[RFC 5/11] fanotify: fastpath to ignore certain in core inodes
From: Eric Paris
Date: Fri Sep 26 2008 - 17:19:48 EST
fanotify: fastpath to ignore certain in core inodes
From: Eric Paris <eparis@xxxxxxxxxx>
Allows a process to ignore certain events for the given in core inodes.
This does not preclude the need to keep a userspace or xattr based
cache if a fanotifier listener wants something with a different lifetime
than in core inode.
Signed-off-by: Eric Paris <eparis@xxxxxxxxxx>
---
fs/inode.c | 6 +
fs/notify/Makefile | 2
fs/notify/fanotify.c | 22 +++++
fs/notify/fanotify.h | 43 +++++++++
fs/notify/fastpath.c | 204 +++++++++++++++++++++++++++++++++++++++++++++
fs/notify/fastpath_user.c | 159 +++++++++++++++++++++++++++++++++++
fs/notify/group.c | 9 ++
include/linux/fanotify.h | 5 +
include/linux/fs.h | 5 +
include/linux/fsnotify.h | 8 ++
10 files changed, 462 insertions(+), 1 deletions(-)
create mode 100644 fs/notify/fastpath.c
create mode 100644 fs/notify/fastpath_user.c
diff --git a/fs/inode.c b/fs/inode.c
index 0487ddb..cc802ae 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -21,6 +21,7 @@
#include <linux/cdev.h>
#include <linux/bootmem.h>
#include <linux/inotify.h>
+#include <linux/fsnotify.h>
#include <linux/mount.h>
/*
@@ -183,6 +184,10 @@ static struct inode *alloc_inode(struct super_block *sb)
}
inode->i_private = NULL;
inode->i_mapping = mapping;
+#ifdef CONFIG_FANOTIFY
+ INIT_LIST_HEAD(&inode->fastpath_entries);
+ rwlock_init(&inode->fastpath_rwlock);
+#endif
}
return inode;
}
@@ -191,6 +196,7 @@ void destroy_inode(struct inode *inode)
{
BUG_ON(inode_has_buffers(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
diff --git a/fs/notify/Makefile b/fs/notify/Makefile
index 01915f7..8fac822 100644
--- a/fs/notify/Makefile
+++ b/fs/notify/Makefile
@@ -3,4 +3,4 @@ obj-$(CONFIG_INOTIFY_USER) += inotify_user.o
obj-$(CONFIG_DNOTIFY) += dnotify.o
-obj-$(CONFIG_FANOTIFY) += fanotify.o notification.o notification_user.o group.o group_user.o info_user.o
+obj-$(CONFIG_FANOTIFY) += fanotify.o notification.o notification_user.o group.o group_user.o info_user.o fastpath.o fastpath_user.o
diff --git a/fs/notify/fanotify.c b/fs/notify/fanotify.c
index d3ee8d1..c7aa1ad 100644
--- a/fs/notify/fanotify.c
+++ b/fs/notify/fanotify.c
@@ -35,6 +35,15 @@
struct dentry *fanotify_fs_root;
+void fanotify_inode_delete(struct inode *inode)
+{
+ if (likely(list_empty(&groups)))
+ return;
+
+ fanotify_fastpath_clear(inode);
+}
+EXPORT_SYMBOL_GPL(fanotify_inode_delete);
+
void fanotify(struct file *file, unsigned int mask)
{
struct fanotify_group *group;
@@ -52,6 +61,8 @@ void fanotify(struct file *file, unsigned int mask)
if (!S_ISREG(inode->i_mode))
return;
+ if (mask & FAN_MODIFY)
+ fanotify_fastpath_clear(inode);
/*
* SRCU!! the groups list is very very much read only and the path is
* very hot (assuming something is using fanotify) Not blocking while
@@ -67,6 +78,8 @@ void fanotify(struct file *file, unsigned int mask)
idx = srcu_read_lock(&groups_srcu_struct);
list_for_each_entry_rcu(group, &groups, group_list) {
if (mask & group->mask) {
+ if (is_fastpath(file, mask, group))
+ continue;
if (!event) {
event = create_event(file, mask);
/* shit, we OOM'd and now we can't tell, lets hope something else blows up */
@@ -112,6 +125,15 @@ static __init int fanotify_init(void)
return rc;
}
+ rc = fastpath_init();
+ if (rc) {
+ fanotify_notification_uninit();
+ fanotify_register_uninit();
+ securityfs_remove(fanotify_fs_root);
+ fanotify_fs_root = NULL;
+ return rc;
+ }
+
return 0;
}
__initcall(fanotify_init);
diff --git a/fs/notify/fanotify.h b/fs/notify/fanotify.h
index 78d8be0..2a00668 100644
--- a/fs/notify/fanotify.h
+++ b/fs/notify/fanotify.h
@@ -21,10 +21,15 @@ struct fanotify_group {
struct list_head notification_list; /* list of event_holder this group needs to send to userspace */
wait_queue_head_t notification_waitq; /* read() on the notification file blocks on this waitq */
+ /* stores all fastapth entries assoc with this group so they can be cleaned on unregister */
+ struct mutex fastpath_mutex; /* protect fastpath_entries list */
+ struct list_head fastpath_entries; /* all fastpath entries for this group */
+
char *name; /* group name used for register/unregister matching */
struct dentry *subdir; /* pointer to fanotify/name dentry */
struct dentry *notification; /* pointer to fanotify/name/notification dentry */
struct dentry *info; /* pointer to fanotify/name/info dentry */
+ struct dentry *fastpath; /* pointer to fanotify/name/fastpath dentry */
};
/*
@@ -59,6 +64,33 @@ struct fanotify_event {
atomic_t refcnt; /* how many groups still are using/need to send this event */
};
+
+
+/*
+ * a fastpath is simply an entry attached to an in core inode which allows an
+ * fanotify listener to indicate they are no longer interested in events of
+ * a type matching mask.
+ *
+ * these are flushed when an inode is evicted from core or when the inode is
+ * modified (as seen by fsnotify_access)
+ */
+struct fanotify_fastpath_entry {
+ struct fanotify_group *group; /* group this fastpath entry is for */
+ unsigned int mask; /* mask this fastpath entry is for */
+ struct inode *inode; /* inode this entry is associated with */
+ /*
+ * killcnt starts at 0 and is incremented when this entry will be killed
+ * since multiple processes my try to kill it at the same time.
+ * One might try to kill it from walking the i_list when an inode is being
+ * kicked out of cache and one might try to kill it when an fanotify group
+ * is being unregistered. The last one to finish cleaning everything up
+ * and decrement it to 0 will do the actual killing
+ */
+ atomic_t killcnt;
+ struct list_head i_list; /* list of fastpath_entries by inode->fastpath_entries */
+ struct list_head g_list; /* list of fastpath_entries by group->fastpath_entries */
+};
+
extern struct dentry *fanotify_fs_root;
extern struct mutex groups_mutex;
extern struct srcu_struct groups_srcu_struct;
@@ -84,6 +116,17 @@ extern void destroy_event_holder(struct fanotify_event_holder *holder);
extern __init int fanotify_notification_init(void);
extern __init int fanotify_notification_uninit(void);
+extern int fanotify_fastpath_user_create(struct fanotify_group *group);
+extern int fanotify_fastpath_user_destroy(struct fanotify_group *group);
+
+extern void fanotify_fastpath_get(struct fanotify_fastpath_entry *entry);
+extern void fanotify_fastpath_put(struct fanotify_fastpath_entry *entry);
+extern int is_fastpath(struct file *file, unsigned int mask, struct fanotify_group *group);
+extern int fanotify_fastpath_clear(struct inode *inode);
+extern int fanotify_fastpath_add(struct fanotify_group *group, int fd, unsigned int mask);
+extern __init int fastpath_init(void);
+extern __init int fastpath_uninit(void);
+
extern void fanotify_get_group(struct fanotify_group *group);
extern void fanotify_put_group(struct fanotify_group *group);
extern int fanotify_register_group(char *name, unsigned int mask);
diff --git a/fs/notify/fastpath.c b/fs/notify/fastpath.c
new file mode 100644
index 0000000..c134426
--- /dev/null
+++ b/fs/notify/fastpath.c
@@ -0,0 +1,204 @@
+/*
+ * Copyright (C) 2008 Red Hat, Inc., Eric Paris <eparis@xxxxxxxxxx>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2, or (at your option)
+ * any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; see the file COPYING. If not, write to
