Re: RTL8156(A|B) chip requires r8156 to be force loaded to operate

On Mon, Dec 27, 2021 at 06:21:24PM -0800, Jakub Kicinski wrote:

On Fri, 24 Dec 2021 21:30:18 +0100 Ryan Lahfa wrote:

quoted

Hi all,

I recently bought an USB-C 2.5Gbps external network card, which shows in
`lsusb` as:

quoted

Bus 002 Device 003: ID 0bda:8156 Realtek Semiconductor Corp. USB 10/100/1G/2.5G LAN  

By default, on my distribution (NixOS "21.11pre319254.b5182c214fa")'s
latest kernel (`pkgs.linuxPackages_latest`) which shows in `uname -nar`
as:

quoted

Linux $machine 5.15.10 #1-NixOS SMP Fri Dec 17 09:30:17 UTC 2021 x86_64 GNU/Linux  

The network card is loaded with `cdc_ncm` driver and is unable to detect
any carrier even when one is actually plugged in, I tried multiple
things, I confirmed independently that the carrier is working.

Through further investigations and with the help of a user on
Libera.Chat #networking channel, we blacklisted `cdc_ncm`, but nothing
get loaded in turn.

Then, I forced the usage of r8152 for the device 0bda:8156 using `echo
0bda 8156 > /sys/bus/usb/drivers/r8152/new_id`, and... miracle.
Everything just worked.

I am uncertain whether this falls in kernel's responsibility or not, it
seems indeed that my device is listed for r8152: https://github.com/torvalds/linux/blob/master/drivers/net/usb/r8152.c#L9790 introduced by this commit https://github.com/torvalds/linux/commit/195aae321c829dd1945900d75561e6aa79cce208 if I understand well, which is tagged for 5.15.

I am curious to see how difficult would that be to write a patch for
this and fix it, meanwhile, here is my modest contribution with this bug
report, hopefully, this is the right place for them.

Can you please share the output of lsusb -d '0bda:8156' -vv ?

Here it is, attached.

Kind regards,
-- 
Ryan Lahfa