From: Waiman Long <Waiman.Long@hpe.com>
To: "Theodore Ts'o" <tytso@mit.edu>,
Andreas Dilger <adilger.kernel@dilger.ca>,
Tejun Heo <tj@kernel.org>, Christoph Lameter <cl@linux.com>
Cc: linux-ext4@vger.kernel.org, linux-kernel@vger.kernel.org,
Scott J Norton <scott.norton@hpe.com>,
Douglas Hatch <doug.hatch@hpe.com>,
Toshimitsu Kani <toshi.kani@hpe.com>,
Waiman Long <Waiman.Long@hpe.com>
Subject: [PATCH v2 4/4] ext4: Make cache hits/misses per-cpu counts
Date: Fri, 8 Apr 2016 12:16:22 -0400 [thread overview]
Message-ID: <1460132182-11690-5-git-send-email-Waiman.Long@hpe.com> (raw)
In-Reply-To: <1460132182-11690-1-git-send-email-Waiman.Long@hpe.com>
This patch changes the es_stats_cache_hits and es_stats_cache_misses
statistics counts to per-cpu variables to reduce cacheline contention
issues whem multiple threads are trying to update those counts
simultaneously. It uses the new per-cpu stats APIs provided by the
percpu_stats.h header file.
With a 38-threads fio I/O test with 2 shared files (on DAX-mount
NVDIMM) running on a 4-socket Haswell-EX server with 4.6-rc1 kernel,
the aggregated bandwidths before and after the patch were:
Test W/O patch With patch % change
---- --------- ---------- --------
Read-only 10173MB/s 16141MB/s +58.7%
Read-write 2830MB/s 4315MB/s +52.5%
Signed-off-by: Waiman Long <Waiman.Long@hpe.com>
---
fs/ext4/extents_status.c | 22 +++++++++++++---------
fs/ext4/extents_status.h | 11 +++++++++--
2 files changed, 22 insertions(+), 11 deletions(-)
diff --git a/fs/ext4/extents_status.c b/fs/ext4/extents_status.c
index e38b987..bfa8d63 100644
--- a/fs/ext4/extents_status.c
+++ b/fs/ext4/extents_status.c
@@ -825,9 +825,9 @@ out:
es->es_pblk = es1->es_pblk;
if (!ext4_es_is_referenced(es1))
ext4_es_set_referenced(es1);
- stats->es_stats_cache_hits++;
+ percpu_stats_inc(&stats->es_stats, es_stats_cache_hits);
} else {
- stats->es_stats_cache_misses++;
+ percpu_stats_inc(&stats->es_stats, es_stats_cache_misses);
}
read_unlock(&EXT4_I(inode)->i_es_lock);
@@ -1113,9 +1113,9 @@ int ext4_seq_es_shrinker_info_show(struct seq_file *seq, void *v)
seq_printf(seq, "stats:\n %lld objects\n %lld reclaimable objects\n",
percpu_counter_sum_positive(&es_stats->es_stats_all_cnt),
percpu_counter_sum_positive(&es_stats->es_stats_shk_cnt));
- seq_printf(seq, " %lu/%lu cache hits/misses\n",
- es_stats->es_stats_cache_hits,
- es_stats->es_stats_cache_misses);
+ seq_printf(seq, " %llu/%llu cache hits/misses\n",
+ percpu_stats_sum(&es_stats->es_stats, es_stats_cache_hits),
+ percpu_stats_sum(&es_stats->es_stats, es_stats_cache_misses));
if (inode_cnt)
seq_printf(seq, " %d inodes on list\n", inode_cnt);
@@ -1142,8 +1142,6 @@ int ext4_es_register_shrinker(struct ext4_sb_info *sbi)
sbi->s_es_nr_inode = 0;
spin_lock_init(&sbi->s_es_lock);
sbi->s_es_stats.es_stats_shrunk = 0;
- sbi->s_es_stats.es_stats_cache_hits = 0;
- sbi->s_es_stats.es_stats_cache_misses = 0;
sbi->s_es_stats.es_stats_scan_time = 0;
sbi->s_es_stats.es_stats_max_scan_time = 0;
