSW_PN_REG_IET_CTRL); +} + +/* enable common IET_ENABLE only if at least 1 port has rx IET enabled. + * UAPI doesn't allow tx enable without rx enable. + */ +void am65_cpsw_iet_common_enable(struct am65_cpsw_common *common) +{ + struct am65_cpsw_port *port; + bool rx_enable = false; + u32 val; + int i; + + for (i = 0; i < common->port_num; i++) { + port = &common->ports[i]; + val = readl(port->port_base + AM65_CPSW_PN_REG_CTL); + rx_enable = !!(val & AM65_CPSW_PN_CTL_IET_PORT_EN); + if (rx_enable) + break; + } + + val = readl(common->cpsw_base + AM65_CPSW_REG_CTL); + + if (rx_enable) + val |= AM65_CPSW_CTL_IET_EN; + else + val &= ~AM65_CPSW_CTL_IET_EN; + + writel(val, common->cpsw_base + AM65_CPSW_REG_CTL); + common->iet_enabled = rx_enable; +} + +/* CPSW does not have an IRQ to notify changes to the MAC Merge TX status + * (active/inactive), but the preemptible traffic classes should only be + * committed to hardware once TX is active. Resort to polling. + */ +void am65_cpsw_iet_commit_preemptible_tcs(struct am65_cpsw_port *port) +{ + u8 preemptible_tcs; + int err; + u32 val; + + if (port->qos.link_speed == SPEED_UNKNOWN) + return; + + val = readl(port->port_base + AM65_CPSW_PN_REG_CTL); + if (!(val & AM65_CPSW_PN_CTL_IET_PORT_EN)) + return; + + /* update common IET enable */ + am65_cpsw_iet_common_enable(port->common); + + /* update verify count */ + err = am65_cpsw_iet_set_verify_timeout_count(port); + if (err) { + netdev_err(port->ndev, "couldn't set verify count: %d\n", err); + return; + } + + val = readl(port->port_base + AM65_CPSW_PN_REG_IET_CTRL); + if (!(val & AM65_CPSW_PN_IET_MAC_DISABLEVERIFY)) { + err = am65_cpsw_iet_verify_wait(port); + if (err) + return; + } + + preemptible_tcs = port->qos.iet.preemptible_tcs; + am65_cpsw_iet_set_preempt_mask(port, preemptible_tcs); +} + +static void am65_cpsw_iet_change_preemptible_tcs(struct am65_cpsw_port *port, u8 preemptible_tcs) +{ + struct am65_cpsw_ndev_priv *priv = am65_ndev_to_priv(port->ndev); + + port->qos.iet.preemptible_tcs = preemptible_tcs; + mutex_lock(&priv->mm_lock); + am65_cpsw_iet_commit_preemptible_tcs(port); + mutex_unlock(&priv->mm_lock); +} + +static void am65_cpsw_iet_link_state_update(struct net_device *ndev) +{ + struct am65_cpsw_ndev_priv *priv = am65_ndev_to_priv(ndev); + struct am65_cpsw_port *port = am65_ndev_to_port(ndev); + + mutex_lock(&priv->mm_lock); + am65_cpsw_iet_commit_preemptible_tcs(port); + mutex_unlock(&priv->mm_lock); +} + static int am65_cpsw_port_est_enabled(struct am65_cpsw_port *port) { return port->qos.est_oper || port->qos.est_admin; @@ -737,6 +908,7 @@ static int am65_cpsw_taprio_replace(struct net_device *ndev,

devm_kfree(&ndev->dev, port->qos.est_admin); port->qos.est_admin = est_new; + am65_cpsw_iet_change_preemptible_tcs(port, taprio->mqprio.preemptible_tcs);

return 0;

@@ -1116,6 +1288,7 @@ void am65_cpsw_qos_link_up(struct net_device *ndev, int link_speed)

port->qos.link_speed = link_speed; am65_cpsw_tx_pn_shaper_apply(port); + am65_cpsw_iet_link_state_update(ndev);

am65_cpsw_est_link_up(ndev, link_speed); port->qos.link_down_time = 0; @@ -1127,6 +1300,7 @@ void am65_cpsw_qos_link_down(struct net_device *ndev)

port->qos.link_speed = SPEED_UNKNOWN; am65_cpsw_tx_pn_shaper_apply(port); + am65_cpsw_iet_link_state_update(ndev);

if (!port->qos.link_down_time) port->qos.link_down_time = ktime_get(); diff --git a/drivers/net/ethernet/ti/am65-cpsw-qos.h b/drivers/net/ethernet/ti/am65-cpsw-qos.h index b54c34c0aa7f..b328e56c5b2b 100644 --- a/drivers/net/ethernet/ti/am65-cpsw-qos.h +++ b/drivers/net/ethernet/ti/am65-cpsw-qos.h @@ -23,6 +23,12 @@ struct am65_cpsw_mqprio { bool shaper_en; };

