From: Maksim Kulis <bokuno@picodata.io>
To: tarantool-patches@dev.tarantool.org
Subject: [Tarantool-patches] [PATCH] small: unite the oscillation cache of all mempools
Date: Wed, 29 Jan 2020 04:48:16 +0300 [thread overview]
Message-ID: <20200129014816.14248-1-bokuno@picodata.io> (raw)
The task is to unite the oscillation cache of all mempools.
In the function slab_put_with_order() we check if there is any
slab of the largest size except the current one, in order to
avoid oscillation.
https://github.com/tarantool/small/blob/master/small/slab_cache.c#L349
---
small/mempool.c | 22 ++++------------------
small/mempool.h | 6 ------
2 files changed, 4 insertions(+), 24 deletions(-)
diff --git a/small/mempool.c b/small/mempool.c
index b20a416..54bd58d 100644
--- a/small/mempool.c
+++ b/small/mempool.c
@@ -126,18 +126,9 @@ mslab_free(struct mempool *pool, struct mslab *slab, void *ptr)
}
mslab_tree_remove(&pool->hot_slabs, slab);
slab->in_hot_slabs = false;
- if (pool->spare > slab) {
- slab_list_del(&pool->slabs, &pool->spare->slab,
- next_in_list);
- slab_put_with_order(pool->cache, &pool->spare->slab);
- pool->spare = slab;
- } else if (pool->spare) {
- slab_list_del(&pool->slabs, &slab->slab,
- next_in_list);
- slab_put_with_order(pool->cache, &slab->slab);
- } else {
- pool->spare = slab;
- }
+ slab_list_del(&pool->slabs, &slab->slab,
+ next_in_list);
+ slab_put_with_order(pool->cache, &slab->slab);
}
}
@@ -153,7 +144,6 @@ mempool_create_with_order(struct mempool *pool, struct slab_cache *cache,
mslab_tree_new(&pool->hot_slabs);
pool->first_hot_slab = NULL;
rlist_create(&pool->cold_slabs);
- pool->spare = NULL;
pool->objsize = objsize;
pool->slab_order = order;
/* Total size of slab */
@@ -179,11 +169,7 @@ mempool_alloc(struct mempool *pool)
{
struct mslab *slab = pool->first_hot_slab;
if (slab == NULL) {
- if (pool->spare) {
- slab = pool->spare;
- pool->spare = NULL;
-
- } else if ((slab = (struct mslab *)
+ if ((slab = (struct mslab *)
slab_get_with_order(pool->cache,
pool->slab_order))) {
mslab_create(slab, pool);
diff --git a/small/mempool.h b/small/mempool.h
index 15d85a4..9d0e162 100644
--- a/small/mempool.h
+++ b/small/mempool.h
@@ -149,12 +149,6 @@ struct mempool
* tree is empty or the allocator runs out of memory.
*/
struct rlist cold_slabs;
- /**
- * A completely empty slab which is not freed only to
- * avoid the overhead of slab_cache oscillation around
- * a single element allocation.
- */
- struct mslab *spare;
/**
* The size of an individual object. All objects
* allocated on the pool have the same size.
--
2.17.1
next reply other threads:[~2020-01-29 1:48 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-01-29 1:48 Maksim Kulis [this message]
2020-01-29 10:28 ` Kirill Yukhin
2020-01-29 21:46 ` Konstantin Osipov
2020-01-30 8:02 ` Kirill Yukhin
2020-01-30 8:34 ` Konstantin Osipov
2020-01-30 11:18 ` Alexander Turenko
2020-01-30 12:23 ` Konstantin Osipov
2020-01-30 13:05 ` Alexander Turenko
2020-01-30 14:47 ` Konstantin Osipov
2020-01-30 12:20 ` Kirill Yukhin
2020-01-30 12:36 ` Konstantin Osipov
2020-04-10 8:36 Aleksandr Lyapunov
[not found] ` <CAPZPwLrNbdCzX7LR+_P8xrZ+unDpw-k2EP1Jj5mAsGOTWz2frA@mail.gmail.com>
2020-04-10 12:44 ` Aleksandr Lyapunov
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200129014816.14248-1-bokuno@picodata.io \
--to=bokuno@picodata.io \
--cc=tarantool-patches@dev.tarantool.org \
--subject='Re: [Tarantool-patches] [PATCH] small: unite the oscillation cache of all mempools' \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox