[v2] tracefs: avoid changing i_mode to a temp value

Commit Message

Sishuai Gong Aug. 18, 2023, midnight UTC

Right now inode->i_mode is updated twice to reach the desired value
in tracefs_apply_options(). Because there is no lock protecting the two
writes, other threads might read the intermediate value of inode->i_mode.

Thread-1			Thread-2
// tracefs_apply_options()	//e.g., acl_permission_check
inode->i_mode &= ~S_IALLUGO;
				unsigned int mode = inode->i_mode;
inode->i_mode |= opts->mode;

I think there is no need to introduce a lock but it is better to
only update inode->i_mode ONCE, so the readers will either see the old
or latest value, rather than an intermediate/temporary value.

Signed-off-by: Sishuai Gong <[e-mail]>
---
 fs/tracefs/inode.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

Patch

diff --git a/fs/tracefs/inode.c b/fs/tracefs/inode.c
index 57ac8aa4a724..0d49922f1127 100644
--- a/fs/tracefs/inode.c
+++ b/fs/tracefs/inode.c
@@ -290,6 +290,7 @@  static int tracefs_apply_options(struct super_block *sb, bool remount)
 	struct tracefs_fs_info *fsi = sb->s_fs_info;
 	struct inode *inode = d_inode(sb->s_root);
 	struct tracefs_mount_opts *opts = &fsi->mount_opts;
+	umode_t tmp_mode;
 
 	/*
 	 * On remount, only reset mode/uid/gid if they were provided as mount
@@ -297,8 +298,9 @@  static int tracefs_apply_options(struct super_block *sb, bool remount)
 	 */
 
 	if (!remount || opts->opts & BIT(Opt_mode)) {
-		inode->i_mode &= ~S_IALLUGO;
-		inode->i_mode |= opts->mode;
+		tmp_mode = READ_ONCE(inode->i_mode) & ~S_IALLUGO;
+		tmp_mode |= opts->mode;
+		WRITE_ONCE(inode->i_mode, tmp_mode);
 	}
 
 	if (!remount || opts->opts & BIT(Opt_uid))