+struct am65_cpsw_iet { + u8 preemptible_tcs; + u32 original_max_blks; + int verify_time_ms; +}; + struct am65_cpsw_ale_ratelimit { unsigned long cookie; u64 rate_packet_ps; @@ -34,6 +40,7 @@ struct am65_cpsw_qos { ktime_t link_down_time; int link_speed; struct am65_cpsw_mqprio mqprio; + struct am65_cpsw_iet iet;

struct am65_cpsw_ale_ratelimit ale_bc_ratelimit; struct am65_cpsw_ale_ratelimit ale_mc_ratelimit; @@ -93,6 +100,8 @@ void am65_cpsw_qos_link_up(struct net_device *ndev, int link_speed); void am65_cpsw_qos_link_down(struct net_device *ndev); int am65_cpsw_qos_ndo_tx_p0_set_maxrate(struct net_device *ndev, int queue, u32 rate_mbps); void am65_cpsw_qos_tx_p0_rate_init(struct am65_cpsw_common *common); +void am65_cpsw_iet_commit_preemptible_tcs(struct am65_cpsw_port *port); +void am65_cpsw_iet_common_enable(struct am65_cpsw_common *common); #else static inline int am65_cpsw_qos_ndo_setup_tc(struct net_device *ndev, enum tc_setup_type type, @@ -117,6 +126,102 @@ static inline int am65_cpsw_qos_ndo_tx_p0_set_maxrate(struct net_device *ndev,

static inline void am65_cpsw_qos_tx_p0_rate_init(struct am65_cpsw_common *common) { } +static inline void am65_cpsw_iet_commit_preemptible_tcs(struct am65_cpsw_port *port) +{ } +static inline void am65_cpsw_iet_common_enable(struct am65_cpsw_common *common) +{ } #endif