err = percpu_counter_init(&sbi->s_es_stats.es_stats_all_cnt, 0, GFP_KERNEL);
@@ -1153,15 +1151,20 @@ int ext4_es_register_shrinker(struct ext4_sb_info *sbi)
if (err)
goto err1;
+ err = percpu_stats_init(&sbi->s_es_stats.es_stats, es_stats_cnt, 0);
+ if (err)
+ goto err2;
+
sbi->s_es_shrinker.scan_objects = ext4_es_scan;
sbi->s_es_shrinker.count_objects = ext4_es_count;
sbi->s_es_shrinker.seeks = DEFAULT_SEEKS;
err = register_shrinker(&sbi->s_es_shrinker);
if (err)
- goto err2;
+ goto err3;
return 0;
-
+err3:
+ percpu_stats_destroy(&sbi->s_es_stats.es_stats);
err2:
percpu_counter_destroy(&sbi->s_es_stats.es_stats_shk_cnt);
err1:
@@ -1173,6 +1176,7 @@ void ext4_es_unregister_shrinker(struct ext4_sb_info *sbi)
{
percpu_counter_destroy(&sbi->s_es_stats.es_stats_all_cnt);
percpu_counter_destroy(&sbi->s_es_stats.es_stats_shk_cnt);
+ percpu_stats_destroy(&sbi->s_es_stats.es_stats);
unregister_shrinker(&sbi->s_es_shrinker);
}
diff --git a/fs/ext4/extents_status.h b/fs/ext4/extents_status.h
index f7aa24f..c163189 100644
--- a/fs/ext4/extents_status.h
+++ b/fs/ext4/extents_status.h
@@ -11,6 +11,8 @@
#ifndef _EXT4_EXTENTS_STATUS_H
#define _EXT4_EXTENTS_STATUS_H
+#include <linux/percpu_stats.h>
+
/*
* Turn on ES_DEBUG__ to get lots of info about extent status operations.
*/
@@ -67,10 +69,15 @@ struct ext4_es_tree {
struct extent_status *cache_es; /* recently accessed extent */
};
+enum ext4_es_stat_type {
+ es_stats_cache_hits,
+ es_stats_cache_misses,
+ es_stats_cnt,
+};
+
struct ext4_es_stats {
unsigned long es_stats_shrunk;
- unsigned long es_stats_cache_hits;
- unsigned long es_stats_cache_misses;
+ struct percpu_stats es_stats;
u64 es_stats_scan_time;
u64 es_stats_max_scan_time;
struct percpu_counter es_stats_all_cnt;
--
1.7.1
prev parent reply other threads:[~2016-04-08 16:16 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-04-08 16:16 [PATCH v2 0/4] ext4: Improve parallel I/O performance on NVDIMM Waiman Long
2016-04-08 16:16 ` [PATCH v2 1/4] percpu_stats: Simple per-cpu statistics count helper functions Waiman Long
2016-04-08 16:49 ` kbuild test robot
2016-04-08 17:45 ` kbuild test robot
2016-04-08 16:16 ` [PATCH v2 2/4] percpu_stats: Enable 64-bit counts in 32-bit architectures Waiman Long
2016-04-08 16:47 ` Tejun Heo
2016-04-08 17:32 ` Waiman Long
2016-04-08 17:46 ` Tejun Heo
2016-04-08 18:45 ` Waiman Long
2016-04-11 22:17 ` Tejun Heo
2016-04-12 18:15 ` Waiman Long
2016-04-08 16:16 ` [PATCH v2 3/4] ext4: Pass in DIO_SKIP_DIO_COUNT flag if inode_dio_begin() called Waiman Long
2016-04-08 16:16 ` Waiman Long [this message]
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=1460132182-11690-5-git-send-email-Waiman.Long@hpe.com \
--to=waiman.long@hpe.com \
--cc=adilger.kernel@dilger.ca \
--cc=cl@linux.com \
--cc=doug.hatch@hpe.com \
--cc=linux-ext4@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=scott.norton@hpe.com \
--cc=tj@kernel.org \
--cc=toshi.kani@hpe.com \
--cc=tytso@mit.edu \
/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
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).