lkml.org 
[lkml]   [2009]   [Mar]   [19]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    Patch in this message
    /
    From
    Subject[PATCH 07/13] fsnotify: generic notification queue and waitq
    Date
    inotify needs to do asyc notification in which event information is stored
    on a queue until the listener is ready to receive it. This patch
    implements a generic notification queue for inotify (and later fanotify) to
    store events to be sent at a later time.

    Signed-off-by: Eric Paris <eparis@redhat.com>
    ---

    fs/notify/fsnotify.h | 2
    fs/notify/group.c | 9 ++
    fs/notify/notification.c | 209 ++++++++++++++++++++++++++++++++++++--
    include/linux/fsnotify_backend.h | 35 ++++++
    4 files changed, 244 insertions(+), 11 deletions(-)

    diff --git a/fs/notify/fsnotify.h b/fs/notify/fsnotify.h
    index be842e2..4ae0741 100644
    --- a/fs/notify/fsnotify.h
    +++ b/fs/notify/fsnotify.h
    @@ -15,6 +15,8 @@ extern struct srcu_struct fsnotify_grp_srcu;
    extern struct list_head fsnotify_groups;
    extern __u32 fsnotify_mask;

    +extern void fsnotify_flush_notif(struct fsnotify_group *group);
    +
    extern void fsnotify_clear_marks_by_group(struct fsnotify_group *group);
    extern void fsnotify_clear_marks_by_inode(struct inode *inode);
    #endif /* _LINUX_FSNOTIFY_PRIVATE_H */
    diff --git a/fs/notify/group.c b/fs/notify/group.c
    index db7ddf5..e30c58d 100644
    --- a/fs/notify/group.c
    +++ b/fs/notify/group.c
    @@ -94,6 +94,9 @@ static void fsnotify_destroy_group(struct fsnotify_group *group)
    /* clear all inode mark entries for this group */
    fsnotify_clear_marks_by_group(group);

    + /* clear the notification queue of all events */
    + fsnotify_flush_notif(group);
    +
    if (group->ops->free_group_priv)
    group->ops->free_group_priv(group);

    @@ -181,6 +184,12 @@ struct fsnotify_group *fsnotify_obtain_group(unsigned int priority, unsigned int
    group->group_num = group_num;
    group->mask = mask;

    + mutex_init(&group->notification_mutex);
    + INIT_LIST_HEAD(&group->notification_list);
    + init_waitqueue_head(&group->notification_waitq);
    + group->q_len = 0;
    + group->max_events = UINT_MAX;
    +
    spin_lock_init(&group->mark_lock);
    INIT_LIST_HEAD(&group->mark_entries);

    diff --git a/fs/notify/notification.c b/fs/notify/notification.c
    index eb23a69..be98511 100644
    --- a/fs/notify/notification.c
    +++ b/fs/notify/notification.c
    @@ -33,6 +33,21 @@
    #include "fsnotify.h"

    static struct kmem_cache *event_kmem_cache;
    +static struct kmem_cache *event_holder_kmem_cache;
    +/*
    + * this is a magic event we sent when the q is too full. since it doesn't
    + * hold real event information we just keep one system wide and use it any time
    + * it is needed. It's refcnt is set 1 at kernel init time and will never
    + * get set to 0 so it will never get 'freed'
    + */
    +static struct fsnotify_event q_overflow_event;
    +
    +/* return 1 if something is available, return 0 otherwise */
    +int fsnotify_check_notif_queue(struct fsnotify_group *group)
    +{
    + BUG_ON(!mutex_is_locked(&group->notification_mutex));
    + return !list_empty(&group->notification_list);
    +}

    void fsnotify_get_event(struct fsnotify_event *event)
    {
    @@ -45,26 +60,175 @@ void fsnotify_put_event(struct fsnotify_event *event)
    return;

    if (atomic_dec_and_test(&event->refcnt)) {
    - if (event->data_type == FSNOTIFY_EVENT_PATH) {
    + if (event->data_type == FSNOTIFY_EVENT_PATH)
    path_put(&event->path);
    - event->path.dentry = NULL;
    - event->path.mnt = NULL;
    - }
    + kmem_cache_free(event_kmem_cache, event);
    + }
    +}

    - event->mask = 0;
    +struct fsnotify_event_holder *alloc_event_holder(void)
    +{
    + return kmem_cache_alloc(event_holder_kmem_cache, GFP_KERNEL);
    +}

    - kmem_cache_free(event_kmem_cache, event);
    +void fsnotify_destroy_event_holder(struct fsnotify_event_holder *holder)
    +{
    + kmem_cache_free(event_holder_kmem_cache, holder);
    +}
    +
    +/*
    + * check if 2 events contain the same information.
    + */
    +static inline int event_compare(struct fsnotify_event *old, struct fsnotify_event *new)
    +{
    + if ((old->mask == new->mask) &&
    + (old->to_tell == new->to_tell) &&
    + (old->data_type == new->data_type)) {
    + switch (old->data_type) {
    + case (FSNOTIFY_EVENT_INODE):
    + if (old->inode == new->inode)
    + return 1;
    + break;
    + case (FSNOTIFY_EVENT_PATH):
    + if ((old->path.mnt == new->path.mnt) &&
    + (old->path.dentry == new->path.dentry))
    + return 1;
    + case (FSNOTIFY_EVENT_NONE):
    + return 1;
    + };
    }
    + return 0;
    }

    -struct fsnotify_event *fsnotify_create_event(struct inode *to_tell, __u32 mask, void *data, int data_type)
    +/*
    + * Add an event to the group notification queue. The group can later pull this
    + * event off the queue to deal with.
    + */
    +int fsnotify_add_notif_event(struct fsnotify_group *group, struct fsnotify_event *event)
    +{
    + struct fsnotify_event_holder *holder;
    + struct list_head *list = &group->notification_list;
    + struct fsnotify_event_holder *last_holder;
    + struct fsnotify_event *last_event;
    +
    + /*
    + * holder locking
    + *
    + * only this task is going to be adding this event to lists, thus only
    + * this task can add the in event holder to a list.
    + *
    + * other tasks may be removing this event from some other group's
    + * notification_list.
    + *
    + * those other tasks will del_init() the embedded holder.event_list under
    + * the event->lock. If we see it empty we know that once we
    + * get that lock the embedded holder will be ok for us to (re)use
    + */
    + if (list_empty(&event->holder.event_list))
    + holder = (struct fsnotify_event_holder *)event;
    + else
    + holder = alloc_event_holder();
    +
    + if (!holder)
    + return -ENOMEM;
    +
    + mutex_lock(&group->notification_mutex);
    +
    + if (group->q_len >= group->max_events)
    + event = &q_overflow_event;
    +
    + spin_lock(&event->lock);
    +
    + if (!list_empty(list)) {
    + last_holder = list_entry(list->prev, struct fsnotify_event_holder, event_list);
    + last_event = last_holder->event;
    + if (event_compare(last_event, event)) {
    + spin_unlock(&event->lock);
    + mutex_unlock(&group->notification_mutex);
    + if (holder != (struct fsnotify_event_holder *)event)
    + fsnotify_destroy_event_holder(holder);
    + return 0;
    + }
    + }
    +
    + group->q_len++;
    + holder->event = event;
    +
    + fsnotify_get_event(event);
    + list_add_tail(&holder->event_list, list);
    + spin_unlock(&event->lock);
    + mutex_unlock(&group->notification_mutex);
    +
    + wake_up(&group->notification_waitq);
    + return 0;
    +}
    +
    +/*
    + * remove and return the first event from the notification list. There is a
    + * reference held on this event since it was on the list. It is the responsibility
    + * of the caller to drop this reference.
    + */
    +struct fsnotify_event *fsnotify_remove_notif_event(struct fsnotify_group *group)
    {
    struct fsnotify_event *event;
    + struct fsnotify_event_holder *holder;

    - event = kmem_cache_alloc(event_kmem_cache, GFP_KERNEL);
    - if (!event)
    - return NULL;
    + BUG_ON(!mutex_is_locked(&group->notification_mutex));
    +
    + holder = list_first_entry(&group->notification_list, struct fsnotify_event_holder, event_list);
    +
    + event = holder->event;
    +
    + spin_lock(&event->lock);
    + holder->event = NULL;
    + list_del_init(&holder->event_list);
    + spin_unlock(&event->lock);
    +
    + /* event == holder means we are referenced through the in event holder */
    + if (event != (struct fsnotify_event *)holder)
    + fsnotify_destroy_event_holder(holder);
    +
    + group->q_len--;

    + return event;
    +}
    +
    +/*
    + * this will not remove the event, that must be done with fsnotify_remove_notif_event()
    + */
    +struct fsnotify_event *fsnotify_peek_notif_event(struct fsnotify_group *group)
    +{
    + struct fsnotify_event *event;
    + struct fsnotify_event_holder *holder;
    +
    + BUG_ON(!mutex_is_locked(&group->notification_mutex));
    +
    + holder = list_first_entry(&group->notification_list, struct fsnotify_event_holder, event_list);
    + event = holder->event;
    +
    + return event;
    +}
    +
    +/*
    + * called when a group is being torn down to clean up any outstanding
    + * event notifications.
    + */
    +void fsnotify_flush_notif(struct fsnotify_group *group)
    +{
    + struct fsnotify_event *event;
    +
    + mutex_lock(&group->notification_mutex);
    + while (fsnotify_check_notif_queue(group)) {
    + event = fsnotify_remove_notif_event(group);
    + fsnotify_put_event(event);
    + }
    + mutex_unlock(&group->notification_mutex);
    +}
    +
    +static void initialize_event(struct fsnotify_event *event)
    +{
    + event->holder.event = NULL;
    + INIT_LIST_HEAD(&event->holder.event_list);
    atomic_set(&event->refcnt, 1);

    spin_lock_init(&event->lock);
    @@ -72,7 +236,28 @@ struct fsnotify_event *fsnotify_create_event(struct inode *to_tell, __u32 mask,
    event->path.dentry = NULL;
    event->path.mnt = NULL;
    event->inode = NULL;
    + event->data_type = FSNOTIFY_EVENT_NONE;
    +
    + event->to_tell = NULL;
    +}
    +
    +/*
    + * create a new event to send to all interested fsnotify implementors.
    + * @to_tell the inode which is supposed to receive the event (sometimes a
    + * parent of the inode to which the event happened.
    + * @mask what actually happened.
    + * @data pointer to the object which was actually affected
    + * @data_type flag indication if the data is a file, path, inode, nothing...
    + */
    +struct fsnotify_event *fsnotify_create_event(struct inode *to_tell, __u32 mask, void *data, int data_type)
    +{
    + struct fsnotify_event *event;

    + event = kmem_cache_alloc(event_kmem_cache, GFP_KERNEL);
    + if (!event)
    + return NULL;
    +
    + initialize_event(event);
    event->to_tell = to_tell;

    switch (data_type) {
    @@ -109,6 +294,10 @@ struct fsnotify_event *fsnotify_create_event(struct inode *to_tell, __u32 mask,
    __init int fsnotify_notification_init(void)
    {
    event_kmem_cache = kmem_cache_create("fsnotify_event", sizeof(struct fsnotify_event), 0, SLAB_PANIC, NULL);
    + event_holder_kmem_cache = kmem_cache_create("fsnotify_event_holder", sizeof(struct fsnotify_event_holder), 0, SLAB_PANIC, NULL);
    +
    + initialize_event(&q_overflow_event);
    + q_overflow_event.mask = FS_Q_OVERFLOW;

    return 0;
    }
    diff --git a/include/linux/fsnotify_backend.h b/include/linux/fsnotify_backend.h
    index 5c815b4..e5b674e 100644
    --- a/include/linux/fsnotify_backend.h
    +++ b/include/linux/fsnotify_backend.h
    @@ -95,6 +95,13 @@ struct fsnotify_group {

    const struct fsnotify_ops *ops; /* how this group handles things */

    + /* needed to send notification to userspace */
    + struct mutex notification_mutex; /* protect the notification_list */
    + 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 */
    + unsigned int q_len; /* events on the queue */
    + unsigned int max_events; /* maximum events allowed on the list */
    +
    /* stores all fastapth entries assoc with this group so they can be cleaned on unregister */
    spinlock_t mark_lock; /* protect mark_entries list */
    struct list_head mark_entries; /* all inode mark entries for this group */
    @@ -108,11 +115,32 @@ struct fsnotify_group {
    };

    /*
    + * A single event can be queued in multiple group->notification_lists.
    + *
    + * each group->notification_list will point to an event_holder which in turns points
    + * to the actual event that needs to be sent to userspace.
    + *
    + * Seemed cheaper to create a refcnt'd event and a small holder for every group
    + * than create a different event for every group
    + *
    + */
    +struct fsnotify_event_holder {
    + struct fsnotify_event *event;
    + struct list_head event_list;
    +};
    +
    +/*
    * all of the information about the original object we want to now send to
    * a group. If you want to carry more info from the accessing task to the
    * listener this structure is where you need to be adding fields.
    */
    struct fsnotify_event {
    + /*
    + * If we create an event we are also likely going to need a holder
    + * to link to a group. So embed one holder in the event. Means only
    + * one allocation for the common case where we only have one group
    + */
    + struct fsnotify_event_holder holder;
    spinlock_t lock; /* protection for the associated event_holder and private_list */
    struct inode *to_tell;
    /*
    @@ -124,6 +152,7 @@ struct fsnotify_event {
    struct inode *inode;
    };
    /* when calling fsnotify tell it if the data is a path or inode */
    +#define FSNOTIFY_EVENT_NONE 0
    #define FSNOTIFY_EVENT_PATH 1
    #define FSNOTIFY_EVENT_INODE 2
    #define FSNOTIFY_EVENT_FILE 3
    @@ -203,7 +232,11 @@ extern void fsnotify_put_group(struct fsnotify_group *group);

    extern void fsnotify_get_event(struct fsnotify_event *event);
    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);
    +
    +extern int fsnotify_add_notif_event(struct fsnotify_group *group, struct fsnotify_event *event);
    +extern int fsnotify_check_notif_queue(struct fsnotify_group *group);
    +extern struct fsnotify_event *fsnotify_peek_notif_event(struct fsnotify_group *group);
    +extern struct fsnotify_event *fsnotify_remove_notif_event(struct fsnotify_group *group);

    /* functions used to manipulate the marks attached to inodes */
    extern void fsnotify_recalc_inode_mask(struct inode *inode);


    \
     
     \ /
      Last update: 2009-03-19 19:11    [W:0.043 / U:92.220 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site