Banana Pi BPI-R2’s U-boot & Linux 4.4 Source Code & MediaTek MT7623N Datasheet Released

Banana Pi BPI-R2 is a multimedia router board powered by MediaTek MT7623N quad core processor with 2GB RAM, 5 Gigabit Ethernet ports, up to two SATA ports, two USB 3.0 ports, HDMI output, and I/O headers. The board is not for sale yet, but the company has recently released the source code with U-boot and Linux 4.4.70, as well as a datasheet for MediaTek MT7623N processor.

The source code can be found on Github, so let’s see if we can build it:


After a couple of minutes, the build would end with:


For the very last step, it asks you to login as root / sudoer, which it should not do… But we end up with the images, so at least it builds:


MediaTek has also been active by committing patchsets for MT7623 to the Linux Kernel Mailing List, so mainline Linux is an eventual possibility for BPI-R2 board. We just don’t have a clear view of what works and what doesn’t with mainline.

Mediatek MT7623N PCIe Subsystem Block Diagram

The datasheet is a 1,235-page document, but the name “MT7623N Datasheet for Development Board” implies that it may actually be a subset of another larger and more complete datasheet. Nevertheless, it looks to have enough information to control peripherals like GPIOs, I2C, PWM, UART, timers, GMAC, USB, etc… You’ll also find BPI-R2 schematics (PDF only) in the board’s Wiki.

39
Leave a Reply

avatar
39 Comment threads
0 Thread replies
10 Followers
 
Most reacted comment
Hottest comment thread
12 Comment authors
lionwangLinAdminmdeljim stPaul Recent comment authors
  Subscribe  
newest oldest most voted
Notify of
Willy
Guest
Willy

Good. Maybe it will be time that I replace my home firewall (14 years-old Via C3-533, 5*100 Mbps, 128MB RAM, 32 MB flash, kernel 2.4, last rebooted 3 years ago to move to my new house). I just strongly doubt such a board will be as reliable and will last that long :-/ But at least with a mainline kernel I could hope to get issues fixed over time.

tkaiser
Guest
tkaiser

Willy :
But at least with a mainline kernel I could hope to get issues fixed over time.

You should keep in mind that what happened above is the result of an absolutely unnecessary and time wasting procedure: poking those ignorant Sinovoip guys over and over again to get such basic stuff like schematics, sources or even a simple boot log. But they don’t get it, their level of ignorance is simply to high.

Wrt mainline kernel and MTK there are interesting insights available: https://forum.armbian.com/index.php?/topic/4567-board-bring-up-banana-pi-r2/#comment-34726

Mum
Guest
Mum

Hopefully the microUSB port is there for OTG (or USB uart?) and not for power 😉

tkaiser
Guest
tkaiser

Mum :
Hopefully the microUSB port is there for OTG (or USB uart?) and not for power

Please visit the official documentation of this device: bananapi.gitbooks.io/banana-pi-bpi-r2-open-source-smart-router/content/bpi-r2-quick-start.html

Scroll down to the bottom to read ‘Finally, at the very left of the bottom edge is the USB power connector. Plug in a regulated power supply that is rated at 5V ±5% / 2000mA (or 2A). Any number bigger than 700 mA will also work.’

The official documentation tells you can use Micro USB to power the board and 5V/700mA are fine. They know that this is wrong, they have been told several times that this is wrong, they don’t give a shit, they even laugh at you when you try to help them improving. Welcome to the world of SinoVoip.

