Posts by fabiobassa

MIKMXIII

does the Samba problem appear onto wifi or ethernet cable too ?
If on wifi , have a look at post 83 and 85

If in any situation , well let's wait ilmich answer

And is not true that all mxq pro are 3229, we catched some 3228 ( not even 3228a )

oneillb
compliments Barry for such detailled work and for your goals

Yours must be a very particular board that needs to be tweaked very much but as already ilmich answered, we shouldn't forgot about temperature and other factors of stability . I am still oriented in low response from vcc switches due poor design in ripple filtering

BTW increasing voltage and thus conseguentially current , especially at low running speed where the cpu should " relax and take breath" could lead to an increasing soc TEMP and this is a factor to be carefully investigated .

And also don't forget that we are asking so much from a box that now costs around or less than 20 dollars  

Great great job Barry !!

oneillb

I own many 329q ( all ddr3 ) and some of those boxes are the basis of all investigations made in the last years since starting studies up to actual days. jock2 himself and me have studied on one of these ! you can check here the time date and from here all has born (december 2019 !!!!! )

https://forum.armbian.com/topic/12401-lo…k3229-rockchip/
ilmich owns a board IDENTICAL to yours and he is not suffering any of those problems you show here.

Don't forget that those cheap tvbox are made with the last rubbish components found in china offices or plants so I wouldn't be surprised thinking about faulty components ( ram ? defective capacitors on power switching thus conseguentially vcc ripple,? bad solder joins that under heat/stress go faulty ? and the list could go growing).

The fact that under certains conditions is working and under stress or/and idle moments the board crashes are the most evil problems to investigate, but I must say you many thanks because with your feedbacks and your logs you helping ourselves to better investigating

oneillb

This is for general culture and it is how basically works the boot flow in rockchip. Biggest limitation is that is not possible ON NANDS write under 0x2000 with new tools unless you use a old way to burn idbloader ( and multitool is capable of do it)

http://opensource.rock-chips.com/wiki_Boot_option

Reinstalling android with windows AndroidTool ( AndroidTool_Release_v2.1.7z ) also could be a good starting point and after android is in place again do the procedure as ilmich suggests. Unfortunately nands are a very pita !!

DISCLAIMER : android uses some internal " partitions " called rkparameters...... consider this when going to install libreelec because it is different from master boot record and even gpt. Reinstall android and came back here to ask enlightment to ilmic or jock

chewitt
Tanix is only a "brand" company.

I can give for sure to you that I have a Tanix that inside is rockchip , the very old and familiar for me rk322x , so the " roulette buying" is:

"I order TANIX TX3 !!"

What the hell will I get ? And spin the roulette

Flecky
I misanderstood your request then my bad

since this 3ad is for 322x maybe you want open a new post with this hybrid mode ( multitool for armbian for flashing libreeelc , it sounds good anyway )
Just one question : the version of libreeelc for 3318 you are using does support nand ?
for example here , with 322x , for reading nand you must remain on LEGACY KERNEL because no drivers for nand on mainline kernel
When you boot libreelec from sd on 3318 can you at least see the internal nand ?? ( open ssh and do a lsblk or a blkid . do you see the internal nand ? )

EDIT
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

are you sure this 3318 has a NAND ? It should be a emmc

Flecky
Maybe I am wrong or I misanderstood your problem, but...........
Are you trying to burn a libreelec image made for 3228 with multitool 3318 on a 3318 device ?

I suppose it is quite obvious your device is bricked after , they are two totally different devices and platforms
322x ( 3228 3228a 3229 ) are 32 bit
3318 is 64 bit
diferent ddrbin different uboot different ALL

If I misanderstood your problem, then sorry !

ericmartin971

Hello eric, you speaking of two different things. In reality there are two drivers , one is ssv6051 other is ssv6x5x and they are different but... but..since the reference in pid vid is the same... the kernel loads sometimes the wrong one. For this reason is necessary blacklist the offending one
If i can suggest identify the right chip you have, if 5ghz it COULD or could-not be ssv6256P . In this case you also need ssv6x5x-sw.bin and ssv6x5x-wifi.cfg in the right places, usually under /etc/firmware

As chewitt already said first you have to REMAIN on 4.4.194 , second if you will you can have a peek in the armbian forum ( thanks chewitt for mentioning it) where jock and myself are " fightening" against those different situations

Lajos
Absolutely yes , it is one of the first box on which has been experimented a lot.
Give a try to any 3228 with DDR2 on the first post

goolpone

prego, you are welcome. It is a stick dongle and I remember uart pins should be exposed there.
Any way UNFORTUNATELY for all of us enthusiastic people of these boards is veryyyy difficult debug without a uart because the steps involved in booting are a big chain and error is around the corner ( idbloader, trust_os, uboot )
So it worths have a usb/ttl on the desk

goolpone

are you try to boot from SD card or from internal storage ?
I give for sure your android is booting fine.

