armbian
sunxi-6.10: Add armbian patches
Description
- Add armbian patches from sunxi-6.7
- Adapt the patch series to the new v6.10.5 kernel version
- Delete unused patches and add newly extracted patches.
How Has This Been Tested?
- [x] Build for arm64 & test work on board
____ ____ _ __ __ __ _ _
| __ )| _ \(_) | \/ |/ /_ | || |
| _ \| |_) | | | |\/| | '_ \| || |_
| |_) | __/| | | | | | (_) |__ _|
|____/|_| |_| |_| |_|\___/ |_|
Welcome to Armbian_community 24.8.0-trunk Bookworm with bleeding edge Linux 6.10.5-edge-sunxi64
No end-user support: built from trunk
System load: 2% Up time: 11 min
Memory usage: 6% of 1.94G IP: 192.168.100.101
CPU temp: 26°C Usage of /: 18% of 7.2G
Time CPU load %cpu %sys %usr %nice %io %irq Tcpu C.St.
15:14:27 1008 MHz 0.00 5% 0% 0% 0% 3% 0% 23,3 °C 0/7
15:14:33 1008 MHz 0.00 3% 2% 0% 0% 0% 0% 22,8 °C 0/7
15:14:38 1008 MHz 0.00 2% 2% 0% 0% 0% 0% 22,2 °C 0/7
Proposing to remove the bigtreetech cb1 related patches that use the old pre-mainline device trees, for 6.10 the bigtreetech-pi device tree will be used with different patches.
Proposing to remove the bigtreetech cb1 related patches that use the old pre-mainline device trees, for 6.10 the
bigtreetech-pidevice tree will be used with different patches.
I only have bananpi-m64 available. I thoughtlessly adapted all the other patches to the new version and did not test them. It is assumed that someone else will change this.
At this stage, the compilation is error-free for arm64, but with a lot of warnings for wifi drivers.
I'm done with this.
@JohnTheCoolingFan You can just continue. I can give you access to my repository.
Getting an error trying to do kernel-dtb for bigtreetech-cb1 or bananapim4zero edge, on commit c54ab52ac077aa35fa4c978d2132c4ac47b177b7:
arch/arm64/boot/dts/allwinner/sun50i-h616.dtsi:708.21-723.5: ERROR (phandle_references): /soc/spi@5011000: Reference to non-existent node or label "spi1_cs1_pin"
also defined at arch/arm64/boot/dts/allwinner/sun50i-h618-bananapi-m4-zero.dts:235.7-248.3
Getting an error trying to do
kernel-dtbforbigtreetech-cb1orbananapim4zeroedge
It looks like I haven't made a commit with the bugs fixed. This is my inattention.
Simply changed some file names so that the cb1 dtsi enhancements would be applied and other dts files won't be created, trying to use the bigtreetech-pi dts. No Ethernet and no HDMI output, though... Not sure why.
No Ethernet and no HDMI output, though... Not sure why.
What does dmesg report? UART console.
What does
dmesgreport? UART console.
