fsnotify: add marks to inodes so groups can interpret how to handle those inodes

This patch creates a way for fsnotify groups to attach marks to inodes.
These marks have little meaning to the generic fsnotify infrastructure
and thus their meaning should be interpreted by the group that attached
them to the inode's list.

dnotify and inotify  will make use of these markings to indicate which
inodes are of interest to their respective groups.  But this implementation
has the useful property that in the future other listeners could actually
use the marks for the exact opposite reason, aka to indicate which inodes
it had NO interest in.
Signed-off-by: Eric Paris <eparis@redhat.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>

fs/inode.c

@@ -22,6 +22,7 @@
 #include <linux/cdev.h>
 #include <linux/bootmem.h>
 #include <linux/inotify.h>
+#include <linux/fsnotify.h>
 #include <linux/mount.h>
 #include <linux/async.h>
 
@@ -189,6 +190,10 @@ struct inode *inode_init_always(struct super_block *sb, struct inode *inode)
 	inode->i_private = NULL;
 	inode->i_mapping = mapping;
 
+#ifdef CONFIG_FSNOTIFY
+	inode->i_fsnotify_mask = 0;
+#endif
+
 	return inode;
 
 out_free_security:
@@ -221,6 +226,7 @@ void destroy_inode(struct inode *inode)
 	BUG_ON(inode_has_buffers(inode));
 	ima_inode_free(inode);
 	security_inode_free(inode);
+	fsnotify_inode_delete(inode);
 	if (inode->i_sb->s_op->destroy_inode)
 		inode->i_sb->s_op->destroy_inode(inode);
 	else
@@ -252,6 +258,9 @@ void inode_init_once(struct inode *inode)
 	INIT_LIST_HEAD(&inode->inotify_watches);
 	mutex_init(&inode->inotify_mutex);
 #endif
+#ifdef CONFIG_FSNOTIFY
+	INIT_HLIST_HEAD(&inode->i_fsnotify_mark_entries);
+#endif
 }
 EXPORT_SYMBOL(inode_init_once);
 

fs/notify/Makefile

-obj-$(CONFIG_FSNOTIFY)		+= fsnotify.o notification.o group.o
+obj-$(CONFIG_FSNOTIFY)		+= fsnotify.o notification.o group.o inode_mark.o
 
 obj-y			+= dnotify/
 obj-y			+= inotify/

fs/notify/fsnotify.c

@@ -25,6 +25,15 @@
 #include <linux/fsnotify_backend.h>
 #include "fsnotify.h"
 
+/*
+ * Clear all of the marks on an inode when it is being evicted from core
+ */
+void __fsnotify_inode_delete(struct inode *inode)
+{
+	fsnotify_clear_marks_by_inode(inode);
+}
+EXPORT_SYMBOL_GPL(__fsnotify_inode_delete);
+
 /*
  * 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
@@ -43,6 +52,8 @@ void fsnotify(struct inode *to_tell, __u32 mask, void *data, int data_is)
 	if (!(mask & fsnotify_mask))
 		return;
 
+	if (!(mask & to_tell->i_fsnotify_mask))
+		return;
 	/*
 	 * SRCU!!  the groups list is very very much read only and the path is
 	 * very hot.  The VAST majority of events are not going to need to do
@@ -51,6 +62,8 @@ void fsnotify(struct inode *to_tell, __u32 mask, void *data, int data_is)
 	idx = srcu_read_lock(&fsnotify_grp_srcu);
 	list_for_each_entry_rcu(group, &fsnotify_groups, group_list) {
 		if (mask & group->mask) {
+			if (!group->ops->should_send_event(group, to_tell, mask))
+				continue;
 			if (!event) {
 				event = fsnotify_create_event(to_tell, mask, data, data_is);
 				/* shit, we OOM'd and now we can't tell, maybe

fs/notify/fsnotify.h

@@ -12,4 +12,9 @@ extern struct srcu_struct fsnotify_grp_srcu;
 extern struct list_head fsnotify_groups;
 /* all bitwise OR of all event types (FS_*) for all fsnotify_groups */
 extern __u32 fsnotify_mask;
+
+/* final kfree of a group */
+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);
 #endif	/* __FS_NOTIFY_FSNOTIFY_H_ */

fs/notify/group.c

@@ -54,6 +54,29 @@ void fsnotify_recalc_global_mask(void)
 	fsnotify_mask = mask;
 }
 
