New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Bump to linux-4.x.y #3
Comments
Thank you for your attention. In general we focus on a kernel which gets support from chip vendors. If chip vendors release a new kernel usually we will catch up with it otherwise we may not do it. |
Thanks for the comment. It would bring about a lot of nice things like having updated device tree support. |
+1 |
@friendlyarm is there any word from the chip vendor (Nexell I believe) about when they are going to release a new version? Supporting 4.x.y on some of your products would be a enormous win. |
We just added 4.6.7 and 4.7.2 for H3 based boards, first one is Nanopi Neo: http://www.armbian.com/nanopi-neo/ |
This issue is specifically referring to linux-3.4.y boards i.e. Nexell boards. |
With Nexell I haven't done much of a research yet but moving from "old android kernel" toward recent is a heavy project even for experienced crews, especially if documentation is poor, if we need to face blobs and (c) issues. How far and what was already done, don't know and can't really comment. Out of all dev boards, what is known to me, good mainline support is on: Rpi, most popular Allwinners, some imx6, ... , rk3188, odroid xu4, ... Even Samsung chips have issues on mainline kernel what prevents usage in productive environment, while you can use first three from the list mostly without any serious problems. |
Yup, the request is intended to be addressed to FriendlyArm and Nexell to complete the port as vendors of the boards rather than a request for the larger community to unbreak the codebase and forward port :) |
Exactly @ndoo. I have attempted to email the contact noted in It seems to me that someone at @friendlyarm must have a contact at Nexell who can at least provide a timeline. |
Have a look here for the latest news: Nexell:
FriendlyARM:
|
Oh, it's 2017 |
They have totally gave up this project... never an update... 🙁
… Il giorno 08 apr 2017, alle ore 16:50, daiaji ***@***.***> ha scritto:
Oh, it's 2017
No new kernel messages
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub, or mute the thread.
|
@spacecdr |
Dear Metro94, dear jensen, dear hmz007 Dear Friendly-arm community, I would like to use the kernel from the A710/v4.4 branch of https://github.com/SamsungARTIK/linux-artik on the Nanopi-M3. With gcc-linaro-6.3.1-2017.02-x86_64_aarch64-linux-gnu.tar.xz toolchain from Linaro, I get : ./arch/arm64/boot/dts/nexell/s5p6818-artik710-raptor-rev01.dtb and 211 modules. (I will see later for the three modules bcmdhd.ko nx_vpu.ko vr.ko, for now, I just need a reasonable recent kernel able to start an android init service or even just a linux elf binary or bash script) and a image arch/arm64/boot/Image Is it possible to load it ? -with the nanopi2-lollipop-mr1 branch of https://github.com/friendlyarm/uboot_nanopi2.git ? if not -with the A710/v2016.01 branch of https://github.com/SamsungARTIK/u-boot-artik.git or the NanoPi_M3 branch of https://github.com/metro94/NanoPi_M3-U-boot.git ? if not -with a mixture of NanoPi_M3 branch of https://github.com/metro94/NanoPi_M3-2ndboot.git + one of the previous mentionned uboot if not something even more strange ;) ??? What is the current status of Samsung/Nexell/friendlyarm collaboration on these reasonable recents uboot + 64bits-linux/drivers ??? I mean... come on... guys, it's May 2017, Jonathan Corbet announced yesterday on lwn.net The 4.11.2, 4.10.17, 4.9.29, 4.4.69, and 3.18.54 stable kernel updates have all been released with the usual set of important fixes. Note that this is the final update for the 4.10 kernel. and Greg KH announced in September 2016 on lwn.net the 3.14.y LTS kernel series is now end-of-life ... Please Friendly-Arm, update the wiki and give us a way to have reasonable recent uboot/linux/{ubuntu/android} software !! and not a 32bits 3.4.* kernel thas has already reached it's Projected EOL (https://www.kernel.org/category/releases.html) We can help, test, debug, but the bare-metal stuff have to be done by friendly-arm / nexell / samsung ... Thank you all in advance ! |
Actually we tried code from https://github.com/SamsungARTIK/boot-firmwares-artik710 and https://github.com/SamsungARTIK/u-boot-artik and made an SD card. Unfortunately that SD card couldn't boot our board. We may have to wait for Nexell to release its latest BSP. We can do development based on u-boot-artik and https://github.com/SamsungARTIK/linux-artik but without a workable boot-firmware we cannot boot our board and test an image. |
Have you tried this alternative to 2ndboot.bin ? https://github.com/metro94/s5p6818_bootloader it boots ... |
compiled in DEBUG=1 mode, one gets :
|
I have compiled the bl1 firmware from sources bl-artik710 with BOARD set to DRONE and it works. I have used NSIH based on drone-sd-64.txt file stored in reference-nsih in this project. Below is the script used by me to embed the compiled bl1 and u-boot in SD card. Note that first partition on the SD card should start at 2048 sector. The script should be updated before use, the environment variables with SD device and location of the bl1 and u-boot sources shall be set appropriately.
