Re: [PATCH] locking/mutex: Disable preemption in __mutex_unlock_slowpath()

[Waiman Long <llong@redhat.com>]

On 7/9/25 2:28 PM, Linus Torvalds wrote:
> On Wed, 9 Jul 2025 at 11:21, Linus Torvalds
> <torvalds@linux-foundation.org> wrote:
>> And I suspect it would involve the exact *opposite* of your patch:
>> make mutex_unlock() actively cause preemption after it has released
>> the lock but before it has done the final accesses.
> .. sadly, I suspect we have a ton of mutex_unlock() users in atomic
> contexts, so we probably can't do that. It's not like you *should* do
> it, but I don't think we've ever disallowed it.
>
> You can't use mutex_unlock from interrupts etc, but you can use it
> while holding a spinlock.

I have just sent out another mutex patch to enforce a context switch in 
__mutex_unlock_slowpath() under the right context.

As for this one, you are right that hiding it may not be the best idea. 
So I am going to drop it.

Cheers,
Longman