Banana Pi BPI-M4-Zero (sun50i-h618) EMAC1 internal PHY not detected (“No PHY found!”)

[GeEg]

armbianmonitor logs: https://paste.armbian.com/rolisaboru

 

I’m running Armbian 25.2 on a Banana Pi BPI-M4-Zero (Allwinner sun50i-h618). EMAC1 (ethernet@5030000) with its internal PHY never comes up:

sunxi-gmac 5030000.ethernet end0: renamed from eth0
… No PHY found! … phy init again… No PHY found! … phy init failed

 

So far I have tried:
 

1. Directly patching full.dts

Enabled /soc/ethernet@5030000 (status = “okay”)

Added phy-mode = "rmii", pinctrl-0 = <&rmii_pins>, phy-handle, phy-supply etc.

Defined a simple ethernet-phy@0 under MDIO.

Recompiled to sun50i-h618-bananapi-m4-zero.dtb and rebooted.

 

2. User overlay (.dts → .dtbo)

Created fragments to:

Turn on /soc/ethernet@5030000/mdio

Inject an ethernet-phy@0 node

Enable EMAC1, assign RMII pins, MDIO and PHY handle

Compiled with dtc -@ and loaded via user_overlays=bpi-m4z-eth in armbianEnv.txt.

3. Following Armbian build PR #6617
Tried the patch for H618’s internal AC300 PHY from https://github.com/armbian/build/pull/6617

No change: PHY still not detected, no power-up.

4. Troubleshooting notes

Overlay apply errors: FDT_ERR_NOTFOUND

dtc compile errors on duplicate or missing phandle references

Verified pinctrl entries (rmii-pins) and alias names match base DTS

I’ve collected a full hardware/software report via armbianmonitor—see the logs here: https://paste.armbian.com/rolisaboru
 

What I’m looking for

A known-good snippet of device-tree (base DTS or overlay) that brings up the internal EMAC1 PHY on sun50i-h618

Advice on any missing power-or-clock settings or MFD nodes required for the AC300 PHY

Confirmation whether additional kernel/config patches are needed

Thanks in advance for any pointers!

 

[going]

3 часа назад, GeEg сказал:

I’ve collected a full hardware/software report via armbianmonitor—see the logs here: https://paste.armbian.com/rolisaboru
 

What I’m looking for

A known-good snippet of device-tree (base DTS or overlay) that brings up the internal EMAC1 PHY on sun50i-h618

Advice on any missing power-or-clock settings or MFD nodes required for the AC300 PHY

 

> sunxi-gmac 5030000.ethernet end0: No PHY found!

The first question. What brand of chip is used in your case?

 

The second question. Is this an external device?

[  341.101053] usb 2-1: new high-speed USB device number 2 using ehci-platform
[  341.257994] usb 2-1: New USB device found, idVendor=0b95, idProduct=772b, bcdDevice= 0.01
[  341.258024] usb 2-1: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[  341.258033] usb 2-1: Product: AX88772B
[  341.258039] usb 2-1: Manufacturer: ASIX Elec. Corp.
[  341.258045] usb 2-1: SerialNumber: E5A1DD
[  341.902647] asix 2-1:1.0 (unnamed net_device) (uninitialized): PHY [usb-002:002:10] driver [Asix Electronics AX88772C] (irq=POLL)
[  341.905987] Asix Electronics AX88772C usb-002:002:10: attached PHY driver (mii_bus:phy_addr=usb-002:002:10, irq=POLL)

 

In order for the ethernet-phy to work, it is necessary to turn on the PWM on the internal pin of the processor.

As example see: arch/arm64/boot/dts/allwinner/sun50i-h618-orangepi-zero2w.dts

and arch/arm64/boot/dts/allwinner/sun50i-h618-orangepi-zero2w.dts#L521-L539

[GeEg]

Quote

> sunxi-gmac 5030000.ethernet end0: No PHY found!

The first question. What brand of chip is used in your case?

It is Allwinner H618, Quad-core ARM Cortex™-A53 processor.

 

Quote

The second question. Is this an external device?

[ 341.101053] usb 2-1: new high-speed USB device number 2 using ehci-platform [ 341.257994] usb 2-1: New USB device found, idVendor=0b95, idProduct=772b, bcdDevice= 0.01 [ 341.258024] usb 2-1: New USB device strings: Mfr=1, Product=2, SerialNumber=3 [ 341.258033] usb 2-1: Product: AX88772B [ 341.258039] usb 2-1: Manufacturer: ASIX Elec. Corp. [ 341.258045] usb 2-1: SerialNumber: E5A1DD [ 341.902647] asix 2-1:1.0 (unnamed net_device) (uninitialized): PHY [usb-002:002:10] driver [Asix Electronics AX88772C] (irq=POLL) [ 341.905987] Asix Electronics AX88772C usb-002:002:10: attached PHY driver (mii_bus:phy_addr=usb-002:002:10, irq=POLL)

Yes it is external USB network card,  which I used just to be able to send logs made via armbianmonitor to the internet. 


The state You see in the logs is just latest stage of many different attempts: with custom overlays and/or fixes on the sun50i-h618-bananapi-m4-zero.dtb (decompile, fix, compile, boot). In addition have tried some patches using H616 files. 
I also found problem is due to, PHY isn't properly powered on, based on topic: https://github.com/armbian/build/pull/6617 but after plenty of tries didnt manage to make it work ;(

[going]

13 часов назад, GeEg сказал:

Цитата

> sunxi-gmac 5030000.ethernet end0: No PHY found!

The first question. What brand of chip is used in your case?

It is Allwinner H618, Quad-core ARM Cortex™-A53 processor.

I meant which Eth chip is used in your case

Bananapi-m4-berry:
h618 ==> Eth chip rtl8211f ==> connecting socket

Bananapi-m4-zero:
h618 ==> Eth chip ????  ==> connecting socket

As far as I understand, this is some kind of expansion card for the 26 pin connector.
Can you post a diagram of this expansion board here.
The brand of the Ethernet chip matters.

[GeEg]

Quote

I meant which Eth chip is used in your case
 

Bananapi-m4-berry:
h618 ==> Eth chip rtl8211f ==> connecting socket

Bananapi-m4-zero:
h618 ==> Eth chip ???? ==> connecting socket
 

As far as I understand, this is some kind of expansion card for the 26 pin connector.
Can you post a diagram of this expansion board here.
The brand of the Ethernet chip matters.

Ach, OK. On Banana Pi M4 Zero (H618), the Ethernet PHY is integrated into the SoC, specifically within the AC200 block. There is no external Ethernet PHY chip like RTL8211F — the EMAC1 interface uses RMII and talks directly to the internal PHY.

The issue I’m facing is likely due to incompatibility with kernel 6.x.x. I haven’t been able to get any overlay to work properly. However, with a kernel from the 5.4.x series, Ethernet works out-of-the-box on the same hardware.

Any insights on proper overlay structure or required patches for AC200 support in newer kernels would be highly appreciated. Thanks for your help!

[c0rnelius]

This is being worked on https://lore.kernel.org/linux-sunxi/20250526002924.2567843-1-james.hilliard1@gmail.com/T/#t

 

There is a version of this that was ported over from a vendor kernel to Armbian. https://github.com/armbian/build/pull/6617