The Linux kernel is much more problematic, but I have managed to boot the linux-artik, branch A710v4.4. To be precise, I have entered shell within initial ramdisk, through serial port:
I have used defconfig for drone, too. To be honest, I don't remember all the changes that I have made in the configuration. As I remember, I have disabled some drivers because they have compilation errors. At least DS1307 driver I have disabled. There were also some important changes in Linux source code. First change I have made in file head.S (arch/arm64/kernel/head.S). The two lines below should be added somewhere at startup, e.g. in el2_setup:
Second change: in file s5p6818.dtsi (arch/arm64/boot/dts/nexell/s5p6818.dtsi) the "psci" device description I have removed (commented out). Otherwise system panics at smc instruction call. I have also removed all "enable-method=psci" from all processor descriptions. Third change: a bug in axp228-mfd.c file (drivers/mfd/axp228-mfd.c). Below is the patch. The bug causes memory corruption and kernel panic little later. From unknown reason the driver does not find axp228 chip and returns with error. Before return it releases memory, but bad function is used for that.
I got stuck in CPU configuration - the kernel sees only one processor. In dts file the processor "enable-method" is set to "psci", but I don't know how this might work. The smc instruction is disabled at exception level 3 and not intercepted by exception level 2. Even if it would be enabled, the boot loader lacks of handling code. The bootloader provided by metro94 has one bug in trustzone configuration, but after fix the bootloader works also. It may be used instead of bl1-artik loader. Below is the needed patch:
|
Nice work! |
Actually I'm trying to work with vanilla kernel 4.11.6. I have some trouble when all 8 cpu's are enabled. System behaves unstable, but with 4 processors seems to work pretty. At least ethernet, display, usb and serial port are working. I'm writing this post from NanoPi with the 64-bit kernel. Device drivers I have moved from Samsung Artik kernel 4.4. Here is my desktop: Graphics driver is nice, because screen resolution may be changed on the fly. No need to compile kernel to change resolution. I can upload prebuilt boot loader and kernel image. But work is in progress. |
coooool!
rafaello7 <notifications@github.com>于2017年7月6日周四 03:51写道:
… Actually I'm trying to work with vanilla kernel 4.11.6. I have some
trouble when all 8 cpu's are enabled. System behaves unstable, but with 4
processors seems to work pretty. At least ethernet, display, usb and serial
port are working. I'm writing this post from NanoPi with the 64-bit kernel.
Device drivers I have moved from Samsung Artik kernel 4.4. Here is my
desktop:
[image: screenshot_2017-07-05_19-30-56]
<https://user-images.githubusercontent.com/2477169/27881537-8d87cf0e-61b8-11e7-8ace-22630032e7a3.png>
Graphics driver is nice, because screen resolution may be changed on the
fly. No need to compile kernel to change resolution.
I can upload prebuilt boot loader and kernel image. But work is in
progress.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#3 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AARmIHj31O3HbS1BGiqlc7Odd1sn9V5Hks5sK-lOgaJpZM4JJ7vz>
.
|
Nice work rafaello7 !!! |
All my work is on github: Additionally I'm using nanopi-load tool to upload images to device through USB port instead of embedding in SD card. But booting from SD card is also possible. I provide some instructions soon, how to build and run all together. Maybe in README on bl1-nanopi-m3 project. |
@rafaello7 coooooool! |
Yes. |
I have added build instructions on bl1-nanopi-m3 page. |
Sorry for ignoring the issue of @rafaello7 for a long time. He has made a great progress on it. Actually my NanoPi has been sold, so I'm afraid not to maintain s5p6818_bootloader and s5pxx18_load from now on. |
@metro94 no problem ;) As of Linux 4.11.6, there are stability problems with this kernel when all 8 CPU's are brought up. Kernel boots correctly, but just after boot I'm performing simple test: I'm starting a few "yes" processes:
Kernel hangs just after 6th process start. I have no idea how to fix the problem. Kernel 4.4 from Samsung Artik does not have the problem, but I have no idea how to find the change that makes difference. I will try to prepare the Artik v4.4 kernel for NanoPi. Kernel 4.11 goes aside for now. |
I have found the problem reason. In fact, these were two reasons: two long extension cables between USB socket on my PC and micro USB cable connected to NanoPi. I think you can guess the rest. After shortening the distance NanoPi seems to work stable with 8 CPU's. |
Excellent - that's exciting news! |
So, now I change location to SD card. Installed fixed Release 1.2. this |
I tried it /dev/mmcblk2p1 instead of uuid for /boot |
can you give me contact? |
Please install release 1.3. I have added possibility to specify boot device in command line:
Are you booting from SD card or from EMMC? You should install updated bl1 on SD if you are booting from it. |
@rafaello7
|
@PVlado I don't know about any such datasheets. I looked at bl1 source code. |
@rafaello7 CPU Wakeup done! WFI is expected. U-Boot 2016.01 (May 16 2018 - 00:08:21 +0800) DRAM: 2 GiB Starting kernel ... [ 0.000000] Booting Linux on physical CPU 0x0 |