+#define AM65_CPSW_REG_CTL 0x004 +#define AM65_CPSW_PN_REG_CTL 0x004 +#define AM65_CPSW_PN_REG_MAX_BLKS 0x008 +#define AM65_CPSW_PN_REG_TX_PRI_MAP 0x018 +#define AM65_CPSW_PN_REG_RX_PRI_MAP 0x020 +#define AM65_CPSW_PN_REG_IET_CTRL 0x040 +#define AM65_CPSW_PN_REG_IET_STATUS 0x044 +#define AM65_CPSW_PN_REG_IET_VERIFY 0x048 +#define AM65_CPSW_PN_REG_FIFO_STATUS 0x050 +#define AM65_CPSW_PN_REG_EST_CTL 0x060 +#define AM65_CPSW_PN_REG_PRI_CIR(pri) (0x140 + 4 * (pri)) +#define AM65_CPSW_PN_REG_PRI_EIR(pri) (0x160 + 4 * (pri)) + +/* AM65_CPSW_REG_CTL register fields */ +#define AM65_CPSW_CTL_IET_EN BIT(17) +#define AM65_CPSW_CTL_EST_EN BIT(18) + +/* AM65_CPSW_PN_REG_CTL register fields */ +#define AM65_CPSW_PN_CTL_IET_PORT_EN BIT(16) +#define AM65_CPSW_PN_CTL_EST_PORT_EN BIT(17) + +/* AM65_CPSW_PN_REG_EST_CTL register fields */ +#define AM65_CPSW_PN_EST_ONEBUF BIT(0) +#define AM65_CPSW_PN_EST_BUFSEL BIT(1) +#define AM65_CPSW_PN_EST_TS_EN BIT(2) +#define AM65_CPSW_PN_EST_TS_FIRST BIT(3) +#define AM65_CPSW_PN_EST_ONEPRI BIT(4) +#define AM65_CPSW_PN_EST_TS_PRI_MSK GENMASK(7, 5) + +/* AM65_CPSW_PN_REG_IET_CTRL register fields */ +#define AM65_CPSW_PN_IET_MAC_PENABLE BIT(0) +#define AM65_CPSW_PN_IET_MAC_DISABLEVERIFY BIT(2) +#define AM65_CPSW_PN_IET_MAC_LINKFAIL BIT(3) +#define AM65_CPSW_PN_IET_MAC_MAC_ADDFRAGSIZE_MASK GENMASK(10, 8) +#define AM65_CPSW_PN_IET_MAC_MAC_ADDFRAGSIZE_OFFSET 8 +#define AM65_CPSW_PN_IET_MAC_PREMPT_MASK GENMASK(23, 16) +#define AM65_CPSW_PN_IET_MAC_PREMPT_OFFSET 16 + +#define AM65_CPSW_PN_IET_MAC_SET_ADDFRAGSIZE(n) (((n) << AM65_CPSW_PN_IET_MAC_MAC_ADDFRAGSIZE_OFFSET) & \ + AM65_CPSW_PN_IET_MAC_MAC_ADDFRAGSIZE_MASK) +#define AM65_CPSW_PN_IET_MAC_GET_ADDFRAGSIZE(n) (((n) & AM65_CPSW_PN_IET_MAC_MAC_ADDFRAGSIZE_MASK) >> \ + AM65_CPSW_PN_IET_MAC_MAC_ADDFRAGSIZE_OFFSET) +#define AM65_CPSW_PN_IET_MAC_SET_PREEMPT(n) (((n) << AM65_CPSW_PN_IET_MAC_PREMPT_OFFSET) & \ + AM65_CPSW_PN_IET_MAC_PREMPT_MASK) +#define AM65_CPSW_PN_IET_MAC_GET_PREEMPT(n) (((n) & AM65_CPSW_PN_IET_MAC_PREMPT_MASK) >> \ + AM65_CPSW_PN_IET_MAC_PREMPT_OFFSET) + +/* AM65_CPSW_PN_REG_IET_STATUS register fields */ +#define AM65_CPSW_PN_MAC_STATUS GENMASK(3, 0) +#define AM65_CPSW_PN_MAC_VERIFIED BIT(0) +#define AM65_CPSW_PN_MAC_VERIFY_FAIL BIT(1) +#define AM65_CPSW_PN_MAC_RESPOND_ERR BIT(2) +#define AM65_CPSW_PN_MAC_VERIFY_ERR BIT(3) + +/* AM65_CPSW_PN_REG_IET_VERIFY register fields */ +#define AM65_CPSW_PN_MAC_VERIFY_CNT_MASK GENMASK(23, 0) +#define AM65_CPSW_PN_MAC_GET_VERIFY_CNT(n) ((n) & AM65_CPSW_PN_MAC_VERIFY_CNT_MASK) +/* 10 msec converted to NSEC */ +#define AM65_CPSW_IET_VERIFY_CNT_MS (10) +#define AM65_CPSW_IET_VERIFY_CNT_NS (AM65_CPSW_IET_VERIFY_CNT_MS * \ + NSEC_PER_MSEC) + +/* AM65_CPSW_PN_REG_FIFO_STATUS register fields */ +#define AM65_CPSW_PN_FST_TX_PRI_ACTIVE_MSK GENMASK(7, 0) +#define AM65_CPSW_PN_FST_TX_E_MAC_ALLOW_MSK GENMASK(15, 8) +#define AM65_CPSW_PN_FST_EST_CNT_ERR BIT(16) +#define AM65_CPSW_PN_FST_EST_ADD_ERR BIT(17) +#define AM65_CPSW_PN_FST_EST_BUFACT BIT(18) + +/* EST FETCH COMMAND RAM */ +#define AM65_CPSW_FETCH_RAM_CMD_NUM 0x80 +#define AM65_CPSW_FETCH_CNT_MSK GENMASK(21, 8) +#define AM65_CPSW_FETCH_CNT_MAX (AM65_CPSW_FETCH_CNT_MSK >> 8) +#define AM65_CPSW_FETCH_CNT_OFFSET 8 +#define AM65_CPSW_FETCH_ALLOW_MSK GENMASK(7, 0) +#define AM65_CPSW_FETCH_ALLOW_MAX AM65_CPSW_FETCH_ALLOW_MSK + +/* AM65_CPSW_PN_REG_MAX_BLKS fields for IET and No IET cases */ +/* 7 blocks for pn_rx_max_blks, 13 for pn_tx_max_blks*/ +#define AM65_CPSW_PN_TX_RX_MAX_BLKS_IET 0xD07 + +/* Slave IET Stats. register offsets */ +#define AM65_CPSW_STATN_IET_RX_ASSEMBLY_ERROR 0x140 +#define AM65_CPSW_STATN_IET_RX_ASSEMBLY_OK 0x144 +#define AM65_CPSW_STATN_IET_RX_SMD_ERROR 0x148 +#define AM65_CPSW_STATN_IET_RX_FRAG 0x14c +#define AM65_CPSW_STATN_IET_TX_HOLD 0x150 +#define AM65_CPSW_STATN_IET_TX_FRAG 0x154 + +/* number of priority queues per port FIFO */ +#define AM65_CPSW_PN_FIFO_PRIO_NUM 8 + #endif /* AM65_CPSW_QOS_H_ */