LJB Posted April 25, 2019 Posted April 25, 2019 Armbianmonitor: http://ix.io/1H9v Hi, I just got a NanoPi Neo Plus2 for which I must prepare an image for our software. Basically I must have a working Ethernet-over-USB, Wifi and NIC. I started with your current Debian image compiled with uses Armbian v8.75 and Linux kernel v4.19.20. In this image Wifi works, but loading the g_ether module does not seem to work. I assume I need to patch the kernel to make this happen, which means I need to build my own images. So I created a Virtualbox environment as described in the documentation, and compiled an image with all the defaults (apart from a date-tag I add to the kernel name for identification). I installed the image to the eMMC and booted. In this image there is no wifi interface and g_ether also does not work. The armbianmonitor output for my new image is attached to this query (http://ix.io/1H9v), but I also created an armbianmonitor output for your Debian image (http://ix.io/1H9u) in which the Wifi interface does appear. Wifi: Comparing the dmesg output, it is obvious that firmware fails to load into the AP6212 wifi unit on my image. I've downloaded and installed firmware from several sites (official and other) to try and fix, but nothing solves the problem. Any ideas? Is this a regression in the Linux kernel (4.19.20 -> 4.19.36)? OTG: Configuration of the USB-OTG for the Ethernet-over-USB gadget is usually done in the DTS files on ARM processors (I think - a bit rusted). I've noticed that there are several patches applied to the kernel from other boards such as the OrangePi, so maybe someone already has a patch for another H5 based board which will work for the NanoPiNeo+2? Regards, LJB
Tido Posted April 25, 2019 Posted April 25, 2019 Hi LJB, If you need 2 x NIC and WiFi, why didn't go for: http://www.orangepi.org/OrangePiR1/ ?
LJB Posted April 25, 2019 Author Posted April 25, 2019 47 minutes ago, Tido said: Hi LJB, If you need 2 x NIC and WiFi, why didn't go for: http://www.orangepi.org/OrangePiR1/ ? The client has a device with only a USB port which must be able to connect to this board
Tido Posted April 25, 2019 Posted April 25, 2019 3 minutes ago, LJB said: with only a USB port 13 Pins Header, with 2x USB, IR pin, AUDIO(MIC, AV)
LJB Posted April 25, 2019 Author Posted April 25, 2019 I was able to fix the Ethernet-over-USB problem with the help of another (recent) post. I found the DTS file (after adding some oil on the rusty parts) for the my board @ "./cache/sources/linux-mainline/orange-pi-4.19/arch/arm64/boot/dts/allwinner/sun50i-h5-nanopi-neo-plus2.dts". After changing the USB setting to "otg", I created a patch file and added it to the build/patch/kernel/sunxi-{dev,next} directories. This great build system automatically applied it and I now see a usb0 interface after loading the g_ether module. I've attached my patch file which should work for any v4.19 kernel. So now only the Wifi issue remain board-nanopineoplus2-enable-otg.patch
LJB Posted April 25, 2019 Author Posted April 25, 2019 21 minutes ago, Tido said: 13 Pins Header, with 2x USB, IR pin, AUDIO(MIC, AV) The client require TCP access to this board via USB - I think they want to get debug info and remote syslog via this port. Sorry I am not the SE on this system so that is beyond my scope of worries.
LJB Posted April 25, 2019 Author Posted April 25, 2019 2 hours ago, LJB said: I was able to fix the Ethernet-over-USB problem with the help of another (recent) post. I found the DTS file (after adding some oil on the rusty parts) for the my board @ "./cache/sources/linux-mainline/orange-pi-4.19/arch/arm64/boot/dts/allwinner/sun50i-h5-nanopi-neo-plus2.dts". After changing the USB setting to "otg", I created a patch file and added it to the build/patch/kernel/sunxi-{dev,next} directories. This great build system automatically applied it and I now see a usb0 interface after loading the g_ether module. I've attached my patch file which should work for any v4.19 kernel. So now only the Wifi issue remain board-nanopineoplus2-enable-otg.patch The USB OTG is not working - the microUSB connector data lines are not connected to the SoC (based on schematic rev1704) so Ethernet-over-USB will never work !!!
Igor Posted April 25, 2019 Posted April 25, 2019 1 hour ago, LJB said: The USB OTG is not working https://github.com/friendlyarm/linux/blob/sunxi-4.14.y/arch/arm/boot/dts/sun8i-h3-nanopi.dtsi#L518-L523 It's also software problem.
LJB Posted April 25, 2019 Author Posted April 25, 2019 I've escalated my problem - I guess the SE will select another board. For curiosity sake, does anybody have any idea(s) about why the WIFI firmware is not working in linux kernel 4.19.36 (but in 4.19.20)?
Tido Posted April 25, 2019 Posted April 25, 2019 1 hour ago, LJB said: For curiosity sake just use the searchbox, right hand, top corner and look for: AP6212
Igor Posted April 25, 2019 Posted April 25, 2019 33 minutes ago, LJB said: For curiosity sake, does anybody have any idea(s) about why the WIFI firmware is not working in linux kernel 4.19.36 (but in 4.19.20)? Because its a mess with those wireless chips. There are different versions of them and each needs different firmware settings ... 33 minutes ago, LJB said: I've escalated my problem - I guess the SE will select another board. Which one you have in mind?
LJB Posted April 25, 2019 Author Posted April 25, 2019 2 hours ago, Tido said: just use the searchbox, right hand, top corner and look for: AP6212 I followed all the help I could find in the forum as well as on the net. None however was specific to the Nanopi Neo Plus 2, so I thought I'd ask.
LJB Posted April 25, 2019 Author Posted April 25, 2019 6 minutes ago, Igor said: Because its a mess with those wireless chips. There are different versions of them and each needs different firmware settings ... Which one you have in mind? Haha - SE's worries. I get paid if she waists my time or not.
5kft Posted April 27, 2019 Posted April 27, 2019 The problem with AP6212 appears to be a regression caused by this commit: https://github.com/armbian/build/commit/52d82c921d02366f5fe7fa5e6a754227ce9e6913. I added a comment to the commit regarding this... 2
Recommended Posts