linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jeff Layton <jlayton@kernel.org>
To: Alexander Viro <viro@zeniv.linux.org.uk>,
	 Christian Brauner <brauner@kernel.org>, Jan Kara <jack@suse.cz>,
	 Andrew Morton <akpm@linux-foundation.org>
Cc: Josef Bacik <josef@toxicpanda.com>,
	linux-fsdevel@vger.kernel.org,  linux-kernel@vger.kernel.org,
	Jeff Layton <jlayton@kernel.org>
Subject: [PATCH RFC 4/4] fs: try an opportunistic lookup for O_CREAT opens too
Date: Fri, 02 Aug 2024 17:45:05 -0400	[thread overview]
Message-ID: <20240802-openfast-v1-4-a1cff2a33063@kernel.org> (raw)
In-Reply-To: <20240802-openfast-v1-0-a1cff2a33063@kernel.org>

Today, when opening a file we'll typically do a fast lookup, but if
O_CREAT is set, the kernel always takes the exclusive inode lock. I'm
sure this was done with the expectation that O_CREAT being set means
that we expect to do the create, but that's often not the case. Many
programs set O_CREAT even in scenarios where the file already exists.

This patch rearranges the pathwalk-for-open code to also attempt a
fast_lookup in the O_CREAT case.  Have the code always do a fast_lookup
(unless O_EXCL is set), and return that without taking the inode_lock
when a positive dentry is found in the O_CREAT codepath.

Signed-off-by: Jeff Layton <jlayton@kernel.org>
---
 fs/namei.c | 43 ++++++++++++++++++++++++++++++++++++-------
 1 file changed, 36 insertions(+), 7 deletions(-)

diff --git a/fs/namei.c b/fs/namei.c
index b9bdb8e6214a..1793ed090314 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -3538,7 +3538,7 @@ static const char *open_last_lookups(struct nameidata *nd,
 	struct dentry *dir = nd->path.dentry;
 	int open_flag = op->open_flag;
 	bool got_write = false;
-	struct dentry *dentry;
+	struct dentry *dentry = NULL;
 	const char *res;
 
 	nd->flags |= op->intent;
@@ -3549,28 +3549,57 @@ static const char *open_last_lookups(struct nameidata *nd,
 		return handle_dots(nd, nd->last_type);
 	}
 
-	if (!(open_flag & O_CREAT)) {
-		if (nd->last.name[nd->last.len])
+	/*
+	 * We _can_ be in RCU mode here. For everything but O_EXCL case, do a
+	 * fast lookup for the dentry first. For O_CREAT case, we are only
+	 * interested in positive dentries. If nothing suitable is found,
+	 * fall back to locked codepath.
+	 */
+	if ((open_flag & (O_CREAT | O_EXCL)) != (O_CREAT | O_EXCL)) {
+		/* Trailing slashes? */
+		if (unlikely(nd->last.name[nd->last.len]))
 			nd->flags |= LOOKUP_FOLLOW | LOOKUP_DIRECTORY;
-		/* we _can_ be in RCU mode here */
+
 		dentry = lookup_fast(nd);
 		if (IS_ERR(dentry))
 			return ERR_CAST(dentry);
+	}
+
+	if (!(open_flag & O_CREAT)) {
 		if (likely(dentry))
 			goto finish_lookup;
 
 		if (WARN_ON_ONCE(nd->flags & LOOKUP_RCU))
 			return ERR_PTR(-ECHILD);
 	} else {
-		/* create side of things */
+		/* If negative dentry was found earlier,
+		 * discard it as we'll need to use the slow path anyway.
+		 */
 		if (nd->flags & LOOKUP_RCU) {
-			if (!try_to_unlazy(nd))
+			bool unlazied;
+
+			/* discard negative dentry if one was found */
+			if (dentry && !dentry->d_inode)
+				dentry = NULL;
+
+			unlazied = dentry ? try_to_unlazy_next(nd, dentry) :
+					    try_to_unlazy(nd);
+			if (!unlazied)
 				return ERR_PTR(-ECHILD);
+		} else if (dentry && !dentry->d_inode) {
+			/* discard negative dentry if one was found */
+			dput(dentry);
+			dentry = NULL;
 		}
 		audit_inode(nd->name, dir, AUDIT_INODE_PARENT);
+
 		/* trailing slashes? */
-		if (unlikely(nd->last.name[nd->last.len]))
+		if (unlikely(nd->last.name[nd->last.len])) {
+			dput(dentry);
 			return ERR_PTR(-EISDIR);
+		}
+		if (dentry)
+			goto finish_lookup;
 	}
 
 	if (open_flag & (O_CREAT | O_TRUNC | O_WRONLY | O_RDWR)) {

-- 
2.45.2


  parent reply	other threads:[~2024-08-02 21:45 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-08-02 21:45 [PATCH RFC 0/4] fs: try an opportunistic lookup for O_CREAT opens too Jeff Layton
2024-08-02 21:45 ` [PATCH RFC 1/4] fs: remove comment about d_rcu_to_refcount Jeff Layton
2024-08-02 21:45 ` [PATCH RFC 2/4] fs: add a kerneldoc header over lookup_fast Jeff Layton
2024-08-02 21:45 ` [PATCH RFC 3/4] lockref: rework CMPXCHG_LOOP to handle contention better Jeff Layton
2024-08-03  4:44   ` Mateusz Guzik
2024-08-03  9:09     ` Mateusz Guzik
2024-08-03 10:59       ` Jeff Layton
2024-08-03 11:21         ` Mateusz Guzik
2024-08-03 11:32           ` Jeff Layton
2024-08-05 11:44             ` Christian Brauner
2024-08-05 12:52               ` Jeff Layton
2024-08-06 11:36                 ` Christian Brauner
2024-08-03 10:55     ` Jeff Layton
2024-08-02 21:45 ` Jeff Layton [this message]
2024-08-05 10:46 ` [PATCH RFC 0/4] fs: try an opportunistic lookup for O_CREAT opens too Christian Brauner
2024-08-05 11:55   ` Jeff Layton

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=20240802-openfast-v1-4-a1cff2a33063@kernel.org \
    --to=jlayton@kernel.org \
    --cc=akpm@linux-foundation.org \
    --cc=brauner@kernel.org \
    --cc=jack@suse.cz \
    --cc=josef@toxicpanda.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@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 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).