mtk_eth_soc phylink updates
Subsystems (4)
cover mtk_eth_soc phylink updates
Hi,
This series ultimately updates mtk_eth_soc to use phylink_pcs, with some fixes along the way.
Previous attempts to update this driver (which is now marked as legacy) have failed due to lack of testing. I am hoping that this time will be different; Marek can test RGMII modes, but not SGMII. So all that we know is that this patch series probably doesn't break RGMII.
1) remove unused mac_mode and sgmii flags members from structures.
2) remove unnecessary interpretation of speed when configuring 1000 and 2500 Base-X
3) move configuration of SGMII duplex setting from mac_config() to link_up()
4) only pass in interface mode to mtk_sgmii_setup_mode_force()
5) move decision about which mtk_sgmii_setup_mode_*() function to call into mtk_sgmii.c
6) add a fixme comment for RGMII explaning why the call to mtk_gmac0_rgmii_adjust() is completely wrong - this needs to be addressed by someone who has the hardware and can test an appropriate fix. This fixme means that the driver still can't become non-legacy.
7) move gmac setup from mac_config() to mac_finish() - this preserves the order that we write to the hardware when we eventually convert to phylink_pcs()
8) move configuration of syscfg0 in SGMII/802.3z mode to mac_finish() for the same reasons as (7).
9) convert mtk_sgmii.c code structure and the mtk_sgmii structure to suit conversion to phylink_pcs
10) finally convert to phylink_pcs
As there has been no feedback from mtk_eth_soc maintainers to my RFC on April 6th, not my reminder on April 11th, so it's now time to merge this anyway. Mediatek code seems to be submitted to the kernel and then the maintainers scarper...
drivers/net/ethernet/mediatek/mtk_eth_soc.c | 103 +++++++++------- drivers/net/ethernet/mediatek/mtk_eth_soc.h | 40 +++---- drivers/net/ethernet/mediatek/mtk_sgmii.c | 176 ++++++++++++++++------------ 3 files changed, 186 insertions(+), 133 deletions(-)
Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/ FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!
PATCH 1/12 net: mtk_eth_soc: remove unused mac->mode
mac->mode is only ever written to in one location, and is thus superflous. Remove it.
--- drivers/net/ethernet/mediatek/mtk_eth_soc.c | 1 - drivers/net/ethernet/mediatek/mtk_eth_soc.h | 1 - 2 files changed, 2 deletions(-)
drivers/net/ethernet/mediatek/mtk_eth_soc.c +0 -1
@@ -2982,7 +2982,6 @@ static int mtk_add_mac(struct mtk_eth *eth, struct device_node *np)/* mac config is not set */ mac->interface =PHY_INTERFACE_MODE_NA ; - mac->mode =MLO_AN_PHY ; mac->speed =SPEED_UNKNOWN ; mac->phylink_config.dev = ð->netdev[id]->dev;
drivers/net/ethernet/mediatek/mtk_eth_soc.h +0 -1
@@ -999,7 +999,6 @@ struct mtk_eth {struct mtk_mac {int id;phy_interface_t interface; -unsigned int mode;int speed;struct device_node *of_node;struct phylink *phylink;
Signature · 1 line
2.30.2
PATCH 2/12 net: mtk_eth_soc: remove unused sgmii flags
In reply to Russell King (Oracle)The "flags" member of struct mtk_sgmii appears to be unused, as are the MTK_SGMII_PHYSPEED_* and MTK_HAS_FLAGS() macros. Remove them.
--- drivers/net/ethernet/mediatek/mtk_eth_soc.h | 8 -------- 1 file changed, 8 deletions(-)
drivers/net/ethernet/mediatek/mtk_eth_soc.h +0 -8
@@ -867,23 +867,15 @@ struct mtk_soc_data {/* currently no SoC has more than 2 macs */ #define MTK_MAX_DEVS 2 -#define MTK_SGMII_PHYSPEED_AN BIT(31) -#define MTK_SGMII_PHYSPEED_MASK GENMASK(2, 0) -#define MTK_SGMII_PHYSPEED_1000 BIT(0) -#define MTK_SGMII_PHYSPEED_2500 BIT(1) -#define MTK_HAS_FLAGS (flags, _x) (((flags) & (_x)) == (_x)) -/* struct mtk_sgmii - This is the structure holding sgmii regmap and its * characteristics * @regmap: The register map pointing at the range used to setup * SGMII modes -* @flags: The enum refers to which mode the sgmii wants to run on * @ana_rgc3: The offset refers to register ANA_RGC3 related to regmap */ struct mtk_sgmii {struct regmap *regmap[MTK_MAX_DEVS ]; -u32 flags[MTK_MAX_DEVS ];u32 ana_rgc3; };
Signature · 1 line
2.30.2
PATCH 3/12 net: mtk_eth_soc: add mask and update PCS speed definitions
In reply to Russell King (Oracle)The PCS speed setting is a two bit field, but it is defined as two separate bits. Add a bitfield mask for the speed definitions, an
use the FIELD_PREP() macro to define each PCS speed.
