Section: Linux Programmer's Manual (7)
Updated: 2014-04-24
Index Return to Main Contents


fanotify - monitoring filesystem events  


The fanotify API provides notification and interception of filesystem events. Use cases include virus scanning and hierarchical storage management. Currently, only a limited set of events is supported. In particular there is no support for create, delete, and move events.

Additional capabilities compared to the inotify(7) API are monitoring of complete mounts, access permission decisions, and the possibility to read or modify files before access by other applications.

The following system calls are used with this API: fanotify_init(2), fanotify_mark(2), poll(2), ppoll(2), read(2), write(2), and close(2).

fanotify_init(2) creates and initializes an fanotify notification group and returns a file descriptor referring to it.

An fanotify notification group is an internal object of the kernel which holds a list of files, directories, and mount points for which events shall be created.

For each entry in an fanotify notification group, two bit masks exist. One mask (the mark mask) defines file activities for which an event shall be created. Another mask (the ignore mask) defines activities for which no event shall be generated. Having these two types of masks permits a mount point or directory to be marked for receiving events, while at the same time ignoring events for specific objects under that mount point or directory.

A possible usage of the ignore mask is for a file cache. Events of interest for a file cache are modification of a file and closing of the same. Hence, the cached directory or mount point is to be marked to receive these events. After receiving the first event informing that a file has been modified, the corresponding cache entry will be invalidated. No further modification events for this file are of interest until the file is closed. Hence, the modify event can be added to the ignore mask. Upon receiving the closed event, the modify event can be removed from the ignore mask and the file cache entry can be updated.

The entries in the fanotify notification groups refer to files and directories via their inode number and to mounts via their mount ID. If files or directories are renamed or moved, the respective entries survive. If files or directories are deleted or mounts are unmounted, the corresponding entries are deleted.

Two types of events exist: notification events and permission events. Notification events are only informative and require no action to be taken by the receiving application except for closing the file descriptor passed in the event. Permission events are requests to the receiving application to decide whether permission for a file access shall be granted. For these events, the recipient must write a response which decides whether access is granted or not.

When all file descriptors referring to the fanotify notification group are closed, the fanotify group is released and its resources are freed for reuse by the kernel.

fanotify_mark(2) adds a file, directory, or mount to the group and specifies which events shall be reported (or ignored), or removes or modifies such an entry.

When an fanotify event occurs, the fanotify file descriptor indicates as readable when passed to epoll(7), poll(2), or select(2).

All events for an fanotify group are collected in a queue. Consecutive events for the same filesystem object and originating from the same process may be merged into a single event, with the exception that two permission events are never merged into one queue entry. Queue entries for notification events are removed when the event has been read. Queue entries for permission events are removed when the permission decision has been taken by writing to the fanotify file descriptor.

Calling read(2) for the file descriptor returned by fanotify_init(2) blocks (if the flag FAN_NONBLOCK is not specified in the call to fanotify_init(2)) until either a file event occurs or the call is interrupted by a signal (see signal(7)).

The return value of read(2) is the length of the filled buffer, or -1 in case of an error. After a successful read(2), the read buffer contains one or more of the following structures:

struct fanotify_event_metadata {
    __u32 event_len;
    __u8 vers;
    __u8 reserved;
    __u16 metadata_len;
    __aligned_u64 mask;
    __s32 fd;
    __s32 pid;

This is the length of the data for the current event and the offset to the next event in the buffer. In the current implementation the value of event_len is always FAN_EVENT_METADATA_LEN. In principal the API design would allow to return variable length structures. Therefore, and for performance reasons, it is recommended to use a larger buffer size when reading, for example 4096 bytes.
This field holds a version number for the structures. It must be compared to FANOTIFY_METADATA_VERSION to verify that the structures at runtime match the structures at compile time. In case of a mismatch, the application should abandon trying to use the fanotify file descriptor.
This field is not used.
This is the length of the structure. The field was introduced to facilitate the implementation of optional headers per event type. No such optional headers exist in the current implementation.
This is a bit mask describing the event.
This is an open file descriptor for the object being accessed, or FAN_NOFD if a queue overflow occurred. The file descriptor can be used to access the contents of the monitored file or directory. It has internal the flag FMODE_NONOTIFY set. This flag suppresses fanotify event generation. Hence, when the receiver of the fanotify event accesses the notified file or directory using this file descriptor, no additional events will be created. The reading application is responsible for closing the file descriptor.
This is the ID of the process that caused the event. A program listening to fanotify events can compare this PID to the PID returned by getpid(2), to determine whether the event is caused by the listener itself, or is due to a file access by another program.

