net: dsa: add driver for MaxLinear GSW1xx switch family
Subsystems (3)
RFC PATCH 22/23 net: dsa: add driver for MaxLinear GSW1xx switch family
Add driver for the MaxLinear GSW1xx family of Ethernet switch ICs which are based on the same IP as the Lantiq/Intel GSWIP found in the Lantiq VR9 and Intel GRX MIPS router SoCs. The main difference is that instead of using memory-mapped I/O to communicate with the host CPU these ICs are connected via MDIO (or SPI, which isn't supported by this driver). Implement the regmap API to access the switch registers over MDIO to allow reusing lantiq_gswip_common for all core functionality.
The GSW1xx also comes with a SerDes port capable of 1000Base-X, SGMII and 2500Base-X, which can either be used to connect an external PHY or SFP cage, or as the CPU port. Support for the SerDes interface is implemented in this driver using the phylink_pcs interface.
--- drivers/net/dsa/Kconfig | 12 + drivers/net/dsa/Makefile | 1 + drivers/net/dsa/mxl-gsw1xx.c | 710 +++++++++++++++++++++++++++++++ drivers/net/dsa/mxl-gsw1xx_pce.h | 160 +++++++ 4 files changed, 883 insertions(+) create mode 100644 drivers/net/dsa/mxl-gsw1xx.c create mode 100644 drivers/net/dsa/mxl-gsw1xx_pce.h
drivers/net/dsa/Kconfig +12 -0
@@ -39,6 +39,18 @@ config NET_DSA_LANTIQ_GSWIP This enables support for the Lantiq / Intel GSWIP 2.1 found in the xrx200 / VR9 SoC. +config NET_DSA_MXL_GSW1XX + tristate "MaxLinear GSW1xx Ethernet switch support" + select NET_DSA_TAG_MXL_GSW1XX + select NET_DSA_LANTIQ_COMMON + help + This enables support for the MaxLinear GSW1xx family of 1GE switches + GSW120 4 port, 2 PHYs, RGMII & SGMII/2500Base-X + GSW125 4 port, 2 PHYs, RGMII & SGMII/2500Base-X, industrial temperature + GSW140 6 port, 4 PHYs, RGMII & SGMII/2500Base-X + GSW141 6 port, 4 PHYs, RGMII & SGMII + GSW145 6 port, 4 PHYs, RGMII & SGMII/2500Base-X, industrial temperature + config NET_DSA_MT7530 tristate "MediaTek MT7530 and MT7531 Ethernet switch support" select NET_DSA_TAG_MTK
drivers/net/dsa/Makefile +1 -0
@@ -7,6 +7,7 @@ obj-$(CONFIG_FIXED_PHY) += dsa_loop_bdinfo.o endif obj-$(CONFIG_NET_DSA_LANTIQ_GSWIP) += lantiq_gswip.o obj-$(CONFIG_NET_DSA_LANTIQ_COMMON) += lantiq_gswip_common.o +obj-$(CONFIG_NET_DSA_MXL_GSW1XX) += mxl-gsw1xx.o obj-$(CONFIG_NET_DSA_MT7530) += mt7530.o obj-$(CONFIG_NET_DSA_MT7530_MDIO) += mt7530-mdio.o obj-$(CONFIG_NET_DSA_MT7530_MMIO) += mt7530-mmio.o
drivers/net/dsa/mxl-gsw1xx.c new file +710 -0
710 lines not on this page · raw message
drivers/net/dsa/mxl-gsw1xx_pce.h new file +160 -0
160 lines not on this page · raw message
Signature · 1 line
2.50.1
Hi Daniel,
Quoted 12 lines · Daniel Golle
Add driver for the MaxLinear GSW1xx family of Ethernet switch ICs which are based on the same IP as the Lantiq/Intel GSWIP found in the Lantiq VR9 and Intel GRX MIPS router SoCs. The main difference is that instead of using memory-mapped I/O to communicate with the host CPU these ICs are connected via MDIO (or SPI, which isn't supported by this driver). Implement the regmap API to access the switch registers over MDIO to allow reusing
lantiq_gswip_commonfor all core functionality.The GSW1xx also comes with a SerDes port capable of 1000Base-X, SGMII and 2500Base-X, which can either be used to connect an external PHY or SFP cage, or as the CPU port. Support for the SerDes interface is implemented in this driver using the
phylink_pcsinterface.
...
