On Fri 2023-03-31 12:42:09, Sergey Senozhatsky wrote:
> On (23/03/30 09:23), Linus Torvalds wrote:
> > Although we also do have some other issues - I think down_trylock() is
> > ok in irq contexts, but mutex_trylock() is not. Maybe that's why
> > printk uses semaphores? I forget.
>
> Yes, correct. IIRC we also cannot safely call mutex_unlock() from IRQ
> context because it takes some internal mutex spin_lock in a non-IRQ-safe
> manner. Semaphore is OK in this regard, both semaphore try_lock() and
> unlock() can be called from IRQ.
One more reason is that mutex must be released in the same context
that took it. And printk() tries to pass console_sem() to another context.
It was added by the commit dbdda842fe96 ("printk: Add console owner
and waiter logic to load balance console writes"). It was relatively
effective in reducing the risk of soft lockups.
Best Regards,
Petr
