[PATCH 04/12] fsnotify: parent event notification

From: Eric Paris
Date: Mon Apr 27 2009 - 17:55:27 EST

inotify and dnotify both use a similar parent notification mechanism. We
add a generic parent notification mechanism to fsnotify for both of these
to use. This new machanism also adds the dentry flag optimization which
exists for inotify to dnotify.

Signed-off-by: Eric Paris <eparis@xxxxxxxxxx>
---

fs/notify/fsnotify.c | 91 ++++++++++++++++++++++++++++++++++++++
fs/notify/fsnotify.h | 5 ++
fs/notify/inode_mark.c | 17 +++++++
include/linux/dcache.h | 3 +
include/linux/fsnotify.h | 34 ++++++++++----
include/linux/fsnotify_backend.h | 64 +++++++++++++++++++++++++++
6 files changed, 204 insertions(+), 10 deletions(-)

diff --git a/fs/notify/fsnotify.c b/fs/notify/fsnotify.c
index d565462..2dbab3c 100644
--- a/fs/notify/fsnotify.c
+++ b/fs/notify/fsnotify.c
@@ -35,6 +35,97 @@ void __fsnotify_inode_delete(struct inode *inode)
EXPORT_SYMBOL_GPL(__fsnotify_inode_delete);

/*
+ * Given an inode, first check if we care what happens to our children. Inotify
+ * and dnotify both tell their parents about events. If we care about any event
+ * on a child we run all of our children and set a dentry flag saying that the
+ * parent cares. Thus when an event happens on a child it can quickly tell if
+ * if there is a need to find a parent and send the event to the parent.
+ */
+void __fsnotify_update_child_dentry_flags(struct inode *inode)
+{
+ struct dentry *alias;
+ int watched;
+
+ if (!S_ISDIR(inode->i_mode))
+ return;
+
+ /* determine if the children should tell inode about their events */
+ watched = fsnotify_inode_watches_children(inode);
+
+ spin_lock(&dcache_lock);
+ /* run all of the dentries associated with this inode. Since this is a
+ * directory, there damn well better only be one item on this list */
+ list_for_each_entry(alias, &inode->i_dentry, d_alias) {
+ struct dentry *child;
+
+ /* run all of the children of the original inode and fix their
+ * d_flags to indicate parental interest (their parent is the
+ * original inode) */
+ list_for_each_entry(child, &alias->d_subdirs, d_u.d_child) {
+ if (!child->d_inode)
+ continue;
+
+ spin_lock(&child->d_lock);
+ if (watched)
+ child->d_flags |= DCACHE_FSNOTIFY_PARENT_WATCHED;
+ else
+ child->d_flags &= ~DCACHE_FSNOTIFY_PARENT_WATCHED;
+ spin_unlock(&child->d_lock);
+ }
+ }
+ spin_unlock(&dcache_lock);
+}
+
+/* Notify this dentry's parent about a child's events. */
+void __fsnotify_parent(struct dentry *dentry, __u32 mask)
+{
+ struct dentry *parent;
+ struct inode *p_inode;
+ bool send = false;
+ bool should_update_children = false;
+
+ if (!(dentry->d_flags | DCACHE_FSNOTIFY_PARENT_WATCHED))
+ return;
+
+ spin_lock(&dentry->d_lock);
+ parent = dentry->d_parent;
+ p_inode = parent->d_inode;
+
+ if (fsnotify_inode_watches_children(p_inode)) {
+ if (p_inode->i_fsnotify_mask & mask) {
+ dget(parent);
+ send = true;
+ }
+ } else {
+ /*
+ * The parent doesn't care about events on it's children but
+ * at least one child thought it did. We need to run all the
+ * children and update their d_flags to let them know p_inode
+ * doesn't care about them any more.
+ */
+ dget(parent);
+ should_update_children = true;
+ }
+
+ spin_unlock(&dentry->d_lock);
+
+ if (send) {
+ /* we are notifying a parent so come up with the new mask which
+ * specifies these are events which came from a child. */
+ mask |= FS_EVENT_ON_CHILD;
+
+ fsnotify(p_inode, mask, dentry->d_inode, FSNOTIFY_EVENT_INODE);
+ dput(parent);
+ }
+
+ if (unlikely(should_update_children)) {
+ __fsnotify_update_child_dentry_flags(p_inode);
+ dput(parent);
+ }
+}
+EXPORT_SYMBOL_GPL(__fsnotify_parent);
+
+/*
* This is the main call to fsnotify. The VFS calls into hook specific functions
* in linux/fsnotify.h. Those functions then in turn call here. Here will call
* out to all of the registered fsnotify_group. Those groups can then use the
diff --git a/fs/notify/fsnotify.h b/fs/notify/fsnotify.h
index 8ebcbe8..83b8ec0 100644
--- a/fs/notify/fsnotify.h
+++ b/fs/notify/fsnotify.h
@@ -17,4 +17,9 @@ extern __u32 fsnotify_mask;
extern void fsnotify_final_destroy_group(struct fsnotify_group *group);
/* run the list of all marks associated with inode and flag them to be freed */
extern void fsnotify_clear_marks_by_inode(struct inode *inode);
+/*
+ * update the dentry->d_flags of all of inode's children to indicate if inode cares
+ * about events that happen to its children.
+ */
+extern void __fsnotify_update_child_dentry_flags(struct inode *inode);
#endif /* __FS_NOTIFY_FSNOTIFY_H_ */
diff --git a/fs/notify/inode_mark.c b/fs/notify/inode_mark.c
index cdc1541..a395348 100644
--- a/fs/notify/inode_mark.c
+++ b/fs/notify/inode_mark.c
@@ -131,6 +131,8 @@ void fsnotify_recalc_inode_mask(struct inode *inode)
spin_lock(&inode->i_lock);
fsnotify_recalc_inode_mask_locked(inode);
spin_unlock(&inode->i_lock);
+
+ __fsnotify_update_child_dentry_flags(inode);
}

/*
@@ -190,6 +192,19 @@ void fsnotify_destroy_mark_by_entry(struct fsnotify_mark_entry *entry)
group->ops->freeing_mark(entry, group);

/*
+ * __fsnotify_update_child_dentry_flags(inode);
+ *
+ * I really want to call that, but we can't, we have no idea if the inode
+ * still exists the second we drop the entry->lock.
+ *
+ * The next time an event arrive to this inode from one of it's children
+ * __fsnotify_parent will see that the inode doesn't care about it's
+ * children and will update all of these flags then. So really this
+ * is just a lazy update (and could be a perf win...)
+ */
+
+
+ /*
* it's possible that this group tried to destroy itself, but this
* this mark was simultaneously being freed by inode. If that's the
* case, we finish freeing the group here.
@@ -323,6 +338,8 @@ int fsnotify_add_mark(struct fsnotify_mark_entry *entry,
if (lentry) {
ret = -EEXIST;
fsnotify_put_mark(lentry);
+ } else {
+ __fsnotify_update_child_dentry_flags(inode);
}

return ret;
diff --git a/include/linux/dcache.h b/include/linux/dcache.h
index 8e4d98c..2f1d7b0 100644
--- a/include/linux/dcache.h
+++ b/include/linux/dcache.h
@@ -180,7 +180,8 @@ d_iput: no no no yes
#define DCACHE_REFERENCED 0x0008 /* Recently used, don't discard. */
#define DCACHE_UNHASHED 0x0010

-#define DCACHE_INOTIFY_PARENT_WATCHED 0x0020 /* Parent inode is watched */
+#define DCACHE_INOTIFY_PARENT_WATCHED 0x0020 /* Parent inode is watched by inotify */
+#define DCACHE_FSNOTIFY_PARENT_WATCHED 0x0040 /* Parent inode is watched by some fsnotify listener */

#define DCACHE_COOKIE 0x0040 /* For use by dcookie subsystem */

diff --git a/include/linux/fsnotify.h b/include/linux/fsnotify.h
index 5625905..fa51966 100644
--- a/include/linux/fsnotify.h
+++ b/include/linux/fsnotify.h
@@ -23,15 +23,31 @@
static inline void fsnotify_d_instantiate(struct dentry *entry,
struct inode *inode)
{
+ __fsnotify_d_instantiate(entry, inode);
+
inotify_d_instantiate(entry, inode);
}