Quoted 2 lines
drivers/net/dsa/mxl-gsw1xx.c new file +0
-0
...
Quoted 4 lines
static int gsw1xx_sgmii_pcs_config(struct
phylink_pcs*pcs,drivers/net/dsa/mxl-gsw1xx.c new file +3
-0+unsigned int neg_mode, + phy_interface_t interface, + const unsigned long *advertising,
drivers/net/dsa/mxl-gsw1xx.c line 167
Quoted 10 lines
drivers/net/dsa/mxl-gsw1xx.c new file +10
-0+ bool permit_pause_to_mac) +{ +struct gsw1xx_priv *priv =sgmii_pcs_to_gsw1xx (pcs); +bool sgmii_mac_mode =dsa_is_user_port (priv->gswip.ds,GSW1XX_SGMII_PORT ); +u16 txaneg, anegctl, val, nco_ctrl; +int ret; + +/* Assert and deassert SGMII shell reset */ + ret =regmap_set_bits (priv->shell,GSW1XX_SHELL_RST_REQ , +GSW1XX_RST_REQ_SGMII_SHELL );
Can this be moved into gsw1xx_probe() maybe?
The thing is, if the switch is bootstrapped in
"Self-start Mode: Managed Switch Sub-Mode", SGMII will be already brought out of reset (by bootloader?) (GSWIP_CFG register), refer
to "Table 12 Registers Configuration for Self-start Mode: Managed Switch Sub-Mode" in datasheet. And nobody would disable SGMII if it's unused otherwise.
Quoted 14 lines
drivers/net/dsa/mxl-gsw1xx.c new file +14
-0+if (ret <0 ) +return ret; + + ret =regmap_clear_bits (priv->shell,GSW1XX_SHELL_RST_REQ , +GSW1XX_RST_REQ_SGMII_SHELL ); +if (ret <0 ) +return ret; + +/* Hardware Bringup FSM Enable */ + ret =regmap_write (priv->sgmii,GSW1XX_SGMII_PHY_HWBU_CTRL , +GSW1XX_SGMII_PHY_HWBU_CTRL_EN_HWBU_FSM | +GSW1XX_SGMII_PHY_HWBU_CTRL_HW_FSM_EN ); +if (ret <0 ) +return ret;
drivers/net/dsa/mxl-gsw1xx.c line 191
Quoted 10 lines
drivers/net/dsa/mxl-gsw1xx.c new file +10
-0+ +/* Configure SGMII PHY Receiver */ + val =FIELD_PREP (GSW1XX_SGMII_PHY_RX0_CFG2_EQ , +GSW1XX_SGMII_PHY_RX0_CFG2_EQ_DEF ) | +GSW1XX_SGMII_PHY_RX0_CFG2_LOS_EN | +GSW1XX_SGMII_PHY_RX0_CFG2_TERM_EN | +FIELD_PREP (GSW1XX_SGMII_PHY_RX0_CFG2_FILT_CNT , +GSW1XX_SGMII_PHY_RX0_CFG2_FILT_CNT_DEF ); + +// if (!priv->dts.sgmii_rx_invert)
^^
There is still a room for some cleanup ;-)
Quoted 6 lines
drivers/net/dsa/mxl-gsw1xx.c new file +6
-0+ val |=GSW1XX_SGMII_PHY_RX0_CFG2_INVERT ; + + ret =regmap_write (priv->sgmii,GSW1XX_SGMII_PHY_RX0_CFG2 , val); +if (ret <0 ) +return ret; +
...
