[Banana BPI-R4] all related to MTK-SDK

[woziwrt]

EDIT:
Instead of this release, I recommend downloading and testing the BPI-R4_MLO_beta2_upd-1.1 release, which is in post (76) of this thread and which has some additional improvements.

---

Here you can download the release of mainline openwrt-24.10 (commit 4cc1da1) with MLO alpha version with

• some patches updates and improvement
• fix WPS WPA3 connection

BPI-R4_filogic_09012025

Here you can download the release of build with modemmanager support of MBIM USB3 modems and some additional packages
BPI-R4_filogic_09012025_mm

NOTES:

all builds are build out with @rmandrad patch so jumbo frames are supported

it is still the first alpha version under development that supports

• Single Wiphy Architecture
• 320 MHz bandwidth
• 4096-QAM MCS12, MCS13, MCS14, MCS15
• WPA3 key management (AKM24)
• MLO Basic Functionality (Advertisement/Discovery/Setup)

and doesn't support

• Hardware Peak Performance
• MLO Post-Setup Features

wed_enable=Yand sr_scene_detect=Y enabled by default

WPA3 SAE password for SSID MT76_AP_MLD is 12345678

builds are not fully compatible with LuCI network, wireless, dashboard...

If you would like to build out your own build you can download and unpack the
builder_09012025
folder and run ./build-mlo-alpha-version.sh script. Due incompatibility of u-boot .config version with kernel is required to press enter several times to confirm RX buffer size (SERIAL_RX_BUFFER_SIZE) [256].

After successful completion of the autobuild script you can build your own build as follows

cd openwrt
make menuconfig

Change Target Profile from Multiple devices to Bananapi BPi-R4 or Bananapi BPi-R4 2.5GE and modify whatever you want and then
make -j1 V=sc

[glassdoor]

cool stuff.

just to add on, mtk-sdk git has also rolled out quite a few patches on kernel 6.6 over the last few months which can be applied cleanly to current 24.10 and mainline openwrt
-HQoS
-trng
-eip197
-numerous wifi related stuff

start your search here:
https://git01.mediatek.com/plugins/gitiles/openwrt/feeds/mtk-openwrt-feeds/

[Aivaras-s]

just compiled and tried - same issues as on snapshot without mtk-sdk , some devices have slow wifi speed on 2.4ghz and 5ghz (20Mbps)

Similar thread on banana pi forum https://forum.banana-pi.org/t/whats-the-best-firmware-for-bpi-r4-now-extreme-slow-wifi-7-speed-on-openwrt-snapshot/19528/8

[glassdoor]

no idea on wireless. i adopt mtk patches purely from wired standpoint. Maybe others can chip in.

[zerojj]

Surprisingly stable build for an alpha version. Tested with a windows 11 client with qcncm865. Combinations of either 5Ghz+6Ghz or 2.4Ghz+6Ghz work. I have not seen all 3 frequencies in MLO at the same time.

There is a very serious problem - signal power. It is impossible to get more than 7-12dbm in any region or at any width. As I understand it, the normal value should be 20-23 dbi for 5Ghz and 6Ghz radio.

Perhaps I have a new revision of the board or hardware problems? On all OpenWrt firmware except the official one from SinoVopi with a proprietary driver, the power does not rise more than 7dbm for the 5Ghz channel.

And sorry for strange question, but jumbo packets is not the same as jumbo frames?

[shaggyrogers]

I noticed this as well, so presumably this rules out hardware as a potential issue - at least on 24.10-rc4, transmission power (as reported in luci) doesn't seem to exceed 7dBm. Range is also noticeably better for me using the MP4.0 SinoVoip image.

[bruda]

Good afternoon, thanks @woziwrt for the great work you are doing. The 3 antennas in be mode work almost perfectly. Now I will put images of everything, test carried out on your image from 9-1-2025 on SD..

My wifi 7 card used for testing is WiFi 7 QCNCM865 qualcoom

First I want to help with the antenna powers. Now I put a patch and the way to put it in the image you have sent us. Whoever makes their images can use it too.

The patch puts all the antennas at maximum power in all countries of the world. The patch is not mine but I want to share it. It is in the 4pda forum.

