Any way to combine Omega WiFi with LEDE builds?



  • The standard WiFi driver for the MT7688 currently included in LEDE apparently neither works reliably, nor with acceptable throughput at this time (That driver is true open source work in progress, contributed by very dedicated LEDE developers in a very hostile field, so I'm not criticizing it at all - it's just not ready yet!)

    Onion, for this reason, is using another (proprietary?) driver for the official Omega2 firmware, which works fine.

    But using the Omega2 with a LEDE build currently means essentially having no WiFi.

    That's how I happily used the Omega2 for in my projects of the last few months (always using wired Ethernet).

    But now I'd like to mix the work of these months (e.g. the ledchain kernel module) with WiFi.

    Easiest would be if Onion would fully publish their LEDE setup, but unfortunately all forum threads regarding this (for example here, here, here and here) and even contributing just stalled.

    Now I wonder if it would be possible to somehow just extract the working wifi driver and use it with otherwise standard LEDE? Any ideas?



  • I'd be very interested in this as well. Anyone with this knowledge?



  • I am interested too for building my own images of LEDE with wireless support.



  • The driver exists. I got it already working. The driver can be found here: https://github.com/WRTnode/openwrt/tree/master/package/wrtnode-wifi/ralink-wifi-mt76x8.

    An alternative approach is to download a custom LEDE source from WereCatf. It contains a lot of open-source packages from Onion including the WiFi driver, ready to be compiled as an image to use on the Omega2(+). Source: https://github.com/WereCatf/source.
    I think this is the one you are looking for.

    So far I got no problems with the WiFi driver in my custom LEDE image.



  • The open source mt76 driver in LEDE master has received some bugfixes today that apparently make it work reliably now! At least my current testing indicates no more need to mess with kernel version dependent proprietary binary blob drivers - mt76 for the first time just works stable! #knockonwood ;-)



  • jackdaw and luz,

    what is the concensus? should we use the proprietary driver on the WRTnode repository, or the updated open MT76 driver?

    I just need this to work as a reliable access point with WPA2 encryption, i do NOT need any station mode. (a separate radio is best for that in my opinion)

    please share what kernel version you are running too, thanks!

    thanks,
    Mike



  • Better go for the updated open MT76 driver. Like Luz recommended. At the time the MT76 driver didnt work so I used the mt76x8 one. I used the default kernel version from original LEDE source.



  • @Jackdaw said in Any way to combine Omega WiFi with LEDE builds?:

    Better go for the updated open MT76 driver. Like Luz recommended. At the time the MT76 driver didnt work so I used the mt76x8 one. I used the default kernel version from original LEDE source.

    The latest MT76 driver pulled by the Lede Master branch appears to work great at first, but then the data stops after a few minutes. The client must then disassociate and re-associate with the access point to get data to flow again, so something is going on.

    I will try the Mt76x8 driver now, I assume you mean the one from the WRTnode repository...

    Edit: I can't get the mt76x8 driver fromt he WRTnode repository to show up in my make menuconfig (the ralink category shows up now, but it's empty).



  • @mike-tedeschi unfortunately, after a seemingly perfect start with the patches from 20 days ago, the mt76 again behaves erratically for me as well.

    Some days it works fine without a hitch for a long time, and then other days (like yesterday) performance is extremely poor and there are a lot of lengthy connection dropouts - without any change in the setup or wifi environment (as far as I can tell).

    Still, I'm hoping they will eventually find the problem in the mt76 (psyborg and Daniel Golle apparently are working on it, e.g. see their slightly heated discussion over at the LEDE forum ;-)

    Because I hate the hassles with scraping binaries from some other projects and then being tied to a specific kernel version, which prevents going with other LEDE improvements etc.

    It would be different if for example Onion itself (hello @Lazar-Demin) would provide a daily bot-built binary blob of their driver matching the latest LEDE release, not only their specific Omega firmware…



  • Hi @Jackdaw . How did you make the mt76x8 driver working? I compiled a firmware image using WereCatf's sources and enabled Onion -> Ralink -> ralink-wifi-mt76x8 driver, but I always get a kernel panic during boot when that's enabled. The mt76 driver works as an AP, but I can't get it to connect as client and the speed is awful.

    Thanks in advance.



  • I also got a kernel panic using WereCatf's source. Not sure why it does that. I used the original source from LEDE (https://github.com/lede-project/source) , imported the mt76x8 driver and compiled the whole thing.



  • @luz said

    unfortunately, after a seemingly perfect start with the patches from 20 days ago, the mt76 again behaves erratically for me as well.

    Some days it works fine without a hitch for a long time, and then other days (like yesterday) performance is extremely poor and there are a lot of lengthy connection dropouts - without any change in the setup or wifi environment (as far as I can tell).

    This is a guess based on a hunch ... looking at a discussion at the Linux wireless patch, referenced by the LEDE developers you sited - I see some code that changes a value based on the wifi channel. Perhaps some of these channel dependent values are bad code ...? Or, I have often seen Omegas jump to another wifi channel in the boot process when wifi is initialized. Here I used the free Android app Wifi Analyzer and sometimes see the Omega2 change it's AP wifi channel during the boot. Have you observed if firmware with the Mt76 patch might work fine on some wifi channel and not well on another? Perhaps there is a bug here, could that channel related value be wrong if and when the Omega2 switches the wifi channel?!

    If you see some merit in this hypothesis, during the Omega boot you can observe the wifi channel behavior and assignment using an app like Wifi Analyzer, running on your phone or other tablet.



  • I managed to reproduce the firmware almost the same as the omega, same packages. Only in newer versions and with Wifi (package) without kernel panic occur. But I can not start the wifi.

    I'm using the Omega drive but I can not start it.

    When running the wifi command:

    ifconfig: SIOCGIFFLAGS: No such device
    ra0       no private ioctls.
    
    ra0       no private ioctls.
    
    ra0       no private ioctls.
    
    ra0       no private ioctls.
    
    ra0       no private ioctls.
    
    ra0       no private ioctls.
    
    ra0       no private ioctls.
    
    ra0       no private ioctls.
    
    ra0       no private ioctls.
    
    ifconfig: SIOCGIFFLAGS: No such device
    ifconfig: SIOCGIFFLAGS: No such device
    ra0       no private ioctls.
    
    ra0       no private ioctls.
    
    setting apcli
    apcli0    no private ioctls.
    
    apcli0    no private ioctls.
    
    apcli0    no private ioctls.
    
    apcli0    no private ioctls.
    
    apcli0    no private ioctls.
    
    apcli0    no private ioctls.
    
    ra0       no private ioctls.
    
    ifconfig: SIOCGIFFLAGS: No such device
    brctl: iface ra0: No such device
    ra0       no private ioctls.
    
    ra0(ralink): enable failed
    


Looks like your connection to Onion Community was lost, please wait while we try to reconnect.