Quoted 118 lines
drivers/net/dsa/mxl-gsw1xx.c new file +118
-0+static int gsw1xx_probe (struct mdio_device *mdiodev) +{ +struct device *dev = &mdiodev->dev; +struct gsw1xx_priv *priv; +struct dsa_switch *ds; +int ret; + + priv =devm_kzalloc (dev,sizeof (*priv),GFP_KERNEL ); +if (!priv) +return -ENOMEM ; + + priv->mdio_dev = mdiodev; + priv->smdio_badr =GSW1XX_SMDIO_BADR_UNKNOWN ; + + priv->gswip.dev = dev; + priv->gswip.hw_info =of_device_get_match_data (dev); +if (!priv->gswip.hw_info) +return -EINVAL ; + +mutex_init (&priv->gswip.pce_table_lock); + + priv->gswip.gswip =gsw1xx_regmap_init (priv,"switch" , +GSW1XX_SWITCH_BASE ,0xfff ); +if (IS_ERR (priv->gswip.gswip)) +return PTR_ERR (priv->gswip.gswip); + + priv->gswip.mdio =gsw1xx_regmap_init (priv,"mdio" ,GSW1XX_MMDIO_BASE , +0xff ); +if (IS_ERR (priv->gswip.mdio)) +return PTR_ERR (priv->gswip.mdio); + + priv->gswip.mii =gsw1xx_regmap_init (priv,"mii" ,GSW1XX_RGMII_BASE , +0xff ); +if (IS_ERR (priv->gswip.mii)) +return PTR_ERR (priv->gswip.mii); + + priv->sgmii =gsw1xx_regmap_init (priv,"sgmii" ,GSW1XX_SGMII_BASE , +0xfff ); +if (IS_ERR (priv->sgmii)) +return PTR_ERR (priv->sgmii); + + priv->gpio =gsw1xx_regmap_init (priv,"gpio" ,GSW1XX_GPIO_BASE , +0xff ); +if (IS_ERR (priv->gpio)) +return PTR_ERR (priv->gpio); + + priv->clk =gsw1xx_regmap_init (priv,"clk" ,GSW1XX_CLK_BASE ,0xff ); +if (IS_ERR (priv->clk)) +return PTR_ERR (priv->clk); + + priv->shell =gsw1xx_regmap_init (priv,"shell" ,GSW1XX_SHELL_BASE , +0xff ); +if (IS_ERR (priv->shell)) +return PTR_ERR (priv->shell); + + priv->sgmii_pcs.ops = &gsw1xx_sgmii_pcs_ops; + priv->sgmii_pcs.poll =1 ; +__set_bit (PHY_INTERFACE_MODE_SGMII , + priv->sgmii_pcs.supported_interfaces); +__set_bit (PHY_INTERFACE_MODE_1000BASEX , + priv->sgmii_pcs.supported_interfaces); +if (priv->gswip.hw_info->supports_2500m) +__set_bit (PHY_INTERFACE_MODE_2500BASEX , + priv->sgmii_pcs.supported_interfaces); + + ret =regmap_read (priv->gswip.gswip,GSWIP_VERSION , &priv->gswip.version); +if (ret <0 ) +return ret; + + ds =devm_kzalloc (dev,sizeof (*ds),GFP_KERNEL ); +if (!ds) +return -ENOMEM ; + + priv->gswip.ds = ds; + + ret =gswip_allocate_vlans (&priv->gswip); +if (ret) +return ret; + + ds->dev = dev; + ds->num_ports =GSW1XX_PORTS ; + ds->priv = &priv->gswip; + ds->ops = &gswip_switch_ops; + ds->phylink_mac_ops = &gswip_phylink_mac_ops; + ds->fdb_isolation = true; + +/* configure GPIO pin-mux for MMDIO in case of external PHY connected to +* SGMII or RGMII as slave interface +*/ +regmap_set_bits (priv->gpio,GPIO_ALTSEL0 ,3 ); +regmap_set_bits (priv->gpio,GPIO_ALTSEL1 ,3 ); + +dev_set_drvdata (dev, ds); + + ret =dsa_register_switch (ds); +if (ret) { +if (ret != -EPROBE_DEFER ) +dev_err (dev,"%s: Error %d register DSA switch\n" , + __func__, ret); +return ret; + } + + ret =gswip_validate_cpu_port (ds); +if (ret) + goto disable_switch; + +dev_info (dev,"probed GSWIP version %lx mod %lx\n" , + (priv->gswip.version &GSWIP_VERSION_REV_MASK ) >>GSWIP_VERSION_REV_SHIFT , + (priv->gswip.version &GSWIP_VERSION_MOD_MASK ) >>GSWIP_VERSION_MOD_SHIFT ); + +return 0 ; + +disable_switch: +gswip_disable_switch (&priv->gswip); +dsa_unregister_switch (ds); + +return ret; +}
...