--- drivers/net/ethernet/mediatek/mtk_eth_soc.h | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-)
drivers/net/ethernet/mediatek/mtk_eth_soc.h +5 -3
@@ -17,6 +17,7 @@#include <linux/phylink.h> #include <linux/rhashtable.h> #include <linux/dim.h> +#include <linux/bitfield.h> #include "mtk_ppe.h" #define MTK_QDMA_PAGE_SIZE 2048
@@ -493,9 +494,10 @@#define SGMSYS_SGMII_MODE 0x20#define SGMII_IF_MODE_BIT0 BIT(0)#define SGMII_SPEED_DUPLEX_AN BIT(1) -#define SGMII_SPEED_10 0x0 -#define SGMII_SPEED_100 BIT(2) -#define SGMII_SPEED_1000 BIT(3) +#define SGMII_SPEED_MASK GENMASK(3, 2) +#define SGMII_SPEED_10 FIELD_PREP(SGMII_SPEED_MASK, 0) +#define SGMII_SPEED_100 FIELD_PREP(SGMII_SPEED_MASK, 1) +#define SGMII_SPEED_1000 FIELD_PREP(SGMII_SPEED_MASK, 2)#define SGMII_DUPLEX_FULL BIT(4)#define SGMII_IF_MODE_BIT5 BIT(5)#define SGMII_REMOTE_FAULT_DIS BIT(8)
Signature · 1 line
2.30.2
PATCH 4/12 net: mtk_eth_soc: correct 802.3z speed setting
In reply to Russell King (Oracle)Phylink does not guarantee that state->speed will be set correctly in the mac_config() call, so it's a bug that the driver makes use of it. Moreover, it is making use of it in a function that is only ever called for 1000BASE-X and 2500BASE-X which operate at a fixed speed which happens to be the same setting irrespective of the interface mode. We can simply remove the switch statement and just set the SGMII interface speed.
--- drivers/net/ethernet/mediatek/mtk_sgmii.c | 18 +++++------------- 1 file changed, 5 insertions(+), 13 deletions(-)
drivers/net/ethernet/mediatek/mtk_sgmii.c +5 -13
@@ -34,6 +34,7 @@ int mtk_sgmii_init(struct mtk_sgmii *ss, struct device_node *r, u32 ana_rgc3)return 0 ; } +/* For SGMII interface mode */ int mtk_sgmii_setup_mode_an (struct mtk_sgmii *ss,int id) {unsigned int val;
@@ -60,6 +61,9 @@ int mtk_sgmii_setup_mode_an(struct mtk_sgmii *ss, int id)return 0 ; } +/* For 1000BASE-X and 2500BASE-X interface modes, which operate at a +* fixed speed. +*/ int mtk_sgmii_setup_mode_force (struct mtk_sgmii *ss,int id,const struct phylink_link_state *state) {
@@ -82,19 +86,7 @@ int mtk_sgmii_setup_mode_force(struct mtk_sgmii *ss, int id,/* SGMII force mode setting */ regmap_read (ss->regmap[id],SGMSYS_SGMII_MODE , &val); val &= ~SGMII_IF_MODE_MASK ; - -switch (state->speed) { -case SPEED_10 : - val |=SGMII_SPEED_10 ; -break ; -case SPEED_100 : - val |=SGMII_SPEED_100 ; -break ; -case SPEED_2500 : -case SPEED_1000 : - val |=SGMII_SPEED_1000 ; -break ; - } + val |=SGMII_SPEED_1000 ;if (state->duplex ==DUPLEX_FULL ) val |=SGMII_DUPLEX_FULL ;
Signature · 1 line
2.30.2
PATCH 5/12 net: mtk_eth_soc: correct 802.3z duplex setting
In reply to Russell King (Oracle)Phylink does not guarantee that state->duplex will be set correctly in the mac_config() call, so it's a bug that the driver makes use of it.
Move the 802.3z PCS duplex configuration to mac_link_up().
