Full agree with you, it’s a development board was never sold as a end consumer product
I don’t understand something about this modern world, why would MTK release a closed version of the driver if its final version will be open?
1 Like
For MediaTek, “proprietary” drivers are their main drivers and the ones they provide to companies like Asus, Linksys, NETGEAR, etc.
“Open source” drivers are the ones they are FORCED to share with the Linux community because they are using their kernel or maybe they share them voluntarily. 
The bottom line with drivers is that companies want the ”proprietary” drivers so most of the work goes into improving that garbage.
“Open source” drivers are the redheaded stepchild in this scenario.
That’s not to say they won’t work good but ”proprietary” drivers is going to get most of the love.
I’m glad to see I’m not the only one that feels that this project has been utterly mishandled.
I’ve said the following over and over again, but it seems to fall on mostly deaf ears (with a few selective replies (sorry, Frank, but your answers do not satisfy).
- If the product is not available yet, don’t advertise it for sale
- If the product features are not yet available (and if the software does not exist, you cannot claim the product will work, because you obviously don’t know yet) then BE HONEST and say the features are not available yet, both on the wiki and in the places you sell it.
- Don’t use misleading photographs in your advert that show hardware that will not be shipped to the customer
- If this is a development board, PROVIDE THE SDK or be honest, and say that the SDK does not exist. I cannot obtain an SDK for an open driver (regardless of completeness or functionality)
- Provide instructions for a replicatable build for all supported OSes (Debian and OpenWRT are supported at a minimum, so that should not be hard.
- Communicate well and do not tell upset customers that they are idiots, or “can’t read”.
My background - I have been an open source developer for over 30 years. I worked to separate the ARM Linux port from the arm26 port in the late 90’s - the foundation codebase that all you newbie devs are building on top of - You’re Welcome, by the way - without the work I did there, you’d still be picking the #IFDEF ARCH_ARM26 mov pc,lr … function return assembly out of all your arm assembly code, because nobody else was willing to undertake such an enormous task (this was before the days of git, btw - so imagine patching an entire CPU architectures low-level assembly code using just diff and patch for a moment, if you will. For a time, I was the Linux/arm26 architecture maintainer, but life pressures eventually meant I coudn’t continue to do that in the end.
I ported ARM Linux to the Toshiba e570, e580, e400 and e800, along with a couple of other models that are now so old their model numbers escape my memory. I wrote the w100 video driver /from scratch/ (later alongside Richard) by direct reverse engineering - without reference to any software, closed OR open, and with no documentation available at all - the entire work was done by peeking and poking the hadrware after chainbooting from WinCE. I think you’ll still find my contributions from more recently in one of Broadcom’s WiFi drivers (I forget which, but it supports both SD and PCIe. It’s in the mainline kernel if anyone cares), so I know how irritating WiFi device manufacturers are to deal with.
I wrote the tmio-mmc driver by reverse-engineering those devices WindowsCE system images, and studying the asic3 driver. This driver is still current today, and in use in millions of devices globally, including many laptops.
Following this most of my work has been in industry so I am not credited (I dont own the copyright to that work) and, sadly, much of it was never released because the parent companies involved abandoned the projects, and my employers promises that I would be allowed to release the sources were not true. I’ve worked on video decoders, comms hardware, and written an entire NTP server (that actually works so well I can tell if the lid of the device is closed or merely resting on top of it, purely from the tiny change in heat dissipation affecting the clock drift), from scratch. I found a bug that (TTBOMK) is still unfixed in ntpd whereby the non-kernel time conditioning code has a 1/x that should be x/1 in the PLL/FLL loop calculations. Go have a look if you like. I tried to submit a fix, but the ntpd project is dead due to the gatekeeping nature of the devs (check the email lists - they weren’t even interested in my patches unless I bought their £100 book). I could go on.
As you can see - I know my shit when it comes to low level coding.
I would love to join in the development of the BPI-r4 and the BE14000 WiFi drivers, but I’m not going to reverse engineer a badly broken closed source driver (btw. - reverse engineering is actually illegal where I live) in order to do so.
I bought the BPI-r4 and be14000 because we were promised an open source platform.
I did not expect everything to work perfectly out of the box.
I did not expect it to (necessarily) even boot out of the box.
What I expected was:
- Hardware which has passed at least some testing (the BPI-r4 baseboard did actually manage this, to give credit.)
- OPEN software images that boot to at least the most recent development environment, if 1) has been done (again, the baseboard did actually deliver in this regard, as both the Debian and OpenWRT mainline images boot from SD and eMMC)
- A fully documented boot process - block diagram and hardware specifications publicly available. These I have been unable to find because the project Wiki is a mess.
- Full documentation for the entire platform and all components therein.
- An up to date project wiki with valid links to current images AND sourcecode (this has not been the case, especially with the BE14000)
- A welcoming development community where the developers are happy to assist newcomers with issues with the above (this very much is not the case)
- Honesty - both in the projects goals, and in any items sold by the project. Clear indications of the project status, with dates.
What I found was:
- An unwelcoming and very confusing forum, with a patronising “guide” that couldn’t easily be skipped
- A badly organised, out of date Wiki, with terrible documentation, links and information about things that functionally don’t exist (BE19000) that aren’t labelled as historic, and no datestamps
- A bad attitude towards any criticism of the project.
Look - you guys, from my perspective, YOU are the newbies at this. I’ve been doing it for 3 decades. Instead of telling me I’m an idiot, try LEARNING. Listen to criticism. Improve the documentation. Stop lying in your adverts and promising things you have no idea will even work at this stage and have not secured guarantees of being possible to release in an open manner. Help newcomers to the project replicate your work so that they can join in.
