Re: [tip:core/locking] x86/smp: Move waiting on contended ticketlock out of line

[Rik van Riel]

On 02/27/2013 11:49 PM, Linus Torvalds wrote:
> On Wed, Feb 27, 2013 at 8:06 PM, Davidlohr Bueso <davidlohr.bueso@xxxxxx> wrote:
> > The attached file shows how the amount of sys time used by the ipc lock
> > for a 4 and 8 socket box.
>
> I have to say, even with the improvements, that looks pretty
> disgusting. It really makes me wonder if that thing couldn't be done
> better some way. Is it the SysV semaphores that this all ends up
> using, or what?
>
> That said, I think the IPC layer is just about the perfect candidate
> for things like this, because I'm afraid that nobody is ever going to
> fix it.

There's more to it than that. Userspace expects the IPC layer
to provide exclusion and/or serialization. That makes it a
rather poor candidate for parallelism.

Btw, the IPC lock is already fairly fine grained. One ipc
lock is allocated for each set of semaphores allocated through
sys_semget. Looking up those semaphores in the namespace, when
they are used later, is done under RCU.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/