Skip to contentSkip to conversation

problem with duplicate sysfs directories and files

DISCUSSION Greg KH 3 participants 7 messages Mar 8, 2006 – Mar 10, 2006

Subsystems (2)
lore
Share & tools
lore

Greg KH

author
#0 lore raw

Hi,

I spent some time tonight trying to track down how to fix the issue of duplicate sysfs files and/or directories. This happens when you try to create a kobject with the same name in the same directory. The creation of the second kobject will fail, but the directory will remain in sysfs.

Now I know this isn't a normal operation, but it would be good to fix this eventually. I traced the issue down to fs/sysfs/dir.c:create_dir() and the check for:

	if (error && (error != -EEXIST)) {

Problem is, error is set to -EEXIST, so we don't clean up properly. Now I know we can't just not check for this, as if you do that error cleanup, the original kobject's sysfs entry gets very messed up (ls -l does not like it at all...)

But I can't seem to figure out what exactly we need to do to clean up properly here.

Do you, or anyone else, have any pointers or ideas?

thanks,

greg k-h

#1 lore raw
Quoted 18 lines · Greg KH

Hi,

I spent some time tonight trying to track down how to fix the issue of duplicate sysfs files and/or directories. This happens when you try to create a kobject with the same name in the same directory. The creation of the second kobject will fail, but the directory will remain in sysfs.

Now I know this isn't a normal operation, but it would be good to fix this eventually. I traced the issue down to fs/sysfs/dir.c:create_dir() and the check for:

	if (error && (error != -EEXIST)) {

Problem is, error is set to -EEXIST, so we don't clean up properly. Now I know we can't just not check for this, as if you do that error cleanup, the original kobject's sysfs entry gets very messed up (ls -l does not like it at all...)

Quoted 5 lines

But I can't seem to figure out what exactly we need to do to clean up properly here.

Do you, or anyone else, have any pointers or ideas?

Emit a loud warning and don't bother cleaning up - leave the current behaviour as-is. Whatever takes the least amount code and has the minimum end-user impact, IMO.

author
#2 lore raw
Quoted 22 lines · Andrew Morton
Quoted 21 lines · Greg KH

Hi,

I spent some time tonight trying to track down how to fix the issue of duplicate sysfs files and/or directories. This happens when you try to create a kobject with the same name in the same directory. The creation of the second kobject will fail, but the directory will remain in sysfs.

Now I know this isn't a normal operation, but it would be good to fix this eventually. I traced the issue down to fs/sysfs/dir.c:create_dir() and the check for:

	if (error && (error != -EEXIST)) {

Problem is, error is set to -EEXIST, so we don't clean up properly. Now I know we can't just not check for this, as if you do that error cleanup, the original kobject's sysfs entry gets very messed up (ls -l does not like it at all...)

But I can't seem to figure out what exactly we need to do to clean up properly here.

Quoted 6 lines
Quoted 2 lines

Do you, or anyone else, have any pointers or ideas?

Emit a loud warning and don't bother cleaning up - leave the current behaviour as-is. Whatever takes the least amount code and has the minimum end-user impact, IMO.

Yeah, we do give a warning. Well, we used to always do it, but now that more people are not using kobject_register() but only kobject_add(), it doesn't always show up...

Here's a patch that I've added to my queue that fixes the warn issue, but it still would be nice to fix up sysfs someday.

