On Wed 2017-03-29 18:25:06, Sergey Senozhatsky wrote:
> Offload printing of printk_deferred() messages from IRQ context
> to a schedulable printing kthread, when possible (the same way
> we do it in vprintk_emit()). Otherwise, console_unlock() can
> force the printing CPU to spend unbound amount of time flushing
> kernel messages from IRQ context.
>
> Signed-off-by: Sergey Senozhatsky <sergey.senozhat...@gmail.com>
> ---
> kernel/printk/printk.c | 12 ++++++++++--
> 1 file changed, 10 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
> index ab6b3b2a68c6..1927b5cb5cbe 100644
> --- a/kernel/printk/printk.c
> +++ b/kernel/printk/printk.c
> @@ -2741,8 +2741,16 @@ static void wake_up_klogd_work_func(struct irq_work
> *irq_work)
> * If trylock fails, someone else is doing the printing.
> * PRINTK_PENDING_OUTPUT bit is cleared by console_unlock().
> */
> - if (console_trylock())
> - console_unlock();
> + if (printk_kthread_enabled()) {
> + wake_up_process(printk_kthread);
Note that the relation between printk_kthread_enabled()
and wake_up_process() is racy. The conditions might change
between these two calls. It looks fine here, well almost.
The critical point is in vprintk_emit(). It must use the emergency
mode (call the consoles directly) when it is called from a process
that started the emergency mode.
We could be more relaxed here. IMHO, the only sensitive situation
is if printk_deferred() is used in the emergency context.
We might want to use the emergency mode here as well but
it is not guaranteed. printk_emergency might get cleared
in the meantime.
A solution might be to add one more bit, e.g.
PRINTK_PENDING_EMERGENCY_OUTPUT. We should force the emergency mode
here when it is set. It should be cleared together with the normal
PRINTK_PENDING_OUTPUT.
Or do you think that this is a corner case that we could
ignore for now?
Otherwise, it looks fine to me.
Best Regards,
Petr