The patch must be applied when the image starts and we can connect to it with winscp, before connecting to any wireless network.

db.zip

Unzip the zip and put the downloaded files in /lib/firmwares, then by telnet put the following commands one by one.

iw reg reload
wifi down
wifi up
reboot

we will reboot and we will be able to work normally

once we do this we will have all the antennas at maximum power in all countries

now I put 2 speeds of 6g and 5g from the image of @woziwrt

I put the wireless connections to both antennas

now I put if I connect to wifi 6 first and then to mlo, the mlo speeds are not bad but not what we all want, rather they are similar to wifi 5 speeds

but if on the contrary I connect to wifi 5 and then to mlo, the speeds are ridiculous I put images of connection to 5g mlo, and the speeds.

but ultimately it is doing a great job excellent and my congratulations, we already have the 3 antennas in be mode working as they practically should work.

as mlo says it is still in alpha, but I thank you for your great work.

Thank you very much for your time

[woziwrt]

Here is another release of MLO alpha version:

• kernel update to 6.6.71
• mt76 update to Git HEAD 2025-01-04
• some patches refactored and improvement
• frames

mac80211 is still 2024-10-25 but the build out with 6.12.6 is on the way...

BPI-R4_filogic_14012025.zip
builder_14012025.zip
and release with modem manager mbim usb3 without
BPI-R4_filogic_14012025_mm.zip

[bruda]

Good evening and thank you very much for your work and dedication. Tomorrow we will do some tests.
Regards

[bruda]

Good morning @woziwrt, I'll give you my first impressions, as always using your SD version.

This time I haven't applied any patch.

First I'll put 5G, antenna power and speeds, everything seems to work correctly, the antenna power is below the patch I put yesterday but only 4 decibels.

However, when we connect to 5G, it connects to the 2 5G and 6G antennas, but the speeds are very low. I'll put images for you.

In 6g it has been impossible for me to connect to give me internet even though it connected to the antenna, the power is 12 dbm, well below the patch I put yesterday, but first connecting to 6g even though it didn't give me internet, and then connecting mlo did give me internet, but it only connected to the 5g antenna, and gave 5g speeds, and the power of the 3 antennas were 5g at 160 MHz.

Now I have to go, I will continue later doing tests applying the power patch.

I will post the results since with the patch you can always connect to any antenna and frequency in any country in the world, always of course if that country has wifi 7 you can connect if it doesn't you can't, but you can connect to all the other bands and all the channels.

Is there any way to put it in your compilations directly?

Thanks for your great work and good luck.

Then I will post the results applying the patch

[bruda]

Good news, once the power patch has been applied, I'll give you the previews.

I can connect to the 3 antennas without any problem, more than decent speeds in all of them.

mlo in 6g connects at 160 mhz, on the other hand mlo in 5g connects at 320 mhz, and in both we have decent speeds but not spectacular,

I'll give you images of everything.

I've sent you the antenna power repositories privately, in case it can help you.

we started with 6g

we continue now with 5g

If I connect to the 5g antenna first and then to mlo, it changes to 320 mhz, but yesterday it gave very low speeds, and now it gives more than decent speeds.

If I connect to 6g, which I can do with the patch and have internet, then I switch to mlo, it changes to 160mhz, and the speeds are decent.

Thanks for your great work

[woziwrt]

Thank you for your testing. Nice work

[bruda]

Good morning, I'm leaving you the test I've done. The tests were carried out with the Qualcomm FastConnect 7800 Wi-Fi 7 High Band Simultaneous (HBS) Network Adapter card . I've also posted the interface page so you can see that it works with Wireguard and the speeds of the same. The speeds in the 3 bands in wifi 7 be mode are optimal, the power of the antennas are at maximum in all countries and you can connect to any frequency that will do it without problem.

MLO works correctly although it still lacks a bit of speed. I'm sure that in future works you'll get it to work at the optimal speed.

MLO always takes the 6g and 5g bands together even if you first select 2g and then go to it, or 5g and then go to it and the same as 6g. For the rest, it works great.

Wireguard works correctly.

I'll put all the images.

6g band in mode be

5g band in mode be

2g band in mode be

interfaces wireguard

speeds of the 3 bands in be mode with wireguard

It is a version that is still in beta, but it can be used all day long and it works just great.

Thank you for your time and dedication and your great contribution to the OpenWRT community.

We look forward to your next work.

Have a nice weekend

[kusmi]

Sorry for my noob question: I see a lot of BPI-R4 packages here - but why should I install the ones here vs. the official snapshots from OpenWrt?

Is this here basically a testbed for the newest & greatest patches which have not yet found the way into official OpenWrt? Or is it a basic setup with sensible default config, to get started quickly?

