Frequent `rcu: INFO: rcu_sched detected stalls on CPUs/tasks` events logged.

[Myron]

This started happening since accepting the kernel when I used apt to upgrade the BananaPi Pro sbc I've got.

 

armbianmonitor report:

https://paste.armbian.com/izuboducul

 

@teknoid are you aware of this?

[Banana.31415]

I can confirm this issue on an old BananaPi 1 from Lemaker. It began with Kernel 6.6.

After returning to the older 6.1 the stalls disappeared.

 

Apr 13 23:02:02 eisbaer kernel: rcu: INFO: rcu_sched self-detected stall on CPU
Apr 13 23:02:02 eisbaer kernel: rcu:         0-....: (5249 ticks this GP) idle=56dc/1/0x40000002 softirq=5480265/5480265 fqs=2606
Apr 13 23:02:02 eisbaer kernel: rcu:         (t=5250 jiffies g=9317845 q=1167 ncpus=2)
Apr 13 23:02:02 eisbaer kernel: CPU: 0 PID: 27606 Comm: htop Tainted: G         C         6.6.16-current-sunxi #1
Apr 13 23:02:02 eisbaer kernel: Hardware name: Allwinner sun7i (A20) Family
Apr 13 23:02:02 eisbaer kernel: PC is at stmmac_get_stats64+0x26/0x128
Apr 13 23:02:02 eisbaer kernel: LR is at 0xc2e6b000
Apr 13 23:02:02 eisbaer kernel: pc : [<c078c5ce>]    lr : [<c2e6b000>]    psr: 80010033
Apr 13 23:02:02 eisbaer kernel: sp : f1621c68  ip : c2e68000  fp : 00000001
Apr 13 23:02:02 eisbaer kernel: r10: f1621e68  r9 : c2ee3c48  r8 : c2e68000
Apr 13 23:02:02 eisbaer kernel: r7 : 00000000  r6 : 00000001  r5 : 00000000  r4 : 80000000
Apr 13 23:02:02 eisbaer kernel: r3 : 07c697cd  r2 : c2e6ae48  r1 : f1621d28  r0 : c2e68000
Apr 13 23:02:02 eisbaer kernel: Flags: Nzcv  IRQs on  FIQs on  Mode SVC_32  ISA Thumb  Segment none
Apr 13 23:02:02 eisbaer kernel: Control: 50c5387d  Table: 46a2c06a  DAC: 00000051
Apr 13 23:02:02 eisbaer kernel:  stmmac_get_stats64 from dev_get_stats+0x27/0xd0
Apr 13 23:02:02 eisbaer kernel:  dev_get_stats from dev_seq_printf_stats+0x21/0x124
Apr 13 23:02:02 eisbaer kernel:  dev_seq_printf_stats from dev_seq_show+0x11/0x24
Apr 13 23:02:02 eisbaer kernel:  dev_seq_show from seq_read_iter+0x281/0x35c
Apr 13 23:02:02 eisbaer kernel:  seq_read_iter from seq_read+0x61/0x84
Apr 13 23:02:02 eisbaer kernel:  seq_read from proc_reg_read+0x71/0x90
Apr 13 23:02:02 eisbaer kernel:  proc_reg_read from vfs_read+0x75/0x1e4
Apr 13 23:02:02 eisbaer kernel:  vfs_read from ksys_read+0x45/0x9c
Apr 13 23:02:02 eisbaer kernel:  ksys_read from ret_fast_syscall+0x1/0x5c
Apr 13 23:02:02 eisbaer kernel: Exception stack(0xf1621fa8 to 0xf1621ff0)
Apr 13 23:02:02 eisbaer kernel: 1fa0:                   01141aa0 000005e8 00000004 011c7200 00000400 00000001
Apr 13 23:02:02 eisbaer kernel: 1fc0: 01141aa0 000005e8 b6e13888 00000003 0000000a bedcddd4 00000000 00000000
Apr 13 23:02:02 eisbaer kernel: 1fe0: 00000003 bedcdcd0 b6dae2bb b6d27616

 

[Fabitanker]

Hello!
I have a Banana-Pi M1 (Allwinner A20 SoC), where the same issue occurred since March 22nd, 2024.
I upgraded from 6.1.63-current-sunxi to 6.6.16-current-sunxi at this date.
I could observe in the logs how after some booted time the sysstat-collect services started by systemd took 20-40 seconds to complete.

Normally it takes <100ms to do so, and after some time the cpu stalls completely, and the system becomes unresponsive if not rebooted.

After reboot, everything fime for some hours, then issue occurs again.

Switched to armbian edge (6.7.4-sunxi) kernel, issue persisted.

Switched to armbian legacy (6.1.77-sunxi) kernel, issue disappeared.

I hope the culprit can be found out, i am willing to assist in error reporting, since my system is non-critical.

 

Thanks for the error report, this way i know i am not the only one!