From: Dave Chinner <david@fromorbit.com>
To: glommer@parallels.com
Cc: linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org,
linux-mm@kvack.org, xfs@oss.sgi.com
Subject: [PATCH 03/19] dcache: remove dentries from LRU before putting on dispose list
Date: Wed, 28 Nov 2012 10:14:30 +1100 [thread overview]
Message-ID: <1354058086-27937-4-git-send-email-david@fromorbit.com> (raw)
In-Reply-To: <1354058086-27937-1-git-send-email-david@fromorbit.com>
From: Dave Chinner <dchinner@redhat.com>
One of the big problems with modifying the way the dcache shrinker
and LRU implementation works is that the LRU is abused in several
ways. One of these is shrink_dentry_list().
Basically, we can move a dentry off the LRU onto a different list
without doing any accounting changes, and then use dentry_lru_prune()
to remove it from what-ever list it is now on to do the LRU
accounting at that point.
This makes it -really hard- to change the LRU implementation. The
use of the per-sb LRU lock serialises movement of the dentries
between the different lists and the removal of them, and this is the
only reason that it works. If we want to break up the dentry LRU
lock and lists into, say, per-node lists, we remove the only
serialisation that allows this lru list/dispose list abuse to work.
To make this work effectively, the dispose list has to be isolated
from the LRU list - dentries have to be removed from the LRU
*before* being placed on the dispose list. This means that the LRU
accounting and isolation is completed before disposal is started,
and that means we can change the LRU implementation freely in
future.
This means that dentries *must* be marked with DCACHE_SHRINK_LIST
when they are placed on the dispose list so that we don't think that
parent dentries found in try_prune_one_dentry() are on the LRU when
the are actually on the dispose list. This would result in
accounting the dentry to the LRU a second time. Hence
dentry_lru_prune() has to handle the DCACHE_SHRINK_LIST case
differently because the dentry isn't on the LRU list.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
fs/dcache.c | 73 +++++++++++++++++++++++++++++++++++++++++++++++++++--------
1 file changed, 63 insertions(+), 10 deletions(-)
diff --git a/fs/dcache.c b/fs/dcache.c
index e0c97fe..0124a84 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -330,7 +330,6 @@ static void dentry_lru_add(struct dentry *dentry)
static void __dentry_lru_del(struct dentry *dentry)
{
list_del_init(&dentry->d_lru);
- dentry->d_flags &= ~DCACHE_SHRINK_LIST;
dentry->d_sb->s_nr_dentry_unused--;
this_cpu_dec(nr_dentry_unused);
}
@@ -340,6 +339,8 @@ static void __dentry_lru_del(struct dentry *dentry)
*/
static void dentry_lru_del(struct dentry *dentry)
{
+ BUG_ON(dentry->d_flags & DCACHE_SHRINK_LIST);
+
if (!list_empty(&dentry->d_lru)) {
spin_lock(&dentry->d_sb->s_dentry_lru_lock);
__dentry_lru_del(dentry);
@@ -351,28 +352,42 @@ static void dentry_lru_del(struct dentry *dentry)
* Remove a dentry that is unreferenced and about to be pruned
* (unhashed and destroyed) from the LRU, and inform the file system.
* This wrapper should be called _prior_ to unhashing a victim dentry.
+ *
+ * Check that the dentry really is on the LRU as it may be on a private dispose
+ * list and in that case we do not want to call the generic LRU removal
+ * functions. This typically happens when shrink_dcache_sb() clears the LRU in
+ * one go and then try_prune_one_dentry() walks back up the parent chain finding
+ * dentries that are also on the dispose list.
*/
static void dentry_lru_prune(struct dentry *dentry)
{
if (!list_empty(&dentry->d_lru)) {
+
if (dentry->d_flags & DCACHE_OP_PRUNE)
dentry->d_op->d_prune(dentry);
- spin_lock(&dentry->d_sb->s_dentry_lru_lock);
- __dentry_lru_del(dentry);
- spin_unlock(&dentry->d_sb->s_dentry_lru_lock);
+ if ((dentry->d_flags & DCACHE_SHRINK_LIST))
+ list_del_init(&dentry->d_lru);
+ else {
+ spin_lock(&dentry->d_sb->s_dentry_lru_lock);
+ __dentry_lru_del(dentry);
+ spin_unlock(&dentry->d_sb->s_dentry_lru_lock);
+ }
+ dentry->d_flags &= ~DCACHE_SHRINK_LIST;
}
}
static void dentry_lru_move_list(struct dentry *dentry, struct list_head *list)
{
+ BUG_ON(dentry->d_flags & DCACHE_SHRINK_LIST);
+
spin_lock(&dentry->d_sb->s_dentry_lru_lock);
if (list_empty(&dentry->d_lru)) {
list_add_tail(&dentry->d_lru, list);
- dentry->d_sb->s_nr_dentry_unused++;
- this_cpu_inc(nr_dentry_unused);
} else {
list_move_tail(&dentry->d_lru, list);
+ dentry->d_sb->s_nr_dentry_unused--;
+ this_cpu_dec(nr_dentry_unused);
}
spin_unlock(&dentry->d_sb->s_dentry_lru_lock);
}
@@ -840,12 +855,18 @@ static void shrink_dentry_list(struct list_head *list)
}
/*
+ * The dispose list is isolated and dentries are not accounted
+ * to the LRU here, so we can simply remove it from the list
+ * here regardless of whether it is referenced or not.
+ */
+ list_del_init(&dentry->d_lru);
+
+ /*
* We found an inuse dentry which was not removed from
- * the LRU because of laziness during lookup. Do not free
- * it - just keep it off the LRU list.
+ * the LRU because of laziness during lookup. Do not free it.
*/
if (dentry->d_count) {
- dentry_lru_del(dentry);
+ dentry->d_flags &= ~DCACHE_SHRINK_LIST;
spin_unlock(&dentry->d_lock);
continue;
}
@@ -897,6 +918,8 @@ relock:
} else {
list_move_tail(&dentry->d_lru, &tmp);
dentry->d_flags |= DCACHE_SHRINK_LIST;
+ this_cpu_dec(nr_dentry_unused);
+ sb->s_nr_dentry_unused--;
spin_unlock(&dentry->d_lock);
if (!--count)
break;
@@ -910,6 +933,28 @@ relock:
shrink_dentry_list(&tmp);
}
+/*
+ * Mark all the dentries as on being the dispose list so we don't think they are
+ * still on the LRU if we try to kill them from ascending the parent chain in
+ * try_prune_one_dentry() rather than directly from the dispose list.
+ */
+static void
+shrink_dcache_list(
+ struct list_head *dispose)
+{
+ struct dentry *dentry;
+
+ rcu_read_lock();
+ list_for_each_entry_rcu(dentry, dispose, d_lru) {
+ spin_lock(&dentry->d_lock);
+ dentry->d_flags |= DCACHE_SHRINK_LIST;
+ this_cpu_dec(nr_dentry_unused);
+ spin_unlock(&dentry->d_lock);
+ }
+ rcu_read_unlock();
+ shrink_dentry_list(dispose);
+}
+
/**
* shrink_dcache_sb - shrink dcache for a superblock
* @sb: superblock
@@ -924,8 +969,16 @@ void shrink_dcache_sb(struct super_block *sb)
spin_lock(&sb->s_dentry_lru_lock);
while (!list_empty(&sb->s_dentry_lru)) {
list_splice_init(&sb->s_dentry_lru, &tmp);
+
+ /*
+ * account for removal here so we don't need to handle it later
+ * even though the dentry is no longer on the lru list.
+ */
+ this_cpu_sub(nr_dentry_unused, sb->s_nr_dentry_unused);
+ sb->s_nr_dentry_unused = 0;
+
spin_unlock(&sb->s_dentry_lru_lock);
- shrink_dentry_list(&tmp);
+ shrink_dcache_list(&tmp);
spin_lock(&sb->s_dentry_lru_lock);
}
spin_unlock(&sb->s_dentry_lru_lock);
--
1.7.10
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2012-11-27 23:14 UTC|newest]
Thread overview: 57+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-11-27 23:14 [RFC, PATCH 00/19] Numa aware LRU lists and shrinkers Dave Chinner
2012-11-27 23:14 ` [PATCH 01/19] dcache: convert dentry_stat.nr_unused to per-cpu counters Dave Chinner
2012-11-27 23:14 ` [PATCH 02/19] dentry: move to per-sb LRU locks Dave Chinner
2012-11-27 23:14 ` Dave Chinner [this message]
2012-11-27 23:14 ` [PATCH 04/19] mm: new shrinker API Dave Chinner
2012-11-27 23:14 ` [PATCH 05/19] shrinker: convert superblock shrinkers to new API Dave Chinner
2012-12-20 11:06 ` Glauber Costa
2012-12-21 1:46 ` Dave Chinner
2012-12-21 10:17 ` Glauber Costa
2012-11-27 23:14 ` [PATCH 06/19] list: add a new LRU list type Dave Chinner
2012-11-28 16:10 ` Christoph Hellwig
2012-11-27 23:14 ` [PATCH 07/19] inode: convert inode lru list to generic lru list code Dave Chinner
2012-11-27 23:14 ` [PATCH 08/19] dcache: convert to use new lru list infrastructure Dave Chinner
2012-11-27 23:14 ` [PATCH 09/19] list_lru: per-node " Dave Chinner
2012-12-20 11:21 ` Glauber Costa
2012-12-21 1:54 ` Dave Chinner
2013-01-16 19:21 ` Glauber Costa
2013-01-16 22:55 ` Dave Chinner
2013-01-17 0:35 ` Glauber Costa
2013-01-17 4:22 ` Dave Chinner
2013-01-17 18:21 ` Glauber Costa
2013-01-18 0:10 ` Dave Chinner
2013-01-18 0:14 ` Glauber Costa
2013-01-18 8:11 ` Dave Chinner
2013-01-18 19:10 ` Glauber Costa
2013-01-19 0:10 ` Dave Chinner
2013-01-19 0:13 ` Glauber Costa
2013-01-18 0:51 ` Glauber Costa
2013-01-18 8:08 ` Dave Chinner
2013-01-18 19:01 ` Glauber Costa
2012-11-27 23:14 ` [PATCH 10/19] shrinker: add node awareness Dave Chinner
2012-11-27 23:14 ` [PATCH 11/19] fs: convert inode and dentry shrinking to be node aware Dave Chinner
2012-11-27 23:14 ` [PATCH 12/19] xfs: convert buftarg LRU to generic code Dave Chinner
2012-11-27 23:14 ` [PATCH 13/19] xfs: Node aware direct inode reclaim Dave Chinner
2012-11-27 23:14 ` [PATCH 14/19] xfs: use generic AG walk for background " Dave Chinner
2012-11-27 23:14 ` [PATCH 15/19] xfs: convert dquot cache lru to list_lru Dave Chinner
2012-11-28 16:17 ` Christoph Hellwig
2012-11-27 23:14 ` [PATCH 16/19] fs: convert fs shrinkers to new scan/count API Dave Chinner
2012-11-27 23:14 ` [PATCH 17/19] drivers: convert shrinkers to new count/scan API Dave Chinner
2012-11-28 1:13 ` Chris Wilson
2012-11-28 3:17 ` Dave Chinner
2012-11-28 8:21 ` Glauber Costa
2012-11-28 21:28 ` Dave Chinner
2012-11-29 10:29 ` Glauber Costa
2012-11-29 22:02 ` Dave Chinner
2013-06-07 13:37 ` Konrad Rzeszutek Wilk
2012-11-27 23:14 ` [PATCH 18/19] shrinker: convert remaining shrinkers to " Dave Chinner
2012-11-27 23:14 ` [PATCH 19/19] shrinker: Kill old ->shrink API Dave Chinner
2012-11-29 19:02 ` [RFC, PATCH 00/19] Numa aware LRU lists and shrinkers Andi Kleen
2012-11-29 22:09 ` Dave Chinner
2012-12-20 11:45 ` Glauber Costa
2012-12-21 2:50 ` Dave Chinner
2012-12-21 10:41 ` Glauber Costa
2013-01-21 16:08 ` Glauber Costa
2013-01-21 23:21 ` Dave Chinner
2013-01-23 14:36 ` Glauber Costa
2013-01-23 23:46 ` Dave Chinner
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=1354058086-27937-4-git-send-email-david@fromorbit.com \
--to=david@fromorbit.com \
--cc=glommer@parallels.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=xfs@oss.sgi.com \
/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).