[bruda]

Good afternoon colleagues, this is a job in which the 3 bands work perfectly, things that in the official repository only the 2g and 5g bands work well in be mode, in ax mode the 3 bands work well in the official ones, also in this one it is clear

it is neither better nor worse than the official one, only that in this version that the colleague @woziwrt works on, you will have wifi 7 in the three bands including 6g.

he is also working hard to make mlo work as it should with the correct speeds, he almost has it ready as you have seen in my review that I have made, the speeds in mlo are more than decent.

I suppose that the colleagues who know more about this could join the colleague @woziwrt and work together.

you can also create your own image the colleague uploaded in the first post how to do it.

It is a snapshot image, that means that the packages are available for a while and then they are no longer there, and many of them cannot be installed since it is also working with the latest kernel and they do not work

When it is a finished image, if the openwrt team wants to put it in their repository to download as a final image, they can do so.

At the moment it can only be downloaded in this thread and not in the official repository https://firmware-selector.openwrt.org/ , there you can also download snapshot images, but the 3 bands do not work as I have told you.

Here the colleague will continue working and uploading images so that we can test them

I hope I have helped you and sorry for the long post I have given you.

Anyway, the best thing you can do is try it and see how it works for you and compare it with the official images that are currently available.

If you like it and it works well for you, you can leave it. For now, I have it as the main router diary image, and since I have another banana, I'm trying new things from the official repository and from the work in this thread.

This image is still in beta, but it's fantastic work in my opinion.

Greetings

[bruda]

This was not for here

[MattJCollins]

Just wanted to say, really, THANK you to you all. I've spent a few days poking around and searching for guidance on getting the BPI-R4 working for 6g, and this fixed all my problems at once. Really appreciate it. I was on Snapshot but for whatever reason even with reg set, etc, I wasnt getting any valid channels for 6g

Again - THANK YOU. Let me know if I can assist with testing in any way. Network is a BPI R4, SFP+ DAC10gb and some scattered APs using Orbi RBR/RBS50;s (cheap - 30$ on ebay for AC aps in low usage locations)

My remaining challenge - if anyone has any thoughts - is that iw list shows 2.4ghz limited to 6dbm and 5ghz limited to 7dbm, regardless of reg domain. This is true even of a freshly burnt snapshot or 24-RC5, or the beta image here. The 6ghz channel happily runs at 23dbm.

[bruda]

Good evening @MattJCollins, if in 2g and 5g, with this version you have the problem of the antenna power, I think it is one of the boards or they are defective, there are several users with this problem,

I would speak to the store that sold it to you and have it replaced, I think your board is defective.

I'm sorry to tell you but in this thread there are several colleagues who have the same problem as you, I have seen that you have seen the thread, I think your board is defective

Speak to the store that sold it to you and have it replaced or your money returned

Regards

[nvvt]

Thank you for providing this. Could you please share the updated patchset?

I need to recompile as I'm using a Fibocom FM350-GL, which needs a patched dtsi file during compilation.

The pending thermal zone would be why the fan maxes out constantly …?

edit: I'm not sure if it's of any use, but there also is a kernel 6.12 fork explicitly made for BPI-R* devices on github.

[deltakx]

I think there is something off with the way tx_power is calculated, read from the EEPROM, or set in the EEPROM by the manufacturer. It is probably not a hardware defect (maybe if the EEPROM is programmed wrong). Here are some tx_power values I have collected from my own devices and your and MattJCollins device:

• My device eas2: 2.4G: 23 dBm, 5G: 23 dBm, 6G 23 dBm (only tested with OpenWRT snapshot firmware)
• My device eas1: 2.4G: 7 dBm, 5G: 7 dBm, 6G: 23 dBm (OpenWRT snapshot firmware, beta1 FW from this thread)
• bruda with beta1 FW, as per screenshots: 2.4G: 27 dBm, 5G: 27 dBm, 6G: 27 dBm
• MattJCollins with OpenWRT snapshot/beta1 FW from this thread: 2.4G: 6 dBm, 5G: 7 dBm, 6G: 23 dBm
• MattJCollins with June 2024 Sinovoip FW: 2.4G: 30 dBm, 5G: 30 dBm, 6G: 12 dBm

Shouldn't these values be all the same for the same hardware? The mt76 driver issue tracker on github has lots of reports regarding TX power on Mediatek devices. I think this is a driver problem.