Add RK3588 support. Rework the RK3568 support to take advantage of the
PHY mode enum like the new RK3588 code.

ok jmatthew@

Attach dwqe(4) to Intel Elkhart Lake PSE SGMII devices.

Patch by msaitoh@netbsd, who tested both PSE SGMII ports on a Helix 330.

Add initial support for Elkhart Lake ethernet to dwqe(4).

For now, only attach to PSE0/RGMII (device ID 0x4ba0) which is the
only device I have access to for testing.

There is a known problem where Tx throughput is lower than expected.
This is being looked into.

ok kettenis@

Fiv the value written to dwqe(4) MAC_1US_TIC_CTR register.

The calculation of this value is supposed to involve a clock frequency
but we were using a clock ID in the range 0-7 instead.

ok kettenis, patrick

allow dwqe.c to build on architectures that do not have machine/fdt.h

Move struct if_device to a new fdt-specific softc struct, along with
the gmac_id field which is only used by if_dwqe_fdt.c at present.
This avoids the need to include any fdt header files in dwqe.c.

ok kettenis@

handle fixed-link configuration in the device tree.

if fixed-link is present, populate the interface baudrate and link
status (full duplex or half duplex), and then call the statch handler
to apply that config to the MAC. if fixed-link is specified then
do not attach a phy.

note that phy lookup and reset still occurs in case the device tree
still uses the deprecated snps,reset-gpio properties. the fixed
link port on the bpi r2 pro is connected to a switch chip (which
is not really a phy) that needs needs a reset provided by the
snps,reset-gpio handling. an improved device tree would have the
reset properties on the switch node so it could do its own reset,
but we can't have nice things can we?

tested on a nanopi r5s, which has a phy but no fixed-link config,
and a banana pi bpi-r2 pro, which has both (cos it has two dwqes,
one with a phy and one without).

rename sc_tick to sc_phy_tick. no functional change.

register a mapping of dwqe interfaces to ofw nodes/phandles.

i forgot to commit this as part of src/sys/dev/fdt/if_dwqe_fdt.c r1.7

found by mglocker@

Determine PHY mode and pass the appropriate flags down to the PHY when we
attach it.

ok dlg@

Use a task to switch clocks on RK3568 as the clock API needs process context.

ok jmatthew@

Add dwqe(4), a driver for the Synopsis DesignWare Ethernet QoS controller
used on the NXP i.MX8MP, the Rockchip RK35xx series and Intel Elkhart Lake.

For now it is committed with FDT-attachment only and still in rough shape,
but it's time for development to occur in-tree.

Discussed with bluhm@
ok kettenis@

Attach dwqe(4) to Intel Elkhart Lake PSE SGMII devices.

Patch by msaitoh@netbsd, who tested both PSE SGMII ports on a Helix 330.

Add initial support for Elkhart Lake ethernet to dwqe(4).

For now, only attach to PSE0/RGMII (device ID 0x4ba0) which is the
only device I have access to for testing.

There is a known problem where Tx throughput is lower than expected.
This is being looked into.

ok kettenis@

Fiv the value written to dwqe(4) MAC_1US_TIC_CTR register.

The calculation of this value is supposed to involve a clock frequency
but we were using a clock ID in the range 0-7 instead.

ok kettenis, patrick

allow dwqe.c to build on architectures that do not have machine/fdt.h

Move struct if_device to a new fdt-specific softc struct, along with
the gmac_id field which is only used by if_dwqe_fdt.c at present.
This avoids the need to include any fdt header files in dwqe.c.

ok kettenis@

handle fixed-link configuration in the device tree.

if fixed-link is present, populate the interface baudrate and link
status (full duplex or half duplex), and then call the statch handler
to apply that config to the MAC. if fixed-link is specified then
do not attach a phy.

note that phy lookup and reset still occurs in case the device tree
still uses the deprecated snps,reset-gpio properties. the fixed
link port on the bpi r2 pro is connected to a switch chip (which
is not really a phy) that needs needs a reset provided by the
snps,reset-gpio handling. an improved device tree would have the
reset properties on the switch node so it could do its own reset,
but we can't have nice things can we?

tested on a nanopi r5s, which has a phy but no fixed-link config,
and a banana pi bpi-r2 pro, which has both (cos it has two dwqes,
one with a phy and one without).

rename sc_tick to sc_phy_tick. no functional change.

register a mapping of dwqe interfaces to ofw nodes/phandles.

i forgot to commit this as part of src/sys/dev/fdt/if_dwqe_fdt.c r1.7

found by mglocker@

Determine PHY mode and pass the appropriate flags down to the PHY when we
attach it.

ok dlg@

Use a task to switch clocks on RK3568 as the clock API needs process context.

ok jmatthew@

Add dwqe(4), a driver for the Synopsis DesignWare Ethernet QoS controller
used on the NXP i.MX8MP, the Rockchip RK35xx series and Intel Elkhart Lake.

For now it is committed with FDT-attachment only and still in rough shape,
but it's time for development to occur in-tree.

Discussed with bluhm@
ok kettenis@

Add initial support for Elkhart Lake ethernet to dwqe(4).

For now, only attach to PSE0/RGMII (device ID 0x4ba0) which is the
only device I have access to for testing.

There is a known problem where Tx throughput is lower than expected.
This is being looked into.

ok kettenis@

Fiv the value written to dwqe(4) MAC_1US_TIC_CTR register.

The calculation of this value is supposed to involve a clock frequency
but we were using a clock ID in the range 0-7 instead.

ok kettenis, patrick

allow dwqe.c to build on architectures that do not have machine/fdt.h

Move struct if_device to a new fdt-specific softc struct, along with
the gmac_id field which is only used by if_dwqe_fdt.c at present.
This avoids the need to include any fdt header files in dwqe.c.

ok kettenis@

handle fixed-link configuration in the device tree.

if fixed-link is present, populate the interface baudrate and link
status (full duplex or half duplex), and then call the statch handler
to apply that config to the MAC. if fixed-link is specified then
do not attach a phy.

note that phy lookup and reset still occurs in case the device tree
still uses the deprecated snps,reset-gpio properties. the fixed
link port on the bpi r2 pro is connected to a switch chip (which
is not really a phy) that needs needs a reset provided by the
snps,reset-gpio handling. an improved device tree would have the
reset properties on the switch node so it could do its own reset,
but we can't have nice things can we?

tested on a nanopi r5s, which has a phy but no fixed-link config,
and a banana pi bpi-r2 pro, which has both (cos it has two dwqes,
one with a phy and one without).

rename sc_tick to sc_phy_tick. no functional change.

register a mapping of dwqe interfaces to ofw nodes/phandles.

i forgot to commit this as part of src/sys/dev/fdt/if_dwqe_fdt.c r1.7

found by mglocker@

Determine PHY mode and pass the appropriate flags down to the PHY when we
attach it.

ok dlg@

Use a task to switch clocks on RK3568 as the clock API needs process context.

ok jmatthew@

Add dwqe(4), a driver for the Synopsis DesignWare Ethernet QoS controller
used on the NXP i.MX8MP, the Rockchip RK35xx series and Intel Elkhart Lake.

For now it is committed with FDT-attachment only and still in rough shape,
but it's time for development to occur in-tree.

Discussed with bluhm@
ok kettenis@