Date:	Wed, 11 Apr 2007 13:15:15 +0900
From:	Tejun Heo <htejun@...il.com>
To:	gregkh@...e.de, maneesh@...ibm.com, dmitry.torokhov@...il.com,
	cornelia.huck@...ibm.com, oneukum@...e.de, rpurdie@...ys.net,
	James.Bottomley@...elEye.com, stern@...land.harvard.edu,
	linux-kernel@...r.kernel.org, akpm@...ux-foundation.org
Subject: [PATCH 12/14 UPDATED] sysfs: implement sysfs_dirent active reference and immediate disconnect

[PATCH] sysfs: implement sysfs_dirent active reference and immediate disconnect

Opening a sysfs node references its associated kobject, so userland
can arbitrarily prolong lifetime of a kobject which complicates
lifetime rules in drivers.  This patch implements active reference and
makes the association between kobject and sysfs immediately breakable.

Now each sysfs_dirent has two reference counts - s_count and s_active.
s_count is a regular reference count which guarantees that the
containing sysfs_dirent is accessible.  As long as s_count reference
is held, all sysfs internal fields in sysfs_dirent are accessible
including s_parent and s_name.

The newly added s_active is active reference count.  This is acquired
by invoking sysfs_get_active() and it's the caller's responsibility to
ensure sysfs_dirent itself is accessible (should be holding s_count
one way or the other).  Dereferencing sysfs_dirent to access objects
out of sysfs proper requires active reference.  This includes access
to the associated kobjects, attributes and ops.

The active references can be drained and denied by calling
sysfs_deactivate().  All active sysfs_dirents must be deactivated
after deletion but before the default reference is dropped.  This
enables immediate disconnect of sysfs nodes.  Once a sysfs_dirent is
deleted, it won't access any entity external to sysfs proper.

Because attr/bin_attr ops access both the node itself and its parent
for kobject, they need to hold active references to both.
sysfs_get/put_active_two() helpers are provided to help grabbing both
references.  Parent's is acquired first and released last.

Unlike other operations, mmapped area lingers on after mmap() is
finished and the module implement implementing it and kobj need to
stay referenced till all the mapped pages are gone.  This is
accomplished by holding one set of active references to the bin_attr
and its parent if there have been any mmap during lifetime of an
openfile.  The references are dropped when the openfile is released.

This change makes sysfs lifetime rules independent from both kobject's
and module's.  It not only fixes several race conditions caused by
sysfs not holding onto the proper module when referencing kobject, but
also helps fixing and simplifying lifetime management in driver model
and drivers by taking sysfs out of the equation.

Please read the following message for more info.

  http://article.gmane.org/gmane.linux.kernel/510293

Signed-off-by: Tejun Heo <htejun@...il.com>
---
Cornelia, this should fix the problem you reported.  It's caused by
doing up_write() on sysfs_dirent which is used for readdir cursor
which is not deactivated before being released.
release_sysfs_dirent() is udpated such that it does
down_write_trylock() on s_active before doing up_write().  This also
covers error paths where allocated sysfs_dirent is put due to errors
during initialization.

Andrew, please use this patch instead of the original one.

Thanks.

 fs/sysfs/bin.c   |   95 +++++++++++++++++++++++++--------------
 fs/sysfs/dir.c   |   22 +++++++--
 fs/sysfs/file.c  |  132 ++++++++++++++++++++++++++++++++-----------------------
 fs/sysfs/inode.c |    8 ++-
 fs/sysfs/sysfs.h |  107 +++++++++++++++++++++++++++++++++++---------
 5 files changed, 250 insertions(+), 114 deletions(-)

Index: work/fs/sysfs/dir.c
===================================================================
--- work.orig/fs/sysfs/dir.c
+++ work/fs/sysfs/dir.c
@@ -22,6 +22,13 @@ void release_sysfs_dirent(struct sysfs_d
  repeat:
 	parent_sd = sd->s_parent;
 
