3 ms·
> you've moved the fast path of the semaphore implementation into user space. I see now why your original comment was a bit inflammatory. I should have been mo
by preshing 12y ago
> you've moved the fast path of the semaphore implementation into user space.
I see now why your original comment was a bit inflammatory. I should have been more clear in the post that by "lightweight", I meant exactly that: "fast path in user space". I guess not everyone shares this vocabulary. I'll improve the post.
You're right that this lightweight mutex is a semaphore, of course. But not every semaphore is a lightweight mutex. So the technique isn't pointless.