[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [RFC] kern: simple futex for gnumach
From: |
Marin Ramesa |
Subject: |
Re: [RFC] kern: simple futex for gnumach |
Date: |
Sun, 22 Sep 2013 10:24:25 +0200 |
On 22.09.2013 01:01:31, Samuel Thibault wrote:
> You have missed an important thing here: it's userland which provides
> the value we are supposed to check for. The idea is that there is a
> window between when userland sees the futex as seeming locked, and
> here. In the meanwhile some other thread might have unlocked the
> futex. And we don't want to hardcode in the kernel what a "locked
> futex" means. That's why it's userland which provides the value the
> kernel should check for.
>
> You can read "futexes are tricky" from Ulrich Drepper, for more
> details.
>
> > +/* Check if there is an existing futex at the address given. */
>
> Better use a hash table :)
>
> Samuel
Samuel, thank you for the comments. I will continue the work in that
direction.