+	/* If @sd is being released after deletion, s_active is write
+	 * locked.  If @sd is cursor for directory walk or being
+	 * released prematurely, s_active has no reader or writer.
+	 */
+	down_write_trylock(&sd->s_active);
+	up_write(&sd->s_active);
+
 	if (sd->s_type & SYSFS_KOBJ_LINK)
 		sysfs_put(sd->s_elem.symlink.target_sd);
 	if (sd->s_type & SYSFS_COPY_NAME)
@@ -69,6 +76,7 @@ struct sysfs_dirent *sysfs_new_dirent(co
 
 	atomic_set(&sd->s_count, 1);
 	atomic_set(&sd->s_event, 1);
+	init_rwsem(&sd->s_active);
 	INIT_LIST_HEAD(&sd->s_children);
 	INIT_LIST_HEAD(&sd->s_sibling);
 
@@ -316,7 +324,6 @@ static void remove_dir(struct dentry * d
 	d_delete(d);
 	sd = d->d_fsdata;
  	list_del_init(&sd->s_sibling);
-	sysfs_put(sd);
 	if (d->d_inode)
 		simple_rmdir(parent->d_inode,d);
 
@@ -325,6 +332,9 @@ static void remove_dir(struct dentry * d
 
 	mutex_unlock(&parent->d_inode->i_mutex);
 	dput(parent);
+
+	sysfs_deactivate(sd);
+	sysfs_put(sd);
 }
 
 void sysfs_remove_subdir(struct dentry * d)
@@ -335,6 +345,7 @@ void sysfs_remove_subdir(struct dentry *
 
 static void __sysfs_remove_dir(struct dentry *dentry)
 {
+	LIST_HEAD(removed);
 	struct sysfs_dirent * parent_sd;
 	struct sysfs_dirent * sd, * tmp;
 
@@ -348,12 +359,17 @@ static void __sysfs_remove_dir(struct de
 	list_for_each_entry_safe(sd, tmp, &parent_sd->s_children, s_sibling) {
 		if (!sd->s_type || !(sd->s_type & SYSFS_NOT_PINNED))
 			continue;
-		list_del_init(&sd->s_sibling);
+		list_move(&sd->s_sibling, &removed);
 		sysfs_drop_dentry(sd, dentry);
-		sysfs_put(sd);
 	}
 	mutex_unlock(&dentry->d_inode->i_mutex);
 
+	list_for_each_entry_safe(sd, tmp, &removed, s_sibling) {
+		list_del_init(&sd->s_sibling);
+		sysfs_deactivate(sd);
+		sysfs_put(sd);
+	}
+
 	remove_dir(dentry);
 	/**
 	 * Drop reference from dget() on entrance.
Index: work/fs/sysfs/sysfs.h
===================================================================
--- work.orig/fs/sysfs/sysfs.h
+++ work/fs/sysfs/sysfs.h
@@ -14,8 +14,14 @@ struct sysfs_elem_bin_attr {
 	struct bin_attribute	* bin_attr;
 };
 
+/*
+ * As long as s_count reference is held, the sysfs_dirent itself is
+ * accessible.  Dereferencing s_elem or any other outer entity
+ * requires s_active reference.
+ */
 struct sysfs_dirent {
 	atomic_t		s_count;
+	struct rw_semaphore	s_active;
 	struct sysfs_dirent	* s_parent;
 	struct list_head	s_sibling;
 	struct list_head	s_children;
@@ -85,43 +91,102 @@ struct sysfs_buffer_collection {
 	struct list_head	associates;
 };
 
-static inline struct kobject * to_kobj(struct dentry * dentry)
+static inline struct sysfs_dirent * sysfs_get(struct sysfs_dirent * sd)
 {
-	struct sysfs_dirent * sd = dentry->d_fsdata;
-	return sd->s_elem.dir.kobj;
+	if (sd) {
+		WARN_ON(!atomic_read(&sd->s_count));
+		atomic_inc(&sd->s_count);
+	}
+	return sd;
 }
 
