1fanotify_init(2) System Calls Manual fanotify_init(2)
3
7 fanotify_init - create and initialize fanotify group
8
10 Standard C library (libc, -lc)
11
13 #include <fcntl.h> /* Definition of O_* constants */
14 #include <sys/fanotify.h>
15 int fanotify_init(unsigned int flags, unsigned int event_f_flags);
17
19 For an overview of the fanotify API, see fanotify(7).
20 fanotify_init() initializes a new fanotify group and returns a file de‐
22 scriptor for the event queue associated with the group.
23 The file descriptor is used in calls to fanotify_mark(2) to specify the
25 files, directories, mounts, or filesystems for which fanotify events
26 shall be created. These events are received by reading from the file
27 descriptor. Some events are only informative, indicating that a file
28 has been accessed. Other events can be used to determine whether an‐
29 other application is permitted to access a file or directory. Permis‐
30 sion to access filesystem objects is granted by writing to the file de‐
31 scriptor.
32 Multiple programs may be using the fanotify interface at the same time
34 to monitor the same files.
35 The number of fanotify groups per user is limited. See fanotify(7) for
37 details about this limit.
38 The flags argument contains a multi-bit field defining the notification
40 class of the listening application and further single bit fields speci‐
41 fying the behavior of the file descriptor.
42 If multiple listeners for permission events exist, the notification
44 class is used to establish the sequence in which the listeners receive
45 the events.
46 Only one of the following notification classes may be specified in
48 flags:
49 FAN_CLASS_PRE_CONTENT
51 This value allows the receipt of events notifying that a file
52 has been accessed and events for permission decisions if a file
53 may be accessed. It is intended for event listeners that need
54 to access files before they contain their final data. This no‐
55 tification class might be used by hierarchical storage managers,
56 for example. Use of this flag requires the CAP_SYS_ADMIN capa‐
57 bility.
58 FAN_CLASS_CONTENT
60 This value allows the receipt of events notifying that a file
61 has been accessed and events for permission decisions if a file
62 may be accessed. It is intended for event listeners that need
63 to access files when they already contain their final content.
64 This notification class might be used by malware detection pro‐
65 grams, for example. Use of this flag requires the CAP_SYS_ADMIN
66 capability.
67 FAN_CLASS_NOTIF
69 This is the default value. It does not need to be specified.
70 This value only allows the receipt of events notifying that a
71 file has been accessed. Permission decisions before the file is
72 accessed are not possible.
73 Listeners with different notification classes will receive events in
75 the order FAN_CLASS_PRE_CONTENT, FAN_CLASS_CONTENT, FAN_CLASS_NOTIF.
76 The order of notification for listeners in the same notification class
77 is undefined.
78 The following bits can additionally be set in flags:
80 FAN_CLOEXEC
82 Set the close-on-exec flag (FD_CLOEXEC) on the new file descrip‐
83 tor. See the description of the O_CLOEXEC flag in open(2).
84 FAN_NONBLOCK
86 Enable the nonblocking flag (O_NONBLOCK) for the file descrip‐
87 tor. Reading from the file descriptor will not block. Instead,
88 if no data is available, read(2) fails with the error EAGAIN.
89 FAN_UNLIMITED_QUEUE
91 Remove the limit on the number of events in the event queue.
92 See fanotify(7) for details about this limit. Use of this flag
93 requires the CAP_SYS_ADMIN capability.
94 FAN_UNLIMITED_MARKS
96 Remove the limit on the number of fanotify marks per user. See
97 fanotify(7) for details about this limit. Use of this flag re‐
98 quires the CAP_SYS_ADMIN capability.
99 FAN_REPORT_TID (since Linux 4.20)
101 Report thread ID (TID) instead of process ID (PID) in the pid
102 field of the struct fanotify_event_metadata supplied to read(2)
103 (see fanotify(7)). Use of this flag requires the CAP_SYS_ADMIN
104 capability.
105 FAN_ENABLE_AUDIT (since Linux 4.15)
107 Enable generation of audit log records about access mediation
108 performed by permission events. The permission event response
109 has to be marked with the FAN_AUDIT flag for an audit log record
110 to be generated. Use of this flag requires the CAP_AUDIT_WRITE
111 capability.
112 FAN_REPORT_FID (since Linux 5.1)
114 This value allows the receipt of events which contain additional
115 information about the underlying filesystem object correlated to
116 an event. An additional record of type FAN_EVENT_INFO_TYPE_FID
117 encapsulates the information about the object and is included
118 alongside the generic event metadata structure. The file de‐
119 scriptor that is used to represent the object correlated to an
120 event is instead substituted with a file handle. It is intended
121 for applications that may find the use of a file handle to iden‐
122 tify an object more suitable than a file descriptor. Addition‐
123 ally, it may be used for applications monitoring a directory or
124 a filesystem that are interested in the directory entry modifi‐
125 cation events FAN_CREATE, FAN_DELETE, FAN_MOVE, and FAN_RENAME,
126 or in events such as FAN_ATTRIB, FAN_DELETE_SELF, and
127 FAN_MOVE_SELF. All the events above require an fanotify group
128 that identifies filesystem objects by file handles. Note that
129 without the flag FAN_REPORT_TARGET_FID, for the directory entry
130 modification events, there is an information record that identi‐
131 fies the modified directory and not the created/deleted/moved
132 child object. The use of FAN_CLASS_CONTENT or
133 FAN_CLASS_PRE_CONTENT is not permitted with this flag and will
134 result in the error EINVAL. See fanotify(7) for additional de‐
135 tails.