CampGareth
Guest

Still 32-bit so no good for my storage needs, but if it’s cheaper than the £100 low end Intel processor models you see floating around it could see some success and would be a nice replacement for my current overkill pfsense box… heck at that power consumption it’d be a good replacement for a cheapo 5 port switch, except it’d be managed too.

tkaiser
Guest
tkaiser

CampGareth :
pfsense box…

The pfSense folks are working on a single ARM device worth the efforts: EspressoBin. And that’s for three simple reasons: sources available long before product launched, correct documentation available long before product launched, a manufacturer listening and caring. That also pretty much describes the main difference to SinoVoip here.

willy
Guest
willy

tkaiser :
You should keep in mind that what happened above is the result of an absolutely unnecessary and time wasting procedure

Yes, I’ve read some of your insisting posts on the subject quite a few times. But let’s try to stay positive, ultimately it worked. I know that sinovoip is ignorant about what opensource and community-driven development is, I just hope that they’ll sell more of their open products than the closed ones so that they start to realize how much more profitable they could be by being open and by pressuring their SoC vendors on this (ie: simply not consider the closed ones). Others went through this a decade ago.

tkaiser
Guest
tkaiser

willy :
start to realize how much more profitable they could be by being open and by pressuring their SoC vendors on this (ie: simply not consider the closed ones)

Huh? Exactly the opposite has happened here. MTK opens themselves currently, SinoVoip got sources from MTK 4 months ago, nothing relevant except a few fixes and increasing kernel number has happened since then. SinoVoip did not provide any real information (I really don’t know why they’re not able to understand that correct information/specifications matter) and only told ‘wait and see’, ‘when finish we github’. Not the SoC vendor is the problem but SinoVoip sitting in between and ignoring everything that happens. That actively hinder open source communities trying to help them and also suffer from understanding this.

This @lionwang guy seems even to be proud of what they’re doing. ‘See, we’re successful even if we behave absolutely stupid’. What’s the purpose of ignoring this pull request for example: https://github.com/BPI-SINOVOIP/BPI-M2U-bsp/pull/3

Willy, you maintain 3.10 LTS kernel, you released 3.10.106 last week, Daniel cleaned up SinoVoip’s kernel sources half a year ago, as soon as you release a new version he patches their sources up to latest version. They ignored this all the time. Now even one step further: they get all your fixes through Daniel’s work for free. It’s one single click on Github taking one second: ‘Merge pull request’. What is SinoVoip doing? Nothing. Why? Since they believe they’re doing right.

Next month they will start to sell their BPi M2 Berry with a shitty Micro USB connector, a 2.5 year old 3.10.65 kernel and this all will be a success since users realize too late what’s happening and the vendor himself being confident he’s doing everything right. Ignorance in every aspect.

cnxsoft :
@tkaiser
@willy
They’ve not done too bad with BPI-R2 considering the board has not launched yet.

Huh? I will not recall how much time and efforts I wasted to poke them to get to this point. All of this (releasing most basic stuff) was only reactions to something they consider ‘tkaiser attacking us for no reason again and again’. And with blog posts like this now this turns even into a SinoVoip success story. Weird and sad since they’re not able to learn from this.

Did they release BPi M2 Berry schematics yet (important to get a clue about the most basic question: can Micro USB be avoided to power the board or are they using step-up converters to allow a SATA disk be powered properly?). No they did not. And if not someone from the outside is stupid enough wasting his time to poke them over and over again they won’t release it timely as usual. But why should any external waste his time to get them doing the most basic stuff? At least this was my last try since they really proof that they’re not able to understand why improving is better than ignoring. They’re even too ignorant/stupid to understand why correct schematics matter.

Mum
Guest
Mum

@cnxsoft
Well they released the schematics, right? No one said the schematics had to be useful… 😉

@tkaiser: I was actually being sarcastic about USB for power in a “oh no not again” way. I agree it’s completely stupid. At least they seem to know this, because all the photos show it powered via the barrel connector. Too bad for the person who buys it and tries to power with microUSB and has endless problems…

tkaiser
Guest
tkaiser

Mum :
@cnxsoft
Well they released the schematics, right? No one said the schematics had to be useful…

Yes, exactly that happened and is another proof that they do not understand why someone is asking for schematics, for sources (sources always also are documentation) or for correct(!) information/documentation and why they should provide this stuff. They are not able to understand that this is the most basic requirement to be able to help them. Since they’re too stupid/ignorant to ever fix this copy&paste crap collected in their ‘technical documentation’ wiki I really hope they were smart enough to add some protection circuitry to the Micro USB port to prevent the OTG port being fried by users inserting a power brick there.

Of course users who want to know how to power this board are still lost since the ‘technical documentation’ lists a 4.4/1.6mm barrel plug while it’s 5.5/2.1mm in reality. And also of course SinoVoip does not get why that might be a problem that needs to be fixed immediately. What puzzles me most is their inability to understand why all of this matters (same eg with providing MD5 checksums for downloads: it took us an insane amount of time convincing them to provide such hashes with their downloads but since they refuse to get the reason why — checking for download integrity — now they provide MD5 sums almost all the time… but sometimes wrong ones)

I knew that you were sarcastic about USB for power but couldn’t resist to point out that Micro USB is also an issue here (but only wrong documentation). And while we’re talking about that stuff at the same time SinoVoip is preparing for the next great shit show called BPi M2 Berry to be powered through Micro USB and prepared to power even a 2.5″ SATA HDD by the board.

BIRD TECHSTEP
Guest
BIRD TECHSTEP

~85$ Next week on Aliexpress.

I waiting.

willy
Guest
willy

@tkaiser
I agree with you. I *thought* they started to understand. The simple fact that they removed values from the schematics and reject PRs for security updates proves the opposite and is lamentable.

Such companies solely exist thanks to review sites like cnxsoft. Maybe it would be time to get some pages to vote for good/bad/improving behaviour for each vendor. End users value the price and features first, expecting to meet little trouble. When you see 90% down vote due to trouble, you tend to avoid that vendor.

Mum
Guest
Mum

willy :
Such companies solely exist thanks to review sites like cnxsoft. Maybe it would be time to get some pages to vote for good/bad/improving behaviour for each vendor. End users value the price and features first, expecting to meet little trouble. When you see 90% down vote due to trouble, you tend to avoid that vendor.

Sorry but I don’t agree. They would still exist even without websites blindly re-posting their press releases. The reason is that AliExpress has brought every poorly designed product within the reach of someone with an internet connection and a means to pay.

It doesn’t matter that they have every new product written about on cnxsoft. There will always be someone with a blog writing about these for the page views.

Most of their sales will come from being one of the cheapest available options with X features on AliExpress. Users don’t do research before buying, they just go to the cheapest thing with moderately good reviews (and most Ali users just give 5 stars if it arrives and isn’t obviously DoA).

So what needs to happen is this: blog writers need to take a bit more time and include statements like “I cannot recommend buying this product until the manufacturer fixes their source/schematics” to the FRONT of the blog post (because people will not read to the end) and people reviewing it poorly on AliExpress.

Sure, maybe adding a huge “DO NOT BUY” warning by cnxsoft will piss off Sinovoip and they won’t send samples anymore, but I doubt it, since it’s usually the distributor who sends it to cnxsoft and not the manufacturer directly. The distributor doesn’t care if the product is crap, they make their commission anyway.

tkaiser
Guest
tkaiser

willy :
Such companies solely exist thanks to review sites like cnxsoft.

I don’t agree here. While I think this specific blog post was not necessary (since creating the impression SinoVoip would’ve released anything on their own while in reality again community wasted huge amounts of time and efforts to poke them) the average Banana customer doesn’t read blogs like CNX.

The problem is SinoVoip’s level of ignorance, they simply don’t get what’s going on. They’ve even a good reputation since community did their work turning their paperweights into something useful. The R2 predecessor R1 is the best example. At the time released literally nothing worked. No SATA, no Ethernet switch, not a single Linux OS image (the only image working was an Android!!! For a ‘smart home server router’). Community stepped in fighting against SinoVoip’s stupidity (same problems as still: not releasing schematics, not providing correct documentation) but finally suceeding and assembling working OS images for their paperweight.

Community developers know SinoVoip’s history and why it’s important to get this stuff from them (you get from every responsible board maker without any discussion) but how does customer perception looks like? When I was asking constantly in SinoVoip’s forum for most basic requirements (schematics, correct information/documentation, sources) I was told by other users to shut up since ‘SinoVoip doesn’t owe us anything’ and ‘they did a great job’. The average Banana customer is absolutely clueless and attributes community work to the board maker. And that’s why community is at fault.

If we would have reacted to board maker behaviour (providing no schematics, no correct information/documentation, no sources) correctly we simply would not have tried to fix their mistakes. The R1 would be the most laughable board in history or SinoVoip folks would have stopped to act as brain damaged retards already (changing their relationship to open source community and improving instead of ignoring). But we fixed their mistakes, customers don’t notice, the Banana brand is renowned for no reason and they can literally sell shit if they want. On top of that their level of ignorance prevents them also from even taking notice. That’s the worst part of this all.

Paul
Guest
Paul

willy :
@tkaiser
Maybe it would be time to get some pages to vote for good/bad/improving behaviour for each vendor. End users value the price and features first, expecting to meet little trouble. When you see 90% down vote due to trouble, you tend to avoid that vendor.

I second that, but fear there maybe few boardmakers left, unlazy enough to publish full and accurate specification of their products. e.g. they may just say it has 802.11ac but not give the data rate, so when the purchaser then tells the seller it only does 100 bytes/sec, they can say your money is now in China, what else do you want ?
I see much input here about power consumption; having read cnx for few years now, thats another favourite for makers to limit their spec to just some capability of the PSU, not the typical average consumption under typical usages.

Willy
Guest
Willy

While I think this specific blog post was not necessary (since creating the impression SinoVoip would’ve released anything on their own while in reality again community wasted huge amounts of time and efforts to poke them) the average Banana customer doesn’t read blogs like CNX.

Well, you hardly find information on such products in your local store. They only exist on the net. RPi is almost the only exception.

The average Banana customer is absolutely clueless and attributes community work to the board maker. And that’s why community is at fault.

On this one I totally agree. That’s why I’m never trying to help those who ostensibly don’t play well. Nice players like mqmaker, friendlyarm or solidrun easily get patches or even just suggestion emails. But the orange and yellow fruits made my time so painful they’ll hardly get anything back. And helping the poor guy who spent $25 on a brick is not a good idea because ultimately it’ll work and he’ll recommend the device to his friends with your fixes. It’s sad but true. The only thing is that at the beginning you never know if a vendor will play nicely. Some can start on the wrong way, understand and evolve. SinoVoip certainly has proven enough that they were not worth investing time fixing their designs, and their customers need to remind the pain they get after buying a device on which nothing works.

jim st
Guest
jim st

A paper with the components on it is not a schematic. Don’t give them credit for that.

Also the DXF which might have had more information is missing, no BOM. So not very open source. One could take a sample board and fix the schematic if one were to be trying to duplicate it as in an open design, but the only ones they screw with what they have up now are buyers, not potential competition.

Puzzling why they did this.

I’ve got an R1 which I played with a bit, but won’t invest more in it till something better than openwrt runs on it with real support.

Paul
Guest
Paul

jim st :
I’ve got an R1 which I played with a bit, but won’t invest more in it till something better than openwrt runs on it with real support.

and what might that ‘better than openwrt’ be ?

tkaiser
Guest
tkaiser

@Willy
Lion Wang, the SinoVoip CEO, gave some insights into why his company isn’t able to meet basic requirements over in comments thread for BPi M2 Berry.

I’ve never seen this level of ignorance, they really don’t get at which low level they’re failing and show no will to improve on such basic things. And the M2 pull request with tons of security fixes is still not merged. That’s such an easy benchmark…

tkaiser
Guest
tkaiser

Willy :
Well, you hardly find information on such products in your local store. They only exist on the net. RPi is almost the only exception.

At least in Germany that’s different, the Banana brand is available at various local stores lying next to Raspberries. And the ‘infomation’ shown there is partially the result of SinoVoip having hired a retard being responsible for ‘technical documentation’ while only able to do copy&paste without understanding. That way their next product will be advertised with ‘RAM: 1GB DDR3 SDRAM with 733 MHz’ while it’s 576 MHz max in reality (the 733MHz are ‘copy&paste gone wrong’ from an older other Banana, I told them over in their forum maybe 20 times already but as usual they don’t give a shit).

Interestingly instead of fixing such mistakes (that happen all the time, none of their Gitbook pages is correct, everything is only copy&paste crappiness) SinoVoip complains about ‘annoying aggression’ when being notified of such obvious mistakes again and again. And now both their German distributor and local resellers will soon have to deal with ‘irreführender Werbung’ (misleading advertisements) and costly stuff like that all just because a company’s CEO in China decides to remain as ignorant as possible not fixing obvious mistakes. 🙂

lionwang
Guest

@tkaiser
@tkaiser
why we not merging pull requests on github ?? for allwinner have update tinaLinux kernel ,we need to update our code from allwinner , we must do this work at first . and facat have build a github for BPI-M2 Ultra ,we have all place at https://bananapi.gitbooks.io/bpi-m2-ultra-open-source-single-board-computer/content/bpi-m2ultrasourcecodeongithub.html
why we use micro USB for power ? when we design hardware , we all compatibility DC and micro USB interface. if you are maker ,you can easy to find micro USB adapter , if you want to do a real product ,you can use DC port.Different applications, different choices.
open souce community, Need understanding and tolerance. We can’t get everything done in the first place. I never say you’re trash, but you always say banana pi is trash. you lose your basic position and respect.You can look on, but you can’t lose basic morals.

again : you see why we always ignore you. In addition to attacking, you don’t have any help with banana PI

mdel
Guest
mdel

@tkaiser

i’m only commenting based on your comments so so it may be out of place, but seeing how you describe that company as being unresponsive or willing to ignore community work, why do you force yourself to get involved in their products ?

I’d understand that you like providing a “proper linux experience” for every device that comes along, but at some point it would be more valuable to pick and choose your targets, especially when you identify a company that is not willing to help in any way.

And i would advise taking a step back when writing your comments, no matter how infuriated you are, using insults will only undermine the point you’re making and give the other side ammos to discredit you.

– Finally something maybe a little more constructive :

having your own armbian distro now, that i assume is getting popular and well respected by the community, you could simply showcase devices that are not supported and why that is, making a respectful bad publicity for devices coming from companies that do not wish to involve in community work..

lionwang
Guest

@mdel

I fully respect all the work that armbian does.Thanks to the developers who contributed to armbain.I also contact igor to provide all the samples and are willing to work closely with them. They are ahead of us in open source.We are learning from each other.
What we can do now is work closely with chip makers.Design hardware, consummate drive, this is our basic job.It is also for the better development of the open source community.So there’s definitely a lot of demand that’s not going to satisfy the mainstream open source community, but our capabilities can only be done here.
But tkaiser, from start to now, in addition to the attack, I don’t think he what is the value of our project, he more than once on behalf of armbian, said armbian official will not support new hardware version of banana PI, I respect anyone’s choice and decision.But there is no need to be biased and offensive.

tkaiser
Guest
tkaiser

@mdel
I agree with you in many points and you can be assured that I will from now on switch to ignore mode. Since in SinoVoip forum posts get censored it was somewhat important to get the whole level of ignorance publicly documented which is done now. 🙂

This is a proof that you can’t help them since they consider being asked for most basic stuff an attack. Same with information, they simply aren’t able to understand why it matters to provide correct documentation. Now that’s done simply let’s forget about them. At least in Germany they’ll have to deal with ‘Irreführende Werbung’ soon maybe then they’ll understand what’s wrong with ignorance.

CampGareth
Guest

@lionwang
Could you add some information to your READMEs on github? I should be able to at a glance figure out what the repository is about, how to use its contents etc. but I can only make guesses and poke bits of it to see what they do.