--- drivers/net/ethernet/mediatek/mtk_eth_soc.c | 16 +++++++++++---- drivers/net/ethernet/mediatek/mtk_eth_soc.h | 1 + drivers/net/ethernet/mediatek/mtk_sgmii.c | 22 +++++++++++++++------ 3 files changed, 29 insertions(+), 10 deletions(-)
drivers/net/ethernet/mediatek/mtk_eth_soc.c +12 -4
@@ -485,8 +485,18 @@ static void mtk_mac_link_up(struct phylink_config *config, { struct mtk_mac *mac = container_of(config, struct mtk_mac, phylink_config); -u32 mcr =mtk_r32 (mac->hw,MTK_MAC_MCR (mac->id)); +u32 mcr; + +if (phy_interface_mode_is_8023z (interface)) { +struct mtk_eth *eth = mac->hw; + +/* Decide how GMAC and SGMIISYS be mapped */ +int sid = (MTK_HAS_CAPS (eth->soc->caps,MTK_SHARED_SGMII )) ? +0 : mac->id; +mtk_sgmii_link_up (eth->sgmii, sid, speed, duplex); + } + mcr =mtk_r32 (mac->hw,MTK_MAC_MCR (mac->id)); mcr &= ~(MAC_MCR_SPEED_100 |MAC_MCR_SPEED_1000 |MAC_MCR_FORCE_DPX |MAC_MCR_FORCE_TX_FC |MAC_MCR_FORCE_RX_FC );
@@ -2986,9 +2996,7 @@ static int mtk_add_mac(struct mtk_eth *eth, struct device_node *np) mac->phylink_config.dev = ð->netdev[id]->dev; mac->phylink_config.type =PHYLINK_NETDEV ; -/* This driver makes use of state->speed/state->duplex in -* mac_config -*/ +/* This driver makes use of state->speed in mac_config */ mac->phylink_config.legacy_pre_march2020 = true; mac->phylink_config.mac_capabilities =MAC_ASYM_PAUSE |MAC_SYM_PAUSE |MAC_10 |MAC_100 |MAC_1000 |MAC_2500FD ;
drivers/net/ethernet/mediatek/mtk_eth_soc.h +1 -0
@@ -1017,6 +1017,7 @@ int mtk_sgmii_init(struct mtk_sgmii *ss, struct device_node *np,int mtk_sgmii_setup_mode_an (struct mtk_sgmii *ss,int id);int mtk_sgmii_setup_mode_force (struct mtk_sgmii *ss,int id,const struct phylink_link_state *state); +void mtk_sgmii_link_up (struct mtk_sgmii *ss,int id,int speed,int duplex);void mtk_sgmii_restart_an (struct mtk_eth *eth,int mac_id);int mtk_gmac_sgmii_path_setup (struct mtk_eth *eth,int mac_id);
drivers/net/ethernet/mediatek/mtk_sgmii.c +16 -6
@@ -83,14 +83,10 @@ int mtk_sgmii_setup_mode_force(struct mtk_sgmii *ss, int id, val &= ~SGMII_AN_ENABLE ;regmap_write (ss->regmap[id],SGMSYS_PCS_CONTROL_1 , val); -/* SGMII force mode setting */ +/* Set the speed etc but leave the duplex unchanged */ regmap_read (ss->regmap[id],SGMSYS_SGMII_MODE , &val); - val &= ~SGMII_IF_MODE_MASK ; + val &=SGMII_DUPLEX_FULL | ~SGMII_IF_MODE_MASK ; val |=SGMII_SPEED_1000 ; - -if (state->duplex ==DUPLEX_FULL ) - val |=SGMII_DUPLEX_FULL ; -regmap_write (ss->regmap[id],SGMSYS_SGMII_MODE , val);/* Release PHYA power down state */
@@ -101,6 +97,20 @@ int mtk_sgmii_setup_mode_force(struct mtk_sgmii *ss, int id,return 0 ; } +/* For 1000BASE-X and 2500BASE-X interface modes */ +void mtk_sgmii_link_up (struct mtk_sgmii *ss,int id,int speed,int duplex) +{ +unsigned int val; + +/* SGMII force duplex setting */ +regmap_read (ss->regmap[id],SGMSYS_SGMII_MODE , &val); + val &= ~SGMII_DUPLEX_FULL ; +if (duplex ==DUPLEX_FULL ) + val |=SGMII_DUPLEX_FULL ; + +regmap_write (ss->regmap[id],SGMSYS_SGMII_MODE , val); +} +void mtk_sgmii_restart_an (struct mtk_eth *eth,int mac_id) {struct mtk_sgmii *ss = eth->sgmii;
Signature · 1 line
2.30.2
PATCH 6/12 net: mtk_eth_soc: stop passing phylink state to sgmii setup
In reply to Russell King (Oracle)Now that mtk_sgmii_setup_mode_force() only uses the interface mode from the phylink state, pass just the interface mode into this function.