-static inline struct kobject *sysfs_get_kobject(struct dentry *dentry)
+static inline void sysfs_put(struct sysfs_dirent * sd)
 {
-	struct kobject * kobj = NULL;
-
-	spin_lock(&dcache_lock);
-	if (!d_unhashed(dentry)) {
-		struct sysfs_dirent * sd = dentry->d_fsdata;
-
-		if (sd->s_type & SYSFS_KOBJ_LINK)
-			sd = sd->s_elem.symlink.target_sd;
+	if (atomic_dec_and_test(&sd->s_count))
+		release_sysfs_dirent(sd);
+}
 
-		kobj = kobject_get(sd->s_elem.dir.kobj);
+/**
+ *	sysfs_get_active - get an active reference to sysfs_dirent
+ *	@sd: sysfs_dirent to get an active reference to
+ *
+ *	Get an active reference of @sd.  This function is noop if @sd
+ *	is NULL.
+ *
+ *	RETURNS:
+ *	Pointer to @sd on success, NULL on failure.
+ */
+static inline struct sysfs_dirent *sysfs_get_active(struct sysfs_dirent *sd)
+{
+	if (sd) {
+		if (unlikely(!down_read_trylock(&sd->s_active)))
+			sd = NULL;
 	}
-	spin_unlock(&dcache_lock);
+	return sd;
+}
 
-	return kobj;
+/**
+ *	sysfs_put_active - put an active reference to sysfs_dirent
+ *	@sd: sysfs_dirent to put an active reference to
+ *
+ *	Put an active reference to @sd.  This function is noop if @sd
+ *	is NULL.
+ */
+static inline void sysfs_put_active(struct sysfs_dirent *sd)
+{
+	if (sd)
+		up_read(&sd->s_active);
 }
 
-static inline struct sysfs_dirent * sysfs_get(struct sysfs_dirent * sd)
+/**
+ *	sysfs_get_active_two - get active references to sysfs_dirent and parent
+ *	@sd: sysfs_dirent of interest
+ *
+ *	Get active reference to @sd and its parent.  Parent's active
+ *	reference is grabbed first.  This function is noop if @sd is
+ *	NULL.
+ *
+ *	RETURNS:
+ *	Pointer to @sd on success, NULL on failure.
+ */
+static inline struct sysfs_dirent *sysfs_get_active_two(struct sysfs_dirent *sd)
 {
 	if (sd) {
-		WARN_ON(!atomic_read(&sd->s_count));
-		atomic_inc(&sd->s_count);
+		if (sd->s_parent && unlikely(!sysfs_get_active(sd->s_parent)))
+			return NULL;
+		if (unlikely(!sysfs_get_active(sd))) {
+			sysfs_put_active(sd->s_parent);
+			return NULL;
+		}
 	}
 	return sd;
 }
 
-static inline void sysfs_put(struct sysfs_dirent * sd)
+/**
+ *	sysfs_put_active_two - put active references to sysfs_dirent and parent
+ *	@sd: sysfs_dirent of interest
+ *
+ *	Put active references to @sd and its parent.  This function is
+ *	noop if @sd is NULL.
+ */
+static inline void sysfs_put_active_two(struct sysfs_dirent *sd)
 {
-	if (atomic_dec_and_test(&sd->s_count))
-		release_sysfs_dirent(sd);
+	if (sd) {
+		sysfs_put_active(sd);
+		sysfs_put_active(sd->s_parent);
+	}
+}
+
+/**
+ *	sysfs_deactivate - deactivate sysfs_dirent
+ *	@sd: sysfs_dirent to deactivate
+ *
+ *	Deny new active references and drain existing ones.  s_active
+ *	will be unlocked when the sysfs_dirent is released.
+ */
+static inline void sysfs_deactivate(struct sysfs_dirent *sd)
+{
+	down_write(&sd->s_active);
 }
 