Well, there's no eth0 interface and the only snippet I could find about hdmi is this:
[ 0.033135] platform 3001000.clock: Fixed dependency cycle(s) with /soc/rtc@7000000
[ 0.036404] platform 6000000.hdmi: Fixed dependency cycle(s) with /soc/tcon-top@6510000
[ 0.036813] platform 6000000.hdmi: Fixed dependency cycle(s) with /soc/tcon-top@6510000
[ 0.036922] platform 6510000.tcon-top: Fixed dependency cycle(s) with /soc/hdmi@6000000
[ 0.036979] platform 6510000.tcon-top: Fixed dependency cycle(s) with /soc/lcd-controller@6515000
[ 0.036996] platform 6510000.tcon-top: Fixed dependency cycle(s) with /soc/bus@1000000/mixer@100000
[ 0.037163] platform 6510000.tcon-top: Fixed dependency cycle(s) with /soc/lcd-controller@6515000
[ 0.037232] platform 6515000.lcd-controller: Fixed dependency cycle(s) with /soc/tcon-top@6510000
[ 0.037477] platform 7000000.rtc: Fixed dependency cycle(s) with /soc/clock@7010000
[ 0.037582] platform 7000000.rtc: Fixed dependency cycle(s) with /soc/clock@7010000
[ 0.037679] platform 7010000.clock: Fixed dependency cycle(s) with /soc/rtc@7000000
[ 0.038781] platform 6000000.hdmi: Fixed dependency cycle(s) with /connector
[ 0.038861] platform connector: Fixed dependency cycle(s) with /soc/hdmi@6000000
I suspect that there might be something about the device tree not loading properly, but the patches to explicitly enable hdmi and ethernet are applied with no errors, perhaps I overlooked something? A warning during compile, or changes in drivers from 6.6... Will have to do quite a bit of investigation on that one. At least the board boots properly, wifi works fine at least.
I suspect that there might be something about the device tree not loading properly, but the patches to explicitly enable hdmi and ethernet are applied with no errors, perhaps I overlooked something? A warning during compile, or changes in drivers from 6.6...
After I adapted the patches and began to build, the compiler reported an error of duplicate nodes in the DTS many times. I had to remove some of them. I could have confused something. If you have a working image (with a different core), will you be able to compare the HDMI nodes in the DTS. Not the files, but the ones that were applied on a working device? Using the command:
dtc --sort -I fs -O dts /sys/firmware/devicetree/base > dts-out.txt
Maybe this will give you a clue.
As for Ethernet eth0. I realized that the problem is with the pwm driver. @chraac has kindly agreed to add a pwm driver to this pull request. He seems to have a working option here #7139. Let's wait a bit.
Well, here's the output file from that command: dts-out.txt
As for ethernet, yes, it is dependent on sunxi pwm, but also from what I remember u-boot also plays a role in initializing the ethernet PHY.
Hi @The-going , regarding the sunxi's pwm driver and gmac driver, created a PR at you fork, please have a look: https://github.com/The-going/armbian-build/pull/25
Well, here's the output file from that command: dts-out.txt
To be honest, this information doesn't tell me anything. I don't have another such file extracted from a well-functioning device.
I canceled your changes. As far as I understand, the sun50i-h616-bigtreetech-cb1-emc.dts and the sun50i-h616-bigtreetech-cb1-sd.dts are different devices. If someone added them, does it mean that someone needs them?
As far as I understand, the sun50i-h616-bigtreetech-cb1-emc.dts and the sun50i-h616-bigtreetech-cb1-sd.dts are different devices. If someone added them, does it mean that someone needs them?
These files are added by BigTreeTech. These are two revisions of the CB1 module: with eMMC and without, the latter needing an sd card to boot.
But 6.10 mainline has a bigtreetech-pi device tree, which can be used for BigTreeTech Pi or CB1 module in a PI4B adapter, and there is another device tree for a CB1 installed into a Manta carrier board, which simply integrates a 3d printer main board and a compute module slot for CB1/CB2.
Armbian has just one board representing that: bigtreetech-cb1. And so I assumed the safest case of a BigTreeTech Pi or CB1+PI4B. The only device tree used by armbian before bigtreetech-pi was added to mainline is the sd variant. I think it might be possible to support an emmc CB1 version using a device tree overlay. As for Manta case, it seems we are going to just ignore it and shove it under the same board config: bigtreetech-cb1.
CB2/Pi2 has a similar problem in my eyes, but it's out of scope for discussion in this PR as it's a rockchip-based board.
These files are added by BigTreeTech.
Thanks for the detailed explanations.
I'm wondering on which device did you check the performance of HDMI?
model = "BigTreeTech Pi"?
I've checked on a BigTreeTech Pi v1.2, that's the only version of the "first gen" I have.
