[PATCH v3 06/13] vfs: make vfs_create break delegations on parent directory

Jeff Layton posted 13 patches 3 months, 2 weeks ago
There is a newer version of this series
[PATCH v3 06/13] vfs: make vfs_create break delegations on parent directory
Posted by Jeff Layton 3 months, 2 weeks ago
In order to add directory delegation support, we need to break
delegations on the parent whenever there is going to be a change in the
directory.

Add a delegated_inode parameter to vfs_create. Most callers are
converted to pass in NULL, but do_mknodat() is changed to wait for a
delegation break if there is one.

Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: NeilBrown <neil@brown.name>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
---
 fs/ecryptfs/inode.c      |  2 +-
 fs/namei.c               | 26 +++++++++++++++++++-------
 fs/nfsd/nfs3proc.c       |  2 +-
 fs/nfsd/vfs.c            |  3 +--
 fs/open.c                |  2 +-
 fs/overlayfs/overlayfs.h |  2 +-
 fs/smb/server/vfs.c      |  2 +-
 include/linux/fs.h       |  2 +-
 8 files changed, 26 insertions(+), 15 deletions(-)

diff --git a/fs/ecryptfs/inode.c b/fs/ecryptfs/inode.c
index 88631291b32535f623a3fbe4ea9b6ed48a306ca0..661709b157ce854c3bfdfdb13f7c10435fad9756 100644
--- a/fs/ecryptfs/inode.c
+++ b/fs/ecryptfs/inode.c
@@ -189,7 +189,7 @@ ecryptfs_do_create(struct inode *directory_inode,
 	rc = lock_parent(ecryptfs_dentry, &lower_dentry, &lower_dir);
 	if (!rc)
 		rc = vfs_create(&nop_mnt_idmap, lower_dir,
-				lower_dentry, mode, true);
+				lower_dentry, mode, true, NULL);
 	if (rc) {
 		printk(KERN_ERR "%s: Failure to create dentry in lower fs; "
 		       "rc = [%d]\n", __func__, rc);
diff --git a/fs/namei.c b/fs/namei.c
index 786f42bd184b5dbf6d754fa1fb6c94c0f75429f2..7510942e0249de19df4363b92f813b3acdfc2254 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -3460,11 +3460,12 @@ static inline umode_t vfs_prepare_mode(struct mnt_idmap *idmap,
 
 /**
  * vfs_create - create new file
- * @idmap:	idmap of the mount the inode was found from
- * @dir:	inode of the parent directory
- * @dentry:	dentry of the child file
- * @mode:	mode of the child file
- * @want_excl:	whether the file must not yet exist
+ * @idmap:		idmap of the mount the inode was found from
+ * @dir:		inode of the parent directory
+ * @dentry:		dentry of the child file
+ * @mode:		mode of the child file
+ * @want_excl:		whether the file must not yet exist
+ * @delegated_inode:	returns parent inode, if the inode is delegated.
  *
  * Create a new file.
  *
@@ -3475,7 +3476,8 @@ static inline umode_t vfs_prepare_mode(struct mnt_idmap *idmap,
  * raw inode simply pass @nop_mnt_idmap.
  */
 int vfs_create(struct mnt_idmap *idmap, struct inode *dir,
-	       struct dentry *dentry, umode_t mode, bool want_excl)
+	       struct dentry *dentry, umode_t mode, bool want_excl,
+	       struct inode **delegated_inode)
 {
 	int error;
 
@@ -3488,6 +3490,9 @@ int vfs_create(struct mnt_idmap *idmap, struct inode *dir,
 
 	mode = vfs_prepare_mode(idmap, dir, mode, S_IALLUGO, S_IFREG);
 	error = security_inode_create(dir, dentry, mode);
+	if (error)
+		return error;
+	error = try_break_deleg(dir, delegated_inode);
 	if (error)
 		return error;
 	error = dir->i_op->create(idmap, dir, dentry, mode, want_excl);
@@ -4365,6 +4370,7 @@ static int do_mknodat(int dfd, struct filename *name, umode_t mode,
 	struct path path;
 	int error;
 	unsigned int lookup_flags = 0;
+	struct inode *delegated_inode = NULL;
 