Here’s an example of a better README (though by no means the best): https://github.com/mholt/caddy

It has a project description, key features, instructions for installing and running, links to further documentation, instructions for contributing…

I appreciate that you’re new to open source and that it basically isn’t a thing behind the great firewall, but what you’ve done is generally considered to be the bare minimum.

tkaiser
Guest
tkaiser

@CampGareth
Nice try to get SinoVoip improving on Github usage. Better look through this https://github.com/BPI-SINOVOIP/BPI-M3-bsp/issues/3 and especially all the other still open issues over there to get the idea what to expect. Issue #2 is also telling since of course the only documentation ever provided was wrong as usual (‘copy&paste gone wrong’ as every time… and the person responsible for this mess is still in place destroying all the hard work the few brave souls inside this company or soon MTK staff do)

Besides that the great firewall or being in China is no excuse to fail at this low level, just look at FriendlyELEC or other Chinese vendors who do not give a shit about everything 🙂

CampGareth
Guest

@tkaiser
/me gives self the “an attempt was made” sticker.

Wondered how they’d respond to a kinder voice, the answer… remains to be seen. The next attempt will swear at them every other word, while attempting to be constructive, much like “what the f*** should I have for dinner” etc. Maybe we’ll find the balance ;D

LinAdmin
Guest
LinAdmin

lionwang :
@tkaiser
We can’t get everything done in the first place.

I agree that it all needs quite some time and effort. However, could you please explain why SinoVoip publishes a schematic where component values are deleted? Can you give a date when by latest a complete schematic will be published?
Thank you!

willy
Guest
willy

@LinAdmin
… and why people who kindly propose pull requests to update their outdated kernels to the latest security fixes are simply ignored, leaving all their customers exposed to random bugs and even critical flaws ? That’s the most insulting attitude to have towards people who provide quality work on their spare time.

lionwang
Guest

@CampGareth

thank you , i will let my colleague to do this work .

tkaiser
Guest
tkaiser

@willy
This is most probably the reason why they feel the need to ignore your and all of our work: https://github.com/tinalinux/linux-3.10/blob/r18-v0.9/Makefile

LinAdmin
Guest
LinAdmin

@lionwang:

Just wondering why you preferred not to answer my polite question?

Willy
Guest
Willy

@tkaiser
I don’t understand your point regarding this makefile. What did you mean ? That they’re reusing someone else’s kernel or that they have some binary dependencies on this outdated one ?

tkaiser
Guest
tkaiser

@Willy
The Banana folks said they’re not able to merge community pull requests with security fixes (bringing their smelly OS images for their M2 Ultra from 3.10.65 to 3.10.106) since ‘for allwinner have update tinaLinux kernel ,we need to update our code from allwinner , we must do this work at first’. The above is this new tinalinux kernel release (for another SoC but same business unit being responsible for). And Allwinner chose to provide something still being a 3.10.65 version with some stuff being fixed (eg. Dirty COW is fixed in this new 3.10.65 as it has been in your 3.10.104 LTS release). I would call this a mess but SinoVoip ‘need to update our code from allwinner’ for whatever reasons.

To me it seems pretty obvious that they neither understand the meaning of ‘open source’ nor ‘security’. And even if MTK will now perform way better than Allwinner here I really would avoid software from a company sitting in between and not caring about security at all on a device like the R2 that is supposed to be used as firewall or router.

willy
Guest
willy

@tkaiser
Got it. Indeed it simply shows they are only able to route PCBs and place components on a board, but have zero clue about software, which is a bit sad for a board vendor! I can accept that some board vendors solely rely on the SoC vendor’s software, but only if this one acts responsibly (ie: follows updates). That’s definitely not been the case with allwinner, providing outdated BSPs. I doubt it will be much better with MTK given that such vendors generally rely on contractors to port to mainline and on the board vendors to report bugs that have to be fixed.

This simply means it’s better to stay away from Banana-whatever until the device is fully supported in mainline. After all it’s their choice.

lionwang
Guest

@LinAdmin
you can send mail to [email protected] .

lionwang
Guest

Tkaiser ,You are not a god of the open source community, please keep the basic respect for others ,At least let others don’t hate you

Now everyone know why you never says a bad word about orange PI.

http://www.cnx-software.com/2017/07/15/office-factory-business-model-and-ambitious-plans-of-shenzhen-xunlong-software-orange-pi-maker/
You don’t need to take a wrong way to achieve the purpose of you