Device RNDIS: Fix linux enumeration

[ayedm1]

No description provided.

[ayedm1]

#250

[rahmanih]

hello @ayedm1,
could you please update the merge branch to dev?

[yames-smile]

Hello, as stated in #171, should the device really respond with its own MAC address (ux_slave_class_rndis_local_node_id) to a UX_DEVICE_CLASS_RNDIS_OID_802_3_PERMANENT_ADDRESS request ?

On this documentation I cannot see any difference between the permanent address and the current address OIDs, except regarding their lifetime. So I don't think there should have one OID returning ux_slave_class_rndis_remote_node_id and the other returning ux_slave_class_rndis_local_node_id like here.

Also, when I tested, my PC took the returned MAC address as its own address, not the device address.

[fdesbiens]

@rahmanih Can you look at @yames-smile's latest comment, please?

[rahmanih]

Tested on STM32 Device.

• before the fix :

[  +0.000005] usb 1-9: Product: STM32 RNDIS Device
[  +0.000004] usb 1-9: Manufacturer: STMicroelectronics
[  +0.000004] usb 1-9: SerialNumber: RNDIS001
[  +0.004446] rndis_host 1-9:1.0: RNDIS_MSG_QUERY(0x01010101) invalid response - off 16 len 4
[  +0.000015] rndis_host 1-9:1.0: rndis get ethaddr, -33
[  +0.000960] rndis_host: probe of 1-9:1.0 failed with error -33

• after the fix :

[  +0.000005] usb 1-9: Product: STM32 RNDIS Device
[  +0.000004] usb 1-9: Manufacturer: STMicroelectronics
[  +0.000003] usb 1-9: SerialNumber: RNDIS001
[  +0.006858] rndis_host 1-9:1.0 eth1: register 'rndis_host' at usb-0000:00:14.0-9, RNDIS device, 00:02:02:03:00:00
[  +0.017132] rndis_host 1-9:1.0 enx000202030000: renamed from eth1

LGTM 🥇

[rahmanih]

@fdesbiens

this can be merged!

Not Yet , ,need further tests

[yames-smile]

Hello @rahmanih, in your test did you configure 00:02:02:03:00:00 as the host address ? As explained in my previous comment, I had to return ux_slave_class_rndis_remote_node_id instead of ux_slave_class_rndis_local_node_id in order to make my PC take the correct MAC address.

[rahmanih]

Hello @rahmanih, in your test did you configure 00:02:02:03:00:00 as the host address ? As explained in my previous comment, I had to return ux_slave_class_rndis_remote_node_id instead of ux_slave_class_rndis_local_node_id in order to make my PC take the correct MAC address.

missed that! I'll do it right now.

need further tests based latest comments.

[rahmanih]

Hello @rahmanih, in your test did you configure 00:02:02:03:00:00 as the host address ? As explained in my previous comment, I had to return ux_slave_class_rndis_remote_node_id instead of ux_slave_class_rndis_local_node_id in order to make my PC take the correct MAC address.

missed that! I'll do it right now.

I did the test on a Linux Fedora distribution and found that both implementation is working fine, Linux Fedore is automatically self-assigning the MC address.
In the test I used both ux_slave_class_rndis_remote_node_id ux_slave_class_rndis_local_node_id is populated with the same value at init stage.
https://github.com/STMicroelectronics/x-cube-azrtos-h7/blob/c191feed5b39c4ac74763da01b625a065d3d48cf/Projects/STM32H735G-DK/Applications/USBX/Ux_Device_RNDIS/USBX/App/app_usbx_device.c#L177

https://github.com/STMicroelectronics/x-cube-azrtos-h7/blob/c191feed5b39c4ac74763da01b625a065d3d48cf/Projects/STM32H735G-DK/Applications/USBX/Ux_Device_RNDIS/USBX/App/app_usbx_device.c#L188

@ayedm1 I think it is better to use the same ux_slave_class_rndis_remote_node_id for both "CURRENT" and "PERMANENT" OID request

[rahmanih]

let's have this request exactly like the UX_DEVICE_CLASS_RNDIS_OID_802_3_CURRENT_ADDRESS

[ayedm1]

Hi @rahmanih, Fixed

[rahmanih]

LGTM 💯

[rahmanih]

@fdesbiens ready for merge.