vfs: get rid of old '->iterate' directory operation

All users now just use '->iterate_shared()', which only takes the
directory inode lock for reading.

Filesystems that never got convered to shared mode now instead use a
wrapper that drops the lock, re-takes it in write mode, calls the old
function, and then downgrades the lock back to read mode.

This way the VFS layer and other callers no longer need to care about
filesystems that never got converted to the modern era.

The filesystems that use the new wrapper are ceph, coda, exfat, jfs,
ntfs, ocfs2, overlayfs, and vboxsf.

Honestly, several of them look like they really could just iterate their
directories in shared mode and skip the wrapper entirely, but the point
of this change is to not change semantics or fix filesystems that
haven't been fixed in the last 7+ years, but to finally get rid of the
dual iterators.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>

Documentation/filesystems/locking.rst

@@ -551,9 +551,8 @@ mutex or just to use i_size_read() instead.
 Note: this does not protect the file->f_pos against concurrent modifications
 since this is something the userspace has to take care about.
 
-->iterate() is called with i_rwsem exclusive.
+->iterate_shared() is called with i_rwsem held for reading, and with the
-
+file f_pos_lock held exclusively
-->iterate_shared() is called with i_rwsem at least shared.
 
 ->fasync() is responsible for maintaining the FASYNC bit in filp->f_flags.
 Most instances call fasync_helper(), which does that maintenance, so it's

Documentation/filesystems/porting.rst

@@ -537,7 +537,7 @@ vfs_readdir() is gone; switch to iterate_dir() instead
 
 **mandatory**
 
-->readdir() is gone now; switch to ->iterate()
+->readdir() is gone now; switch to ->iterate_shared()
 
 **mandatory**
 
@@ -693,24 +693,19 @@ parallel now.
 
 ---
 
-**recommended**
+**mandatory**
 
-->iterate_shared() is added; it's a parallel variant of ->iterate().
+->iterate_shared() is added.
 Exclusion on struct file level is still provided (as well as that
 between it and lseek on the same struct file), but if your directory
 has been opened several times, you can get these called in parallel.
 Exclusion between that method and all directory-modifying ones is
 still provided, of course.
 
-Often enough ->iterate() can serve as ->iterate_shared() without any
+If you have any per-inode or per-dentry in-core data structures modified
-changes - it is a read-only operation, after all.  If you have any
+by ->iterate_shared(), you might need something to serialize the access
-per-inode or per-dentry in-core data structures modified by ->iterate(),
+to them.  If you do dcache pre-seeding, you'll need to switch to
-you might need something to serialize the access to them.  If you
+d_alloc_parallel() for that; look for in-tree examples.
-do dcache pre-seeding, you'll need to switch to d_alloc_parallel() for
-that; look for in-tree examples.
-
-Old method is only used if the new one is absent; eventually it will
-be removed.  Switch while you still can; the old one won't stay.
 
 ---
 
@@ -930,9 +925,9 @@ should be done by looking at FMODE_LSEEK in file->f_mode.
 filldir_t (readdir callbacks) calling conventions have changed.  Instead of
 returning 0 or -E... it returns bool now.  false means "no more" (as -E... used
 to) and true - "keep going" (as 0 in old calling conventions).  Rationale:
-callers never looked at specific -E... values anyway.  ->iterate() and
+callers never looked at specific -E... values anyway. -> iterate_shared()
-->iterate_shared() instance require no changes at all, all filldir_t ones in
+instances require no changes at all, all filldir_t ones in the tree
-the tree converted.
+converted.
 
 ---
 

fs/ceph/dir.c

@@ -2019,9 +2019,10 @@ unsigned ceph_dentry_hash(struct inode *dir, struct dentry *dn)
 	}
 }
 
+WRAP_DIR_ITER(ceph_readdir) // FIXME!
 const struct file_operations ceph_dir_fops = {
 	.read = ceph_read_dir,
-	.iterate = ceph_readdir,
+	.iterate_shared = shared_ceph_readdir,
 	.llseek = ceph_dir_llseek,
 	.open = ceph_open,
 	.release = ceph_release,
@@ -2033,7 +2034,7 @@ const struct file_operations ceph_dir_fops = {
 };
 
 const struct file_operations ceph_snapdir_fops = {
-	.iterate = ceph_readdir,
+	.iterate_shared = shared_ceph_readdir,
 	.llseek = ceph_dir_llseek,
 	.open = ceph_open,
 	.release = ceph_release,

fs/coda/dir.c

@@ -429,21 +429,14 @@ static int coda_readdir(struct file *coda_file, struct dir_context *ctx)
 	cfi = coda_ftoc(coda_file);
 	host_file = cfi->cfi_container;
 