Well a tricky way to have an INITIAL serial out put is connect a usb/ttl to pin 4 on sd reader counting from where the switch of sd is ( if big sd reader you must count from right to left , pin n°4 ).
OF COURSE if you have sd inserted after a while it comes garbage from that pin , but if you boot android you will see every thing till the end.
This my affermation could and could NOT work on your board and don't ask me why there is a serial output there, but it is, and it is very usefull.

Would you upload photos of your board somewhere, please?

Ty

gianni253

ciao gianni, buongiorno, good morning.

Trying to answer to your questions is not easy .Those rk322x boards comes in such great variety ( 3228 3238a 3229 nand emmc ddr2 ddr3 ) and each of those has a specific boot process. The "worst" ( let me use this term ) are 3228 ddr3 nand because they use a very specific idbloader ( very very first stage ) , a common u-boot and a very specific trust-os to get the dmc ( ddr scaling ) function on

knaerzche did a great great job to cover most possibilities in booting on different platforms but still evil could happen

1) yes the process tries to boot from sd first, then from usb ONLY if you use a modified uboot ( the one present in multitool ) .

And then the process take in charge kernel and his drivers that finally recognize internal nand ( with emmc is easier)
But if you erased everything well the boot process will follow other path because your nand now is an empty container

2) grab your original android stock firmware and burn it with AndroidTool_Release_v2.67.zip ( google for it) . it is for window and it will put the right things in the right places ( idbloader, u-boot, trust etc etc ) and you will have android back on your box
If you did a ADB backup You still need a piece called parameters.txt to " teach" the tool to put the various pieces in the right places.
better flash WHOLE stock firmware.
But why go back on android if you want to approach LIBREELEC ???

about point 3 I suppose jock can answer better since the early stage of the boot process is similar on armbian and libreelec

ccs1664
when you say

"Really, an image from armbian booted "
is exactly what I was trying to explain with my previous post , but I am sorry you didn't even read it.

Commercial names are nothing so no help indeed to say why your board doesn't start without additional infos given

If you can boot to armbian and have a shell, would you provide the output of:

lsblk
free

hexdump -C /sys/bus/nvmem/devices/rockchip-efuse0/nvmem

cat /sys/class/devfreq/dmc/trans_stat ( probabily here no output )

Thanks

ccs1664

well sometimes reading whole bunch of a forum is annoying and timewasting , but also is a great resource to find the answers
For example if you go to on this same forum
[unoffical] LE-9.2/9.80-Images for RK3229/RK3228
there is a statement exactly on mxq pro 4k that works with v88 Mars II and MK 809 IV

Now the REAL problem with this kind of board is that commercial name mean NOTHING, and when i say nothin I mean NOTHING

I have seen boards claiming to be 3229 and instead efuse revealed being 3228a , claiming have 64 giga rom and were just 16 or saying 4 gb ram and have only one

So commercial name is equivalent to say mickey mouse or goofy, thank to producers that use the first avaible chipset they have and mark with the best name they invent

In "general" any rom for lower chip ( 3228 ) and ddr 2 should boot quite every where, even on the bigger brother 3229 with ddr3 and then you can try to tune the whole thing

@DS1

this is a well known board, shishion or andowl brand !

it is a REAL 3229 but with ddr2 and emmc storage

so ANY of images on the first post with this characteristic will surely boot !

here it goes the dtb extracted from the image provided
hope can be usefull

tin01
beg pardon , but from reading your link , it is speaking of

This tool is written by Team CoreELEC and exclusive for Amlogic devices running CoreELEC

and also:

Currently supported Amlogic SoC’s: etc etct etc

In this 3ad we are on Rockchip devices.

Is this bl 301 applicable even to rockchip ?

creited

You are able to use Linux this is great because you can try armbian starting from SD!

Take your libreelec SD and substitute all the content of fat part and rootfs ext4 part with content from armbian.

In other words, don't touch miniloader-trustos-boot of libreelec mars II and use kernel init and rootfs from Jock compilation

Give it a try