The bit mask in mask signals which events have occurred for a single filesystem object. More than one of the following flags can be set at once in the bit mask.

A file or a directory (but see BUGS) was accessed (read).
A file or a directory was opened.
A file was modified.
A file that was opened for writing (O_WRONLY or O_RDWR) was closed.
A file that was only opened for reading (O_RDONLY) or a directory was closed.
The event queue exceeded the limit of 16384 entries. This limit can be overridden in the call to fanotify_init(2) by setting the flag FAN_UNLIMITED_QUEUE.
An application wants to read a file or directory, for example using read(2) or readdir(2). The reader must write a response that determines whether the permission to access the filesystem object shall be granted.
An application wants to open a file or directory. The reader must write a response that determines whether the permission to open the filesystem object shall be granted.

To check for any close event, the following bit mask may be used:


The following macros are provided to iterate over a buffer containing fanotify event metadata returned by read(2) from an fanotify file descriptor.

FAN_EVENT_OK(meta, len)
This macro checks the remaining length len of the buffer meta against the length of the metadata structure and the event_len field of the first metadata structure in the buffer.
FAN_EVENT_NEXT(meta, len)
This macro sets the pointer meta to the next metadata structure using the length indicated in the event_len field of the metadata structure and reduces the remaining length of the buffer

For permission events, the application must write(2) a structure of the following form to the fanotify file descriptor:

struct fanotify_response {
    __s32 fd;
    __u32 response;

This is the file descriptor from the structure fanotify_event_metadata.
This field indicates whether or not the permission is to be granted. Its value must be either FAN_ALLOW to allow the file operation or FAN_DENY to deny the file operation.

If access has been denied, the requesting application call will receive an error EPERM.

To end listening, it is sufficient to close(2) the fanotify file descriptor. The outstanding permission events will be set to allowed, and all resources will be returned to the kernel.

The file /proc/<pid>/fdinfo/<fd> contains information about fanotify marks for file descriptor fd of process pid. See Documentation/filesystems/proc.txt for details.  


In addition to the usual errors for read(2), the following errors can occur when reading from the fanotify file descriptor:
The buffer is too short to hold the event.
The per-process limit on the number of open files has been reached. See the description of RLIMIT_NOFILE in getrlimit(2).
The system-wide limit on the number of open files has been reached. See /proc/sys/fs/file-max in proc(5).
A write enabled file descriptor shall be created for a file that is executing. This error is returned by read(2), if O_RDWR or O_WRONLY was specified in the event_f_flags argument when calling fanotify_init(2) and the event occured for a monitored file that is currently being execuded.

In addition to the usual errors for write(2), the following errors can occur when writing to the fanotify file descriptor:

Fanotify access permissions are not enabled in the kernel configuration or the value of response in the response structure is not valid.
The file descriptor fd in the response structure is not valid. This might occur because the file was already deleted by another thread or process.


The fanotify API was introduced in version 2.6.36 of the Linux kernel and enabled in version 2.6.37. Fdinfo support was added in version 3.8.  


The fanotify API is Linux-specific.  


The fanotify API is available only if the kernel was built with the CONFIG_FANOTIFY configuration option enabled. In addition, fanotify permission handling is available only if the CONFIG_FANOTIFY_ACCESS_PERMISSIONS configuration option is enabled.  

Limitations and caveats

Fanotify reports only events that a user-space program triggers through the filesystem API. As a result, it does not catch remote events that occur on network filesystems.

The fanotify API does not report file accesses and modifications that may occur because of mmap(2), msync(2), and munmap(2).

Events for directories are created only if the directory itself is opened, read, and closed. Adding, removing, or changing children of a marked directory does not create events for the monitored directory itself.

Fanotify monitoring of directories is not recursive: to monitor subdirectories under a directory, additional marks must be created. (But note that the fanotify API provides no way of detecting when a subdirectory has been created under a marked directory, which makes recursive monitoring difficult.) Monitoring mounts offers the capability to monitor a whole directory tree.

The event queue can overflow. In this case, events are lost.  


As of Linux 3.15, the following bugs existed:
readdir(2) does not create a FAN_ACCESS event.
When an event is generated, no check is made to see whether the user ID of the receiving process has authorization to read or write the file before passing a file descriptor for that file in This poses a security risk, when the CAP_SYS_ADMIN capability is set for programs executed by unprivileged users.


The following program demonstrates the usage of the fanotify API. It marks the mount passed as argument and waits for events of type FAN_PERM_OPEN and FAN_CLOSE_WRITE. When a permission event occurs, a FAN_ALLOW response is given.

The following output was recorded while editing file /home/user/temp/notes. Before the file was opened, a FAN_OPEN_PERM event occurred. After the file was closed, a FAN_CLOSE_WRITE event occurred. Execution of the program ends when the user presses the ENTER key.  

Example output

# ./fanotify_example /home
Press enter key to terminate.
Listening for events.
FAN_OPEN_PERM: File /home/user/temp/notes
FAN_CLOSE_WRITE: File /home/user/temp/notes

Listening for events stopped.

Program source

#define _GNU_SOURCE // needed for O_LARGEFILE
#include <errno.h>
#include <fcntl.h>
#include <limits.h>
#include <poll.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/fanotify.h>
#include <unistd.h>

/* Read all available fanotify events from the file descriptor 'fd' */

handle_events(int fd)
    const struct fanotify_event_metadata *metadata;
    char buf[4096];
    ssize_t len;
    char path[PATH_MAX];
    ssize_t path_len;
    char procfd_path[PATH_MAX];
    struct fanotify_response response;

    /* Loop while events can be read from fanotify file descriptor. */

    for(;;) {

        /* Read some events. */

        len = read(fd, (void *) &buf, sizeof(buf));
        if (len == -1 && errno != EAGAIN) {

        /* Check if end of available data reached. */

        if (len <= 0)

        /* Point to the first event in the buffer. */

        metadata = (struct fanotify_event_metadata *) buf;

        /* Loop over all events in the buffer. */

        while (FAN_EVENT_OK(metadata, len)) {

            /* Check that run time and compile time structures
               match. */

            if (metadata->vers != FANOTIFY_METADATA_VERSION) {
                        "Mismatch of fanotify metadata version.\n");

            /* Check that the event contains a file descriptor. */

            if (metadata->fd >= 0) {

                /* Handle open permission event. */

                if (metadata->mask & FAN_OPEN_PERM) {
                    printf("FAN_OPEN_PERM: ");

                    /* Allow file to be opened. */

                    response.fd = metadata->fd;
                    response.response = FAN_ALLOW;
                    write(fd, &response, sizeof(
                              struct fanotify_response));

                /* Handle closing of writable file event. */

                if (metadata->mask & FAN_CLOSE_WRITE) {
                    printf("FAN_CLOSE_WRITE: ");

                /* Determine path of the file accessed. */

                snprintf(procfd_path, sizeof(procfd_path),
                         "/proc/self/fd/%d", metadata->fd);
                path_len = readlink(procfd_path, path,
                                    sizeof(path) - 1);
                if (path_len == -1) {

                path[path_len] = '\0';
                printf("File %s", path);

                /* Close the file descriptor of the event. */


            /* Forward pointer to next event. */

            metadata = FAN_EVENT_NEXT(metadata, len);

main(int argc, char *argv[])
    char buf;
    int fd, poll_num;
    nfds_t nfds;
    struct pollfd fds[2];

    /* Check mount point is supplied. */

    if (argc != 2) {
        printf("Usage: %s MOUNT\n", argv[0]);

    printf("Press enter key to terminate.\n");

    /* Create the file descriptor for accessing the fanotify API. */

                       O_RDONLY | O_LARGEFILE);
    if (fd == -1) {

    /* Mark the mount for
       - permission events before opening files
       - notification events after closing a write enabled
         file descriptor. */

    if (fanotify_mark(fd, FAN_MARK_ADD | FAN_MARK_MOUNT,
                      FAN_OPEN_PERM | FAN_CLOSE_WRITE, -1,
                      argv[1]) == -1) {

    /* Prepare for polling. */

    nfds = 2;

    /* Console input. */

    fds[0].fd = STDIN_FILENO;
    fds[0].events = POLLIN;

    /* Fanotify input. */

    fds[1].fd = fd;
    fds[1].events = POLLIN;

    /* This is the loop to wait for incoming events. */

    printf("Listening for events.\n");
    while (1) {
        poll_num = poll(fds, nfds, -1);
        if (poll_num == -1) {
            if (errno == EINTR)
        if (poll_num > 0) {
            if (fds[0].revents & POLLIN) {

                /* Console input is available. Empty stdin and quit. */

                while (read(STDIN_FILENO, &buf, 1) > 0 && buf != '\n')
            if (fds[1].revents & POLLIN) {

                /* Fanotify events are available. */


    /* Close fanotify file descriptor. */

    printf("Listening for events stopped.\n");
    return EXIT_SUCCESS;


fanotify_init(2), fanotify_mark(2), inotify(7)



Limitations and caveats
Example output
Program source

This document was created by man2html using the manual pages.