Quoted 9 lines
drivers/net/dsa/mxl-gsw1xx.c new file +9
-0+static struct mdio_driver gsw1xx_driver = { + .probe = gsw1xx_probe, + .remove = gsw1xx_remove, + .shutdown = gsw1xx_shutdown, + .mdiodrv.driver = { + .name ="mxl-gsw1xx" , + .of_match_table = gsw1xx_of_match, + }, +};
Signature · 3 lines
Alexander Sverdlin Siemens AG www.siemens.com
Quoted 40 lines · Alexander
Hi Daniel,
Quoted 12 lines · Daniel Golle
Add driver for the MaxLinear GSW1xx family of Ethernet switch ICs which are based on the same IP as the Lantiq/Intel GSWIP found in the Lantiq VR9 and Intel GRX MIPS router SoCs. The main difference is that instead of using memory-mapped I/O to communicate with the host CPU these ICs are connected via MDIO (or SPI, which isn't supported by this driver). Implement the regmap API to access the switch registers over MDIO to allow reusing
lantiq_gswip_commonfor all core functionality.The GSW1xx also comes with a SerDes port capable of 1000Base-X, SGMII and 2500Base-X, which can either be used to connect an external PHY or SFP cage, or as the CPU port. Support for the SerDes interface is implemented in this driver using the
phylink_pcsinterface....
Quoted 2 lines
drivers/net/dsa/mxl-gsw1xx.c new file +0
-0...
Quoted 4 lines
static int gsw1xx_sgmii_pcs_config(struct
phylink_pcs*pcs,drivers/net/dsa/mxl-gsw1xx.c new file +3
-0+unsigned int neg_mode, + phy_interface_t interface, + const unsigned long *advertising,drivers/net/dsa/mxl-gsw1xx.c line 167
Quoted 10 lines
drivers/net/dsa/mxl-gsw1xx.c new file +10
-0+ bool permit_pause_to_mac) +{ +struct gsw1xx_priv *priv =sgmii_pcs_to_gsw1xx (pcs); +bool sgmii_mac_mode =dsa_is_user_port (priv->gswip.ds,GSW1XX_SGMII_PORT ); +u16 txaneg, anegctl, val, nco_ctrl; +int ret; + +/* Assert and deassert SGMII shell reset */ + ret =regmap_set_bits (priv->shell,GSW1XX_SHELL_RST_REQ , +GSW1XX_RST_REQ_SGMII_SHELL );Can this be moved into
gsw1xx_probe()maybe?
Quoted 5 lines
The thing is, if the switch is bootstrapped in
"Self-start Mode: Managed Switch Sub-Mode", SGMII will be already brought out of reset (by bootloader?) (GSWIP_CFGregister), refer
to "Table 12 Registers Configuration for Self-start Mode: Managed Switch Sub-Mode" in datasheet. And nobody would disable SGMII if it's unused otherwise.
What you say is true if the SGMII interface is used as the CPU port or to connect a (1000M/100M/10M) PHY. However, it can also be used to connect SFP modules, which can be hot-plugged. Or a 2500M/1000M/100M/10M PHY which requires switching to 2500Base-X mode in case of a 2500M link on the UTP interface comes up, but uses SGMII for all lower speeds.
We can probably do this similar to drivers/net/pcs/pcs-mtk-lynxi.c and only do a full reconf including reset if there are major changes which actually require that, but as the impact is minimal and the vendor implementation also carries out a reset as the first thing when configuring the SGMII interface, I'd just keep it like that for now. Optimization can come later if actually required.
Another good thing would probably be to implement pcs_enable() and pcs_disable() ops which put the whole SGMII port into a low-power state (stopping clocks, maybe asserting reset as you suggest...) and bring it back up. However, also this can be done after initial support has been merged and verified to work in all cases (I only got the MxL8611x evaluation board on which GSW145 is connected to MxL86111, so I can't really do anything too fancy with the SGMII interface other than making sure it works with that PHY with both, enabled and disabled SGMII in-band negotiation).
