Andrea

@SteeMan this is supposed to be a remote and unattended system, so I would like to control the updates very carefully in order to avoid breaking anything using apt (it happens sometimes...). I know this is quite a challenge, but frameworks like rauc and swupdate are purposely designed to guarantee atomically and safe system upgrades. I would disable apt unattended system upgrades, protect the rootfs using overlayroot and possibly use rauc to upgrade the systems using a full image. I am also evaluating tools like systemd-repart that may repartition the storage (eMMC, or SD) live.

Hi all, has anyone ever tried to use an atomic firmware upgrade system like rauc or swupdate? https://rauc.readthedocs.io/en/latest/ https://sbabic.github.io/swupdate/swupdate.html I am playing with rauc, in order to do that I have to customize the partitioning of the SD/eMMC: - root1 (4GB) - root2 (4GB) - data (8GB, or remaining space) Can someone suggest how to customize partitioning in such way? I played around setting USE_HOOK_FOR_PARTITION=yes, and hooking the create_partition_table but had no luck. Any hint? Thanks in advance! Andrea

Thank you @going, I will evaluate this.

Is it possible just to rebuild the kernel, without fetching kernel updates and reapplying patches? I would find it useful to develop kernel drivers like in a development cycle.

Any success with this? I have the same question

A custom designed board around H616.

Hi all, is it possible to flash the a board's eMMC via USB from another PC? I already checked this but could not figure out exactly how. Thanks in advance, kind regards

Thanks for the advice!

Hi all, Armbian v23.11 build fails while compiling ATF, reporting weird "cannot find cr" message: https://paste.next.armbian.com/evewilodap Can you please help? Thanks in advance!

Charge pump is a very simple topology of DC-DC switching converter, normally driven by SoC PWMs. I have checked the DT yaml, but could not find anything useful. I found a PWM driver for Allwinner H6, but it does not work (I get error -22, EINVAL when loading the driver module).

Is it possible to rebuild the kernel without reapplying all the patches?

Hi all, I would like to use the PWM1 in order to drive a charge pump, but I am struggling to enable PWM using the DTB on Armbian v23.11. Anyone succeeded in configuring PWMs in H616? Thanks in advance!

I confirm that using an updated kernel (6.7.3) the build succeds. Linux armbian 6.7.3-060703-generic #202401311935 SMP PREEMPT_DYNAMIC Thu Feb 1 00:46:47 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux Thank you very much!

Despite the fact the problem is very similar, I am not using WSL2 to build. I am using a Ubuntu VM : Ubuntu 22.04.3 LTS (GNU/Linux 5.15.0-94-generic x86_64) andrea@armbian:~$ uname -a Linux armbian 5.15.0-94-generic #104-Ubuntu SMP Tue Jan 9 15:25:40 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux User adeepn confirms that it happens also on Ubuntu w/kernel 5.15. I will try to upgrade to 6.1 and let you know. Thanks for the hint

Hi @Werner, yes of course: https://paste.next.armbian.com/pakisigigu

Hello all, I am receiving this error while building the image: Error: Partition(s) 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64 on /dev/loop0 have been written, but we have been unable to inform the kernel of the change, probably because it/they are in use. As a result, the old partition(s) will remain in use. You should reboot now before making further changes. I am . This happens in the partitioning phase (lib/functions/image/partitioning.sh line 229, using branch "v23.11" commit bba856bb57823556e12fa180284c7490fdd3708d😞 run_host_command_logged partprobe "${LOOP}" Investigated a lot, tried to unmont the image, losetup -d the loop device but nothing helped. Can you please help me? Kind regards, Andrea

Hi @Igor, thanks for the hint. I thought PREFER_DOCKER=yes was the default, however I will give it a try. Host OS is Fedora 39, kernel 6.5.12-300.fc39.x86_64 and qemu version 2:8.1.1-1.fc39. Kind regards

Hi all, I am trying to bring up a new board that uses both Allwinner H616 emacs, the 1Gbps (emac0) and the 10/100Mbps w/internal phy (emac1). I brought in all patches found about that, but it looks like the emac1 does not work properly. It does not reset, this is what I found in the dmesg: [ 2.228695] dwmac-sun8i 5030000.ethernet: EMAC reset timeout As per the datasheet, the reg EMAC_BASIC_CTL1's LSB (SOFT_RST bit) is a bit that must set to 1 to reset, then automatically resets to 0 when the emac has finished the reset procedure. In my case, it remains 1 causing the kernel to unload emac1 driver. Anyone succeded in support the 10/100Mbps internal phy? Thanks in advance!

@ag123 thanks for the hint, I will give it a try event though I am running Fedora. However the build seems to fail while installing packages, not building the image.

@Werner you are absolutely right. https://paste.next.armbian.com/odemumewas

Hi @christop, sorry for the late reply. I had run into the finest details of the build script: PACKAGE_LIST_ADDITIONAL env var has been removed; PACKAGE_LIST_BOARD may work, but only when it's not set into the board .conf file (as it would override the environment variable set while calling compile.sh); the only reliable way is to use EXTRA_PACKAGES_IMAGE (it will require EXTRA_PACKAGES_IMAGE_REFS to be added as well). Example: EXTRA_PACKAGES_IMAGE="btrfs-progs net-tools screen devmem2" EXTRA_PACKAGES_IMAGE_REFS="build:compile.sh:0 build:compile.sh:0 build:compile.sh:0 build:compile.sh:0"

You may use the "overlayroot" package: all writes to the / root fs are volatile as they're made into RAM. You may use EEPROM or eMMC to store non volatile data.

Hi there, I am trying to build the image using docker, but I get several "Segmentation fault (core dumped)" error messages and as result the build gets interrupted. Would like to stick to docker build (I am running of fedora 39, and it automatically selects docker), as it looks to be far more efficient rather than running it into a VM. Any idea on how could I get rid of this issue? Thanks is advance! Andrea

Hi @jesus2k21 , sorry for the late reply. I came to the same conclusions by other means, but yes it was the way to go!

Hi all, I need to customize the /etc/fstab during the build to mount a mtd partition and also to override some mountpoints with tmpfs in order to reduce wearing on mmc. So, have added the following lines to customize-image.sh: cat <<EOT >> /etc/fstab # Reduce SD wearing by storing log files into ram (no need as overlayroot is used to freeze all the rootfs) tmpfs /tmp tmpfs defaults,noatime,nosuid,nodev 0 0 tmpfs /run tmpfs defaults,noatime,nosuid,nodev 0 0 tmpfs /var/tmp tmpfs defaults,noatime,nosuid,nodev 0 0 tmpfs /var/log tmpfs defaults,noatime,nosuid,nodev,noexec 0 0 tmpfs /var/lib/logrotate tmpfs defaults,noatime,nosuid,nodev,noexec,size=1m,mode=0755 0 0 tmpfs /var/lib/sudo tmpfs defaults,noatime,nosuid,nodev,noexec,size=1m,mode=0700 0 0 # MTD EEPROM memory dev/mtdblock0 /mnt/mtd jffs2 ro,relatime 0 0 EOT However changes in /etc/fstab are overridden somewhere over the build process, and this is the resulting /etc/fstab: UUID=c8167755-b557-4288-9109-9108bc48dd94 / btrfs defaults,noatime,commit=600 0 1 UUID=27ac6cc6-eff3-4736-a7bd-5517e3b150c0 /boot ext4 defaults,commit=600,errors=remount-ro 0 2 tmpfs /tmp tmpfs defaults,nosuid 0 0 Is there a way to safely patch the /etc/fstab file? Thanks very much! Andrea