Device-tree improvements

[summers]

Thats strange, your device tree should have explicitly said that it should load that kernel module! Is it listed in $this->bbcode_second_pass_code('', 'lsmod')? Guess it may just be a very quiet module. Also $this->bbcode_second_pass_code('', 'ls /dev/watchdog') is the sure sign its running ...

[TheSaint]

$this->bbcode_second_pass_quote('summers', 'T')$this->bbcode_second_pass_code('', 'lsmod')?

[summers]

Ta thanks. Intersing that linarno does a monolithic kernel. arch by using generic kernels always have a fair few modules.

And the /dev/watchdog is the sure sigh - says that it does work on your machine, which was what I suspected....

[TheSaint]

I peep into the Armbian file. Interesting to see their arrangement, which is using boot.scr and having modular kernel, with limited numbers of modules. It seems that they didn't included extra modules and the setup is pretty light.
One particular think, that I found, is the miqi,dtb used to start the Tinker Board kernel. The system boots with initramfs first. I should experiment to write their (included) u-boot and see if the Arch will come up to the login.
Wireless module is like ALarm, staging module. Perhaps I may try the TinkerOS if it works better.

Also DiePi si pretty similar to the TinkerOS, just stripped down to the core. They use the same u-boot as TinkerOS, vfat boot partition and miniarm dtb. The rest is a minimized debian install, with a lot of scripting to guide the newbyes like school boys thought the installation.

I suggest you to have a look. Just mount the image with 4 Mbytes offset.

[summers]

Yes I suspect wireless will only move away from the staging module, when someone decides to to the wifi chip properly.

The basic RTL8723 chip has been mainlined https://github.com/torvalds/linux/tree/master/drivers/net/wireless/realtek/rtl8xxxu.

Problem is the sdio interface on RTL8723BS, but that code is also in the kernel. https://github.com/torvalds/linux/blob/master/drivers/mmc/host/dw_mmc-rockchip.c

So should just be a cause of marrying the two. To mind mind it just needs someone that knows the code, to set about making the change ...

[summers]

okay, tinker-board-s dts has landed in linux next:

https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/tree/arch/arm/boot/dts/rk3288-tinker-s.dts

So its on the way, only enables eMMC, but its a start ...

[TheSaint]

Is it like you've prepared? Or has it some improvement ?
I was settling with Armbian, so I've to make a space and try it

EDIT
Oh no, another Gbyte and download the entire kernel repository

[summers]

Ah yes, cloning a linux kernel repository is now about 1GB in size, so downloading is a lot of bandwidth. I have a clone of the rockchip reposity - and needed that to patch against it ...

The device tree that this change produces is about 50% of what I sent to you, e.g. it enables the eMMC, but not the wifi. Code though is what you tested.

Idea behind mainlining, is it will then become standard, so people will be able to install a standard arm arch on a TB(S) and hopefully it should work without patches ....

Patch this time was as much as anything setting up rs3288-tinker-s.dts device tree. As the only known difference between the TB and TBS is the eMMC, I suspect thats the only difference that will go into the kernel, any other changes we make will have to be common for TB and TBS.

[TheSaint]

After all, I'm glad we could promote a mainline for TB(S). Of course there still some other works to come, I hope ASUS will keep up the job, as a matter of market.
Anyway, one is the chip manufacturer and another is the board manufacturer, there could be some hidden story, which won't tell each other
Imaging to capture more buyers because they will have a chance to use any distro
I hope they'll come to set the fiber connection, so no more to pay to 4G downloading huge files, which is a bit high price. Unfortunately, I might face installation denial because out of coverage

[summers]

Yes it would be good for ASUS to consider mainlining to the kernel the set up of TB(S). Alas it seems like neither ASUS nor rockchip see mainlining as important, ASUS just pushed their debian distribution with patches.

Now mainlining is fair more effort, than just getting code to run on a machine. But the effort is worth while, e.g. its the only way to ensure that Arm Arch will run on the board and be maintained.

I'm not sure though what it takes for board to be added to the list of arm arch supported boards, I suspect that without a contribution, it won't be added by the maintainers. This is a pity, as the basic TB(S) machine we are getting close to it just running the plain vanilla arm arch, and have most things basically working, with just a few rough edges.