+/*
+ * Update the group->mask by running all of the marks associated with this
+ * group and finding the bitwise | of all of the mark->mask.  If we change
+ * the group->mask we need to update the global mask of events interesting
+ * to the system.
+ */
+void fsnotify_recalc_group_mask(struct fsnotify_group *group)
+{
+	__u32 mask = 0;
+	__u32 old_mask = group->mask;
+	struct fsnotify_mark_entry *entry;
+
+	spin_lock(&group->mark_lock);
+	list_for_each_entry(entry, &group->mark_entries, g_list)
+		mask |= entry->mask;
+	spin_unlock(&group->mark_lock);
+
+	group->mask = mask;
+
+	if (old_mask != mask)
+		fsnotify_recalc_global_mask();
+}
+
 /*
  * Take a reference to a group so things found under the fsnotify_grp_mutex
  * can't get freed under us
@@ -66,7 +89,7 @@ static void fsnotify_get_group(struct fsnotify_group *group)
 /*
  * Final freeing of a group
  */
-static void fsnotify_destroy_group(struct fsnotify_group *group)
+void fsnotify_final_destroy_group(struct fsnotify_group *group)
 {
 	if (group->ops->free_group_priv)
 		group->ops->free_group_priv(group);
@@ -74,6 +97,24 @@ static void fsnotify_destroy_group(struct fsnotify_group *group)
 	kfree(group);
 }
 
+/*
+ * Trying to get rid of a group.  We need to first get rid of any outstanding
+ * allocations and then free the group.  Remember that fsnotify_clear_marks_by_group
+ * could miss marks that are being freed by inode and those marks could still
+ * hold a reference to this group (via group->num_marks)  If we get into that
+ * situtation, the fsnotify_final_destroy_group will get called when that final
+ * mark is freed.
+ */
+static void fsnotify_destroy_group(struct fsnotify_group *group)
+{
+	/* clear all inode mark entries for this group */
+	fsnotify_clear_marks_by_group(group);
+
+	/* past the point of no return, matches the initial value of 1 */
+	if (atomic_dec_and_test(&group->num_marks))
+		fsnotify_final_destroy_group(group);
+}
+
 /*
  * Remove this group from the global list of groups that will get events
  * this can be done even if there are still references and things still using
@@ -173,6 +214,10 @@ struct fsnotify_group *fsnotify_obtain_group(unsigned int group_num, __u32 mask,
 	group->group_num = group_num;
 	group->mask = mask;
 
+	spin_lock_init(&group->mark_lock);
+	atomic_set(&group->num_marks, 0);
+	INIT_LIST_HEAD(&group->mark_entries);
+
 	group->ops = ops;
 
 	mutex_lock(&fsnotify_grp_mutex);
@@ -188,6 +233,8 @@ struct fsnotify_group *fsnotify_obtain_group(unsigned int group_num, __u32 mask,
 	/* group not found, add a new one */
 	list_add_rcu(&group->group_list, &fsnotify_groups);
 	group->on_group_list = 1;
+	/* being on the fsnotify_groups list holds one num_marks */
+	atomic_inc(&group->num_marks);
 
 	mutex_unlock(&fsnotify_grp_mutex);
 

include/linux/fs.h

@@ -755,6 +755,11 @@ struct inode {
 
 	__u32			i_generation;
 
+#ifdef CONFIG_FSNOTIFY
+	__u32			i_fsnotify_mask; /* all events this inode cares about */
+	struct hlist_head	i_fsnotify_mark_entries; /* fsnotify mark entries */
+#endif
+
 #ifdef CONFIG_DNOTIFY
 	unsigned long		i_dnotify_mask; /* Directory notify events */
 	struct dnotify_struct	*i_dnotify; /* for directory notifications */

include/linux/fsnotify.h

@@ -99,6 +99,14 @@ static inline void fsnotify_move(struct inode *old_dir, struct inode *new_dir,
 	audit_inode_child(new_name, moved, new_dir);
 }
 
+/*
+ * fsnotify_inode_delete - and inode is being evicted from cache, clean up is needed
+ */
+static inline void fsnotify_inode_delete(struct inode *inode)
+{
+	__fsnotify_inode_delete(inode);
+}
+
 /*
  * fsnotify_nameremove - a filename was removed from a directory
  */
@@ -121,6 +129,7 @@ static inline void fsnotify_inoderemove(struct inode *inode)
 	inotify_inode_is_dead(inode);
 
 	fsnotify(inode, FS_DELETE_SELF, inode, FSNOTIFY_EVENT_INODE);
+	__fsnotify_inode_delete(inode);
 }
 
 /*

include/linux/fsnotify_backend.h

@@ -48,17 +48,25 @@
 
 struct fsnotify_group;
 struct fsnotify_event;
+struct fsnotify_mark_entry;
 
 /*
  * Each group much define these ops.  The fsnotify infrastructure will call
  * these operations for each relevant group.
  *
+ * should_send_event - given a group, inode, and mask this function determines
+ *		if the group is interested in this event.
  * handle_event - main call for a group to handle an fs event
  * free_group_priv - called when a group refcnt hits 0 to clean up the private union
+ * freeing-mark - this means that a mark has been flagged to die when everything
+ *		finishes using it.  The function is supplied with what must be a
+ *		valid group and inode to use to clean up.
  */
 struct fsnotify_ops {
+	bool (*should_send_event)(struct fsnotify_group *group, struct inode *inode, __u32 mask);
 	int (*handle_event)(struct fsnotify_group *group, struct fsnotify_event *event);
 	void (*free_group_priv)(struct fsnotify_group *group);
+	void (*freeing_mark)(struct fsnotify_mark_entry *entry, struct fsnotify_group *group);
 };
 