--- drivers/net/ethernet/mediatek/mtk_eth_soc.c | 2 +- drivers/net/ethernet/mediatek/mtk_eth_soc.h | 2 +- drivers/net/ethernet/mediatek/mtk_sgmii.c | 4 ++-- 3 files changed, 4 insertions(+), 4 deletions(-)
drivers/net/ethernet/mediatek/mtk_eth_soc.c +1 -1
@@ -390,7 +390,7 @@ static void mtk_mac_config(struct phylink_config *config, unsigned int mode,/* Setup SGMIISYS with the determined property */ if (state->interface !=PHY_INTERFACE_MODE_SGMII ) err =mtk_sgmii_setup_mode_force (eth->sgmii, sid, - state); + state->interface);else if (phylink_autoneg_inband (mode)) err =mtk_sgmii_setup_mode_an (eth->sgmii, sid);
drivers/net/ethernet/mediatek/mtk_eth_soc.h +1 -1
@@ -1016,7 +1016,7 @@ int mtk_sgmii_init(struct mtk_sgmii *ss, struct device_node *np, u32 ana_rgc3);int mtk_sgmii_setup_mode_an (struct mtk_sgmii *ss,int id);int mtk_sgmii_setup_mode_force (struct mtk_sgmii *ss,int id, -const struct phylink_link_state *state); +phy_interface_t interface);void mtk_sgmii_link_up (struct mtk_sgmii *ss,int id,int speed,int duplex);void mtk_sgmii_restart_an (struct mtk_eth *eth,int mac_id);
drivers/net/ethernet/mediatek/mtk_sgmii.c +2 -2
@@ -65,7 +65,7 @@ int mtk_sgmii_setup_mode_an(struct mtk_sgmii *ss, int id)* fixed speed. */ int mtk_sgmii_setup_mode_force (struct mtk_sgmii *ss,int id, -const struct phylink_link_state *state) +phy_interface_t interface) {unsigned int val;
@@ -74,7 +74,7 @@ int mtk_sgmii_setup_mode_force(struct mtk_sgmii *ss, int id,regmap_read (ss->regmap[id], ss->ana_rgc3, &val); val &= ~RG_PHY_SPEED_MASK ; -if (state->interface ==PHY_INTERFACE_MODE_2500BASEX ) +if (interface ==PHY_INTERFACE_MODE_2500BASEX ) val |=RG_PHY_SPEED_3_125G ;regmap_write (ss->regmap[id], ss->ana_rgc3, val);
Signature · 1 line
2.30.2
PATCH 7/12 net: mtk_eth_soc: provide mtk_sgmii_config()
In reply to Russell King (Oracle)Provide mtk_sgmii_config() to wrap up the decisions about which SGMII configuration will be called.
--- drivers/net/ethernet/mediatek/mtk_eth_soc.c | 7 +------ drivers/net/ethernet/mediatek/mtk_eth_soc.h | 5 ++--- drivers/net/ethernet/mediatek/mtk_sgmii.c | 20 +++++++++++++++++--- 3 files changed, 20 insertions(+), 12 deletions(-)
drivers/net/ethernet/mediatek/mtk_eth_soc.c +1 -6
@@ -388,12 +388,7 @@ static void mtk_mac_config(struct phylink_config *config, unsigned int mode, 0 : mac->id;/* Setup SGMIISYS with the determined property */ -if (state->interface !=PHY_INTERFACE_MODE_SGMII ) - err =mtk_sgmii_setup_mode_force (eth->sgmii, sid, - state->interface); -else if (phylink_autoneg_inband (mode)) - err =mtk_sgmii_setup_mode_an (eth->sgmii, sid); - + err =mtk_sgmii_config (eth->sgmii, sid, mode, state->interface);if (err) goto init_err;
drivers/net/ethernet/mediatek/mtk_eth_soc.h +2 -3
@@ -1014,9 +1014,8 @@ u32 mtk_r32(struct mtk_eth *eth, unsigned reg);int mtk_sgmii_init (struct mtk_sgmii *ss,struct device_node *np,u32 ana_rgc3); -int mtk_sgmii_setup_mode_an (struct mtk_sgmii *ss,int id); -int mtk_sgmii_setup_mode_force (struct mtk_sgmii *ss,int id, -phy_interface_t interface); +int mtk_sgmii_config (struct mtk_sgmii *ss,int id,unsigned int mode, +phy_interface_t interface);void mtk_sgmii_link_up (struct mtk_sgmii *ss,int id,int speed,int duplex);void mtk_sgmii_restart_an (struct mtk_eth *eth,int mac_id);
drivers/net/ethernet/mediatek/mtk_sgmii.c +17 -3
@@ -35,7 +35,7 @@ int mtk_sgmii_init(struct mtk_sgmii *ss, struct device_node *r, u32 ana_rgc3) }/* For SGMII interface mode */ -int mtk_sgmii_setup_mode_an (struct mtk_sgmii *ss,int id) +static int mtk_sgmii_setup_mode_an (struct mtk_sgmii *ss,int id) {unsigned int val;
@@ -64,8 +64,8 @@ int mtk_sgmii_setup_mode_an(struct mtk_sgmii *ss, int id)/* For 1000BASE-X and 2500BASE-X interface modes, which operate at a * fixed speed. */ -int mtk_sgmii_setup_mode_force (struct mtk_sgmii *ss,int id, -phy_interface_t interface) +static int mtk_sgmii_setup_mode_force (struct mtk_sgmii *ss,int id, +phy_interface_t interface) {unsigned int val;
@@ -97,6 +97,20 @@ int mtk_sgmii_setup_mode_force(struct mtk_sgmii *ss, int id,return 0 ; } +int mtk_sgmii_config (struct mtk_sgmii *ss,int id,unsigned int mode, +phy_interface_t interface) +{ +int err =0 ; + +/* Setup SGMIISYS with the determined property */ +if (interface !=PHY_INTERFACE_MODE_SGMII ) + err =mtk_sgmii_setup_mode_force (ss, id, interface); +else if (phylink_autoneg_inband (mode)) + err =mtk_sgmii_setup_mode_an (ss, id); + +return err; +} +/* For 1000BASE-X and 2500BASE-X interface modes */ void mtk_sgmii_link_up (struct mtk_sgmii *ss,int id,int speed,int duplex) {
Signature · 1 line
2.30.2
PATCH 9/12 net: mtk_eth_soc: move MAC_MCR setting to mac_finish()
In reply to Russell King (Oracle)Move the setting of the MTK_MAC_MCR register from the end of mac_config into the phylink mac_finish() method, to keep it as the very last write that is done during configuration.