If people cant join in, because the barrier to entry is too great (and don’t pretend that anyone but yourselves has access to the WiFi firmware source, because I’m not even convinced YOU have it yet), then what you have is merely a personal vanity project that you are scamming people for money to fund, with no intention of having an open development process at all - you all just want to look like gods with The Knowledge, that the rest of usCannot Have Access To.
Knock it off - it’s not big, and it’s not clever. It just makes you all look like dicks.
Open development means that everyone can join in on an equal footing. If that isn’t possible, then you don’t have an open development platform at all.
1 Like
They welcomed you and responded lovingly like this:
1 Like
Good morning @espiro, he says he won’t write again but with a few words like the ones you mentioned, I totally agree with everything.
I’m still waiting for Sinovoip BPI Team Leader to appear once and for all and explain what he has sold us, unless he only wants to make money.
I still have all the messages exchanged with him on Aliexpress
Let’s see if he deigns and with his wisdom enlightens all of us fools who have not been able to understand his project
By the way, I would like to send you some very interesting things about people who are already there in private:
Kernel Version 6.6.51 at least they already have wifi 7, but the speeds are not correct. But at the user level it is an openwrt that you can work correctly since the banana pir r4 card is good.
Through the forum, sending you private messages is filtered and if you send a simple email or pages that are not to their liking, they delete them directly.
No one that I have tried to help with pages where all the information is and so on has received the message, that is why I put it here.
At least you know that there are people working hard, but of course they are not part of the banana team, and they do not want the openwrt community to make them look ridiculous, and to release the project that they call before them.
I leave you an image with wifi 7 and more radios that we did not only have 3, as you will see there is a 4th radio than the wifi 7, but the speed is still not correct
I forgot this is a wifi 7 speed test, as you can see there is still a long way to go.
As you can see, there are people working hard in the OpenWRT community, but of course all this is of no interest to this forum.
Speed test on BANANA 6G
Greetings to you
I think plenty of people tried to tell you that those aren’t finished products many times, most of the people around here really enjoy the them for what they are: good open hardware discussed in the open and often refined according to the community feedback.
The BPI-R4 and the wifi7 board got produced that way. There is some gap in the software support now because the process of upstreaming a fairly complex driver takes time and OpenWRT is in the middle of a quite big overhaul itself.
Right now additional tinkering is required.
The new documentation site seems to contain plenty of information to get started, I wonder if the legacy wiki is out of sync and that lead to some of the grievances.
Addressing people in such confrontational way does not win goodwill anyway.
Incidentally I think this thread happened exactly during the Autumn Festival so I wouldn’t expect official replies during this time or the weekends.
I saw the page you linked several months ago. i made several criticisms of the page at that time, and it looks like nothing at all has changed.
It blows my mind that people making “open hardware” just Don’t seem to get that they need to provide open software too.
A Deb12 image (for example) is nice to have - it allows a quick test etc.
But in and of itself it’s not really any better than a closed source OS - without instructions on replicating the build, it simply cannot be described as open.
How can anyone trust a build that is just dumped on a website, with zero description of what has been done to create it?
Why is it so hard to understand?
PS - with two websites, neither of which have dates etc. on anything, how are you supposed to know what is more up to date?
Just get the docs sorted out. And get rid of the out of date site.
which image do you mean? for my images you can look in the ci-pipelines and repos how it is done…
a bit of introducing information: en:bpi-r4:start [FW-WEB Wiki]
That’s unfair.
They cooperate with larger opensource communities, they provide the sources they use for their builds and usually within a release cycle upstream OpenWRT has support for the board.
You may be annoyed that the code drops aren’t a nice tidy tree with neatly split patches, but you seem quite demanding for what you are paying.
To be honest they shall not be responsible for all of the criticism. MTK did not provide a day 0 open source driver support when they announced the platform. What can hardware manufacturers do is limited. Considering we have had so many blobs since the 90s and most people don’t care even don’t even realise about it. I really doubt if we could have a fully open-sourced and non-obsolete hardware someday.
Well, it seems that we already have a version as it should have been from the beginning and they had announced the banana team.
But it seems that it was not them and we already have the first stable version as far as possible…
You can now choose in 6g, if you want to have BE, or on the contrary choose AX
You also have it in the rest in 5g and in 2g…
both in ax and in be, the speeds are very similar but the truth is that it is incredible.
speed in ax
speed in BE
Thank you to all those who have worked freely on this project that they have brought forward, I am very sorry that it was not the banana team.
Thanks to all those who have worked freely in this project that they have brought forward, I am very sorry that it was not the banana team.
Greetings to everyone, and finally we have what we all expected from the beginning, an openwrt wifi7
It was never meant to be the banana team, it was always the openwrt team same for R3 etc
1 Like
How do you get 6Ghz at 320Mhz working alongside 5Ghz? It fails here unless I set the 6Ghz band to 160Mhz.
I am currently using this build
Same issue here, BTW anyone have those 6g 5g and 2.4g led light up cause I dont
Good morning @Mathieulh, it seems that you have not understood me, the only thing I have said is that all antennas have the option to put them in BE.
You seem to know more than anyone in this world, thanks for your advice, but thanks to the teams that are not from the banana team, we finally have wifi7 on banana pi r4.
If you notice, it seems that you have not read what I have put the images first in AX, and then in BE.
I have said that both 2g and 5g have the same options to put them in ax and in be.
Surely you have not read or understood what I have put.
By the way, Sinovoip BPI Team Leader, there is no need for explanations anymore, the messages from aliexpress will remain between us.
Greetings
Francisco, can you post the image for download? Thanks
1 Like