2013-10-28 10:03:18

by Alexander Gordeev

Subject: [PATCH 2/5] percpu_ida: Move waking up waiters out of atomic contexts

Currently percpu_ida_free() waikes up waiters always with local
interrupts disabled and sometimes with pool->lock held. Yet, it
does not appear there is any reason why it could not be done out
of these atomic contexts.

Signed-off-by: Alexander Gordeev <[email protected]>
---
lib/percpu_ida.c | 11 +++++++----
1 files changed, 7 insertions(+), 4 deletions(-)

diff --git a/lib/percpu_ida.c b/lib/percpu_ida.c
index 96cfacf..9dd8741 100644
--- a/lib/percpu_ida.c
+++ b/lib/percpu_ida.c
@@ -223,6 +223,7 @@ void percpu_ida_free(struct percpu_ida *pool, unsigned tag)
struct percpu_ida_cpu *tags;
unsigned long flags;
unsigned nr_free;
+ bool wake_up = false;

BUG_ON(tag >= pool->nr_tags);

@@ -241,8 +242,8 @@ void percpu_ida_free(struct percpu_ida *pool, unsigned tag)
* Pairs with smp_rmb() in steal_tags()
*/
smp_wmb();
- wake_up(&pool->wait);
- }
+ wake_up = true;
+ }

if (nr_free == pool->percpu_max_size) {
spin_lock(&pool->lock);
@@ -255,13 +256,15 @@ void percpu_ida_free(struct percpu_ida *pool, unsigned tag)
move_tags(pool->freelist, &pool->nr_free,
tags->freelist, &tags->nr_free,
pool->percpu_batch_size);
-
- wake_up(&pool->wait);
+ wake_up = true;
}
spin_unlock(&pool->lock);
}

local_irq_restore(flags);
+
+ if (wake_up)
+ wake_up(&pool->wait);
}
EXPORT_SYMBOL_GPL(percpu_ida_free);

--
1.7.7.6


--
Regards,
Alexander Gordeev
[email protected]

2013-10-28 21:21:44

by Kent Overstreet

Subject: Re: [PATCH 2/5] percpu_ida: Move waking up waiters out of atomic contexts

On Mon, Oct 28, 2013 at 11:05:25AM +0100, Alexander Gordeev wrote:
> Currently percpu_ida_free() waikes up waiters always with local
> interrupts disabled and sometimes with pool->lock held. Yet, it
> does not appear there is any reason why it could not be done out
> of these atomic contexts.

This should be a noticable performance boost, nested irqsave/restore is painful.

Acked-by: Kent Overstreet <[email protected]>