+ * the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include <linux/file.h>
+#include <linux/fs.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/namei.h>
+#include <linux/poll.h>
+#include <linux/rculist.h>
+#include <linux/rcupdate.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/security.h>
+#include <linux/uaccess.h>
+#include <linux/writeback.h>
+
+#include <linux/fanotify.h>
+#include "fanotify.h"
+
+static struct kmem_cache *fastpath_kmem_cache;
+
+static void fanotify_fastpath_kill(struct fanotify_fastpath_entry *entry)
+{
+ entry->group = NULL;
+ entry->inode = NULL;
+ entry->mask = 0;
+ entry->i_list.next = NULL;
+ entry->g_list.next = NULL;
+ kmem_cache_free(fastpath_kmem_cache, entry);
+}
+
+static struct fanotify_fastpath_entry *fanotify_fastpath_alloc(void)
+{
+ struct fanotify_fastpath_entry *entry;
+
+ entry = kmem_cache_alloc(fastpath_kmem_cache, GFP_KERNEL);
+
+ return entry;
+}
+
+void fanotify_fastpath_get(struct fanotify_fastpath_entry *entry)
+{
+ atomic_inc(&entry->killcnt);
+}
+
+void fanotify_fastpath_put(struct fanotify_fastpath_entry *entry)
+{
+ if (atomic_dec_and_test(&entry->killcnt))
+ fanotify_fastpath_kill(entry);
+}
+
+int is_fastpath(struct file *file, unsigned int mask, struct fanotify_group *group)
+{
+ struct inode *inode = file->f_path.dentry->d_inode;
+ struct fanotify_fastpath_entry *entry;
+ int found = 0;
+
+ /* if noone is listening, might as well fastpath this group huh? */
+ if (!atomic_read(&group->num_clients))
+ return 1;
+
+ read_lock(&inode->fastpath_rwlock);
+ list_for_each_entry(entry, &inode->fastpath_entries, i_list) {
+ /* is the entry for this group and is the entry->mask a superset of this event mask? */
+ if ((entry->group == group) && ((entry->mask & mask) == mask)) {
+ found = 1;
+ break;
+ }
+ }
+ read_unlock(&inode->fastpath_rwlock);
+
+ return found;
+}
+
+int fanotify_fastpath_clear(struct inode *inode)
+{
+ struct fanotify_fastpath_entry *entry;
+ struct fanotify_group *group;
+
+ write_lock(&inode->fastpath_rwlock);
+ while (!list_empty(&inode->fastpath_entries)) {
+ entry = list_first_entry(&inode->fastpath_entries, struct fanotify_fastpath_entry, i_list);
+
+ /* make sure the entry survives until off both lists */
+ fanotify_fastpath_get(entry);
+ list_del_init(&entry->i_list);
+
+ group = entry->group;
+
+ write_unlock(&inode->fastpath_rwlock);
+
+ mutex_lock(&group->fastpath_mutex);
+ if (!list_empty(&entry->g_list))
+ list_del_init(&entry->g_list);
+ mutex_unlock(&group->fastpath_mutex);
+
+ /* off both lists, may kill now */
+ fanotify_fastpath_put(entry);
+
+ write_lock(&inode->fastpath_rwlock);
+ }
+ write_unlock(&inode->fastpath_rwlock);
+
+ return 0;
+}
+
+/*
+ * add the fastpath entry to the in core inode referenced by fd
+ */
+int fanotify_fastpath_add(struct fanotify_group *group, int fd, unsigned int mask)
+{
+ struct file *file;
+ int fput_needed;
+ int ret = 0;
+ /* we initialize entry to shut up the compiler in case we just to out... */
+ struct fanotify_fastpath_entry *entry = NULL, *lentry;
+ struct inode *inode;
+ bool used = true;
+
+ file = fget_light(fd, &fput_needed);
+ if (!file)
+ return -EBADF;
+
+ inode = file->f_path.dentry->d_inode;
+
+ if (!S_ISREG(inode->i_mode)) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ /* pre allocate an entry so we can hold the writelock */
+ entry = fanotify_fastpath_alloc();
+ if (!entry) {
+ ret = -ENOMEM;
+ /* be sure that used==TRUE here so it doens't try to free */
+ goto out;
+ }
+
+ /*
+ * this is the only place we hold both the group and the inode lock
+ * we could take them in either order, but we take the mutex first