 /*
@@ -97,7 +105,14 @@ struct fsnotify_group {
 
 	const struct fsnotify_ops *ops;	/* how this group handles things */
 
-	/* prevents double list_del of group_list.  protected by global fsnotify_gr_mutex */
+	/* stores all fastapth entries assoc with this group so they can be cleaned on unregister */
+	spinlock_t mark_lock;		/* protect mark_entries list */
+	atomic_t num_marks;		/* 1 for each mark entry and 1 for not being
+					 * past the point of no return when freeing
+					 * a group */
+	struct list_head mark_entries;	/* all inode mark entries for this group */
+
+	/* prevents double list_del of group_list.  protected by global fsnotify_grp_mutex */
 	bool on_group_list;
 
 	/* groups can define private fields here or use the void *private */
@@ -137,12 +152,38 @@ struct fsnotify_event {
 	__u32 mask;		/* the type of access, bitwise OR for FS_* event types */
 };
 
+/*
+ * a mark is simply an entry attached to an in core inode which allows an
+ * fsnotify listener to indicate they are either no longer interested in events
+ * of a type matching mask or only interested in those events.
+ *
+ * these are flushed when an inode is evicted from core and may be flushed
+ * when the inode is modified (as seen by fsnotify_access).  Some fsnotify users
+ * (such as dnotify) will flush these when the open fd is closed and not at
+ * inode eviction or modification.
+ */
+struct fsnotify_mark_entry {
+	__u32 mask;			/* mask this mark entry is for */
+	/* we hold ref for each i_list and g_list.  also one ref for each 'thing'
+	 * in kernel that found and may be using this mark. */
+	atomic_t refcnt;		/* active things looking at this mark */
+	struct inode *inode;		/* inode this entry is associated with */
+	struct fsnotify_group *group;	/* group this mark entry is for */
+	struct hlist_node i_list;	/* list of mark_entries by inode->i_fsnotify_mark_entries */
+	struct list_head g_list;	/* list of mark_entries by group->i_fsnotify_mark_entries */
+	spinlock_t lock;		/* protect group, inode, and killme */
+	struct list_head free_i_list;	/* tmp list used when freeing this mark */
+	struct list_head free_g_list;	/* tmp list used when freeing this mark */
+	void (*free_mark)(struct fsnotify_mark_entry *entry); /* called on final put+free */
+};
+
 #ifdef CONFIG_FSNOTIFY
 
 /* called from the vfs helpers */
 
 /* main fsnotify call to send events */
 extern void fsnotify(struct inode *to_tell, __u32 mask, void *data, int data_is);
+extern void __fsnotify_inode_delete(struct inode *inode);
 
 
 /* called from fsnotify listeners, such as fanotify or dnotify */
@@ -153,6 +194,8 @@ extern void fsnotify_recalc_global_mask(void);
 extern struct fsnotify_group *fsnotify_obtain_group(unsigned int group_num,
 						    __u32 mask,
 						    const struct fsnotify_ops *ops);
+/* run all marks associated with this group and update group->mask */
+extern void fsnotify_recalc_group_mask(struct fsnotify_group *group);
 /* drop reference on a group from fsnotify_obtain_group */
 extern void fsnotify_put_group(struct fsnotify_group *group);
 
@@ -163,6 +206,22 @@ extern void fsnotify_put_event(struct fsnotify_event *event);
 extern struct fsnotify_event_private_data *fsnotify_get_priv_from_event(struct fsnotify_group *group,
 									struct fsnotify_event *event);
 
+/* functions used to manipulate the marks attached to inodes */
+
+/* run all marks associated with an inode and update inode->i_fsnotify_mask */
+extern void fsnotify_recalc_inode_mask(struct inode *inode);
+extern void fsnotify_init_mark(struct fsnotify_mark_entry *entry, void (*free_mark)(struct fsnotify_mark_entry *entry));
+/* find (and take a reference) to a mark associated with group and inode */
+extern struct fsnotify_mark_entry *fsnotify_find_mark_entry(struct fsnotify_group *group, struct inode *inode);
+/* attach the mark to both the group and the inode */
+extern int fsnotify_add_mark(struct fsnotify_mark_entry *entry, struct fsnotify_group *group, struct inode *inode);
+/* given a mark, flag it to be freed when all references are dropped */
+extern void fsnotify_destroy_mark_by_entry(struct fsnotify_mark_entry *entry);
+/* run all the marks in a group, and flag them to be freed */
+extern void fsnotify_clear_marks_by_group(struct fsnotify_group *group);
+extern void fsnotify_get_mark(struct fsnotify_mark_entry *entry);
+extern void fsnotify_put_mark(struct fsnotify_mark_entry *entry);
+
 /* put here because inotify does some weird stuff when destroying watches */
 extern struct fsnotify_event *fsnotify_create_event(struct inode *to_tell, __u32 mask,
 						    void *data, int data_is);
@@ -170,6 +229,10 @@ 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_inode_delete(struct inode *inode)
+{}
+
 #endif	/* CONFIG_FSNOTIFY */
 
 #endif	/* __KERNEL __ */