Re: [PATCH printk v2 12/18] printk: Add kthread for all legacy consoles

[Petr Mladek <pmladek@suse.com>]

On Mon 2024-07-01 16:50:31, Petr Mladek wrote:
> On Fri 2024-06-28 16:17:13, John Ogness wrote:
> > On 2024-06-28, Petr Mladek <pmladek@suse.com> wrote:
> > > On Fri 2024-06-28 14:28:47, John Ogness wrote:
> > >> On 2024-06-28, Petr Mladek <pmladek@suse.com> wrote:
> > >> >> @@ -1494,7 +1508,9 @@ void nbcon_cpu_emergency_exit(void)
> > >> >>  		 * to handle it.
> > >> >>  		 */
> > >> >>  		do_trigger_flush = true;
> > >> >> -		if (printing_via_unlock && !is_printk_deferred()) {
> > >> >> +		if (!force_printkthreads() &&
> > >> >
> > >> > Is this correct? We still need to flush the messages directly
> > >> > when the legacy kthread is not ready yet.
> > >> 
> > >> No! If force_printkthreads() is set, messages will never flush directly
> > >> except for console_flush_on_panic().
> > >
> > > But this is an _emergency_ context! This would be inconsistent with
> > > the nbcon consoles where the messages are flushed directly.
> > >
> > > Is RT sheduling quaranteed when the system reached emergency state?
> > 
> > No.
> > 
> > > In fact, we probably should not check force_printkthreads() here at
> > > all. It would be only for NBCON_PRIO_NORMAL context.
> > 
> > On PREEMPT_RT, legacy consoles are not allowed to print from
> > non-preemptible contexts because they use spinlocks (rtmutexes).
> 
> Ah, I see.
> 
> > This is a disadvantage for legacy consoles on PREEMPT_RT. If people want
> > a legacy console to gain the reliability attribute on PREEMPT_RT, then
> > that console _must_ be converted to nbcon.
> 
> Makes sense.
> 
> Thanks a lot for explanation.

It is clear that the legacy consoles could not be flushed directly in
nbcon_cpu_emergency_exit() with PREEMPT_RT. I thought whether we
could/should allow this for other preemption modes.

My opinion. Let's keep it simple and behave the same in all preemption
modes. force_printkthreads() is enabled only in PREEMPT_RT by default.
Users would get the "expected" legacy behavior in other preemption
modes by default.

By other words, let's keep "force_printkthreads()" to really simulate
RT behavior also in other preemption modes.

That said, I still suggest to rename the variable/function to
"force_legacy_thread". It does not affect the behavior of
the kthreads for nbcon consoles.

Best Regards,
Petr