Re: [RFC][PATCH v5 1/2] printk: Make printk() completely async

[Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>]

On (03/21/16 09:06), Byungchul Park wrote:
> On Sun, Mar 20, 2016 at 11:13:10PM +0900, Sergey Senozhatsky wrote:
[..]
> > +	if (!sync_print) {
> > +		if (in_sched) {
> > +			/*
> > +			 * @in_sched messages may come too early, when we don't
> > +			 * yet have @printk_kthread. We can't print deferred
> > +			 * messages directly, because this may deadlock, route
> > +			 * them via IRQ context.
> > +			 */
> > +			__this_cpu_or(printk_pending,
> > +					PRINTK_PENDING_OUTPUT);
> > +			irq_work_queue(this_cpu_ptr(&wake_up_klogd_work));
> > +		} else if (printk_kthread && !in_panic) {
> > +			/* Offload printing to a schedulable context. */
> > +			wake_up_process(printk_kthread);
> 
> It will not print the "lockup suspected" message at all, for e.g. rq->lock,
> p->pi_lock and any locks which are used within wake_up_process().

this will switch to old SYNC printk() mode should such a lockup ever
happen, which is a giant advantage over any other implementation; doing
wake_up_process() within the 'we can detect recursive printk() here'
gives us better control.

why
  printk()->IRQ->wake_up_process()->spin_dump()->printk()->IRQ->wake_up_process()->spin_dump()->printk()->IRQ...
is better?


> Furtheremore, any printk() within wake_up_process() cannot work at all, as
> well.

there is printk_deferred() which has LOGLEVEL_SCHED and which must be used
in sched functions.

	-ss