 static inline int sysfs_is_shadowed_inode(struct inode *inode)
Index: work/fs/sysfs/bin.c
===================================================================
--- work.orig/fs/sysfs/bin.c
+++ work/fs/sysfs/bin.c
@@ -23,6 +23,7 @@
 struct bin_buffer {
 	struct mutex	mutex;
 	void		*buffer;
+	int		mmapped;
 };
 
 static int
@@ -30,12 +31,20 @@ fill_read(struct dentry *dentry, char *b
 {
 	struct sysfs_dirent *attr_sd = dentry->d_fsdata;
 	struct bin_attribute *attr = attr_sd->s_elem.bin_attr.bin_attr;
-	struct kobject * kobj = to_kobj(dentry->d_parent);
+	struct kobject *kobj = attr_sd->s_parent->s_elem.dir.kobj;
+	int rc;
+
+	/* need attr_sd for attr, its parent for kobj */
+	if (!sysfs_get_active_two(attr_sd))
+		return -ENODEV;
+
+	rc = -EIO;
+	if (attr->read)
+		rc = attr->read(kobj, buffer, off, count);
 
-	if (!attr->read)
-		return -EIO;
+	sysfs_put_active_two(attr_sd);
 
-	return attr->read(kobj, buffer, off, count);
+	return rc;
 }
 
 static ssize_t
@@ -79,12 +88,20 @@ flush_write(struct dentry *dentry, char 
 {
 	struct sysfs_dirent *attr_sd = dentry->d_fsdata;
 	struct bin_attribute *attr = attr_sd->s_elem.bin_attr.bin_attr;
-	struct kobject *kobj = to_kobj(dentry->d_parent);
+	struct kobject *kobj = attr_sd->s_parent->s_elem.dir.kobj;
+	int rc;
+
+	/* need attr_sd for attr, its parent for kobj */
+	if (!sysfs_get_active_two(attr_sd))
+		return -ENODEV;
+
+	rc = -EIO;
+	if (attr->write)
+		rc = attr->write(kobj, buffer, offset, count);
 
-	if (!attr->write)
-		return -EIO;
+	sysfs_put_active_two(attr_sd);
 
-	return attr->write(kobj, buffer, offset, count);
+	return rc;
 }
 
 static ssize_t write(struct file *file, const char __user *userbuf,
@@ -124,14 +141,24 @@ static int mmap(struct file *file, struc
 	struct bin_buffer *bb = file->private_data;
 	struct sysfs_dirent *attr_sd = file->f_path.dentry->d_fsdata;
 	struct bin_attribute *attr = attr_sd->s_elem.bin_attr.bin_attr;
-	struct kobject *kobj = to_kobj(file->f_path.dentry->d_parent);
+	struct kobject *kobj = attr_sd->s_parent->s_elem.dir.kobj;
 	int rc;
 
-	if (!attr->mmap)
-		return -EINVAL;
-
 	mutex_lock(&bb->mutex);
-	rc = attr->mmap(kobj, attr, vma);
+
+	/* need attr_sd for attr, its parent for kobj */
+	if (!sysfs_get_active_two(attr_sd))
+		return -ENODEV;
+
+	rc = -EINVAL;
+	if (attr->mmap)
+		rc = attr->mmap(kobj, attr, vma);
+
+	if (rc == 0 && !bb->mmapped)
+		bb->mmapped = 1;
+	else
+		sysfs_put_active_two(attr_sd);
+
 	mutex_unlock(&bb->mutex);
 
 	return rc;
@@ -139,58 +166,60 @@ static int mmap(struct file *file, struc
 
 static int open(struct inode * inode, struct file * file)
 {
-	struct kobject *kobj = sysfs_get_kobject(file->f_path.dentry->d_parent);
 	struct sysfs_dirent *attr_sd = file->f_path.dentry->d_fsdata;
 	struct bin_attribute *attr = attr_sd->s_elem.bin_attr.bin_attr;
 	struct bin_buffer *bb = NULL;
-	int error = -EINVAL;
+	int error;
 
-	if (!kobj || !attr)
-		goto Done;
+	/* need attr_sd for attr */
+	if (!sysfs_get_active(attr_sd))
+		return -ENODEV;
 
