WARNING: internal 10|100 PHY requires magjack hardware!

[psychedup]

I'd like to know if anyone has gotten this to work yet.

I bought the magjack breakout from Sparkfun here: https://www.sparkfun.com/products/13021

Then I made these connections:
Pi5 pin 17 (RD+) -> P7
Pi5 pin 18 (RD-) -> P8
Pi5 pin 19 (TX+) -> P1
Pi5 pin 20 (TX-) -> P2

I downloaded and installed ayufan's latest release xenial-minimal-rock64-0.6.31-209-arm64.img.xz. (*I realize this is experimental but the 0.5.15 release didn't seem to have the phy enabled).

The latest images have the phy disabled in the device tree so I used this command to enable it:

Code:

sudo enable_dtoverlay eth1 ethernet@ff550000 okay

After that I see messages about eth1 in dmesg:

Code:

[   56.015672] rk_gmac-dwmac ff550000.ethernet: Looking up phy-supply from device tree
[   56.015787] rk_gmac-dwmac ff550000.ethernet: clock input or output? (output).
[   56.015798] rk_gmac-dwmac ff550000.ethernet: Can not read property: tx_delay.
[   56.016667] rk_gmac-dwmac ff550000.ethernet: set tx_delay to 0x30
[   56.017394] rk_gmac-dwmac ff550000.ethernet: Can not read property: rx_delay.
[   56.018205] rk_gmac-dwmac ff550000.ethernet: set rx_delay to 0x10
[   56.018938] rk_gmac-dwmac ff550000.ethernet: integrated PHY? (yes).
[   56.019139] rk_gmac-dwmac ff550000.ethernet: cannot get clock clk_mac_refout
[   56.019909] rk_gmac-dwmac ff550000.ethernet: cannot get clock clk_mac_speed
[   56.025793] rk_gmac-dwmac ff550000.ethernet: init for RMII
[   56.056195] stmmac - user ID: 0x10, Synopsys ID: 0x35
[   56.056204]  Ring mode enabled
[   56.056209]  DMA HW capability register supported
[   56.056212]  Normal descriptors
[   56.056217]  RX Checksum Offload Engine supported (type 2)
[   56.056221]  TX Checksum insertion supported
[   56.056225]  Wake-Up On Lan supported
[   56.056263]  Enable RX Mitigation via HW Watchdog Timer
[   56.056386] of_get_named_gpiod_flags: can't parse 'snps,reset-gpio' property of node '/ethernet@ff550000[0]'
[   56.157916] libphy: stmmac: probed
[   56.157932] eth%d: PHY ID 1234d400 at 0 IRQ POLL (stmmac-1:00) active
[   56.157939] eth%d: PHY ID 00000000 at 1 IRQ POLL (stmmac-1:01)
[   56.157945] eth%d: PHY ID 00000000 at 2 IRQ POLL (stmmac-1:02)
[   56.157951] eth%d: PHY ID 00000000 at 3 IRQ POLL (stmmac-1:03)
[   56.157957] eth%d: PHY ID 00000000 at 4 IRQ POLL (stmmac-1:04)
[   56.157963] eth%d: PHY ID 00000000 at 5 IRQ POLL (stmmac-1:05)
[   56.157969] eth%d: PHY ID 00000000 at 6 IRQ POLL (stmmac-1:06)
[   56.157975] eth%d: PHY ID 00000000 at 7 IRQ POLL (stmmac-1:07)
[   56.157980] eth%d: PHY ID 00000000 at 8 IRQ POLL (stmmac-1:08)
[   56.157986] eth%d: PHY ID 00000000 at 9 IRQ POLL (stmmac-1:09)
[   56.157992] eth%d: PHY ID 00000000 at 10 IRQ POLL (stmmac-1:0a)
[   56.157998] eth%d: PHY ID 00000000 at 11 IRQ POLL (stmmac-1:0b)
[   56.158004] eth%d: PHY ID 00000000 at 12 IRQ POLL (stmmac-1:0c)
[   56.158010] eth%d: PHY ID 00000000 at 13 IRQ POLL (stmmac-1:0d)
[   56.158016] eth%d: PHY ID 00000000 at 14 IRQ POLL (stmmac-1:0e)
[   56.158021] eth%d: PHY ID 00000000 at 15 IRQ POLL (stmmac-1:0f)
[   56.158027] eth%d: PHY ID 00000000 at 16 IRQ POLL (stmmac-1:10)
[   56.158033] eth%d: PHY ID 00000000 at 17 IRQ POLL (stmmac-1:11)
[   56.158039] eth%d: PHY ID 00000000 at 18 IRQ POLL (stmmac-1:12)
[   56.158045] eth%d: PHY ID 00000000 at 19 IRQ POLL (stmmac-1:13)
[   56.158051] eth%d: PHY ID 00000000 at 20 IRQ POLL (stmmac-1:14)
[   56.158057] eth%d: PHY ID 00000000 at 21 IRQ POLL (stmmac-1:15)
[   56.158170] eth%d: PHY ID 00000000 at 22 IRQ POLL (stmmac-1:16)
[   56.158178] eth%d: PHY ID 00000000 at 23 IRQ POLL (stmmac-1:17)
[   56.158184] eth%d: PHY ID 00000000 at 24 IRQ POLL (stmmac-1:18)
[   56.158191] eth%d: PHY ID 00000000 at 25 IRQ POLL (stmmac-1:19)
[   56.158197] eth%d: PHY ID 00000000 at 26 IRQ POLL (stmmac-1:1a)
[   56.158203] eth%d: PHY ID 00000000 at 27 IRQ POLL (stmmac-1:1b)
[   56.158209] eth%d: PHY ID 00000000 at 28 IRQ POLL (stmmac-1:1c)
[   56.158215] eth%d: PHY ID 00000000 at 29 IRQ POLL (stmmac-1:1d)
[   56.158221] eth%d: PHY ID 00000000 at 30 IRQ POLL (stmmac-1:1e)
[   56.158226] eth%d: PHY ID 00000000 at 31 IRQ POLL (stmmac-1:1f)
[   56.188998] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[   59.247515] rk_gmac-dwmac ff550000.ethernet eth1: Link is Up - 100Mbps/Half - flow control off

