I tried to downgrade to 6.18, unfortunately it froze during the process, which left the kernel and boot loader in an incomplete state. I have decided to re-image the whole thing with 6.18.x and for now it appears to be stable again.

Your previous post reminded me that the board supplier changed the BOM of the ODROID-N2+ by using a different SPI flash vendor. He hacked the support for it into his legacy firmware build without making any further note about it. It took some effort of reverse engineering to figure that out. Mainline hasn't picked up this additional driver activation to this day, but I still keep it in my builds anyway. In my latest build for the ODROID-HC4, I also included this driver to see if they might have gone about it in the same way. But your confirmation shows that this is probably not the case. Since I don't have an ODROId-HC4 with the behavior you described on hand, I can't analyze any further what the cause of it is. These days, I also mostly avoid devices with Amlogic SoCs because of their strict closed-source policy and lack of mainline support. And the board manufacturer isn't much more helpful on this point either. Devices powered by Rockchip are much more appealing objects. So you have to help yourself if you want to find a solution.

OK I see: # dmesg | grep watchdog [ 0.462093] sunxi-wdt 1c20c90.watchdog: Watchdog enabled (timeout=16 sec, nowayout=0) That would mean study 'bark' or so, forgot most of it. I guess the board should restart if IPv6 LL connection is not there for more than 1 minute or so, but in relation to NAS makes then only sense if I move from NBD to iSCSI or so.

The forum returned 'nginx bad gateway' but I see draft of may message is still there. In the meantime, I did a simple 'apt update && apt full-upgrade && reboot' and now I have: U-Boot SPL 2024.01-armbian-2024.01-S866c-P7738-Hb9d3-Vf23c-Bb703-R448a (Jun 21 2025 - 02:53:13 +0000) Kernel: Linux 6.18.35-current-sunxi I might look into HW watchdog things, no clue if it is there and able to work, but used it successfully for Raspberrypi4 to workaround USB3 lockups/crashes.

