Re: pagefaults and hang with 5.15.11

[Rolf Eike Beer <eike-kernel@sf-tec.de>]

[-- Attachment #1: Type: text/plain, Size: 1304 bytes --]

Am Sonntag, 26. Dezember 2021, 18:22:12 CET schrieb John David Anglin:
> On 2021-12-26 11:21 a.m., Rolf Eike Beer wrote:
> > [139181.966881] WARNING: CPU: 1 PID: 0 at kernel/rcu/tree.c:613
> > rcu_eqs_enter.constprop.0+0x8c/0x98
> This is probably not reproducible. You might try this change from Sven

At least this time the git testsuite has finished, but with some errors as 
usual.

> diff --git a/arch/parisc/kernel/smp.c b/arch/parisc/kernel/smp.c
> index cf92ece20b75..0cd97fa004c5 100644
> --- a/arch/parisc/kernel/smp.c
> +++ b/arch/parisc/kernel/smp.c
> @@ -228,11 +228,13 @@ static inline void
>   send_IPI_allbutself(enum ipi_message_type op)
>   {
>          int i;
> -
> +
> +       preempt_disable();
>          for_each_online_cpu(i) {
>                  if (i != smp_processor_id())
>                          send_IPI_single(i, op);
>          }
> +       preempt_enable();
>   }
> 
>   #ifdef CONFIG_KGDB

I'll add this and see what happens.

> and my "[PATCH v3] parisc: Rewrite light-weight syscall and futex code"
> change. Page faults in the LWS code can mess up scheduling.

But that would be nothing new. At least the machine has been quite stable in 
the last time.

> I haven't found 5.15.11 to be stable.

Ok, so I'll keep this in the queue for hppa and try again later.

Eike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 195 bytes --]