-	if (host_file->f_op->iterate || host_file->f_op->iterate_shared) {
+	if (host_file->f_op->iterate_shared) {
 		struct inode *host_inode = file_inode(host_file);
 		ret = -ENOENT;
 		if (!IS_DEADDIR(host_inode)) {
-			if (host_file->f_op->iterate_shared) {
+			inode_lock_shared(host_inode);
-				inode_lock_shared(host_inode);
+			ret = host_file->f_op->iterate_shared(host_file, ctx);
-				ret = host_file->f_op->iterate_shared(host_file, ctx);
+			file_accessed(host_file);
-				file_accessed(host_file);
+			inode_unlock_shared(host_inode);
-				inode_unlock_shared(host_inode);
-			} else {
-				inode_lock(host_inode);
-				ret = host_file->f_op->iterate(host_file, ctx);
-				file_accessed(host_file);
-				inode_unlock(host_inode);
-			}
 		}
 		return ret;
 	}
@@ -585,10 +578,11 @@ const struct inode_operations coda_dir_inode_operations = {
 	.setattr	= coda_setattr,
 };
 
+WRAP_DIR_ITER(coda_readdir) // FIXME!
 const struct file_operations coda_dir_operations = {
 	.llseek		= generic_file_llseek,
 	.read		= generic_read_dir,
-	.iterate	= coda_readdir,
+	.iterate_shared	= shared_coda_readdir,
 	.open		= coda_open,
 	.release	= coda_release,
 	.fsync		= coda_fsync,

fs/exfat/dir.c

@@ -306,10 +306,11 @@ static int exfat_iterate(struct file *file, struct dir_context *ctx)
 	return err;
 }
 
+WRAP_DIR_ITER(exfat_iterate) // FIXME!
 const struct file_operations exfat_dir_operations = {
 	.llseek		= generic_file_llseek,
 	.read		= generic_read_dir,
-	.iterate	= exfat_iterate,
+	.iterate_shared	= shared_exfat_iterate,
 	.unlocked_ioctl = exfat_ioctl,
 #ifdef CONFIG_COMPAT
 	.compat_ioctl = exfat_compat_ioctl,

fs/exportfs/expfs.c

@@ -315,7 +315,7 @@ static int get_name(const struct path *path, char *name, struct dentry *child)
 		goto out;
 
 	error = -EINVAL;
-	if (!file->f_op->iterate && !file->f_op->iterate_shared)
+	if (!file->f_op->iterate_shared)
 		goto out_close;
 
 	buffer.sequence = 0;

fs/jfs/namei.c

@@ -1535,9 +1535,10 @@ const struct inode_operations jfs_dir_inode_operations = {
 #endif
 };
 
+WRAP_DIR_ITER(jfs_readdir) // FIXME!
 const struct file_operations jfs_dir_operations = {
 	.read		= generic_read_dir,
-	.iterate	= jfs_readdir,
+	.iterate_shared	= shared_jfs_readdir,
 	.fsync		= jfs_fsync,
 	.unlocked_ioctl = jfs_ioctl,
 	.compat_ioctl	= compat_ptr_ioctl,

fs/ntfs/dir.c

@@ -1525,10 +1525,11 @@ static int ntfs_dir_fsync(struct file *filp, loff_t start, loff_t end,
 
 #endif /* NTFS_RW */
 
+WRAP_DIR_ITER(ntfs_readdir) // FIXME!
 const struct file_operations ntfs_dir_ops = {
 	.llseek		= generic_file_llseek,	/* Seek inside directory. */
 	.read		= generic_read_dir,	/* Return -EISDIR. */
-	.iterate	= ntfs_readdir,		/* Read directory contents. */
+	.iterate_shared	= shared_ntfs_readdir,	/* Read directory contents. */
 #ifdef NTFS_RW
 	.fsync		= ntfs_dir_fsync,	/* Sync a directory to disk. */
 #endif /* NTFS_RW */

fs/ocfs2/file.c

@@ -2793,10 +2793,11 @@ const struct file_operations ocfs2_fops = {
 	.remap_file_range = ocfs2_remap_file_range,
 };
 
+WRAP_DIR_ITER(ocfs2_readdir) // FIXME!
 const struct file_operations ocfs2_dops = {
 	.llseek		= generic_file_llseek,
 	.read		= generic_read_dir,
-	.iterate	= ocfs2_readdir,
+	.iterate_shared	= shared_ocfs2_readdir,
 	.fsync		= ocfs2_sync_file,
 	.release	= ocfs2_dir_release,
 	.open		= ocfs2_dir_open,
@@ -2842,7 +2843,7 @@ const struct file_operations ocfs2_fops_no_plocks = {
 const struct file_operations ocfs2_dops_no_plocks = {
 	.llseek		= generic_file_llseek,
 	.read		= generic_read_dir,
-	.iterate	= ocfs2_readdir,
+	.iterate_shared	= shared_ocfs2_readdir,
 	.fsync		= ocfs2_sync_file,
 	.release	= ocfs2_dir_release,
 	.open		= ocfs2_dir_open,

fs/overlayfs/readdir.c

@@ -954,10 +954,11 @@ static int ovl_dir_open(struct inode *inode, struct file *file)
 	return 0;
 }
 