Quoted 25 lines
Quoted 14 lines
drivers/net/dsa/mxl-gsw1xx.c new file +14
-0+if (ret <0 ) +return ret; + + ret =regmap_clear_bits (priv->shell,GSW1XX_SHELL_RST_REQ , +GSW1XX_RST_REQ_SGMII_SHELL ); +if (ret <0 ) +return ret; + +/* Hardware Bringup FSM Enable */ + ret =regmap_write (priv->sgmii,GSW1XX_SGMII_PHY_HWBU_CTRL , +GSW1XX_SGMII_PHY_HWBU_CTRL_EN_HWBU_FSM | +GSW1XX_SGMII_PHY_HWBU_CTRL_HW_FSM_EN ); +if (ret <0 ) +return ret;drivers/net/dsa/mxl-gsw1xx.c line 191
Quoted 10 lines
drivers/net/dsa/mxl-gsw1xx.c new file +10
-0+ +/* Configure SGMII PHY Receiver */ + val =FIELD_PREP (GSW1XX_SGMII_PHY_RX0_CFG2_EQ , +GSW1XX_SGMII_PHY_RX0_CFG2_EQ_DEF ) | +GSW1XX_SGMII_PHY_RX0_CFG2_LOS_EN | +GSW1XX_SGMII_PHY_RX0_CFG2_TERM_EN | +FIELD_PREP (GSW1XX_SGMII_PHY_RX0_CFG2_FILT_CNT , +GSW1XX_SGMII_PHY_RX0_CFG2_FILT_CNT_DEF ); + +// if (!priv->dts.sgmii_rx_invert) ^^
Quoted 1 line
There is still a room for some cleanup ;-)
Ooops... I forgot about that, it should become a vendor DT property.
Hello Daniel,
Quoted 43 lines · Daniel Golle
Quoted 37 lines
Quoted 12 lines
Add driver for the MaxLinear GSW1xx family of Ethernet switch ICs which are based on the same IP as the Lantiq/Intel GSWIP found in the Lantiq VR9 and Intel GRX MIPS router SoCs. The main difference is that instead of using memory-mapped I/O to communicate with the host CPU these ICs are connected via MDIO (or SPI, which isn't supported by this driver). Implement the regmap API to access the switch registers over MDIO to allow reusing
lantiq_gswip_commonfor all core functionality.The GSW1xx also comes with a SerDes port capable of 1000Base-X, SGMII and 2500Base-X, which can either be used to connect an external PHY or SFP cage, or as the CPU port. Support for the SerDes interface is implemented in this driver using the
phylink_pcsinterface....
Quoted 2 lines
drivers/net/dsa/mxl-gsw1xx.c new file +0
-0...
Quoted 4 lines
static int gsw1xx_sgmii_pcs_config(struct
phylink_pcs*pcs,drivers/net/dsa/mxl-gsw1xx.c new file +3
-0+ unsigned int neg_mode, + phy_interface_t interface, + const unsigned long *advertising,drivers/net/dsa/mxl-gsw1xx.c line 167
Quoted 10 lines
drivers/net/dsa/mxl-gsw1xx.c new file +10
-0+ bool permit_pause_to_mac) +{ +struct gsw1xx_priv *priv =sgmii_pcs_to_gsw1xx (pcs); +bool sgmii_mac_mode =dsa_is_user_port (priv->gswip.ds,GSW1XX_SGMII_PORT ); +u16 txaneg, anegctl, val, nco_ctrl; +int ret; + +/* Assert and deassert SGMII shell reset */ + ret =regmap_set_bits (priv->shell,GSW1XX_SHELL_RST_REQ , + GSW1XX_RST_REQ_SGMII_SHELL);Can this be moved into
gsw1xx_probe()maybe?Quoted 5 lines
The thing is, if the switch is bootstrapped in
"Self-start Mode: Managed Switch Sub-Mode", SGMII will be already brought out of reset (by bootloader?) (GSWIP_CFGregister), refer
to "Table 12 Registers Configuration for Self-start Mode: Managed Switch Sub-Mode" in datasheet. And nobody would disable SGMII if it's unused otherwise.
Quoted 5 lines
What you say is true if the SGMII interface is used as the CPU port or to connect a (1000M/100M/10M) PHY. However, it can also be used to connect SFP modules, which can be hot-plugged. Or a 2500M/1000M/100M/10M PHY which requires switching to 2500Base-X mode in case of a 2500M link on the UTP interface comes up, but uses SGMII for all lower speeds.
I'm actually concerned about use-cases where SGMII is unused.
In "Self-start Mode" SGMII block is being brought up and driver will never disable it. I'm not proposing to move the de-assertion of the reset, but either
the assertion can be done unconditionally somewhere around probe
or struct dsa_switch_ops::setup callback or the assertion can remain here and be duplicated somewhere around init.
Quoted 6 lines
We can probably do this similar to
drivers/net/pcs/pcs-mtk-lynxi.cand only do a full reconf including reset if there are major changes which actually require that, but as the impact is minimal and the vendor implementation also carries out a reset as the first thing when configuring the SGMII interface, I'd just keep it like that for now. Optimization can come later if actually required.