--- drivers/net/ethernet/mediatek/mtk_eth_soc.c | 33 ++++++++++++++------- 1 file changed, 22 insertions(+), 11 deletions(-)
drivers/net/ethernet/mediatek/mtk_eth_soc.c +22 -11
@@ -269,8 +269,8 @@ static void mtk_mac_config(struct phylink_config *config, unsigned int mode, struct mtk_mac *mac = container_of(config, struct mtk_mac, phylink_config);struct mtk_eth *eth = mac->hw; -u32 mcr_cur, mcr_new, sid, i;int val, ge_mode, err =0 ; +u32 sid, i;/* MT76x8 has no hardware settings between for the MAC */ if (!MTK_HAS_CAPS(eth->soc->caps, MTK_SOC_MT7628) &&
@@ -408,16 +408,6 @@ static void mtk_mac_config(struct phylink_config *config, unsigned int mode,return ; } -/* Setup gmac */ - mcr_cur =mtk_r32 (mac->hw,MTK_MAC_MCR (mac->id)); - mcr_new = mcr_cur; - mcr_new |=MAC_MCR_IPG_CFG |MAC_MCR_FORCE_MODE | -MAC_MCR_BACKOFF_EN |MAC_MCR_BACKPR_EN |MAC_MCR_FORCE_LINK ; - -/* Only update control register when needed! */ -if (mcr_new != mcr_cur) -mtk_w32 (mac->hw, mcr_new,MTK_MAC_MCR (mac->id)); -return ; err_phy:
@@ -430,6 +420,26 @@ static void mtk_mac_config(struct phylink_config *config, unsigned int mode, mac->id, phy_modes(state->interface), err); } +static int mtk_mac_finish (struct phylink_config *config,unsigned int mode, +phy_interface_t interface) +{ + struct mtk_mac *mac = container_of(config, struct mtk_mac, + phylink_config); +u32 mcr_cur, mcr_new; + +/* Setup gmac */ + mcr_cur =mtk_r32 (mac->hw,MTK_MAC_MCR (mac->id)); + mcr_new = mcr_cur; + mcr_new |=MAC_MCR_IPG_CFG |MAC_MCR_FORCE_MODE | +MAC_MCR_BACKOFF_EN |MAC_MCR_BACKPR_EN |MAC_MCR_FORCE_LINK ; + +/* Only update control register when needed! */ +if (mcr_new != mcr_cur) +mtk_w32 (mac->hw, mcr_new,MTK_MAC_MCR (mac->id)); + +return 0 ; +} +static void mtk_mac_pcs_get_state (struct phylink_config *config,struct phylink_link_state *state) {
@@ -534,6 +544,7 @@ static const struct phylink_mac_ops mtk_phylink_ops = { .mac_pcs_get_state = mtk_mac_pcs_get_state, .mac_an_restart = mtk_mac_an_restart, .mac_config = mtk_mac_config, + .mac_finish = mtk_mac_finish, .mac_link_down = mtk_mac_link_down, .mac_link_up = mtk_mac_link_up, };
Signature · 1 line
2.30.2
PATCH 8/12 net: mtk_eth_soc: add fixme comment for state->speed use
In reply to Russell King (Oracle)Add a fixme comment for the last remaining incorrect usage of state->speed in the mac_config() method, which is strangely in a code path which is only run when the PHY interface mode changes.
This means if we are in RGMII mode, changes in state->speed will not cause the INTF_MODE, TRGMII_RCK_CTRL and TRGMII_TCK_CTRL registers to be set according to the speed, nor will the TRGPLL clock be set to the correct value.
--- drivers/net/ethernet/mediatek/mtk_eth_soc.c | 8 ++++++++ 1 file changed, 8 insertions(+)
drivers/net/ethernet/mediatek/mtk_eth_soc.c +8 -0
@@ -327,6 +327,14 @@ static void mtk_mac_config(struct phylink_config *config, unsigned int mode, state->interface)) goto err_phy; }else { +/* FIXME: this is incorrect. Not only does it +* use state->speed (which is not guaranteed +* to be correct) but it also makes use of it +* in a code path that will only be reachable +* when the PHY interface mode changes, not +* when the speed changes. Consequently, RGMII +* is probably broken. +*/ mtk_gmac0_rgmii_adjust (mac->hw, state->interface, state->speed);
Signature · 1 line
2.30.2
PATCH 10/12 net: mtk_eth_soc: move restoration of SYSCFG0 to mac_finish()
In reply to Russell King (Oracle)The SGMIISYS configuration is performed while ETHSYS_SYSCFG0 is in a disabled state. In order to preserve this when we switch to phylink_pcs we need to move the restoration of this register to the mac_finish() callback.
