Helios64 Preempt signal leads to system shutdown

Hi,

I've got a Helios64 device running with OpenMediaVault 6.9.2-1 (Shaitan).

kobol@helios64:~$ uname -a
Linux helios64 6.1.50-current-rockchip64 #3 SMP PREEMPT Wed Aug 30 14:11:13 UTC 2023 aarch64 GNU/Linux

It has been stable until I recently install openmediavault-photoprism 6.0.9-1 plugin provided by OMV6 through a podman container. I want to use PhotoPrism to help categorize my photos.

The issue is my device systematically reboots during photo folder initial scan: the scan actually loads the system a lot, and in this state of activities, a reboot occurs after about 30 minutes to 4 hours.

I first thought of a CPU temperature issue, it seems stabilized at about 64°C. But then, I caught the last journal entry before SSH disconnection:

Oct 06 14:17:37 helios64 kernel: rcu: INFO: rcu_preempt detected stalls on CPUs/tasks:
Oct 06 14:17:37 helios64 kernel: rcu:         4-...0: (0 ticks this GP) idle=5b4c/1/0x4000000000000000 softirq=564112/564112 fqs=3016
Oct 06 14:17:37 helios64 kernel:         (detected by 1, t=15005 jiffies, g=827657, q=216 ncpus=6)
Oct 06 14:17:37 helios64 kernel: Task dump for CPU 4:
Oct 06 14:17:37 helios64 kernel: task:photoprism      state:R  running task     stack:0     pid:4761  ppid:4235   flags:0x00000802
Oct 06 14:17:37 helios64 kernel: Call trace:
Oct 06 14:17:37 helios64 kernel:  __switch_to+0xf0/0x170
Oct 06 14:17:37 helios64 kernel:  0xffff80001186bca8
Oct 06 14:17:37 helios64 kernel: rcu: INFO: rcu_preempt detected stalls on CPUs/tasks:
Oct 06 14:17:37 helios64 kernel: rcu:         4-...0: (0 ticks this GP) idle=5b4c/1/0x4000000000000000 softirq=564112/564112 fqs=12017
Oct 06 14:17:37 helios64 kernel:         (detected by 3, t=60010 jiffies, g=827657, q=445 ncpus=6)
Oct 06 14:17:37 helios64 kernel: Task dump for CPU 4:
Oct 06 14:17:37 helios64 kernel: task:photoprism      state:R  running task     stack:0     pid:4761  ppid:4235   flags:0x00000802
Oct 06 14:17:37 helios64 kernel: Call trace:
Oct 06 14:17:37 helios64 kernel:  __switch_to+0xf0/0x170
Oct 06 14:17:37 helios64 kernel:  0xffff80001186bca8
Oct 06 14:17:37 helios64 kernel: rcu: INFO: rcu_preempt detected stalls on CPUs/tasks:
Oct 06 14:17:37 helios64 kernel: rcu:         4-...0: (0 ticks this GP) idle=5b4c/1/0x4000000000000000 softirq=564112/564112 fqs=21018
Oct 06 14:17:37 helios64 kernel:         (detected by 3, t=105015 jiffies, g=827657, q=681 ncpus=6)
Oct 06 14:17:37 helios64 kernel: Task dump for CPU 4:
Oct 06 14:17:37 helios64 kernel: task:photoprism      state:R  running task     stack:0     pid:4761  ppid:4235   flags:0x00000802
Oct 06 14:17:37 helios64 kernel: Call trace:
Oct 06 14:17:37 helios64 kernel:  __switch_to+0xf0/0x170
Oct 06 14:17:37 helios64 kernel:  0xffff80001186bca8

So, I draw a link (possibly wrong) between PhotoPrism preempting the computing power, and the device shuting down.

The system then starts again on watchdog signal and boots without issue.

My last test was to have the system running with PhotoPrism plugin active, however without scanning activities. At this low workload, the system didn't reboot for the 4 days I let it.

Have you got any clue on how I can confirm such preempt signal may cause a system shutdown?

Edited October 6, 20233 yr by Trillien