Realtek RTL8152

Has anyone got this chipset working with the latest EtherUSB with CDC-ECM devices enabled?

My system will recognise it, and put an EtherUSB entry in Interfaces, but it will not connect successfully. Very mixed responses dependent on the method I use to set up the interface.

I seem to get further with the ROOL stack than the ROD stack, but both are struggling.

If I use DHCP, I get an IP address of 169.254.0.0 which I realise means it’s not connected. ShareFS shows the drives on the network, but trying to connect to them gets me the dreaded red hourglass and notghing beyond that.

If I set a static address system, I cannot connect outside of my home network, but ShareFS works as it should. This works with the ROOL stack, but doesn’t with the ROD stack.

I looked at the USBSetConfig command but the help and documentation I could find for that is minimal. I can easily work out the device element but what do I do with the config element? And would this help?

I notice in both cases, the listed hardware (MAC) address is 00:00:00:00:00:00, (as is the device’s serial number, according to USBinfo) which makes me wonder…

I notice in both cases, the listed hardware (MAC) address is 00:00:00:00:00:00

The prefix 00:00:00 says it’s a XEROX, but I think that the driver is failing to read the device details.

I have had a look into my sources history. From two users I know that the RTL8153 has been tested successfully. It seems that ROOL added the RTL8152 afterwards (in their latest source I am aware of). Perhaps you can examine the device yourself.
*USBDevInfo device_number
and afterwards
*USBConfinfo device_number

According to ROOLs source the configuration should be 2.

Perhaps you should an
*RMReinit EtherUSB

Or set the configuration manually to 2 (Using USBSetConfig) and aftwards doing a RMReinit.

Your description looks as if the device in general works. EtherUSB only creates an according ej driver entry.

*EJInfo

should show the entry. Using DHCP is such a thing.
You must perform it on the just created interface.
Be warned! There is no guarantee that e.g. in case that you are having two EJ devices the numbering will be always the same. In worst case the numbering is swapped and doing a DHCP on a certain EJ number will end up in a desaster. And of course doing a DHCP e.g. on an unconnected network device is also no fun. Sometimes ethernet cables are not having the required “connectivity”. However looking on EJInfo results should tell you whether the device is up.

Also

*ifconfig -a

may help to analyze the situation.

The MAC is obtained inside EtherUSB CDC part. This part may be a bit problematically AFAIR. The according USB control request requires a language. At my devices it will return always the same MAC (12 hexchars) for every language but you must pass a valid language. You should be able to trace this using syslog.

EtherUSB can reserve unit numbers for specific devices based on MAC address. This was added to aid development, but may help if you have multiple ej devices. This may only work for devices where the MAC is known early enough, however it only needs to be done for one of the devices.

https://gitlab.riscosopen.org/RiscOS/Sources/Networking/Ethernet/EtherUSB/-/blob/master/doc/!ReadMe

Okay, here’s some of the information that may help:

ejinfo
EtherUSB driver for USB network adapters, version 0.45, by James Peacock DCI version 4.07, prefix 'ej', maximum 16 units Supported backends: SMSC75xx - SMSC 75xx SMSC78xx - SMSC 78xx SMSC95xx - SMSC 95xx MCS7830 - MosChip MCS7830 Pegasus - Pegasus USB Ethernet Adapter AX88772 - ASIX AX88772 AX88172 - ASIX AX88172 CDC - USB Communications Device Class ECM

Type 8035 (AddrLvl 01, ErrLvl 00) handler=(fc3a2914/300035a4)
Type 0806 (AddrLvl 01, ErrLvl 00) handler=(fc3a2914/300035a4)
Type 0800 (AddrLvl 02, ErrLvl 00) handler=(fc3a2914/300035a4)


ifconfig -a
lo0: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> mtu 16384 inet 127.0.0.1 netmask 0xff000000 ej0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500 inet 192.168.1.112 netmask 0xffffff00 broadcast 192.168.1.255 ether 00:00:00:00:00:00

usbconfinfo 5
Current config : 2

Interface 1.1 class 10.0:0 ‘Ethernet Data’
Endpoint 1, Bulk IN, 512 bytes 0 frames
Endpoint 2, Bulk OUT, 512 bytes 0 frames


usbdevinfo 5
USB release : 0210 Device class : 00 Device subclass : 00 Device protocol : 00 Max packet size : 40 Vendor ID : 0BDA Product ID : 8152 Device ID : 2000

1. of configs : 2
Manufacturer : ‘Realtek’
Product : ‘USB 10/100 LAN’
Serial number : ‘000000000000’
Speed : High


All looks like I would expect it to.

I’m increasingly minded to suspect that, because the serial number is 000000000000, could this be being translated to a MAC address of 00:00:00:00:00:00? And could it be that my router is just rejecting this because it doesn’t think it’s real?

This is borne out by the router not knowing the device is connected, even when using static IPs.

Back in the day, wasn’t there a command to set the MAC address of a device that didn’t have one? Early Pi machines didn’t, did they? Nor Beagleboards and Pandaboards, as I remember.

I agree that the reports are looking good (except the MAC).
If possible you should give logging by !Syslog a try (Set level for EtherUSB and plugin the Realtek afterwards). Perhaps it is also possible to obtain the descriptors of the device.
The serial number has nothing to do with the MAC. I checked it with my VIVANCO adapter. Its serial number is 0000000001 but it has a proper MAC address.
There was a variable to set the MAC address. It was implemented by Rainer Schubert and myself in 2011 but only for the SMSC chip build into the BBxM. The concept will only for one missing MAC.
As ROOL took the device into EtherUSB I assume that someone reported it to work for him. Perhaps the chip may be connected to an I²C ROM which contains in general a MAC. However some products may not contain something like this e.g. to save money. If this should be the case the concept of the BBxM SMSC driver could be extended to USB CDC ECM of course. But as mentioned above the concept is limited. And overall I must say that I can’t remember that CDC ECM specification is providing a MAC configuration command but I would have to examine. This would mean that every paket on the related network would be passed to the computer. I don’t think so.
Has the device containing the chip perhaps any switches or buttons to select a special behaviour?