+ * so we don't sleep holding the rwlock
+ */
+ mutex_lock(&group->fastpath_mutex);
+ write_lock(&inode->fastpath_rwlock);
+ list_for_each_entry(lentry, &inode->fastpath_entries, i_list) {
+ if (lentry->group == group) {
+ lentry->mask |= mask;
+ used = false;
+ goto out_unlock;
+ }
+ }
+
+ atomic_set(&entry->killcnt, 0);
+ entry->group = group;
+ entry->mask = mask;
+ entry->inode = inode;
+
+ list_add(&entry->i_list, &inode->fastpath_entries);
+ list_add(&entry->g_list, &group->fastpath_entries);
+
+out_unlock:
+ write_unlock(&inode->fastpath_rwlock);
+ mutex_unlock(&group->fastpath_mutex);
+out:
+ fput_light(file, fput_needed);
+ if (!used)
+ fanotify_fastpath_kill(entry);
+ return ret;
+}
+
+
+__init int fastpath_uninit(void)
+{
+ kmem_cache_destroy(fastpath_kmem_cache);
+ fastpath_kmem_cache = NULL;
+
+ return 0;
+}
+
+__init int fastpath_init(void)
+{
+ fastpath_kmem_cache = kmem_cache_create("fanotify_fastpath_entry", sizeof(struct fanotify_fastpath_entry), 0, SLAB_PANIC, NULL);
+
+ return 0;
+}
diff --git a/fs/notify/fastpath_user.c b/fs/notify/fastpath_user.c
new file mode 100644
index 0000000..5d28aa3
--- /dev/null
+++ b/fs/notify/fastpath_user.c
@@ -0,0 +1,159 @@
+/*
+ * Copyright (C) 2008 Red Hat, Inc., Eric Paris <eparis@xxxxxxxxxx>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2, or (at your option)
+ * any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; see the file COPYING. If not, write to
+ * the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include <linux/dcache.h>
+#include <linux/file.h>
+#include <linux/fs.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/list.h>
+#include <linux/module.h>
+#include <linux/mount.h>
+#include <linux/mutex.h>
+#include <linux/namei.h>
+#include <linux/poll.h>
+#include <linux/rculist.h>
+#include <linux/rcupdate.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/security.h>
+#include <linux/types.h>
+#include <linux/uaccess.h>
+#include <linux/wait.h>
+
+#include <linux/fanotify.h>
+#include "fanotify.h"
+
+static int fastpath_open(struct inode *inode, struct file *file)
+{
+ struct fanotify_group *group = inode->i_private;
+
+ fanotify_get_group(group);
+ return 0;
+}
+
+static int fastpath_release(struct inode *inode, struct file *file)
+{
+ struct fanotify_group *group = inode->i_private;
+
+ fanotify_put_group(group);
+ return 0;
+}
+
+static ssize_t fastpath_write(struct file *file, const char __user *buf, size_t lenp, loff_t *offset)
+{
+ char *p;
+ int fd;
+ unsigned int mask;
+ int rc = 0;
+ struct fanotify_group *group = file->f_path.dentry->d_inode->i_private;
+
+ if (!group)
+ return -EBADF;
+
+ if (!lenp)
+ return 0;
+
+ if (*offset)
+ return -EINVAL;
+
+ if (lenp > PAGE_SIZE)
+ return -ENOMEM;
+
+ p = (char *)get_zeroed_page(GFP_KERNEL);
+
+ rc = copy_from_user(p, buf, lenp);
+ if (rc)
+ goto out;
+ rc = sscanf(p, "%d %x\n", &fd, &mask);
+ if (rc != 2) {
+ rc = -EPROTO;
+ goto out;
+ }
+
+ rc = fanotify_fastpath_add(group, fd, mask);
+ if (rc)
+ goto out;
+
+ rc = lenp;
+out:
+ free_page((unsigned long)p);
+ return rc;
+}
+
+static struct file_operations fanotify_fastpath_fops = {
+ .open = fastpath_open,
+ .release = fastpath_release,
+ .write = fastpath_write,
+};
+
+
+int fanotify_fastpath_user_destroy(struct fanotify_group *group)
+{
+ struct fanotify_fastpath_entry *entry;
+ struct inode *inode;
+
+ mutex_lock(&group->fastpath_mutex);
+ while (!list_empty(&group->fastpath_entries)) {
+ entry = list_first_entry(&group->fastpath_entries, struct fanotify_fastpath_entry, g_list);
+
+ /* make sure the entry survives until it is off both lists */
+ fanotify_fastpath_get(entry);
+
+ /* remove from g_list */
+ list_del_init(&entry->g_list);
+ mutex_unlock(&group->fastpath_mutex);
+
+ inode = entry->inode;
+
+ /* remove from i_list */
+ write_lock(&inode->fastpath_rwlock);
+ if (!list_empty(&entry->i_list))
+ list_del_init(&entry->i_list);
+ write_unlock(&inode->fastpath_rwlock);
+
+ /* off both lists, may free now */
+ fanotify_fastpath_put(entry);
+
+ mutex_lock(&group->fastpath_mutex);
+ }
+ mutex_unlock(&group->fastpath_mutex);
+
+ securityfs_remove(group->fastpath);
+ group->fastpath = NULL;
+ return 0;
+}
+
+int fanotify_fastpath_user_create(struct fanotify_group *group)
+{
+ struct dentry *parent = group->subdir;
+ struct dentry *d_fastpath;
+
+ d_fastpath = securityfs_create_file("fastpath", S_IRUSR|S_IWUSR, parent, group, &fanotify_fastpath_fops);
+ if (IS_ERR(d_fastpath)) {
+ printk(KERN_ERR "fanotify: failed to create fastpath file: %ld\n", PTR_ERR(d_fastpath));
+ group->fastpath = NULL;
+ return PTR_ERR(d_fastpath);
+ }
+ group->fastpath = d_fastpath;
+
+ mutex_init(&group->fastpath_mutex);
+ INIT_LIST_HEAD(&group->fastpath_entries);
+
+ return 0;
+}
diff --git a/fs/notify/group.c b/fs/notify/group.c
index 926d0a4..dc23127 100644
--- a/fs/notify/group.c
+++ b/fs/notify/group.c
@@ -47,6 +47,8 @@ void fanotify_kill_group(struct fanotify_group *group)
fanotify_notification_user_destroy(group);
fanotify_info_user_destroy(group);
+ fanotify_fastpath_user_destroy(group);
+
securityfs_remove(group->subdir);
group->subdir = NULL;
@@ -112,12 +114,18 @@ int fanotify_register_group(char *name, unsigned int mask)
if (rc)
goto out_clean_notification;
+ rc = fanotify_fastpath_user_create(group);
+ if (rc)
+ goto out_clean_info;
+
/* add it */
list_add_rcu(&group->group_list, &groups);
mutex_unlock(&groups_mutex);
return 0;
+out_clean_info:
+ fanotify_info_user_destroy(group);
out_clean_notification:
fanotify_notification_user_destroy(group);
out_clean_subdir:
@@ -126,6 +134,7 @@ out_free_name:
kfree(group->name);
out:
kfree(group);
+ mutex_unlock(&groups_mutex);
return rc;
}
diff --git a/include/linux/fanotify.h b/include/linux/fanotify.h
index 16ff769..a23ab97 100644
--- a/include/linux/fanotify.h
+++ b/include/linux/fanotify.h
@@ -36,16 +36,21 @@
#ifdef __KERNEL__
#include <linux/fs.h>
+#include <linux/list.h>
#ifdef CONFIG_FANOTIFY
extern void fanotify(struct file *file, unsigned int mask);
+extern void fanotify_inode_delete(struct inode *inode);
#else
static inline void fanotify(struct file *file, unsigned int mask)
{}
+static inline void fanotify_inode_delete(struct inode *inode)
+{}
+
#endif /* CONFIG_FANOTIFY */
#endif /* __KERNEL __ */
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 580b513..2d630fc 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -671,6 +671,11 @@ struct inode {
struct mutex inotify_mutex; /* protects the watches list */
#endif
+#ifdef CONFIG_FANOTIFY
+ struct list_head fastpath_entries; /* fanotify fastpath entries protected by group */
+ rwlock_t fastpath_rwlock; /* protect the fastpath entries list */
+#endif
+
unsigned long i_state;
unsigned long dirtied_when; /* jiffies of first dirtying */
diff --git a/include/linux/fsnotify.h b/include/linux/fsnotify.h
index 50d8d8b..0b03d8a 100644
--- a/include/linux/fsnotify.h
+++ b/include/linux/fsnotify.h
@@ -82,6 +82,14 @@ static inline void fsnotify_nameremove(struct dentry *dentry, int isdir)
}
/*
+ * fsnotify_inode_delete - and inode is being evicted from cache, clean up is needed
+ */
+static inline void fsnotify_inode_delete(struct inode *inode)
+{
+ fanotify_inode_delete(inode);
+}
+
+/*
* fsnotify_inoderemove - an inode is going away
*/
static inline void fsnotify_inoderemove(struct inode *inode)
--
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/