Yes, I have similar issues with 7.0.12-edge-sunxi kernel, see serial dump/log below: [ 133.382506] Internal error: Oops - undefined instruction: 0 [#1] SMP THUMB2 [ 133.389572] Modules linked in: rpcsec_gss_krb5 auth_rpcgss nfsv4 nfs lockd grace netfs sunrpc cp210x binfmt_misc usbserial lima gpu_sched nls_iso8859_1 drm_shmem_help er sun4i_gpadc_iio sunxi_cir axp20x_adc sun4i_ts sunxi_cedrus(C) v4l2_mem2mem videobuf2_dma_contig videobuf2_memops videobuf2_v4l2 videobuf2_common videodev mc display_c onnector cpufreq_dt evdev cfg80211 rfkill nfnetlink ip_tables x_tables autofs4 btrfs xor xor_neon libblake2b raid6_pq pinctrl_axp209 axp20x_usb_power industrialio sun4i_ gpadc sunxi phy_generic realtek [ 133.437258] CPU: 1 UID: 0 PID: 184 Comm: kworker/u8:5 Tainted: G C 7.0.12-edge-sunxi #2 VOLUNTARY [ 133.447746] Tainted: [C]=CRAP [ 133.450738] Hardware name: Allwinner sun7i (A20) Family [ 133.455991] Workqueue: pm pm_runtime_work [ 133.460075] PC is at __rpm_callback+0x64/0x10c [ 133.464565] LR is at rpm_callback+0x51/0x58 [ 133.468789] pc : [<c06d2e40>] lr : [<c06d2f39>] psr: 400f00b3 [ 133.475083] sp : f0a5de90 ip : 00000018 fp : c2ef5200 [ 133.480337] r10: c01761ed r9 : 000f4240 r8 : 00000008 [ 133.485590] r7 : 00000000 r6 : 00000000 r5 : c06d0aad r4 : c1e25410 [ 133.492146] r3 : 60000000 r2 : c1e25410 r1 : c1e25410 r0 : c06d0aad [ 133.498704] Flags: nZcv IRQs off FIQs on Mode SVC_32 ISA Thumb Segment none [ 133.506135] Control: 50c5387d Table: 43eb406a DAC: 00000051 [ 133.511902] Register r0 information: non-slab/vmalloc memory [ 133.517616] Register r1 information: slab kmalloc-1k start c1e25400 pointer offset 16 size 1024 [ 133.526415] Register r2 information: slab kmalloc-1k start c1e25400 pointer offset 16 size 1024 [ 133.535209] Register r3 information: non-paged memory [ 133.540302] Register r4 information: slab kmalloc-1k start c1e25400 pointer offset 16 size 1024 [ 133.549095] Register r5 information: non-slab/vmalloc memory [ 133.554795] Register r6 information: NULL pointer [ 133.559537] Register r7 information: NULL pointer [ 133.564279] Register r8 information: non-paged memory [ 133.569367] Register r9 information: non-paged memory [ 133.574455] Register r10 information: non-slab/vmalloc memory [ 133.580241] Register r11 information: slab task_struct start c2ef5200 pointer offset 0 size 2624 [ 133.589123] Register r12 information: non-paged memory [ 133.594301] Process kworker/u8:5 (pid: 184, stack limit = 0xe8b1571c) [ 133.600779] Stack: (0xf0a5de90 to 0xf0a5e000) [ 133.605178] de80: 60000000 c1e25410 c06d0aad 0000000a [ 133.613397] dea0: 00000000 00000008 000f4240 c06d2f39 c1e25410 c06d0aad 0000000a c06d3f23 [ 133.621617] dec0: c0ab03bf 00000402 c15b5180 c9306c24 c3790705 c34c3c70 c1813000 bf8e21ad [ 133.629836] dee0: c2ef5200 c1813000 01000000 569e35c7 c34f0c70 c1e254f8 c1e254f8 c1b9de00 [ 133.638054] df00: c1813000 01000000 c1b9de05 c1b72670 c1813000 c06d45a7 c2fb5900 c0141a87 [ 133.646273] df20: c3790000 569e35c7 00000000 c2ef5200 c1813000 c2fb5930 f0a5df50 c2fb5900 [ 133.654492] df40: c1813000 c2fb5930 c1603d80 c1813020 c2ef5200 c181305c c1813000 c0142345 [ 133.662711] df60: f0a5df78 c1603d80 f0a9deb8 c36b8fc0 c2ef5200 c0142245 c2fb5900 c2d35c80 [ 133.670929] df80: f0a9deb8 00000000 00000000 c0149e29 00000001 c36b8fc0 c0149d59 00000000 [ 133.679145] dfa0: 00000000 00000000 00000000 c0100155 00000000 00000000 00000000 00000000 [ 133.687359] dfc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 [ 133.695575] dfe0: 00000000 00000000 00000000 00000000 00000013 00000000 00000000 00000000 [ 133.703775] Call trace: [ 133.703812] __rpm_callback from rpm_callback+0x51/0x58 [ 133.711641] rpm_callback from rpm_suspend+0xab/0x514 [ 133.716756] rpm_suspend from pm_runtime_work+0x3f/0x70 [ 133.722045] pm_runtime_work from process_one_work+0x12f/0x390 [ 133.727947] process_one_work from worker_thread+0x101/0x254 [ 133.733667] worker_thread from kthread+0xd1/0xec [ 133.738431] kthread from ret_from_fork+0x11/0x1c [ 133.743189] Exception stack(0xf0a5dfb0 to 0xf0a5dff8) [ 133.748280] dfa0: 00000000 00000000 00000000 00000000 [ 133.756494] dfc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 [ 133.764705] dfe0: 00000000 00000000 00000000 00000000 00000013 00000000 [ 133.771365] Code: f7ff ff7c 4640 f7ef (fd27) f104 [ 133.776186] ---[ end trace 0000000000000000 ]--- [ 133.780834] note: kworker/u8:5[184] exited with irqs disabled [ 154.802648] rcu: INFO: rcu_sched detected stalls on CPUs/tasks: [ 154.808663] rcu: 0-...0: (10 ticks this GP) idle=4c4c/1/0x40000000 softirq=7570/7571 fqs=2625 [ 154.817336] rcu: (detected by 1, t=5252 jiffies, g=14613, q=22 ncpus=2) [ 154.824076] Sending NMI from CPU 1 to CPUs 0: It was 3 weeks ago I did see this while doing some maintenance,upgrades,tests,experiments. Mine is a Bananapi M1, I use sun7i-a20-bananapi.dtb as DTB, and should be same for U-Boot, although I would need to check again what version/build I use. A difference is WiFi and some others, but my guess is there is something wrong/ongoing that is getting worse over time, as also with less newer kernels, it locks up after a few days quite often. This has been the case since december 2024 kernel 6.6.x as far as I see that written in my notes. I noticed that because I have used my Bananapi as NAS (blocklevel NBD). But it were paging issues and never really could find a pattern, always random. This error message is different, but might have same root-cause. Kernel 6.16.x I have run, but it lacked AXP PMU driver for the LiPo battery powering/charging which I use every now and then (is a great UPS like that). 6.18.x had it back again, but also lockups (same paging issue, I have already a kernel log dumping 24/7 to try to figure out what is wrong, but actually have no clue). I also used 6.12.90+deb13.1-armmp, so vanilla Debian kernel, but it has no overlays available and also no AXP PMU module, so that make it less useful. I set it back to 6.12.58-current-sunxi so that at least it runs reliably for a few days and I can use it ad-hoc as a serial console terminal for a Raspberrypi4 that also used to crash randomly due to USB3 stack issues and also for debugging and testing rather complex networking issues. I am using custom partition layout, the whole thing also runs as 32-bit UEFI KVM as well as systemd-nspawn container (same cloned MAC address so I un-power the real Bananapi then). The setup uses custom extlinux and standard grub-efi. I could easily test 7.0.x or later in a KVM, but assume it will run fine. It is really something about a combination of A20, platform HW, U-Boot, DeviceTree I think. It is a pity as this SBC is about the most useful I have due to its onboard SATA and LiPo battery management, although not 64-bit.