All of lore.kernel.org
 help / color / mirror / Atom feed
From: "J. Bruce Fields" <bfields@redhat.com>
To: linux-fsdevel@vger.kernel.org
Cc: Christoph Hellwig <hch@infradead.org>,
	Al Viro <viro@ZenIV.linux.org.uk>,
	linux-nfs@vger.kernel.org, "J. Bruce Fields" <bfields@redhat.com>
Subject: [PATCH 8/8] exportfs: fix quadratic behavior in filehandle lookup
Date: Fri, 25 Oct 2013 16:30:05 -0400	[thread overview]
Message-ID: <1382733005-6006-9-git-send-email-bfields@redhat.com> (raw)
In-Reply-To: <1382733005-6006-1-git-send-email-bfields@redhat.com>

From: "J. Bruce Fields" <bfields@redhat.com>

Suppose we're given the filehandle for a directory whose closest
ancestor in the dcache is its Nth ancestor.

The main loop in reconnect_path searches for an IS_ROOT ancestor of
target_dir, reconnects that ancestor to its parent, then recommences the
search for an IS_ROOT ancestor from target_dir.

This behavior is quadratic in N.  And there's really no need to restart
the search from target_dir each time: once a directory has been looked
up, it won't become IS_ROOT again.  So instead of starting from
target_dir each time, we can continue where we left off.

This simplifies the code and improves performance on very deep directory
heirachies.  (I can't think of any reason anyone should need heirarchies
a hundred or more deep, but the performance improvement may be valuable
if only to limit damage in case of abuse.)

Signed-off-by: J. Bruce Fields <bfields@redhat.com>
---
 fs/exportfs/expfs.c |   66 ++++++++++-----------------------------------------
 1 file changed, 13 insertions(+), 53 deletions(-)

diff --git a/fs/exportfs/expfs.c b/fs/exportfs/expfs.c
index b33b9c4..48a359d 100644
--- a/fs/exportfs/expfs.c
+++ b/fs/exportfs/expfs.c
@@ -69,27 +69,6 @@ find_acceptable_alias(struct dentry *result,
 	return NULL;
 }
 
-/*
- * Find root of a disconnected subtree and return a reference to it.
- */
-static struct dentry *
-find_disconnected_root(struct dentry *dentry)
-{
-	dget(dentry);
-	while (!IS_ROOT(dentry)) {
-		struct dentry *parent = dget_parent(dentry);
-
-		if (!(parent->d_flags & DCACHE_DISCONNECTED)) {
-			dput(parent);
-			break;
-		}
-
-		dput(dentry);
-		dentry = parent;
-	}
-	return dentry;
-}
-
 static bool dentry_connected(struct dentry *dentry)
 {
 	dget(dentry);
@@ -225,45 +204,26 @@ out_reconnected:
 static int
 reconnect_path(struct vfsmount *mnt, struct dentry *target_dir, char *nbuf)
 {
-	int err = -ESTALE;
+	struct dentry *dentry, *parent;
 
-	while (target_dir->d_flags & DCACHE_DISCONNECTED) {
-		struct dentry *dentry = find_disconnected_root(target_dir);
+	dentry = dget(target_dir);
 
+	while (dentry->d_flags & DCACHE_DISCONNECTED) {
 		BUG_ON(dentry == mnt->mnt_sb->s_root);
 
-		if (!IS_ROOT(dentry)) {
-			/* must have found a connected parent - great */
-			clear_disconnected(target_dir);
-			dput(dentry);
+		if (IS_ROOT(dentry))
+			parent = reconnect_one(mnt, dentry, nbuf);
+		else
+			parent = dget_parent(dentry);
+
+		if (!parent)
 			break;
-		} else {
-			struct dentry *parent;
-			/*
-			 * We have hit the top of a disconnected path, try to
-			 * find parent and connect.
-			 */
-			 parent = reconnect_one(mnt, dentry, nbuf);
-			 if (!parent)
-				goto out_reconnected;
-			if (IS_ERR(parent)) {
-				err = PTR_ERR(parent);
-				break;
-			}
-			dput(parent);
-		}
 		dput(dentry);
+		if (IS_ERR(parent))
+			return PTR_ERR(parent);
+		dentry = parent;
 	}
-
-	if (target_dir->d_flags & DCACHE_DISCONNECTED) {
-		/* something went wrong - oh-well */
-		if (!err)
-			err = -ESTALE;
-		return err;
-	}
-
-	return 0;
-out_reconnected:
+	dput(dentry);
 	clear_disconnected(target_dir);
 	return 0;
 }
-- 
1.7.9.5


  parent reply	other threads:[~2013-10-25 20:30 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-10-25 20:29 [PATCH 0/8] simplify reconnecting dentries looked up by filehandle (v2) J. Bruce Fields
2013-10-25 20:29 ` J. Bruce Fields
2013-10-25 20:29 ` [PATCH 1/8] exportfs: BUG_ON in crazy corner case J. Bruce Fields
2013-10-25 20:29 ` [PATCH 2/8] exportfs: more detailed comment for path_reconnect J. Bruce Fields
2013-10-25 20:29   ` J. Bruce Fields
2013-10-25 20:30 ` [PATCH 3/8] exportfs: clear DISCONNECTED on all parents sooner J. Bruce Fields
2013-10-25 20:30   ` J. Bruce Fields
2013-10-25 20:30 ` [PATCH 4/8] exportfs: stop retrying once we race with rename/remove J. Bruce Fields
2013-10-25 20:30   ` J. Bruce Fields
2013-10-27  7:04   ` NeilBrown
2013-10-27  7:04     ` NeilBrown
2013-10-28  8:53     ` Christoph Hellwig
2013-10-28  8:53       ` Christoph Hellwig
2013-10-28 15:08     ` J. Bruce Fields
2013-10-25 20:30 ` [PATCH 5/8] exportfs: eliminate unused "noprogress" counter J. Bruce Fields
2013-10-25 20:30 ` [PATCH 6/8] exportfs: move most of reconnect_path to helper function J. Bruce Fields
2013-10-25 20:30 ` [PATCH 7/8] exportfs: better variable name J. Bruce Fields
2013-10-25 20:30   ` J. Bruce Fields
2013-10-25 20:30 ` J. Bruce Fields [this message]
2013-10-30 16:10 ` [PATCH 0/8] simplify reconnecting dentries looked up by filehandle (v2) Christoph Hellwig
2013-10-30 16:10   ` Christoph Hellwig
2013-10-30 16:38   ` J. Bruce Fields
2013-10-30 16:38     ` J. Bruce Fields

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=1382733005-6006-9-git-send-email-bfields@redhat.com \
    --to=bfields@redhat.com \
    --cc=hch@infradead.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-nfs@vger.kernel.org \
    --cc=viro@ZenIV.linux.org.uk \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.