-	/* Grab the module reference for this attribute if we have one */
+	/* Grab the module reference for this attribute */
 	error = -ENODEV;
-	if (!try_module_get(attr->attr.owner)) 
-		goto Done;
+	if (!try_module_get(attr->attr.owner))
+		goto err_sput;
 
 	error = -EACCES;
 	if ((file->f_mode & FMODE_WRITE) && !(attr->write || attr->mmap))
-		goto Error;
+		goto err_mput;
 	if ((file->f_mode & FMODE_READ) && !(attr->read || attr->mmap))
-		goto Error;
+		goto err_mput;
 
 	error = -ENOMEM;
 	bb = kzalloc(sizeof(*bb), GFP_KERNEL);
 	if (!bb)
-		goto Error;
+		goto err_mput;
 
 	bb->buffer = kmalloc(PAGE_SIZE, GFP_KERNEL);
 	if (!bb->buffer)
-		goto Error;
+		goto err_mput;
 
 	mutex_init(&bb->mutex);
 	file->private_data = bb;
 
-	error = 0;
-	goto Done;
+	/* open succeeded, put active reference and pin attr_sd */
+	sysfs_put_active(attr_sd);
+	sysfs_get(attr_sd);
+	return 0;
 
- Error:
-	kfree(bb);
+ err_mput:
 	module_put(attr->attr.owner);
- Done:
-	if (error)
-		kobject_put(kobj);
+ err_sput:
+	sysfs_put_active(attr_sd);
+	kfree(bb);
 	return error;
 }
 
 static int release(struct inode * inode, struct file * file)
 {
-	struct kobject * kobj = to_kobj(file->f_path.dentry->d_parent);
 	struct sysfs_dirent *attr_sd = file->f_path.dentry->d_fsdata;
 	struct bin_attribute *attr = attr_sd->s_elem.bin_attr.bin_attr;
 	struct bin_buffer *bb = file->private_data;
 
-	kobject_put(kobj);
+	if (bb->mmapped)
+		sysfs_put_active_two(attr_sd);
+	sysfs_put(attr_sd);
 	module_put(attr->attr.owner);
 	kfree(bb->buffer);
 	kfree(bb);
Index: work/fs/sysfs/file.c
===================================================================
--- work.orig/fs/sysfs/file.c
+++ work/fs/sysfs/file.c
@@ -88,8 +88,8 @@ remove_from_collection(struct sysfs_buff
  */
 static int fill_read_buffer(struct dentry * dentry, struct sysfs_buffer * buffer)
 {
-	struct sysfs_dirent * sd = dentry->d_fsdata;
-	struct kobject * kobj = to_kobj(dentry->d_parent);
+	struct sysfs_dirent *attr_sd = dentry->d_fsdata;
+	struct kobject *kobj = attr_sd->s_parent->s_elem.dir.kobj;
 	struct sysfs_ops * ops = buffer->ops;
 	int ret = 0;
 	ssize_t count;
@@ -99,8 +99,15 @@ static int fill_read_buffer(struct dentr
 	if (!buffer->page)
 		return -ENOMEM;
 
-	buffer->event = atomic_read(&sd->s_event);
-	count = ops->show(kobj, sd->s_elem.attr.attr, buffer->page);
+	/* need attr_sd for attr and ops, its parent for kobj */
+	if (!sysfs_get_active_two(attr_sd))
+		return -ENODEV;
+
+	buffer->event = atomic_read(&attr_sd->s_event);
+	count = ops->show(kobj, attr_sd->s_elem.attr.attr, buffer->page);
+
+	sysfs_put_active_two(attr_sd);
+
 	BUG_ON(count > (ssize_t)PAGE_SIZE);
 	if (count >= 0) {
 		buffer->needs_read_fill = 0;
@@ -225,14 +232,23 @@ fill_write_buffer(struct sysfs_buffer * 
  *	passing the buffer that we acquired in fill_write_buffer().
  */
 