+/* Notify this dentry's parent about a child's events. */
+static inline void fsnotify_parent(struct dentry *dentry, __u32 mask)
+{
+ __fsnotify_parent(dentry, mask);
+
+ inotify_dentry_parent_queue_event(dentry, mask, 0, dentry->d_name.name);
+}
+
/*
* fsnotify_d_move - entry has been moved
* Called with dcache_lock and entry->d_lock held.
*/
static inline void fsnotify_d_move(struct dentry *entry)
{
+ /*
+ * On move we need to update entry->d_flags to indicate if the new parent
+ * cares about events from this entry.
+ */
+ __fsnotify_update_dcache_flags(entry);
+
inotify_d_move(entry);
}

@@ -117,7 +133,8 @@ static inline void fsnotify_nameremove(struct dentry *dentry, int isdir)
if (isdir)
mask |= FS_IN_ISDIR;
dnotify_parent(dentry, DN_DELETE);
- inotify_dentry_parent_queue_event(dentry, mask, 0, dentry->d_name.name);
+
+ fsnotify_parent(dentry, mask);
}

/*
@@ -187,9 +204,9 @@ static inline void fsnotify_access(struct dentry *dentry)
mask |= FS_IN_ISDIR;

dnotify_parent(dentry, DN_ACCESS);
- inotify_dentry_parent_queue_event(dentry, mask, 0, dentry->d_name.name);
inotify_inode_queue_event(inode, mask, 0, NULL, NULL);

+ fsnotify_parent(dentry, mask);
fsnotify(inode, mask, inode, FSNOTIFY_EVENT_INODE);
}

@@ -205,9 +222,9 @@ static inline void fsnotify_modify(struct dentry *dentry)
mask |= FS_IN_ISDIR;

dnotify_parent(dentry, DN_MODIFY);
- inotify_dentry_parent_queue_event(dentry, mask, 0, dentry->d_name.name);
inotify_inode_queue_event(inode, mask, 0, NULL, NULL);

+ fsnotify_parent(dentry, mask);
fsnotify(inode, mask, inode, FSNOTIFY_EVENT_INODE);
}

@@ -222,9 +239,9 @@ static inline void fsnotify_open(struct dentry *dentry)
if (S_ISDIR(inode->i_mode))
mask |= FS_IN_ISDIR;

- inotify_dentry_parent_queue_event(dentry, mask, 0, dentry->d_name.name);
inotify_inode_queue_event(inode, mask, 0, NULL, NULL);

+ fsnotify_parent(dentry, mask);
fsnotify(inode, mask, inode, FSNOTIFY_EVENT_INODE);
}

@@ -235,16 +252,15 @@ static inline void fsnotify_close(struct file *file)
{
struct dentry *dentry = file->f_path.dentry;
struct inode *inode = dentry->d_inode;
- const char *name = dentry->d_name.name;
fmode_t mode = file->f_mode;
__u32 mask = (mode & FMODE_WRITE) ? FS_CLOSE_WRITE : FS_CLOSE_NOWRITE;

if (S_ISDIR(inode->i_mode))
mask |= FS_IN_ISDIR;

- inotify_dentry_parent_queue_event(dentry, mask, 0, name);
inotify_inode_queue_event(inode, mask, 0, NULL, NULL);

+ fsnotify_parent(dentry, mask);
fsnotify(inode, mask, file, FSNOTIFY_EVENT_FILE);
}

@@ -259,9 +275,9 @@ static inline void fsnotify_xattr(struct dentry *dentry)
if (S_ISDIR(inode->i_mode))
mask |= FS_IN_ISDIR;

- inotify_dentry_parent_queue_event(dentry, mask, 0, dentry->d_name.name);
inotify_inode_queue_event(inode, mask, 0, NULL, NULL);

+ fsnotify_parent(dentry, mask);
fsnotify(inode, mask, inode, FSNOTIFY_EVENT_INODE);
}