--- drivers/net/ethernet/mediatek/mtk_eth_soc.c | 11 +++++++++-- drivers/net/ethernet/mediatek/mtk_eth_soc.h | 1 + 2 files changed, 10 insertions(+), 2 deletions(-)
drivers/net/ethernet/mediatek/mtk_eth_soc.c +9 -2
@@ -400,8 +400,8 @@ static void mtk_mac_config(struct phylink_config *config, unsigned int mode,if (err) goto init_err; -regmap_update_bits (eth->ethsys,ETHSYS_SYSCFG0 , -SYSCFG0_SGMII_MASK , val); +/* Save the syscfg0 value for mac_finish */ + mac->syscfg0 = val; }else if (phylink_autoneg_inband (mode)) {dev_err (eth->dev,"In-band mode not supported in non SGMII mode!\n" );
@@ -425,8 +425,15 @@ static int mtk_mac_finish(struct phylink_config *config, unsigned int mode, { struct mtk_mac *mac = container_of(config, struct mtk_mac, phylink_config); +struct mtk_eth *eth = mac->hw;u32 mcr_cur, mcr_new; +/* Enable SGMII */ +if (interface ==PHY_INTERFACE_MODE_SGMII || +phy_interface_mode_is_8023z (interface)) +regmap_update_bits (eth->ethsys,ETHSYS_SYSCFG0 , +SYSCFG0_SGMII_MASK , mac->syscfg0); +/* Setup gmac */ mcr_cur =mtk_r32 (mac->hw,MTK_MAC_MCR (mac->id)); mcr_new = mcr_cur;
drivers/net/ethernet/mediatek/mtk_eth_soc.h +1 -0
@@ -1001,6 +1001,7 @@ struct mtk_mac {struct mtk_hw_stats *hw_stats;__be32 hwlro_ip[MTK_MAX_LRO_IP_CNT ];int hwlro_ip_cnt; +unsigned int syscfg0; };/* the struct describing the SoC. these are declared in the soc_xyz.c files */
Signature · 1 line
2.30.2
PATCH 12/12 net: mtk_eth_soc: partially convert to phylink_pcs
In reply to Russell King (Oracle)Partially convert mtk_eth_soc to phylink_pcs, moving the configuration, link up and AN restart over. However, it seems mac_pcs_get_state() doesn't actually get the state from the PCS, so we can't convert that over without a better understanding of the hardware.
--- drivers/net/ethernet/mediatek/mtk_eth_soc.c | 49 ++++++++---------- drivers/net/ethernet/mediatek/mtk_eth_soc.h | 7 ++- drivers/net/ethernet/mediatek/mtk_sgmii.c | 55 +++++++++++---------- 3 files changed, 53 insertions(+), 58 deletions(-)
drivers/net/ethernet/mediatek/mtk_eth_soc.c +21 -28
@@ -263,6 +263,25 @@ static void mtk_gmac0_rgmii_adjust(struct mtk_eth *eth,mtk_w32 (eth, val,TRGMII_TCK_CTRL ); } +static struct phylink_pcs *mtk_mac_select_pcs (struct phylink_config *config, +phy_interface_t interface) +{ + struct mtk_mac *mac = container_of(config, struct mtk_mac, + phylink_config); +struct mtk_eth *eth = mac->hw; +unsigned int sid; + +if (interface ==PHY_INTERFACE_MODE_SGMII || +phy_interface_mode_is_8023z (interface)) { + sid = (MTK_HAS_CAPS (eth->soc->caps,MTK_SHARED_SGMII )) ? +0 : mac->id; + +return mtk_sgmii_select_pcs (eth->sgmii, sid); + } + +return NULL ; +} +static void mtk_mac_config (struct phylink_config *config,unsigned int mode,const struct phylink_link_state *state) {
@@ -270,7 +289,7 @@ static void mtk_mac_config(struct phylink_config *config, unsigned int mode, phylink_config);struct mtk_eth *eth = mac->hw;int val, ge_mode, err =0 ; -u32 sid, i; +u32 i;/* MT76x8 has no hardware settings between for the MAC */ if (!MTK_HAS_CAPS(eth->soc->caps, MTK_SOC_MT7628) &&
@@ -391,15 +410,6 @@ static void mtk_mac_config(struct phylink_config *config, unsigned int mode,SYSCFG0_SGMII_MASK , ~(u32)SYSCFG0_SGMII_MASK); -/* Decide how GMAC and SGMIISYS be mapped */ - sid = (MTK_HAS_CAPS (eth->soc->caps,MTK_SHARED_SGMII )) ? -0 : mac->id; - -/* Setup SGMIISYS with the determined property */ - err =mtk_sgmii_config (eth->sgmii, sid, mode, state->interface); -if (err) - goto init_err; -/* Save the syscfg0 value for mac_finish */ mac->syscfg0 = val; }else if (phylink_autoneg_inband (mode)) {
@@ -479,14 +489,6 @@ static void mtk_mac_pcs_get_state(struct phylink_config *config, state->pause |=MLO_PAUSE_TX ; } -static void mtk_mac_an_restart (struct phylink_config *config) -{ - struct mtk_mac *mac = container_of(config, struct mtk_mac, - phylink_config); - -mtk_sgmii_restart_an (mac->hw, mac->id); -} -static void mtk_mac_link_down (struct phylink_config *config,unsigned int mode,phy_interface_t interface) {
