Re: [RFC][PATCH v3 1/2] printk: Make printk() completely async

[Sergey Senozhatsky]

On (03/14/16 08:11), Jan Kara wrote:
[..]
> > +	if (!sync_print) {
> > +		if (printk_thread && !in_panic) {
> > +			/*
> > +			 * Wakeup the printing kthread and offload printing
> > +			 * to a schedulable context.
> > +			 */
> > +			wake_up(&printing_wait);
> > +		} else if (in_sched) {
> > +			/*
> > +			 * @in_sched messages may come too early, when we don't
> > +			 * yet have @printk_thread. 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));
> 
> You cannot call wake_up() from a scheduler context - i.e., if in_sched is
> true, you have to use irq work unconditionally. You can wake up kthread
> from irq work or just directly print the message there. Otherwise the patch
> looks fine.

oh, thanks for catching this! you are right.

	-ss