Skip to contentSkip to conversation

sysfs: fix race between readdir and lseek

PATCH New Ben Hutchings 1 participant 2 messages Apr 7, 2013 1 message not in the archive

Subsystems (2)
lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw

PATCH 21/74 sysfs: fix race between readdir and lseek

3.2.43-rc1 review patch. If anyone has any objections, please let me know.

------------------

From: Ming Lei <[email protected]>

commit 991f76f837bf22c5bb07261cfd86525a0a96650c upstream.

While readdir() is running, lseek() may set filp->f_pos as zero, then may leave filp->private_data pointing to one sysfs_dirent object without holding its reference counter, so the sysfs_dirent object may be used after free in next readdir().

This patch holds inode->i_mutex to avoid the problem since the lock is always held in readdir path.

[bwh: Backported to 3.2: open-code file_inode() which we don't have]
---
 fs/sysfs/dir.c |   13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)
fs/sysfs/dir.c +12 -1
@@ -1023,10 +1023,21 @@ static int sysfs_readdir(struct file * f
 	return 0;
 }
 
+static loff_t sysfs_dir_llseek(struct file *file, loff_t offset, int whence)
+{
+	struct inode *inode = file->f_path.dentry->d_inode;
+	loff_t ret;
+
+	mutex_lock(&inode->i_mutex);
+	ret = generic_file_llseek(file, offset, whence);
+	mutex_unlock(&inode->i_mutex);
+
+	return ret;
+}
 
 const struct file_operations sysfs_dir_operations = {
 	.read		= generic_read_dir,
 	.readdir	= sysfs_readdir,
 	.release	= sysfs_dir_release,
-	.llseek		= generic_file_llseek,
+	.llseek		= sysfs_dir_llseek,
 };
#2 lore raw

PATCH 22/74 sysfs: handle failure path correctly for readdir()

In reply to a message not in the archive

3.2.43-rc1 review patch. If anyone has any objections, please let me know.

------------------

From: Ming Lei <[email protected]>

commit e5110f411d2ee35bf8d202ccca2e89c633060dca upstream.

In case of 'if (filp->f_pos == 0 or 1)' of sysfs_readdir(), the failure from filldir() isn't handled, and the reference counter of the sysfs_dirent object pointed by filp->private_data will be released without clearing filp->private_data, so use after free bug will be triggered later.

This patch returns immeadiately under the situation for fixing the bug, and it is reasonable to return from readdir() when filldir() fails.

---
 fs/sysfs/dir.c |    4 ++++
 1 file changed, 4 insertions(+)
fs/sysfs/dir.c +4 -0
@@ -985,6 +985,8 @@ static int sysfs_readdir(struct file * f
 		ino = parent_sd->s_ino;
 		if (filldir(dirent, ".", 1, filp->f_pos, ino, DT_DIR) == 0)
 			filp->f_pos++;
+		else
+			return 0;
 	}
 	if (filp->f_pos == 1) {
 		if (parent_sd->s_parent)
@@ -993,6 +995,8 @@ static int sysfs_readdir(struct file * f
 			ino = parent_sd->s_ino;
 		if (filldir(dirent, "..", 2, filp->f_pos, ino, DT_DIR) == 0)
 			filp->f_pos++;
+		else
+			return 0;
 	}
 	mutex_lock(&sysfs_mutex);
 	for (pos = sysfs_dir_pos(ns, parent_sd, filp->f_pos, pos);