@@ -507,15 +509,6 @@ static void mtk_mac_link_up(struct phylink_config *config, phylink_config);u32 mcr; -if (phy_interface_mode_is_8023z (interface)) { -struct mtk_eth *eth = mac->hw; - -/* Decide how GMAC and SGMIISYS be mapped */ -int sid = (MTK_HAS_CAPS (eth->soc->caps,MTK_SHARED_SGMII )) ? -0 : mac->id; -mtk_sgmii_link_up (eth->sgmii, sid, speed, duplex); - } - mcr =mtk_r32 (mac->hw,MTK_MAC_MCR (mac->id)); mcr &= ~(MAC_MCR_SPEED_100 | MAC_MCR_SPEED_1000 | MAC_MCR_FORCE_DPX | MAC_MCR_FORCE_TX_FC |
@@ -548,8 +541,8 @@ static void mtk_mac_link_up(struct phylink_config *config,static const struct phylink_mac_ops mtk_phylink_ops = { .validate = phylink_generic_validate, + .mac_select_pcs = mtk_mac_select_pcs, .mac_pcs_get_state = mtk_mac_pcs_get_state, - .mac_an_restart = mtk_mac_an_restart, .mac_config = mtk_mac_config, .mac_finish = mtk_mac_finish, .mac_link_down = mtk_mac_link_down,
drivers/net/ethernet/mediatek/mtk_eth_soc.h +3 -4
@@ -874,10 +874,12 @@ struct mtk_soc_data {* @regmap: The register map pointing at the range used to setup * SGMII modes * @ana_rgc3: The offset refers to register ANA_RGC3 related to regmap +* @pcs: Phylink PCS structure */ struct mtk_pcs {struct regmap *regmap;u32 ana_rgc3; +struct phylink_pcs pcs; };/* struct mtk_sgmii - This is the structure holding sgmii regmap and its
@@ -1020,12 +1022,9 @@ void mtk_stats_update_mac(struct mtk_mac *mac);void mtk_w32 (struct mtk_eth *eth,u32 val,unsigned reg);u32 mtk_r32 (struct mtk_eth *eth,unsigned reg); +struct phylink_pcs *mtk_sgmii_select_pcs (struct mtk_sgmii *ss,int id);int mtk_sgmii_init (struct mtk_sgmii *ss,struct device_node *np,u32 ana_rgc3); -int mtk_sgmii_config (struct mtk_sgmii *ss,int id,unsigned int mode, -phy_interface_t interface); -void mtk_sgmii_link_up (struct mtk_sgmii *ss,int id,int speed,int duplex); -void mtk_sgmii_restart_an (struct mtk_eth *eth,int mac_id);int mtk_gmac_sgmii_path_setup (struct mtk_eth *eth,int mac_id);int mtk_gmac_gephy_path_setup (struct mtk_eth *eth,int mac_id);
drivers/net/ethernet/mediatek/mtk_sgmii.c +29 -26
113 lines not on this page · raw message
Signature · 1 line
2.30.2
PATCH 11/12 net: mtk_eth_soc: convert code structure to suit split PCS support
In reply to Russell King (Oracle)Provide a mtk_pcs structure which encapsulates everything that the PCS functions need (the regmap and ana_rgc3 offset), and use this in the PCS functions. Provide shim functions to convert from the existing "mtk_sgmii_*" interface to the converted PCS functions.
--- drivers/net/ethernet/mediatek/mtk_eth_soc.h | 15 ++- drivers/net/ethernet/mediatek/mtk_sgmii.c | 123 +++++++++++--------- 2 files changed, 79 insertions(+), 59 deletions(-)
drivers/net/ethernet/mediatek/mtk_eth_soc.h +11 -4
@@ -869,16 +869,23 @@ struct mtk_soc_data {/* currently no SoC has more than 2 macs */ #define MTK_MAX_DEVS 2 -/* struct mtk_sgmii - This is the structure holding sgmii regmap and its -* characteristics +/* struct mtk_pcs - This structure holds each sgmii regmap and associated +* data * @regmap: The register map pointing at the range used to setup * SGMII modes * @ana_rgc3: The offset refers to register ANA_RGC3 related to regmap */ +struct mtk_pcs { +struct regmap *regmap; +u32 ana_rgc3; +}; +/* struct mtk_sgmii - This is the structure holding sgmii regmap and its +* characteristics +* @pcs Array of individual PCS structures +*/ struct mtk_sgmii { -struct regmap *regmap[MTK_MAX_DEVS ]; -u32 ana_rgc3; +struct mtk_pcs pcs[MTK_MAX_DEVS ]; };/* struct mtk_eth - This is the main datasructure for holding the state
drivers/net/ethernet/mediatek/mtk_sgmii.c +68 -55
202 lines not on this page · raw message
Signature · 1 line
2.30.2
Quoted 36 lines · On Wed, 18 May 2022 15:53:19 +0100 "Russell King (Oracle)"…
Hi,
This series ultimately updates
mtk_eth_socto usephylink_pcs, with some fixes along the way.Previous attempts to update this driver (which is now marked as legacy) have failed due to lack of testing. I am hoping that this time will be different; Marek can test RGMII modes, but not SGMII. So all that we know is that this patch series probably doesn't break RGMII.
