[aarch64] INFO: rcu_sched detected expedited stalls on CPUs/tasks

[John Garry]

On 22/06/2022 15:40, Bruno Goncalves wrote:

Hi Bruno,

> With the config change that does not set
> CONFIG_RCU_EXP_CPU_STALL_TIMEOUT the problem seems to be fixed for us.
> The newer Fedora kernels already have the config fixed.
> 

OK, thanks for the info. Well those debug options I enabled didn't cause 
problems previously. I'll see if it is one in particular and go from there.

> 
>> On v5.19-rc3 I just enabled some debug configs on a vanilla kernel and
>> can easily reproduce a RCU stall on boot, as below.
>>
>> CONFIG_RCU_EXP_CPU_STALL_TIMEOUT=0 for me, that being the default.
>>
>>    Table To iBMC Success.
>> GetVariable Status : Not Found.
>> [    0.000000] Booting Linux on physical CPU 0x0000010000 [0x410fd082]
>> [    0.000000] Linux version 5.19.0-rc3-00001-gd8610c1c16e8
>> (john at debian) (gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU
>> Binutils for Debian) 2.37) #187 SMP PREEMPT Wed Jun 22 14:08:56 BST 2022
>> [    0.000000] Machine model: Hisilicon PhosphorV660 Development Board
>> [    0.000000] efi: EFI v2.60 by EDK II
>> [    0.000000] efi: SMBIOS=0x3eff0000 SMBIOS 3.0=0x39aa0000
>> ACPI=0x39b70000 ACPI 2.0=0x39b70014 MEMATTR=0x3b8d0018
>> MEMRESERVE=0x3a002d18

Thanks!