-static int 
+static int
 flush_write_buffer(struct dentry * dentry, struct sysfs_buffer * buffer, size_t count)
 {
 	struct sysfs_dirent *attr_sd = dentry->d_fsdata;
-	struct kobject * kobj = to_kobj(dentry->d_parent);
+	struct kobject *kobj = attr_sd->s_parent->s_elem.dir.kobj;
 	struct sysfs_ops * ops = buffer->ops;
+	int rc;
 
-	return ops->store(kobj, attr_sd->s_elem.attr.attr, buffer->page, count);
+	/* need attr_sd for attr and ops, its parent for kobj */
+	if (!sysfs_get_active_two(attr_sd))
+		return -ENODEV;
+
+	rc = ops->store(kobj, attr_sd->s_elem.attr.attr, buffer->page, count);
+
+	sysfs_put_active_two(attr_sd);
+
+	return rc;
 }
 
 
@@ -276,22 +292,22 @@ out:
 
 static int sysfs_open_file(struct inode *inode, struct file *file)
 {
-	struct kobject *kobj = sysfs_get_kobject(file->f_path.dentry->d_parent);
 	struct sysfs_dirent *attr_sd = file->f_path.dentry->d_fsdata;
 	struct attribute *attr = attr_sd->s_elem.attr.attr;
+	struct kobject *kobj = attr_sd->s_parent->s_elem.dir.kobj;
 	struct sysfs_buffer_collection *set;
 	struct sysfs_buffer * buffer;
 	struct sysfs_ops * ops = NULL;
-	int error = 0;
-
-	if (!kobj || !attr)
-		goto Einval;
+	int error;
 
-	/* Grab the module reference for this attribute if we have one */
-	if (!try_module_get(attr->owner)) {
-		error = -ENODEV;
-		goto Done;
-	}
+	/* need attr_sd for attr and ops, its parent for kobj */
+	if (!sysfs_get_active_two(attr_sd))
+		return -ENODEV;
+
+	/* Grab the module reference for this attribute */
+	error = -ENODEV;
+	if (!try_module_get(attr->owner))
+		goto err_sput;
 
 	/* if the kobject has no ktype, then we assume that it is a subsystem
 	 * itself, and use ops for it.
@@ -306,30 +322,30 @@ static int sysfs_open_file(struct inode 
 	/* No sysfs operations, either from having no subsystem,
 	 * or the subsystem have no operations.
 	 */
+	error = -EACCES;
 	if (!ops)
-		goto Eaccess;
+		goto err_mput;
 
 	/* make sure we have a collection to add our buffers to */
 	mutex_lock(&inode->i_mutex);
 	if (!(set = inode->i_private)) {
-		if (!(set = inode->i_private = kmalloc(sizeof(struct sysfs_buffer_collection), GFP_KERNEL))) {
-			error = -ENOMEM;
-			goto Done;
-		} else {
+		error = -ENOMEM;
+		if (!(set = inode->i_private = kmalloc(sizeof(struct sysfs_buffer_collection), GFP_KERNEL)))
+			goto err_mput;
+		else
 			INIT_LIST_HEAD(&set->associates);
-		}
 	}
 	mutex_unlock(&inode->i_mutex);
 
+	error = -EACCES;
+
 	/* File needs write support.
 	 * The inode's perms must say it's ok, 
 	 * and we must have a store method.
 	 */
 	if (file->f_mode & FMODE_WRITE) {
-
 		if (!(inode->i_mode & S_IWUGO) || !ops->store)
-			goto Eaccess;
-
+			goto err_mput;
 	}
 
 	/* File needs read support.
@@ -338,46 +354,45 @@ static int sysfs_open_file(struct inode 
 	 */
 	if (file->f_mode & FMODE_READ) {
 		if (!(inode->i_mode & S_IRUGO) || !ops->show)
-			goto Eaccess;
+			goto err_mput;
 	}
 