Sure, it goes a bit beyond basic support as it's a power consumption optimization, but I thought I'll bring this up now as the re-spin will happen anyway and if you agree on moving the reset assertion, then later patching will not be required.
Signature · 3 lines
Alexander Sverdlin Siemens AG www.siemens.com
Quoted 52 lines · Alexander
Hello Daniel,
Quoted 43 lines · Daniel Golle
Quoted 37 lines
Quoted 12 lines
Add driver for the MaxLinear GSW1xx family of Ethernet switch ICs which are based on the same IP as the Lantiq/Intel GSWIP found in the Lantiq VR9 and Intel GRX MIPS router SoCs. The main difference is that instead of using memory-mapped I/O to communicate with the host CPU these ICs are connected via MDIO (or SPI, which isn't supported by this driver). Implement the regmap API to access the switch registers over MDIO to allow reusing
lantiq_gswip_commonfor all core functionality.The GSW1xx also comes with a SerDes port capable of 1000Base-X, SGMII and 2500Base-X, which can either be used to connect an external PHY or SFP cage, or as the CPU port. Support for the SerDes interface is implemented in this driver using the
phylink_pcsinterface....
Quoted 2 lines
drivers/net/dsa/mxl-gsw1xx.c new file +0
-0...
Quoted 4 lines
static int gsw1xx_sgmii_pcs_config(struct
phylink_pcs*pcs,drivers/net/dsa/mxl-gsw1xx.c new file +3
-0+ unsigned int neg_mode, + phy_interface_t interface, + const unsigned long *advertising,drivers/net/dsa/mxl-gsw1xx.c line 167
Quoted 10 lines
drivers/net/dsa/mxl-gsw1xx.c new file +10
-0+ bool permit_pause_to_mac) +{ +struct gsw1xx_priv *priv =sgmii_pcs_to_gsw1xx (pcs); +bool sgmii_mac_mode =dsa_is_user_port (priv->gswip.ds,GSW1XX_SGMII_PORT ); +u16 txaneg, anegctl, val, nco_ctrl; +int ret; + +/* Assert and deassert SGMII shell reset */ + ret =regmap_set_bits (priv->shell,GSW1XX_SHELL_RST_REQ , + GSW1XX_RST_REQ_SGMII_SHELL);Can this be moved into
gsw1xx_probe()maybe?Quoted 5 lines
The thing is, if the switch is bootstrapped in
"Self-start Mode: Managed Switch Sub-Mode", SGMII will be already brought out of reset (by bootloader?) (GSWIP_CFGregister), refer
to "Table 12 Registers Configuration for Self-start Mode: Managed Switch Sub-Mode" in datasheet. And nobody would disable SGMII if it's unused otherwise.Quoted 5 lines
What you say is true if the SGMII interface is used as the CPU port or to connect a (1000M/100M/10M) PHY. However, it can also be used to connect SFP modules, which can be hot-plugged. Or a 2500M/1000M/100M/10M PHY which requires switching to 2500Base-X mode in case of a 2500M link on the UTP interface comes up, but uses SGMII for all lower speeds.
Quoted 6 lines
I'm actually concerned about use-cases where SGMII is unused.
In "Self-start Mode" SGMII block is being brought up and driver will never disable it. I'm not proposing to move the de-assertion of the reset, but either
the assertion can be done unconditionally somewhere around probe
or structdsa_switch_ops::setupcallback or the assertion can remain here and be duplicated somewhere around init.
Lets assert the SGMII in the probe() function and let .pcs_enable() and .pcs_disable() handle deassertion and assertion at runtime. That's easy and obvious, and makes sure the SGMII reset is always asserted if the SGMII unit isn't used. We can later optmize more and also stop clocks or do whatever MaxLinear folks are telling us would be good to further reduce power consumption and potentially also EM noise.
Quoted 8 lines
Quoted 7 lines
We can probably do this similar to
drivers/net/pcs/pcs-mtk-lynxi.cand only do a full reconf including reset if there are major changes which actually require that, but as the impact is minimal and the vendor implementation also carries out a reset as the first thing when configuring the SGMII interface, I'd just keep it like that for now. Optimization can come later if actually required.
Quoted 4 lines
Sure, it goes a bit beyond basic support as it's a power consumption optimization, but I thought I'll bring this up now as the re-spin will happen anyway and if you agree on moving the reset assertion, then later patching will not be required.
Convinced me ;)