I've checked on a BigTreeTech Pi v1.2, that's the only version of the "first gen" I have.
Very good. Can you check the performance of sunxi-6.6, sunxi-6.7 for this device?
Can you check the performance of sunxi-6.6, sunxi-6.7 for this device?
I've been getting the device to work with these mainline versions with patches in #6656 What do you mean by "check the performance"? The board worked well on 6.6 and 6.7, but there are problems on 6.10.
The test is at this stage:
____ ____ _ __ __ _____
| __ )| _ \(_) | \/ |___ /
| _ \| |_) | | | |\/| | |_ \
| |_) | __/| | | | | |___) |
|____/|_| |_| |_| |_|____/
Welcome to Armbian_community 24.8.0-trunk Bookworm with bleeding edge Linux 6.10.5-edge-sunxi
No end-user support: built from trunk
System load: 18% Up time: 0 min
Memory usage: 8% of 1.96G IP: 192.168.100.101
CPU temp: 34°C Usage of /: 17% of 7.2G
____ ____ _ __ __ __ _ _
| __ )| _ \(_) | \/ |/ /_ | || |
| _ \| |_) | | | |\/| | '_ \| || |_
| |_) | __/| | | | | | (_) |__ _|
|____/|_| |_| |_| |_|\___/ |_|
Welcome to Armbian_community 24.8.0-trunk Bookworm with bleeding edge Linux 6.10.5-edge-sunxi64
No end-user support: built from trunk
System load: 86% Up time: 1 min
Memory usage: 6% of 1.94G IP: 192.168.100.102
CPU temp: 28°C Usage of /: 22% of 7.2G
What do you mean by "check the performance"?
This is an inaccurate translation. But you answered my question:
The board worked well on 6.6 and 6.7,
Can you check how it works now? After the @chraac has added its changes.
Well, on this branch, 6.6 works fine, and 6.10 works if the FDTFILE is changed to be the same as in 6.6 (I have added a hook that changed it to bigtreetech-pi in #6656 ). Ethernet works, but HDMI doesn't.
Should I commit the hook removal to this branch? It was previously done because there wasn't a usual bigtreetech-cb1 device tree before patches in this PR were added.
Well, on this branch, 6.6 works fine, and 6.10 works if the
FDTFILEis changed to be the same as in 6.6 (I have added a hook that changed it tobigtreetech-piin #6656 ). Ethernet works, but HDMI doesn't. Should I commit the hook removal to this branch? It was previously done because there wasn't a usual bigtreetech-cb1 device tree before patches in this PR were added.
I didn't understand what it was about. What kind of hook are we talking about? The commit number? 816101bc777?
Yes, that one.
What does this code do? It's not clear to me. I really don't understand what this code does.
It is a very bad practice to use any hooks for patches. Instead, it would be preferable to make a new DTS file.
It is also a very bad practice to make symbolic links in a folder for one kernel version to existing patch files in a folder for another kernel version. Patches for a single version live their own lives. They may change and then your links will stop working or your patches will not be able to be applied cleanly.
Should I commit the hook removal to this branch?
Please do this in a separate pull request. And replace symbolic links with real patch files.
What does this code do?
Change BOOT_FDT_FILE to a different one because the original plan was to go along with the mainline dts and adapt patches to build on top of it. Otherwise there would be a bunch of dts files (two from a patch, one in mainline) and only one of them used for the board. others being just a leftover junk.
Please do this in a separate pull request. And replace symbolic links with real patch files.
Will do. A PR to main armbian repo or to the branch of this PR? Just to clarify
BigTreeTech CB1 working on this branch rebased on current main (https://github.com/JohnTheCoolingFan/armbian-build/tree/main-sunxi-6.10-btt-cb1-hdmi), Ethernet works, HDMI doesn't, I don't know the reason, but it's ok for this PR and I will address that in a future PR after this gets merged. As for which device trees to use, also matter of discussion for next PRs.