+WRAP_DIR_ITER(ovl_iterate) // FIXME!
 const struct file_operations ovl_dir_operations = {
 	.read		= generic_read_dir,
 	.open		= ovl_dir_open,
-	.iterate	= ovl_iterate,
+	.iterate_shared	= shared_ovl_iterate,
 	.llseek		= ovl_dir_llseek,
 	.fsync		= ovl_dir_fsync,
 	.release	= ovl_dir_release,

fs/readdir.c

@@ -24,6 +24,53 @@
 
 #include <asm/unaligned.h>
 
+/*
+ * Some filesystems were never converted to '->iterate_shared()'
+ * and their directory iterators want the inode lock held for
+ * writing. This wrapper allows for converting from the shared
+ * semantics to the exclusive inode use.
+ */
+int wrap_directory_iterator(struct file *file,
+			    struct dir_context *ctx,
+			    int (*iter)(struct file *, struct dir_context *))
+{
+	struct inode *inode = file_inode(file);
+	int ret;
+
+	/*
+	 * We'd love to have an 'inode_upgrade_trylock()' operation,
+	 * see the comment in mmap_upgrade_trylock() in mm/memory.c.
+	 *
+	 * But considering this is for "filesystems that never got
+	 * converted", it really doesn't matter.
+	 *
+	 * Also note that since we have to return with the lock held
+	 * for reading, we can't use the "killable()" locking here,
+	 * since we do need to get the lock even if we're dying.
+	 *
+	 * We could do the write part killably and then get the read
+	 * lock unconditionally if it mattered, but see above on why
+	 * this does the very simplistic conversion.
+	 */
+	up_read(&inode->i_rwsem);
+	down_write(&inode->i_rwsem);
+
+	/*
+	 * Since we dropped the inode lock, we should do the
+	 * DEADDIR test again. See 'iterate_dir()' below.
+	 *
+	 * Note that we don't need to re-do the f_pos games,
+	 * since the file must be locked wrt f_pos anyway.
+	 */
+	ret = -ENOENT;
+	if (!IS_DEADDIR(inode))
+		ret = iter(file, ctx);
+
+	downgrade_write(&inode->i_rwsem);
+	return ret;
+}
+EXPORT_SYMBOL(wrap_directory_iterator);
+
 /*
  * Note the "unsafe_put_user() semantics: we goto a
  * label for errors.
@@ -40,39 +87,28 @@
 int iterate_dir(struct file *file, struct dir_context *ctx)
 {
 	struct inode *inode = file_inode(file);
-	bool shared = false;
 	int res = -ENOTDIR;
-	if (file->f_op->iterate_shared)
+
-		shared = true;
+	if (!file->f_op->iterate_shared)
-	else if (!file->f_op->iterate)
 		goto out;
 
 	res = security_file_permission(file, MAY_READ);
 	if (res)
 		goto out;
 
-	if (shared)
+	res = down_read_killable(&inode->i_rwsem);
-		res = down_read_killable(&inode->i_rwsem);
-	else
-		res = down_write_killable(&inode->i_rwsem);
 	if (res)
 		goto out;
 
 	res = -ENOENT;
 	if (!IS_DEADDIR(inode)) {
 		ctx->pos = file->f_pos;
-		if (shared)
+		res = file->f_op->iterate_shared(file, ctx);
-			res = file->f_op->iterate_shared(file, ctx);
-		else
-			res = file->f_op->iterate(file, ctx);
 		file->f_pos = ctx->pos;
 		fsnotify_access(file);
 		file_accessed(file);
 	}
-	if (shared)
+	inode_unlock_shared(inode);
-		inode_unlock_shared(inode);
-	else
-		inode_unlock(inode);
 out:
 	return res;
 }

fs/vboxsf/dir.c

@@ -179,9 +179,10 @@ static int vboxsf_dir_iterate(struct file *dir, struct dir_context *ctx)
 	return 0;
 }
 
+WRAP_DIR_ITER(vboxsf_dir_iterate) // FIXME!
 const struct file_operations vboxsf_dir_fops = {
 	.open = vboxsf_dir_open,
-	.iterate = vboxsf_dir_iterate,
+	.iterate_shared = shared_vboxsf_dir_iterate,
 	.release = vboxsf_dir_release,
 	.read = generic_read_dir,
 	.llseek = generic_file_llseek,

include/linux/fs.h

@@ -1780,7 +1780,6 @@ struct file_operations {
 	ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
 	int (*iopoll)(struct kiocb *kiocb, struct io_comp_batch *,
 			unsigned int flags);
-	int (*iterate) (struct file *, struct dir_context *);
 	int (*iterate_shared) (struct file *, struct dir_context *);
 	__poll_t (*poll) (struct file *, struct poll_table_struct *);
 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
@@ -1817,6 +1816,13 @@ struct file_operations {
 				unsigned int poll_flags);
 } __randomize_layout;
 
+/* Wrap a directory iterator that needs exclusive inode access */
+int wrap_directory_iterator(struct file *, struct dir_context *,
+			    int (*) (struct file *, struct dir_context *));
+#define WRAP_DIR_ITER(x) \
+	static int shared_##x(struct file *file , struct dir_context *ctx) \
+	{ return wrap_directory_iterator(file, ctx, x); }
+
 struct inode_operations {
 	struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
 	const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);