[PATCH] sysfs: Allow sysfs attribute files to be pollable

It works like this:
  Open the file
  Read all the contents.
  Call poll requesting POLLERR or POLLPRI (so select/exceptfds works)
  When poll returns,
     close the file and go to top of loop.
   or lseek to start of file and go back to the 'read'.

Events are signaled by an object manager calling
   sysfs_notify(kobj, dir, attr);

If the dir is non-NULL, it is used to find a subdirectory which
contains the attribute (presumably created by sysfs_create_group).

This has a cost of one int  per attribute, one wait_queuehead per kobject,
one int per open file.

The name "sysfs_notify" may be confused with the inotify
functionality.  Maybe it would be nice to support inotify for sysfs
attributes as well?

This patch also uses sysfs_notify to allow /sys/block/md*/md/sync_action
to be pollable

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

3 files changed, 78 insertions, 0 deletions

diff --git a/fs/sysfs/dir.c b/fs/sysfs/dir.c
index 6cfdc9a..610b5bd 100644
--- a/fs/sysfs/dir.c
+++ b/fs/sysfs/dir.c
@@ -43,6 +43,7 @@ static struct sysfs_dirent * sysfs_new_dirent(struct sysfs_dirent * parent_sd,
 
 	memset(sd, 0, sizeof(*sd));
 	atomic_set(&sd->s_count, 1);
+	atomic_set(&sd->s_event, 0);
 	INIT_LIST_HEAD(&sd->s_children);
 	list_add(&sd->s_sibling, &parent_sd->s_children);
 	sd->s_element = element;
diff --git a/fs/sysfs/file.c b/fs/sysfs/file.c
index f1cb1dd..cf37866 100644
--- a/fs/sysfs/file.c
+++ b/fs/sysfs/file.c
@@ -6,6 +6,7 @@
 #include <linux/fsnotify.h>
 #include <linux/kobject.h>
 #include <linux/namei.h>
+#include <linux/poll.h>
 #include <asm/uaccess.h>
 #include <asm/semaphore.h>
 
@@ -57,6 +58,7 @@ struct sysfs_buffer {
 	struct sysfs_ops	* ops;
 	struct semaphore	sem;
 	int			needs_read_fill;
+	int			event;
 };
 
 
@@ -72,6 +74,7 @@ struct sysfs_buffer {
  */
 static int fill_read_buffer(struct dentry * dentry, struct sysfs_buffer * buffer)
 {
+	struct sysfs_dirent * sd = dentry->d_fsdata;
 	struct attribute * attr = to_attr(dentry);
 	struct kobject * kobj = to_kobj(dentry->d_parent);
 	struct sysfs_ops * ops = buffer->ops;
@@ -83,6 +86,7 @@ static int fill_read_buffer(struct dentry * dentry, struct sysfs_buffer * buffer
 	if (!buffer->page)
 		return -ENOMEM;
 
+	buffer->event = atomic_read(&sd->s_event);
 	count = ops->show(kobj,attr,buffer->page);
 	buffer->needs_read_fill = 0;
 	BUG_ON(count > (ssize_t)PAGE_SIZE);
@@ -348,12 +352,84 @@ static int sysfs_release(struct inode * inode, struct file * filp)
 	return 0;
 }
 
+/* Sysfs attribute files are pollable.  The idea is that you read
+ * the content and then you use 'poll' or 'select' to wait for
+ * the content to change.  When the content changes (assuming the
+ * manager for the kobject supports notification), poll will
+ * return POLLERR|POLLPRI, and select will return the fd whether
+ * it is waiting for read, write, or exceptions.
+ * Once poll/select indicates that the value has changed, you
+ * need to close and re-open the file, as simply seeking and reading
+ * again will not get new data, or reset the state of 'poll'.
+ * Reminder: this only works for attributes which actively support
+ * it, and it is not possible to test an attribute from userspace
+ * to see if it supports poll (Nether 'poll' or 'select' return
+ * an appropriate error code).  When in doubt, set a suitable timeout value.
+ */
+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_dentry->d_parent);
+	struct sysfs_dirent * sd = filp->f_dentry->d_fsdata;
+	int res = 0;
+
+	poll_wait(filp, &kobj->poll, wait);
+
+	if (buffer->event != atomic_read(&sd->s_event)) {
+		res = POLLERR|POLLPRI;
+		buffer->needs_read_fill = 1;
+	}
+
+	return res;
+}
+
+
+static struct dentry *step_down(struct dentry *dir, const char * name)
+{
+	struct dentry * de;
+
+	if (dir == NULL || dir->d_inode == NULL)
+		return NULL;
+
+	mutex_lock(&dir->d_inode->i_mutex);
+	de = lookup_one_len(name, dir, strlen(name));
+	mutex_unlock(&dir->d_inode->i_mutex);
+	dput(dir);
+	if (IS_ERR(de))
+		return NULL;
+	if (de->d_inode == NULL) {
+		dput(de);
+		return NULL;
+	}
+	return de;
+}
+
+void sysfs_notify(struct kobject * k, char *dir, char *attr)
+{
+	struct dentry *de = k->dentry;
+	if (de)
+		dget(de);
+	if (de && dir)
+		de = step_down(de, dir);
+	if (de && attr)
+		de = step_down(de, attr);
+	if (de) {
+		struct sysfs_dirent * sd = de->d_fsdata;
+		if (sd)
+			atomic_inc(&sd->s_event);
+		wake_up_interruptible(&k->poll);
+		dput(de);
+	}
+}
+EXPORT_SYMBOL_GPL(sysfs_notify);
+
 const struct file_operations sysfs_file_operations = {
 	.read		= sysfs_read_file,
 	.write		= sysfs_write_file,
 	.llseek		= generic_file_llseek,
 	.open		= sysfs_open_file,
 	.release	= sysfs_release,
+	.poll		= sysfs_poll,
 };
 
 
diff --git a/fs/sysfs/sysfs.h b/fs/sysfs/sysfs.h
index 32958a7..3651ffb 100644
--- a/fs/sysfs/sysfs.h
+++ b/fs/sysfs/sysfs.h
@@ -11,6 +11,7 @@ extern int sysfs_make_dirent(struct sysfs_dirent *, struct dentry *, void *,
 
 extern int sysfs_add_file(struct dentry *, const struct attribute *, int);
 extern void sysfs_hash_and_remove(struct dentry * dir, const char * name);
+extern struct sysfs_dirent *sysfs_find(struct sysfs_dirent *dir, const char * name);
 
 extern int sysfs_create_subdir(struct kobject *, const char *, struct dentry **);
 extern void sysfs_remove_subdir(struct dentry *);