 	error = may_mknod(mode);
 	if (error)
@@ -4384,7 +4390,8 @@ static int do_mknodat(int dfd, struct filename *name, umode_t mode,
 	switch (mode & S_IFMT) {
 		case 0: case S_IFREG:
 			error = vfs_create(idmap, path.dentry->d_inode,
-					   dentry, mode, true);
+					   dentry, mode, true,
+					   &delegated_inode);
 			if (!error)
 				security_path_post_mknod(idmap, dentry);
 			break;
@@ -4399,6 +4406,11 @@ static int do_mknodat(int dfd, struct filename *name, umode_t mode,
 	}
 out2:
 	end_creating_path(&path, dentry);
+	if (delegated_inode) {
+		error = break_deleg_wait(&delegated_inode);
+		if (!error)
+			goto retry;
+	}
 	if (retry_estale(error, lookup_flags)) {
 		lookup_flags |= LOOKUP_REVAL;
 		goto retry;
diff --git a/fs/nfsd/nfs3proc.c b/fs/nfsd/nfs3proc.c
index ad14b34583bb9fd7cd1e29f5f8676fa3442dd661..dbf70d184c0276d198599a22eb0953a2a1dde2c8 100644
--- a/fs/nfsd/nfs3proc.c
+++ b/fs/nfsd/nfs3proc.c
@@ -344,7 +344,7 @@ nfsd3_create_file(struct svc_rqst *rqstp, struct svc_fh *fhp,
 	status = fh_fill_pre_attrs(fhp);
 	if (status != nfs_ok)
 		goto out;
-	host_err = vfs_create(&nop_mnt_idmap, inode, child, iap->ia_mode, true);
+	host_err = vfs_create(&nop_mnt_idmap, inode, child, iap->ia_mode, true, NULL);
 	if (host_err < 0) {
 		status = nfserrno(host_err);
 		goto out;
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 85afd2fad7e08b66b1a9ce372afdea1df52086be..7eaae44467188fab0909fabec986e103bcd52457 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -1639,8 +1639,7 @@ nfsd_create_locked(struct svc_rqst *rqstp, struct svc_fh *fhp,
 	err = 0;
 	switch (type) {
 	case S_IFREG:
-		host_err = vfs_create(&nop_mnt_idmap, dirp, dchild,
-				      iap->ia_mode, true);
+		host_err = vfs_create(&nop_mnt_idmap, dirp, dchild, iap->ia_mode, true, NULL);
 		if (!host_err)
 			nfsd_check_ignore_resizing(iap);
 		break;
diff --git a/fs/open.c b/fs/open.c
index 3d64372ecc675e4795eb0a0deda10f8f67b95640..4d98f8b52b98bc95e52cb247d14871ff6e4a1b5c 100644
--- a/fs/open.c
+++ b/fs/open.c
@@ -1173,7 +1173,7 @@ struct file *dentry_create(const struct path *path, int flags, umode_t mode,
 
 	error = vfs_create(mnt_idmap(path->mnt),
 			   d_inode(path->dentry->d_parent),
-			   path->dentry, mode, true);
+			   path->dentry, mode, true, NULL);
 	if (!error)
 		error = vfs_open(path, f);
 
diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h
index d215d7349489686b66bb66e939b27046f7d836f6..d3123f5d97e86b58e4c9608cf6ef2abd1fcddbcd 100644
--- a/fs/overlayfs/overlayfs.h
+++ b/fs/overlayfs/overlayfs.h
@@ -235,7 +235,7 @@ static inline int ovl_do_create(struct ovl_fs *ofs,
 				struct inode *dir, struct dentry *dentry,
 				umode_t mode)
 {
-	int err = vfs_create(ovl_upper_mnt_idmap(ofs), dir, dentry, mode, true);
+	int err = vfs_create(ovl_upper_mnt_idmap(ofs), dir, dentry, mode, true, NULL);
 
 	pr_debug("create(%pd2, 0%o) = %i\n", dentry, mode, err);
 	return err;
diff --git a/fs/smb/server/vfs.c b/fs/smb/server/vfs.c
index c5f0f3170d586cb2dc4d416b80948c642797fb82..be278bb6b71bab8aa41aed06a8806e7bc2de4cd3 100644
--- a/fs/smb/server/vfs.c
+++ b/fs/smb/server/vfs.c
@@ -189,7 +189,7 @@ int ksmbd_vfs_create(struct ksmbd_work *work, const char *name, umode_t mode)
 
 	mode |= S_IFREG;
 	err = vfs_create(mnt_idmap(path.mnt), d_inode(path.dentry),
-			 dentry, mode, true);
+			 dentry, mode, true, NULL);
 	if (!err) {
 		ksmbd_vfs_inherit_owner(work, d_inode(path.dentry),
 					d_inode(dentry));
diff --git a/include/linux/fs.h b/include/linux/fs.h
index d8bdaf7c87502ff17775602f5391d375738b4ed8..5fcf64d9cf42ce135c0fbcbf6dfbf8816ae0bcb1 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -2111,7 +2111,7 @@ bool inode_owner_or_capable(struct mnt_idmap *idmap,
  * VFS helper functions..
  */
 int vfs_create(struct mnt_idmap *, struct inode *,
-	       struct dentry *, umode_t, bool);
+	       struct dentry *, umode_t, bool, struct inode **);
 struct dentry *vfs_mkdir(struct mnt_idmap *, struct inode *,
 			 struct dentry *, umode_t, struct inode **);
 int vfs_mknod(struct mnt_idmap *, struct inode *, struct dentry *,

-- 
2.51.0
Re: [PATCH v3 06/13] vfs: make vfs_create break delegations on parent directory
Posted by Christian Brauner 3 months, 1 week ago
On Tue, Oct 21, 2025 at 11:25:41AM -0400, Jeff Layton wrote:
> In order to add directory delegation support, we need to break
> delegations on the parent whenever there is going to be a change in the
> directory.
> 
> Add a delegated_inode parameter to vfs_create. Most callers are
> converted to pass in NULL, but do_mknodat() is changed to wait for a
> delegation break if there is one.
> 
> Reviewed-by: Jan Kara <jack@suse.cz>
> Reviewed-by: NeilBrown <neil@brown.name>
> Signed-off-by: Jeff Layton <jlayton@kernel.org>
> ---
>  fs/ecryptfs/inode.c      |  2 +-
>  fs/namei.c               | 26 +++++++++++++++++++-------
>  fs/nfsd/nfs3proc.c       |  2 +-
>  fs/nfsd/vfs.c            |  3 +--
>  fs/open.c                |  2 +-
>  fs/overlayfs/overlayfs.h |  2 +-
>  fs/smb/server/vfs.c      |  2 +-
>  include/linux/fs.h       |  2 +-
>  8 files changed, 26 insertions(+), 15 deletions(-)
> 
> diff --git a/fs/ecryptfs/inode.c b/fs/ecryptfs/inode.c
> index 88631291b32535f623a3fbe4ea9b6ed48a306ca0..661709b157ce854c3bfdfdb13f7c10435fad9756 100644
> --- a/fs/ecryptfs/inode.c
> +++ b/fs/ecryptfs/inode.c
> @@ -189,7 +189,7 @@ ecryptfs_do_create(struct inode *directory_inode,
>  	rc = lock_parent(ecryptfs_dentry, &lower_dentry, &lower_dir);
>  	if (!rc)
>  		rc = vfs_create(&nop_mnt_idmap, lower_dir,
> -				lower_dentry, mode, true);
> +				lower_dentry, mode, true, NULL);

Starts to look like we should epxlore whether a struct create_args (or
some other name) similar to struct renamedata I did some years ago would
help make the code a bit more legible in the future.
Re: [PATCH v3 06/13] vfs: make vfs_create break delegations on parent directory
Posted by Jeff Layton 3 months, 1 week ago
On Wed, 2025-10-29 at 14:23 +0100, Christian Brauner wrote:
> On Tue, Oct 21, 2025 at 11:25:41AM -0400, Jeff Layton wrote:
> > In order to add directory delegation support, we need to break
> > delegations on the parent whenever there is going to be a change in the
> > directory.
> > 
> > Add a delegated_inode parameter to vfs_create. Most callers are
> > converted to pass in NULL, but do_mknodat() is changed to wait for a
> > delegation break if there is one.
> > 
> > Reviewed-by: Jan Kara <jack@suse.cz>
> > Reviewed-by: NeilBrown <neil@brown.name>
> > Signed-off-by: Jeff Layton <jlayton@kernel.org>
> > ---
> >  fs/ecryptfs/inode.c      |  2 +-
> >  fs/namei.c               | 26 +++++++++++++++++++-------
> >  fs/nfsd/nfs3proc.c       |  2 +-
> >  fs/nfsd/vfs.c            |  3 +--
> >  fs/open.c                |  2 +-
> >  fs/overlayfs/overlayfs.h |  2 +-
> >  fs/smb/server/vfs.c      |  2 +-
> >  include/linux/fs.h       |  2 +-
> >  8 files changed, 26 insertions(+), 15 deletions(-)
> > 
> > diff --git a/fs/ecryptfs/inode.c b/fs/ecryptfs/inode.c
> > index 88631291b32535f623a3fbe4ea9b6ed48a306ca0..661709b157ce854c3bfdfdb13f7c10435fad9756 100644
> > --- a/fs/ecryptfs/inode.c
> > +++ b/fs/ecryptfs/inode.c
> > @@ -189,7 +189,7 @@ ecryptfs_do_create(struct inode *directory_inode,
> >  	rc = lock_parent(ecryptfs_dentry, &lower_dentry, &lower_dir);
> >  	if (!rc)
> >  		rc = vfs_create(&nop_mnt_idmap, lower_dir,
> > -				lower_dentry, mode, true);
> > +				lower_dentry, mode, true, NULL);
> 
> Starts to look like we should epxlore whether a struct create_args (or
> some other name) similar to struct renamedata I did some years ago would
> help make the code a bit more legible in the future.

I like that idea. Let me see if I can rework this patch along those
lines. I'll plan to send an updated set.

Thanks,
-- 
Jeff Layton <jlayton@kernel.org>