 	/* No error? Great, allocate a buffer for the file, and store it
 	 * it in file->private_data for easy access.
 	 */
+	error = -ENOMEM;
 	buffer = kzalloc(sizeof(struct sysfs_buffer), GFP_KERNEL);
-	if (buffer) {
-		INIT_LIST_HEAD(&buffer->associates);
-		init_MUTEX(&buffer->sem);
-		buffer->needs_read_fill = 1;
-		buffer->ops = ops;
-		add_to_collection(buffer, inode);
-		file->private_data = buffer;
-	} else
-		error = -ENOMEM;
-	goto Done;
+	if (!buffer)
+		goto err_mput;
 
- Einval:
-	error = -EINVAL;
-	goto Done;
- Eaccess:
-	error = -EACCES;
+	INIT_LIST_HEAD(&buffer->associates);
+	init_MUTEX(&buffer->sem);
+	buffer->needs_read_fill = 1;
+	buffer->ops = ops;
+	add_to_collection(buffer, inode);
+	file->private_data = buffer;
+
+	/* open succeeded, put active references and pin attr_sd */
+	sysfs_put_active_two(attr_sd);
+	sysfs_get(attr_sd);
+	return 0;
+
+ err_mput:
 	module_put(attr->owner);
- Done:
-	if (error)
-		kobject_put(kobj);
+ err_sput:
+	sysfs_put_active_two(attr_sd);
 	return error;
 }
 
 static int sysfs_release(struct inode * inode, struct file * filp)
 {
-	struct kobject * kobj = to_kobj(filp->f_path.dentry->d_parent);
 	struct sysfs_dirent *attr_sd = filp->f_path.dentry->d_fsdata;
 	struct attribute *attr = attr_sd->s_elem.attr.attr;
 	struct sysfs_buffer * buffer = filp->private_data;
 
 	if (buffer)
 		remove_from_collection(buffer, inode);
-	kobject_put(kobj);
+	sysfs_put(attr_sd);
 	/* After this point, attr should not be accessed. */
 	module_put(attr->owner);
 
@@ -406,18 +421,25 @@ static int sysfs_release(struct inode * 
 static unsigned int sysfs_poll(struct file *filp, poll_table *wait)
 {
 	struct sysfs_buffer * buffer = filp->private_data;
-	struct kobject * kobj = to_kobj(filp->f_path.dentry->d_parent);
-	struct sysfs_dirent * sd = filp->f_path.dentry->d_fsdata;
-	int res = 0;
+	struct sysfs_dirent *attr_sd = filp->f_path.dentry->d_fsdata;
+	struct kobject *kobj = attr_sd->s_parent->s_elem.dir.kobj;
+
+	/* need parent for the kobj, grab both */
+	if (!sysfs_get_active_two(attr_sd))
+		goto trigger;
 
 	poll_wait(filp, &kobj->poll, wait);
 
-	if (buffer->event != atomic_read(&sd->s_event)) {
-		res = POLLERR|POLLPRI;
-		buffer->needs_read_fill = 1;
-	}
+	sysfs_put_active_two(attr_sd);
 
-	return res;
+	if (buffer->event != atomic_read(&attr_sd->s_event))
+		goto trigger;
+
+	return 0;
+
+ trigger:
+	buffer->needs_read_fill = 1;
+	return POLLERR|POLLPRI;
 }
 
 
Index: work/fs/sysfs/inode.c
===================================================================
--- work.orig/fs/sysfs/inode.c
+++ work/fs/sysfs/inode.c
@@ -260,12 +260,16 @@ int sysfs_hash_and_remove(struct dentry 
 		if (!strcmp(sd->s_name, name)) {
 			list_del_init(&sd->s_sibling);
 			sysfs_drop_dentry(sd, dir);
-			sysfs_put(sd);
 			found = 1;
 			break;
 		}
 	}
 	mutex_unlock(&dir->d_inode->i_mutex);
 
-	return found ? 0 : -ENOENT;
+	if (!found)
+		return -ENOENT;
+
+	sysfs_deactivate(sd);
+	sysfs_put(sd);
+	return 0;
 }
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists