Re: [PATCH] printk: Correctly handle preemption in console_unlock()

[Sergey Senozhatsky]

On (01/16/17 22:26), Sergey Senozhatsky wrote:
> On (01/16/17 13:48), Petr Mladek wrote:
> [..]
> > The async printk code looks like this:
> > 
> > vprintk_emit(...)
> > {
> > 
> > 
> > 		if (can_printk_async()) {
> > 			/* Offload printing to a schedulable context. */
> > 			printk_kthread_need_flush_console = true;
> > 			wake_up_process(printk_kthread);
> > 		} else {
> > 			/*
> > 			 * Try to acquire and then immediately release the
> > 			 * console semaphore.  The release will print out
> > 			 * buffers and wake up /dev/kmsg and syslog() users.
> > 			 */
> > 			if (console_trylock())
> > 				console_unlock();
> > 		}
> > 
> > So, there is still the console_trylock() for the sync mode. Or do I
> > see an outdated variant?
> 
> no, it doesn't.

I meant that none of this is happening in vprintk_emit(). there is no
console_sem in vprintk_emit() at all. eveything is done in deferred printk.

	-ss