thanks,

greg k-h

---
 lib/kobject.c |   22 ++++++++++++++--------
 1 file changed, 14 insertions(+), 8 deletions(-)
lib/kobject.c +14 -8
@@ -194,6 +194,17 @@ int kobject_add(struct kobject * kobj)
 		unlink(kobj);
 		if (parent)
 			kobject_put(parent);
+
+		/* be noisy on error issues */
+		if (error == -EEXIST)
+			printk("kobject_add failed for %s with -EEXIST, "
+			       "don't try to register things with the "
+			       "same name in the same directory.\n",
+			       kobject_name(kobj));
+		else
+			printk("kobject_add failed for %s (%d)\n",
+			       kobject_name(kobj), error);
+		dump_stack();
 	}
 
 	return error;
@@ -207,18 +218,13 @@ int kobject_add(struct kobject * kobj)
 
 int kobject_register(struct kobject * kobj)
 {
-	int error = 0;
+	int error = -EINVAL;
 	if (kobj) {
 		kobject_init(kobj);
 		error = kobject_add(kobj);
-		if (error) {
-			printk("kobject_register failed for %s (%d)\n",
-			       kobject_name(kobj),error);
-			dump_stack();
-		} else
+		if (!error)
 			kobject_uevent(kobj, KOBJ_ADD);
-	} else
-		error = -EINVAL;
+	}
 	return error;
 }
 
#3 lore raw
In reply to Greg KH
Quoted 7 lines · Greg KH

Hi,

I spent some time tonight trying to track down how to fix the issue of duplicate sysfs files and/or directories. This happens when you try to create a kobject with the same name in the same directory. The creation of the second kobject will fail, but the directory will remain in sysfs.

Let me understand this. Lets say we have sysfs directory tree like

	/sys/a/b/c

and someone is trying to create one more kobject with name "c" for the parent kobject "b" ?

And are you saying that though the new creation fails but the existing directory remains in sysfs? I think failing the new creation and leaving the exisiting directoy is ok. But there is sysfs_dirent leakage which does need fixing.

Quoted 10 lines

Now I know this isn't a normal operation, but it would be good to fix this eventually. I traced the issue down to fs/sysfs/dir.c:create_dir() and the check for:

	if (error && (error != -EEXIST)) {

Problem is, error is set to -EEXIST, so we don't clean up properly. Now I know we can't just not check for this, as if you do that error cleanup, the original kobject's sysfs entry gets very messed up (ls -l does not like it at all...)

Quoted 5 lines

But I can't seem to figure out what exactly we need to do to clean up properly here.

Do you, or anyone else, have any pointers or ideas?

If you are talking about the example above, to me it appears that except a possible sysfs_dirent leakage, we are some what ok, else there would have been more catastrophic results because of the duplicate directory dentry/inode.

As per the current code

static int create_dir(struct kobject * k, struct dentry * p,
const char * n, struct dentry ** d)

{
        int error;
        umode_t mode = S_IFDIR| S_IRWXU | S_IRUGO | S_IXUGO;

        mutex_lock(&p->d_inode->i_mutex);

        *d = lookup_one_len(n, p, strlen(n));

^^^^ lookup_one_len() will return the existing dentry corresponding to
the last component "c" in "/sys/a/b/c" without any error. Just note that VFS is not going to allocate a new dentry for it. The existing dentry's ref count will be increased by one.

if (!IS_ERR(*d)) {
error = sysfs_make_dirent(p->d_fsdata, *d, k, mode, SYSFS_DIR);

^^^^ we do have problem here, a new sysfs_dirent is allocated which
replaces the existing dentry->d_fsdata and yuk... the old sysfs_dirent is no more linked with the existing directory, there by leaking one sysfs_dirent.

Not only for sysfs_create_dir(), I think the problem of existing sysfs_dirent is also there with sysfs_add_file() and sysfs_add_link(). I am working on a patch to plug this leak.

Thanks
Maneesh

Signature · 52 lines
Maneesh Soni
Linux Technology Center, 
IBM India Software Labs,
Bangalore, India
email: [email protected]
Phone: 91-80-51776416





o Following patch checks for existing sysfs_dirent before allocation new one.

Signed-off-by: Maneesh Soni <[email protected]>
---

 linux-2.6.16-rc5-git10-maneesh/fs/sysfs/dir.c |   17 +++++++++++++++++
 1 files changed, 17 insertions(+)

diff -puN fs/sysfs/dir.c~sysfs-check-existing-dirent fs/sysfs/dir.c
--- linux-2.6.16-rc5-git10/fs/sysfs/dir.c~sysfs-check-existing-dirent	2006-03-08 17:50:00.857712216 +0530
+++ linux-2.6.16-rc5-git10-maneesh/fs/sysfs/dir.c	2006-03-08 17:50:00.864711152 +0530
@@ -50,11 +50,28 @@ static struct sysfs_dirent * sysfs_new_d
 	return sd;
 }
 
+/**
+ * Initialise a newly allocated sysfs_dirent and attach it to
+ * the corresponding dentry if present.
+ *
+ * Return -EEXIST if there is already a sysfs element with the same name for
+ * the same parent.
+ *
+ * called with parent inode's i_mutex held
+ */
 int sysfs_make_dirent(struct sysfs_dirent * parent_sd, struct dentry * dentry,
 			void * element, umode_t mode, int type)
 {
 	struct sysfs_dirent * sd;
 
+	list_for_each_entry(sd, &parent_sd->s_children, s_sibling) {
+		const unsigned char * existing = sysfs_get_name(sd);
+		if (strcmp(existing, new))
+			continue;
+		else
+			return -EEXIST;
+	}
+
 	sd = sysfs_new_dirent(parent_sd, element);
 	if (!sd)
 		return -ENOMEM;
_
author
#4 lore raw
Quoted 9 lines · Maneesh Soni
Quoted 7 lines · Greg KH

Hi,

I spent some time tonight trying to track down how to fix the issue of duplicate sysfs files and/or directories. This happens when you try to create a kobject with the same name in the same directory. The creation of the second kobject will fail, but the directory will remain in sysfs.

Quoted 4 lines

Let me understand this. Lets say we have sysfs directory tree like

	/sys/a/b/c

and someone is trying to create one more kobject with name "c" for the parent kobject "b" ?

Yes.

Quoted 2 lines

And are you saying that though the new creation fails but the existing directory remains in sysfs?

No. The new creation fails, but we end up with two "c" directories under "b".

Quoted 2 lines

I think failing the new creation and leaving the exisiting directoy is ok.

I agree, but that's not what happens :)

Quoted 1 line

But there is sysfs_dirent leakage which does need fixing.

Yes.

Quoted 37 lines
Quoted 10 lines

Now I know this isn't a normal operation, but it would be good to fix this eventually. I traced the issue down to fs/sysfs/dir.c:create_dir() and the check for:

	if (error && (error != -EEXIST)) {

Problem is, error is set to -EEXIST, so we don't clean up properly. Now I know we can't just not check for this, as if you do that error cleanup, the original kobject's sysfs entry gets very messed up (ls -l does not like it at all...)

Quoted 5 lines

But I can't seem to figure out what exactly we need to do to clean up properly here.

Do you, or anyone else, have any pointers or ideas?

If you are talking about the example above, to me it appears that except a possible sysfs_dirent leakage, we are some what ok, else there would have been more catastrophic results because of the duplicate directory dentry/inode.

As per the current code

static int create_dir(struct kobject * k, struct dentry * p,
const char * n, struct dentry ** d)

{
        int error;
        umode_t mode = S_IFDIR| S_IRWXU | S_IRUGO | S_IXUGO;

        mutex_lock(&p->d_inode->i_mutex);

        *d = lookup_one_len(n, p, strlen(n));

^^^^ lookup_one_len() will return the existing dentry corresponding to
the last component "c" in "/sys/a/b/c" without any error. Just note that VFS is not going to allocate a new dentry for it. The existing dentry's ref count will be increased by one.

Quoted 7 lines

if (!IS_ERR(*d)) {
error = sysfs_make_dirent(p->d_fsdata, *d, k, mode, SYSFS_DIR);

^^^^ we do have problem here, a new sysfs_dirent is allocated which
replaces the existing dentry->d_fsdata and yuk... the old sysfs_dirent is no more linked with the existing directory, there by leaking one sysfs_dirent.

Yeah, I just couldn't figure out how to clean it up properly :)

Quoted 3 lines

Not only for sysfs_create_dir(), I think the problem of existing sysfs_dirent is also there with sysfs_add_file() and sysfs_add_link(). I am working on a patch to plug this leak.

That's great, thanks.

I have a test module that shows this at:
http://www.kernel.org/pub/linux/kernel/people/gregkh/gregkh-2.6/patches/gregkh/sysfs-test.patch
just load the 'gregkh' module and look in /sys/class/gregkh/ to see the duplicate "gregkh1" directories.

thanks,

greg k-h

#5 lore raw
Quoted 6 lines · Greg KH

I have a test module that shows this at:
http://www.kernel.org/pub/linux/kernel/people/gregkh/gregkh-2.6/patches/gregkh/sysfs-test.patch
just load the 'gregkh' module and look in /sys/class/gregkh/ to see the duplicate "gregkh1" directories.

Thnaks for the testcase, I could verify the following patch using this. I don't see the duplicate "gregkh1" directories. Please queue it for post 2.6.16 if you are ok with it.

Thanks
Maneesh

o The following patch checks for existing sysfs_dirent before preparing new one while creating sysfs directories and files.

---

 linux-2.6.16-rc5-git12-maneesh/fs/sysfs/dir.c     |   30 +++++++++++++++++++++-
 linux-2.6.16-rc5-git12-maneesh/fs/sysfs/file.c    |    4 ++
 linux-2.6.16-rc5-git12-maneesh/fs/sysfs/symlink.c |    3 +-
 linux-2.6.16-rc5-git12-maneesh/fs/sysfs/sysfs.h   |    1 
 4 files changed, 35 insertions(+), 3 deletions(-)
fs/sysfs/dir.c renamed from fs/sysfs/dir.c~sysfs-check-existing-sysfs_dirent +29 -1
diff -puN fs/sysfs/dir.c~sysfs-check-existing-sysfs_dirent fs/sysfs/dir.c
@@ -50,6 +50,30 @@ static struct sysfs_dirent * sysfs_new_d
 	return sd;
 }
 
+/**
+ *
+ * Return -EEXIST if there is already a sysfs element with the same name for
+ * the same parent.
+ *
+ * called with parent inode's i_mutex held
+ */
+int is_sysfs_dirent_exist(struct sysfs_dirent * parent_sd,
+						const unsigned char * new)
+{
+	struct sysfs_dirent * sd;
+
+	list_for_each_entry(sd, &parent_sd->s_children, s_sibling) {
+		const unsigned char * existing = sysfs_get_name(sd);
+		if (strcmp(existing, new))
+			continue;
+		else
+			return -EEXIST;
+	}
+
+	return 0;
+}
+
+
 int sysfs_make_dirent(struct sysfs_dirent * parent_sd, struct dentry * dentry,
 			void * element, umode_t mode, int type)
 {
@@ -102,7 +126,11 @@ static int create_dir(struct kobject * k
 	mutex_lock(&p->d_inode->i_mutex);
 	*d = lookup_one_len(n, p, strlen(n));
 	if (!IS_ERR(*d)) {
-		error = sysfs_make_dirent(p->d_fsdata, *d, k, mode, SYSFS_DIR);
+ 		if (is_sysfs_dirent_exist(p->d_fsdata, n))
+  			error = -EEXIST;
+  		else
+			error = sysfs_make_dirent(p->d_fsdata, *d, k, mode,
+								SYSFS_DIR);
 		if (!error) {
 			error = sysfs_create(*d, mode, init_dir);
 			if (!error) {
fs/sysfs/file.c renamed from fs/sysfs/file.c~sysfs-check-existing-sysfs_dirent +3 -1
diff -puN fs/sysfs/file.c~sysfs-check-existing-sysfs_dirent fs/sysfs/file.c
@@ -365,7 +365,9 @@ int sysfs_add_file(struct dentry * dir, 
 	int error = 0;
 
 	mutex_lock(&dir->d_inode->i_mutex);
-	error = sysfs_make_dirent(parent_sd, NULL, (void *) attr, mode, type);
+	if (!is_sysfs_dirent_exist(parent_sd, attr->name))
+		error = sysfs_make_dirent(parent_sd, NULL, (void *) attr,
+								mode, type);
 	mutex_unlock(&dir->d_inode->i_mutex);
 
 	return error;
fs/sysfs/symlink.c renamed from fs/sysfs/symlink.c~sysfs-check-existing-sysfs_dirent +2 -1
diff -puN fs/sysfs/symlink.c~sysfs-check-existing-sysfs_dirent fs/sysfs/symlink.c
@@ -87,7 +87,8 @@ int sysfs_create_link(struct kobject * k
 	BUG_ON(!kobj || !kobj->dentry || !name);
 
 	mutex_lock(&dentry->d_inode->i_mutex);
-	error = sysfs_add_link(dentry, name, target);
+	if (!is_sysfs_dirent_exist(dentry->d_fsdata, name))
+		error = sysfs_add_link(dentry, name, target);
 	mutex_unlock(&dentry->d_inode->i_mutex);
 	return error;
 }
fs/sysfs/sysfs.h renamed from fs/sysfs/sysfs.h~sysfs-check-existing-sysfs_dirent +1 -0
diff -puN fs/sysfs/sysfs.h~sysfs-check-existing-sysfs_dirent fs/sysfs/sysfs.h
@@ -5,6 +5,7 @@ extern kmem_cache_t *sysfs_dir_cachep;
 extern struct inode * sysfs_new_inode(mode_t mode, struct sysfs_dirent *);
 extern int sysfs_create(struct dentry *, int mode, int (*init)(struct inode *));
 
+extern int is_sysfs_dirent_exist(struct sysfs_dirent *, const unsigned char * );
 extern int sysfs_make_dirent(struct sysfs_dirent *, struct dentry *, void *,
 				umode_t, int);
 

_

Signature · 6 lines
Maneesh Soni
Linux Technology Center, 
IBM India Software Labs,
Bangalore, India
email: [email protected]
Phone: 91-80-51776416
author
#6 lore raw
Quoted 9 lines · Maneesh Soni
Quoted 6 lines · Greg KH

I have a test module that shows this at:
http://www.kernel.org/pub/linux/kernel/people/gregkh/gregkh-2.6/patches/gregkh/sysfs-test.patch
just load the 'gregkh' module and look in /sys/class/gregkh/ to see the duplicate "gregkh1" directories.

Quoted 3 lines

Thnaks for the testcase, I could verify the following patch using this. I don't see the duplicate "gregkh1" directories. Please queue it for post 2.6.16 if you are ok with it.

Looks good. I tweaked it a bit (returned -EEXIST from functions if the file already existed), and renamed the function to play nicer in the global namespace. Updated patch is below

Thanks a lot for doing this, I appreciate it.

greg k-h

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

Quoted 1 line

From [email protected] Thu Mar 9 06:13:04 2006

Date: Thu, 9 Mar 2006 19:40:14 +0530
From: Maneesh Soni <[email protected]>
To: Greg KH <[email protected]>
Cc: Patrick Mochel <[email protected]>
Subject: sysfs: fix problem with duplicate sysfs directories and files
Message-ID: <[email protected]>
Content-Disposition: inline

The following patch checks for existing sysfs_dirent before preparing new one while creating sysfs directories and files.

---
 fs/sysfs/dir.c     |   30 +++++++++++++++++++++++++++++-
 fs/sysfs/file.c    |    6 ++++--
 fs/sysfs/symlink.c |    5 +++--
 fs/sysfs/sysfs.h   |    1 +
 4 files changed, 37 insertions(+), 5 deletions(-)
fs/sysfs/dir.c +29 -1
@@ -51,6 +51,30 @@ static struct sysfs_dirent * sysfs_new_d
 	return sd;
 }
 
+/**
+ *
+ * Return -EEXIST if there is already a sysfs element with the same name for
+ * the same parent.
+ *
+ * called with parent inode's i_mutex held
+ */
+int sysfs_dirent_exist(struct sysfs_dirent *parent_sd,
+			  const unsigned char *new)
+{
+	struct sysfs_dirent * sd;
+
+	list_for_each_entry(sd, &parent_sd->s_children, s_sibling) {
+		const unsigned char * existing = sysfs_get_name(sd);
+		if (strcmp(existing, new))
+			continue;
+		else
+			return -EEXIST;
+	}
+
+	return 0;
+}
+
+
 int sysfs_make_dirent(struct sysfs_dirent * parent_sd, struct dentry * dentry,
 			void * element, umode_t mode, int type)
 {
@@ -117,7 +141,11 @@ static int create_dir(struct kobject * k
 	mutex_lock(&p->d_inode->i_mutex);
 	*d = lookup_one_len(n, p, strlen(n));
 	if (!IS_ERR(*d)) {
-		error = sysfs_make_dirent(p->d_fsdata, *d, k, mode, SYSFS_DIR);
+ 		if (sysfs_dirent_exist(p->d_fsdata, n))
+  			error = -EEXIST;
+  		else
+			error = sysfs_make_dirent(p->d_fsdata, *d, k, mode,
+								SYSFS_DIR);
 		if (!error) {
 			error = sysfs_create(*d, mode, init_dir);
 			if (!error) {
fs/sysfs/file.c +4 -2
@@ -450,10 +450,12 @@ int sysfs_add_file(struct dentry * dir, 
 {
 	struct sysfs_dirent * parent_sd = dir->d_fsdata;
 	umode_t mode = (attr->mode & S_IALLUGO) | S_IFREG;
-	int error = 0;
+	int error = -EEXIST;
 
 	mutex_lock(&dir->d_inode->i_mutex);
-	error = sysfs_make_dirent(parent_sd, NULL, (void *) attr, mode, type);
+	if (!sysfs_dirent_exist(parent_sd, attr->name))
+		error = sysfs_make_dirent(parent_sd, NULL, (void *)attr,
+					  mode, type);
 	mutex_unlock(&dir->d_inode->i_mutex);
 
 	return error;
fs/sysfs/symlink.c +3 -2
@@ -82,12 +82,13 @@ exit1:
 int sysfs_create_link(struct kobject * kobj, struct kobject * target, const char * name)
 {
 	struct dentry * dentry = kobj->dentry;
-	int error = 0;
+	int error = -EEXIST;
 
 	BUG_ON(!kobj || !kobj->dentry || !name);
 
 	mutex_lock(&dentry->d_inode->i_mutex);
-	error = sysfs_add_link(dentry, name, target);
+	if (!sysfs_dirent_exist(dentry->d_fsdata, name))
+		error = sysfs_add_link(dentry, name, target);
 	mutex_unlock(&dentry->d_inode->i_mutex);
 	return error;
 }
fs/sysfs/sysfs.h +1 -0
@@ -5,6 +5,7 @@ extern kmem_cache_t *sysfs_dir_cachep;
 extern struct inode * sysfs_new_inode(mode_t mode, struct sysfs_dirent *);
 extern int sysfs_create(struct dentry *, int mode, int (*init)(struct inode *));
 
+extern int sysfs_dirent_exist(struct sysfs_dirent *, const unsigned char *);
 extern int sysfs_make_dirent(struct sysfs_dirent *, struct dentry *, void *,
 				umode_t, int);