@@ -310,8 +326,8 @@ static inline void fsnotify_change(struct dentry *dentry, unsigned int ia_valid)
if (S_ISDIR(inode->i_mode))
in_mask |= FS_IN_ISDIR;
inotify_inode_queue_event(inode, in_mask, 0, NULL, NULL);
- inotify_dentry_parent_queue_event(dentry, in_mask, 0,
- dentry->d_name.name);
+
+ fsnotify_parent(dentry, in_mask);
fsnotify(inode, in_mask, inode, FSNOTIFY_EVENT_INODE);
}
}
diff --git a/include/linux/fsnotify_backend.h b/include/linux/fsnotify_backend.h
index cad5c4d..13d2dd5 100644
--- a/include/linux/fsnotify_backend.h
+++ b/include/linux/fsnotify_backend.h
@@ -46,6 +46,17 @@
#define FS_DN_RENAME 0x10000000 /* file renamed */
#define FS_DN_MULTISHOT 0x20000000 /* dnotify multishot */

+/* This inode cares about things that happen to its children. Always set for
+ * dnotify and inotify. */
+#define FS_EVENT_ON_CHILD 0x08000000
+
+/* This is a list of all events that may get sent to a parernt based on fs event
+ * happening to inodes inside that directory */
+#define FS_EVENTS_POSS_ON_CHILD (FS_ACCESS | FS_MODIFY | FS_ATTRIB |\
+ FS_CLOSE_WRITE | FS_CLOSE_NOWRITE | FS_OPEN |\
+ FS_MOVED_FROM | FS_MOVED_TO | FS_CREATE |\
+ FS_DELETE)
+
struct fsnotify_group;
struct fsnotify_event;
struct fsnotify_mark_entry;
@@ -183,8 +194,52 @@ struct fsnotify_mark_entry {

/* main fsnotify call to send events */
extern void fsnotify(struct inode *to_tell, __u32 mask, void *data, int data_is);
+extern void __fsnotify_parent(struct dentry *dentry, __u32 mask);
extern void __fsnotify_inode_delete(struct inode *inode);

+static inline int fsnotify_inode_watches_children(struct inode *inode)
+{
+ /* FS_EVENT_ON_CHILD is set if the inode may care */
+ if (!(inode->i_fsnotify_mask & FS_EVENT_ON_CHILD))
+ return 0;
+ /* this inode might care about child events, does it care about the
+ * specific set of events that can happen on a child? */
+ return inode->i_fsnotify_mask & FS_EVENTS_POSS_ON_CHILD;
+}
+
+/*
+ * Update the dentry with a flag indicating the interest of its parent to receive
+ * filesystem events when those events happens to this dentry->d_inode.
+ */
+static inline void __fsnotify_update_dcache_flags(struct dentry *dentry)
+{
+ struct dentry *parent;
+
+ assert_spin_locked(&dcache_lock);
+ assert_spin_locked(&dentry->d_lock);
+
+ parent = dentry->d_parent;
+ if (fsnotify_inode_watches_children(parent->d_inode))
+ dentry->d_flags |= DCACHE_FSNOTIFY_PARENT_WATCHED;
+ else
+ dentry->d_flags &= ~DCACHE_FSNOTIFY_PARENT_WATCHED;
+}
+
+/*
+ * fsnotify_d_instantiate - instantiate a dentry for inode
+ * Called with dcache_lock held.
+ */
+static inline void __fsnotify_d_instantiate(struct dentry *dentry, struct inode *inode)
+{
+ if (!inode)
+ return;
+
+ assert_spin_locked(&dcache_lock);
+
+ spin_lock(&dentry->d_lock);
+ __fsnotify_update_dcache_flags(dentry);
+ spin_unlock(&dentry->d_lock);
+}

/* called from fsnotify listeners, such as fanotify or dnotify */

@@ -230,9 +285,18 @@ extern struct fsnotify_event *fsnotify_create_event(struct inode *to_tell, __u32
static inline void fsnotify(struct inode *to_tell, __u32 mask, void *data, int data_is)
{}

+static inline void __fsnotify_parent(struct dentry *dentry, __u32 mask)
+{}
+
static inline void __fsnotify_inode_delete(struct inode *inode)
{}

+static inline void __fsnotify_update_dcache_flags(struct dentry *dentry)
+{}
+
+static inline void __fsnotify_d_instantiate(struct dentry *dentry, struct inode *inode)
+{}
+
#endif /* CONFIG_FSNOTIFY */

#endif /* __KERNEL __ */

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/