diff mbox series

[03/10] VFS: move want_write checks into lookup_hash_update()

Message ID 166147984374.25420.3477094952897986387.stgit@noble.brown (mailing list archive)
State New, archived
Headers show
Series Improve scalability of directory operations | expand

Commit Message

NeilBrown Aug. 26, 2022, 2:10 a.m. UTC
mnt_want_write() is always called before lookup_hash_update(), so
we can simplify the code by moving the call into that function.

If lookup_hash_update() succeeds, it now will have claimed the
want_write lock.  If it fails, the want_write lock isn't held.

To allow this, lookup_hash_update() now receives a 'struct path *'
instead of 'struct dentry *'

Note that when creating a name, any error from mnt_want_write() does not
get reported unless there is no other error.  For unlink/rmdir though,
an error from mnt_want_write() is immediately fatal - overriding ENOENT
for example.  This behaviour seems strange, but this patch is careful to
preserve it.

Signed-off-by: NeilBrown <neilb@suse.de>
---
 fs/namei.c |   72 ++++++++++++++++++++++++++----------------------------------
 1 file changed, 31 insertions(+), 41 deletions(-)

Comments

Al Viro Aug. 27, 2022, 3:48 a.m. UTC | #1
On Fri, Aug 26, 2022 at 12:10:43PM +1000, NeilBrown wrote:
> mnt_want_write() is always called before lookup_hash_update(), so
> we can simplify the code by moving the call into that function.
> 
> If lookup_hash_update() succeeds, it now will have claimed the
> want_write lock.  If it fails, the want_write lock isn't held.
> 
> To allow this, lookup_hash_update() now receives a 'struct path *'
> instead of 'struct dentry *'
> 
> Note that when creating a name, any error from mnt_want_write() does not
> get reported unless there is no other error.  For unlink/rmdir though,
> an error from mnt_want_write() is immediately fatal - overriding ENOENT
> for example.  This behaviour seems strange, but this patch is careful to
> preserve it.

Would be a nice idea, if not for the mess with LOOKUP_SILLY_RENAME later
in the series...
diff mbox series

Patch

diff --git a/fs/namei.c b/fs/namei.c
index 09c2d007814a..73c3319a1703 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -1632,12 +1632,22 @@  static struct dentry *__lookup_hash(const struct qstr *name,
  */
 static struct dentry *lookup_hash_update(
 	const struct qstr *name,
-	struct dentry *base, unsigned int flags,
+	struct path *path, unsigned int flags,
 	wait_queue_head_t *wq)
 {
 	struct dentry *dentry;
+	struct dentry *base = path->dentry;
 	struct inode *dir = base->d_inode;
-	int err;
+	int err, err2;
+
+	/* For create, don't fail immediately if it's r/o,
+	 * at least try to report other errors.
+	 * For unlink/rmdir where LOOKUP_REVAl is the only
+	 * flag, fail immediately if r/o.
+	 */
+	err2 = mnt_want_write(path->mnt);
+	if (err2 && (flags & ~LOOKUP_REVAL) == 0)
+		return ERR_PTR(err2);
 
 	if (wq && IS_PAR_UPDATE(dir))
 		inode_lock_shared_nested(dir, I_MUTEX_PARENT);
@@ -1677,6 +1687,13 @@  static struct dentry *lookup_hash_update(
 			goto out_err;
 		}
 	}
+	if (err2) {
+		err = err2;
+		d_lookup_done(dentry);
+		d_unlock_update(dentry);
+		dput(dentry);
+		goto out_err;
+	}
 	return dentry;
 
 out_err:
@@ -1684,6 +1701,8 @@  static struct dentry *lookup_hash_update(
 		inode_unlock_shared(dir);
 	else
 		inode_unlock(dir);
+	if (!err2)
+		mnt_drop_write(path->mnt);
 	return ERR_PTR(err);
 }
 
@@ -1700,7 +1719,7 @@  struct dentry *lookup_hash_update_len(const char *name, int nlen,
 				    path->dentry, nlen, &this);
 	if (err)
 		return ERR_PTR(err);
-	return lookup_hash_update(&this, path->dentry, flags, wq);
+	return lookup_hash_update(&this, path, flags, wq);
 }
 EXPORT_SYMBOL(lookup_hash_update_len);
 
@@ -3891,15 +3910,10 @@  static struct dentry *filename_create_one(struct qstr *last, struct path *path,
 					  unsigned int lookup_flags,
 					  wait_queue_head_t *wq)
 {
-	struct dentry *dentry;
 	bool want_dir = lookup_flags & LOOKUP_DIRECTORY;
 	unsigned int reval_flag = lookup_flags & LOOKUP_REVAL;
 	unsigned int create_flag = LOOKUP_CREATE;
-	int err2;
-	int error;
 
-	/* don't fail immediately if it's r/o, at least try to report other errors */
-	err2 = mnt_want_write(path->mnt);
 	/*
 	 * Do the final lookup.  Suppress 'create' if there is a trailing
 	 * '/', and a directory wasn't requested.
@@ -3910,24 +3924,9 @@  static struct dentry *filename_create_one(struct qstr *last, struct path *path,
 		 * or -EEXIST.
 		 */
 		create_flag = 0;
-	dentry = lookup_hash_update(last, path->dentry,
-				    reval_flag | create_flag | LOOKUP_EXCL, wq);
-	if (IS_ERR(dentry))
-		goto drop_write;
-
-	if (unlikely(err2)) {
-		error = err2;
-		goto fail;
-	}
-	return dentry;
-fail:
-	done_path_update(path, dentry, !!wq);
-	dput(dentry);
-	dentry = ERR_PTR(error);
-drop_write:
-	if (!err2)
-		mnt_drop_write(path->mnt);
-	return dentry;
+	return lookup_hash_update(last, path,
+				  reval_flag | create_flag | LOOKUP_EXCL,
+				  wq);
 }
 
 struct dentry *filename_create_one_len(const char *name, int nlen,
@@ -4300,23 +4299,18 @@  int do_rmdir(int dfd, struct filename *name)
 		goto exit2;
 	}
 
-	error = mnt_want_write(path.mnt);
-	if (error)
-		goto exit2;
-
-	dentry = lookup_hash_update(&last, path.dentry, lookup_flags, &wq);
+	dentry = lookup_hash_update(&last, &path, lookup_flags, &wq);
 	error = PTR_ERR(dentry);
 	if (IS_ERR(dentry))
-		goto exit3;
+		goto exit2;
 	error = security_path_rmdir(&path, dentry);
 	if (error)
-		goto exit4;
+		goto exit3;
 	mnt_userns = mnt_user_ns(path.mnt);
 	error = vfs_rmdir(mnt_userns, path.dentry->d_inode, dentry);
-exit4:
+exit3:
 	done_path_update(&path, dentry, true);
 	dput(dentry);
-exit3:
 	mnt_drop_write(path.mnt);
 exit2:
 	path_put(&path);
@@ -4433,12 +4427,8 @@  int do_unlinkat(int dfd, struct filename *name)
 	if (type != LAST_NORM)
 		goto exit2;
 
-	error = mnt_want_write(path.mnt);
-	if (error)
-		goto exit2;
-
 retry_deleg:
-	dentry = lookup_hash_update(&last, path.dentry, lookup_flags, &wq);
+	dentry = lookup_hash_update(&last, &path, lookup_flags, &wq);
 	error = PTR_ERR(dentry);
 	if (!IS_ERR(dentry)) {
 		struct user_namespace *mnt_userns;
@@ -4457,6 +4447,7 @@  int do_unlinkat(int dfd, struct filename *name)
 exit3:
 		done_path_update(&path, dentry, true);
 		dput(dentry);
+		mnt_drop_write(path.mnt);
 	}
 	if (inode)
 		iput(inode);	/* truncate the inode here */
@@ -4466,7 +4457,6 @@  int do_unlinkat(int dfd, struct filename *name)
 		if (!error)
 			goto retry_deleg;
 	}
-	mnt_drop_write(path.mnt);
 exit2:
 	path_put(&path);
 	if (retry_estale(error, lookup_flags)) {