But it doesn't seem connected to anything:

Code:

rock64@rock64:~$ sudo dhclient -v eth1
Internet Systems Consortium DHCP Client 4.3.3
Copyright 2004-2015 Internet Systems Consortium.
All rights reserved.
For info, please visit https://www.isc.org/software/dhcp/

Listening on LPF/eth1/a6:18:42:9c:f9:05
Sending on   LPF/eth1/a6:18:42:9c:f9:05
Sending on   Socket/fallback
DHCPDISCOVER on eth1 to 255.255.255.255 port 67 interval 3 (xid=0x998c2875)
DHCPDISCOVER on eth1 to 255.255.255.255 port 67 interval 8 (xid=0x998c2875)
[  163.723405] Rockchip integrated EPHY stmmac-1:00: rockchip_integrated_phy_fixup_100M: reset phy
DHCPDISCOVER on eth1 to 255.255.255.255 port 67 interval 8 (xid=0x998c2875)
DHCPDISCOVER on eth1 to 255.255.255.255 port 67 interval 10 (xid=0x998c2875)
DHCPDISCOVER on eth1 to 255.255.255.255 port 67 interval 11 (xid=0x998c2875)
[  190.831400] Rockchip integrated EPHY stmmac-1:00: rockchip_integrated_phy_fixup_100M: reset phy
DHCPDISCOVER on eth1 to 255.255.255.255 port 67 interval 20 (xid=0x998c2875)
[  207.919403] Rockchip integrated EPHY stmmac-1:00: rockchip_integrated_phy_fixup_100M: reset phy

Also my switch doesn't show a link. I'm not sure if (a) my wiring is wrong, (b) I need to do something else to enable/configure it completely, or © the adapter is defective.

One thing that seemed odd to me, when I was checking my wiring with my multimeter set for continuity testing, I discovered that pins 1 and 2 on the magjack breakout are connected to each other. At first I thought that was my fault but it seems that they are connected inside the magjack. I looked at the datasheet and it seems that might be correct, but I'm not sure (I don't know much about transformers or ethernet at that level).

I'd appreciate any feedback on what I may be doing wrong.

Thanks,
Chris