1) remove unused
mac_modeand sgmii flags members from structures.
2) remove unnecessary interpretation of speed when configuring 1000 and 2500 Base-X
3) move configuration of SGMII duplex setting frommac_config()tolink_up()
4) only pass in interface mode tomtk_sgmii_setup_mode_force()
5) move decision about which mtk_sgmii_setup_mode_*() function to call intomtk_sgmii.c
6) add a fixme comment for RGMII explaning why the call tomtk_gmac0_rgmii_adjust()is completely wrong - this needs to be addressed by someone who has the hardware and can test an appropriate fix. This fixme means that the driver still can't become non-legacy.
7) move gmac setup frommac_config()tomac_finish()- this preserves the order that we write to the hardware when we eventually convert tophylink_pcs()
8) move configuration of syscfg0 in SGMII/802.3z mode tomac_finish()for the same reasons as (7).
9) convertmtk_sgmii.ccode structure and themtk_sgmiistructure to suit conversion tophylink_pcs
10) finally convert tophylink_pcsAs there has been no feedback from
mtk_eth_socmaintainers to my RFC on April 6th, not my reminder on April 11th, so it's now time to merge this anyway. Mediatek code seems to be submitted to the kernel and then the maintainers scarper...
Quoted 5 lines
drivers/net/ethernet/mediatek/mtk_eth_soc.c | 103 +++++++++------- drivers/net/ethernet/mediatek/mtk_eth_soc.h | 40 +++---- drivers/net/ethernet/mediatek/mtk_sgmii.c | 176 ++++++++++++++++------------ 3 files changed, 186 insertions(+), 133 deletions(-)
This series was applied to netdev/net-next.git (master) by Jakub Kicinski <[email protected]>:
Hello:
This series was applied to netdev/net-next.git (master) by Jakub Kicinski <[email protected]>:
Quoted 11 lines · On Wed, 18 May 2022 15:53:19 +0100 you wrote:
Hi,
This series ultimately updates
mtk_eth_socto usephylink_pcs, with some fixes along the way.Previous attempts to update this driver (which is now marked as legacy) have failed due to lack of testing. I am hoping that this time will be different; Marek can test RGMII modes, but not SGMII. So all that we know is that this patch series probably doesn't break RGMII.
[...]
Here is the summary with links:
- [net-next,01/12] net: mtk_eth_soc: remove unused mac->mode
https://git.kernel.org/netdev/net-next/c/0600bdde1fae - [net-next,02/12] net: mtk_eth_soc: remove unused sgmii flags https://git.kernel.org/netdev/net-next/c/5a7a2f4b29d7 - [net-next,03/12] net: mtk_eth_soc: add mask and update PCS speed definitions https://git.kernel.org/netdev/net-next/c/bc5e93e0cd22 - [net-next,04/12] net: mtk_eth_soc: correct 802.3z speed setting https://git.kernel.org/netdev/net-next/c/7da3f901f8ec - [net-next,05/12] net: mtk_eth_soc: correct 802.3z duplex setting https://git.kernel.org/netdev/net-next/c/a459187390bb - [net-next,06/12] net: mtk_eth_soc: stop passing phylink state to sgmii setup https://git.kernel.org/netdev/net-next/c/4ce5a0bd3958 - [net-next,07/12] net: mtk_eth_soc: provide mtk_sgmii_config() https://git.kernel.org/netdev/net-next/c/1ec619ee4a05 - [net-next,08/12] net: mtk_eth_soc: add fixme comment for state->speed use https://git.kernel.org/netdev/net-next/c/650a49bc65df - [net-next,09/12] net: mtk_eth_soc: move MAC_MCR setting to mac_finish() https://git.kernel.org/netdev/net-next/c/0e37ad71b2ff - [net-next,10/12] net: mtk_eth_soc: move restoration of SYSCFG0 to mac_finish() https://git.kernel.org/netdev/net-next/c/21089867278d - [net-next,11/12] net: mtk_eth_soc: convert code structure to suit split PCS support https://git.kernel.org/netdev/net-next/c/901f3fbe13c3 - [net-next,12/12] net: mtk_eth_soc: partially convert to phylink_pcs https://git.kernel.org/netdev/net-next/c/14a44ab0330d
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html