@PVlado it looks bad. I mean the oops. Does it appear with 2GB ram only? You can also try to load the 4.14 kernel using loaders from FriendlyARM. |
All img. from FA works OK. |
On my board it happened only with it: |
Hello, |
It is possible since yesterday. armbian/build@86a00ad Images are not yet updated and you need to build on your own (with parameter EXPERT="yes") or wait a few days. |
That's great news, Igor! Thanks a million for your great work on Armbian and the M3/T3/T3+! 🥇 |
You are welcome :) https://twitter.com/armbian/status/1038146549124083715 |
@igorpecovnik Just one more question, if I may: Regarding the old, 1GB, NanoPC-T3 (without "+", but which also already had an 8 GB emmc), when I build Armbian myself using board "nanopim3", does it now also offer the chance to install onto emmc for this old board? Thanks a million one more time! 😄 |
It is possible that 1G version won't boot at all. You might need different boot loader settings (the same as for M3/Fire3). Let's do it this way - I will prepare a CSC build target for the one without + and if it works for you it can have the same status. Boards are nearly identical. |
Great! 👍 Once you've prepared this (did I get you right that it will be a new board name, such as "nanopct3" without plus), just give me a ping! And: Clearly, there's no need to hurry - enjoy your weekend!!! 😄 |
Briefly checked: 2G hacks are tied directly to the + board, eMMC size does not matter and its made universal, ... and that's all that is different. BT does not work yet on + and it will not work here ... I have to dive into that once in the future. Thanks. Enjoy yours as well! |
Thanks again, @igorpecovnik Regarding Bluetooth, I might be able to help, because I got it working back in the very old days when I had just received my first T3 - see this post on the FA forums: Once I found some time to install onto my T3+, I will report back here whether I think that my old BT work will be of any help... |
Bad news: Armbian_5.59_Nanopct3plus_Ubuntu_bionic_next_4.14.68_desktop.7z does NOT work at all for emmc on my T3+ 😞 After copying rootfs onto emmc with nand-sata-install (where I only see option 1 - boot from emmc, system on emmc), when I reboot with SD card removed, it simply hangs at Nano-Pi logo with Armbian u-boot. When I leave the SD card in the SD card slot, it looks like it might boot from emmc u-boot, but it then always mounts SD card from /dev/mmcblk2p1 as the root partition and leaves /dev/mmc0blk alone. 😢 Will continue testing tomorrow and report back about u-boot logs, but I need to reestablish my NanoPC serial console debugging environment... |
BTW: I also seem to have issues with wired eth0 on T3+ kernel 4.14.68 when connected to at least one of my switches - will also report back about this later, sometimes no connection, sometimes a bad connection with very erratic latencies (up to 1300ms) and dropped packets... 😞 |
I guess I must have made some mistake somewhere, serial console is a must ... Have you tried to power cycle? Regarding network ... never experience any troubles. That is odd. |
@igorpecovnik I am wondering why there is a "CRC" error in the U-Boot environment - maybe nand-sata-install fails to correctly write an U-Boot environment? Will try to compare the output to that of a successful FA "FriendlyCore" boot from emmc with kernel 4.4.x and Ubuntu xenial and report back one more time... |
Second update: eth0 wired networking issue seems to be an incompatibility between kernel 4.14 eth0 driver and only one out of three switches that I own. Networking works fine using the other two as well as when connecting directly to the DSL router, so it looks like this is an issue with this particular old GbE switch... |
@awl1 @igorpecovnik Though hardly anyone reported In fact, I think the router of MT7621 should not be old. |
That is not a problem but the problem is wrong default boot environment. I didn't clean it (env default -a) before testing and if I clean it now I get the same error as you get. Will be fixed asap. A network is another story. It works well on both of my switches, gigabit or fast ethernet. |
Fixed. Flash this u-boot to eMMC and it should work. Perhaps you will need to clear environment with env default -a dd if=bootemmc.img of=/dev/mmcblk0 seek=1 status=noxfer |
@igorpecovnik Bingo! Many thanks, Igor! I can confirm that it does work fine now! 🏅 👍 Will try to look into the Bluetooth issue later tonight, could be possible I might be able to pay something back (see above with regards to brcm_patchram_plus)... 😃 Regarding eth0, I tend to think the kernel is most probably fine, as it works ok with all but one of my switches/routers. When I find some time, I might check whether the dev kernel changes something to the better for the affected switch, but that's really low prio... |
Regarding Bluetooth: Hmm - it looks like kernel 4.14.x may still be missing kernel modules for Bluetooth - we need /sys/devices/platform/bt_bcm.0 to be able to access AP6212 Bluetooth device!? |
linux-3.4.y is quite an old release, and the amount of changed lines from upstream kernel is not too much.
Are there plans to work on a newer kernel version? Also it would be great if support for this SoC is upstreamed.
The text was updated successfully, but these errors were encountered: