Re: [PATCH 1/2] percpu-rwsem: use synchronize_sched_expedited

[Oleg Nesterov <oleg@redhat.com>]

On 11/29, Andrew Morton wrote:
>
> On Tue, 27 Nov 2012 22:59:52 -0500 (EST)
> Mikulas Patocka <mpatocka@redhat.com> wrote:
>
> > percpu-rwsem: use synchronize_sched_expedited
> >
> > Use synchronize_sched_expedited() instead of synchronize_sched()
> > to improve mount speed.
> >
> > This patch improves mount time from 0.500s to 0.013s.
> >
> > Note: if realtime people complain about the use
> > synchronize_sched_expedited() and synchronize_rcu_expedited(), I suggest
> > that they introduce an option CONFIG_REALTIME or
> > /proc/sys/kernel/realtime and turn off these *_expedited functions if
> > the option is enabled (i.e. turn synchronize_sched_expedited into
> > synchronize_sched and synchronize_rcu_expedited into synchronize_rcu).
> >
> > Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
>
> So I read through this thread but I really didn't see a clear
> description of why mount() got slower.  The changelog for 4b05a1c74d1
> is spectacularly awful :(
>
>
> Apparently the slowdown occurred because a blockdev mount patch
> 62ac665ff9fc07497ca524 ("blockdev: turn a rw semaphore into a percpu rw
> semaphore") newly uses percpu rwsems, and percpu rwsems are slow on the
> down_write() path.
>
> And using synchronize_sched_expedited() rather than synchronize_sched()
> makes percpu_down_write() somewhat less slow.  Correct?
>
> Why is it OK to use synchronize_sched_expedited() here?  If it's
> faster, why can't we use synchronize_sched_expedited() everywhere and
> zap synchronize_sched()?

synchronize_sched_expedited() is much faster but it is heavyweight,
try_stop_cpus() affects the whole system.

We briefly discussed this possibility with Paul, personally I think it
would be better to add percpu_rw_semaphore->use_expedited or add the
percpu_down_write_expedited().

But this change was already applied. I hope we can do this later.

> Oleg, this has implications for your
> percpu_rw_semaphore-reimplement-to-not-block-the-readers-unnecessarily.patch.
> I've changed that to use synchronize_sched_expedited() everywhere -
> please have a think and a retest.

Thanks a lot Andrew. The fix looks obviously fine, but I'll try to test
anyway.

> Note that elsewhere in this thread,
> percpu_rw_semaphore-reimplement-to-not-block-the-readers-unnecessarily.patch
> was found to slow mount down again somewhat, because it adds an
> additional synchronize_sched[_expedited]() to the percpu_down_write()
> path.

Cough... this patch _removes_ one synchronize_sched[_expedited]() and
removes msleep ;) Although the main point was not-block-the-readers.

And I think we can remove another one, but this is not trivial and
percpu_free_rwsem() should be might_sleep().

Oleg.