136 FAN_REPORT_DIR_FID (since Linux 5.9)
138 Events for fanotify groups initialized with this flag will con‐
139 tain (see exceptions below) additional information about a di‐
140 rectory object correlated to an event. An additional record of
141 type FAN_EVENT_INFO_TYPE_DFID encapsulates the information about
142 the directory object and is included alongside the generic event
143 metadata structure. For events that occur on a non-directory
144 object, the additional structure includes a file handle that
145 identifies the parent directory filesystem object. Note that
146 there is no guarantee that the directory filesystem object will
147 be found at the location described by the file handle informa‐
148 tion at the time the event is received. When combined with the
149 flag FAN_REPORT_FID, two records may be reported with events
150 that occur on a non-directory object, one to identify the non-
151 directory object itself and one to identify the parent directory
152 object. Note that in some cases, a filesystem object does not
153 have a parent, for example, when an event occurs on an unlinked
154 but open file. In that case, with the FAN_REPORT_FID flag, the
155 event will be reported with only one record to identify the non-
156 directory object itself, because there is no directory associ‐
157 ated with the event. Without the FAN_REPORT_FID flag, no event
158 will be reported. See fanotify(7) for additional details.
159 FAN_REPORT_NAME (since Linux 5.9)
161 Events for fanotify groups initialized with this flag will con‐
162 tain additional information about the name of the directory en‐
163 try correlated to an event. This flag must be provided in con‐
164 junction with the flag FAN_REPORT_DIR_FID. Providing this flag
165 value without FAN_REPORT_DIR_FID will result in the error EIN‐
166 VAL. This flag may be combined with the flag FAN_REPORT_FID.
167 An additional record of type FAN_EVENT_INFO_TYPE_DFID_NAME,
168 which encapsulates the information about the directory entry, is
169 included alongside the generic event metadata structure and sub‐
170 stitutes the additional information record of type
171 FAN_EVENT_INFO_TYPE_DFID. The additional record includes a file
172 handle that identifies a directory filesystem object followed by
173 a name that identifies an entry in that directory. For the di‐
174 rectory entry modification events FAN_CREATE, FAN_DELETE, and
175 FAN_MOVE, the reported name is that of the created/deleted/moved
176 directory entry. The event FAN_RENAME may contain two informa‐
177 tion records. One of type FAN_EVENT_INFO_TYPE_OLD_DFID_NAME
178 identifying the old directory entry, and another of type
179 FAN_EVENT_INFO_TYPE_NEW_DFID_NAME identifying the new directory
180 entry. For other events that occur on a directory object, the
181 reported file handle is that of the directory object itself and
182 the reported name is '.'. For other events that occur on a non-
183 directory object, the reported file handle is that of the parent
184 directory object and the reported name is the name of a direc‐
185 tory entry where the object was located at the time of the
186 event. The rationale behind this logic is that the reported di‐
187 rectory file handle can be passed to open_by_handle_at(2) to get
188 an open directory file descriptor and that file descriptor along
189 with the reported name can be used to call fstatat(2). The same
190 rule that applies to record type FAN_EVENT_INFO_TYPE_DFID also
191 applies to record type FAN_EVENT_INFO_TYPE_DFID_NAME: if a non-
192 directory object has no parent, either the event will not be re‐
193 ported or it will be reported without the directory entry infor‐
194 mation. Note that there is no guarantee that the filesystem ob‐
195 ject will be found at the location described by the directory
196 entry information at the time the event is received. See fan‐
197 otify(7) for additional details.
198 FAN_REPORT_DFID_NAME
200 This is a synonym for (FAN_REPORT_DIR_FID|FAN_REPORT_NAME).
201 FAN_REPORT_TARGET_FID (since Linux 5.17)
203 Events for fanotify groups initialized with this flag will con‐
204 tain additional information about the child correlated with di‐
205 rectory entry modification events. This flag must be provided
206 in conjunction with the flags FAN_REPORT_FID, FAN_REPORT_DIR_FID
207 and FAN_REPORT_NAME. or else the error EINVAL will be returned.
208 For the directory entry modification events FAN_CREATE,
209 FAN_DELETE, FAN_MOVE, and FAN_RENAME, an additional record of
210 type FAN_EVENT_INFO_TYPE_FID, is reported in addition to the in‐
211 formation records of type FAN_EVENT_INFO_TYPE_DFID,
212 FAN_EVENT_INFO_TYPE_DFID_NAME,
213 FAN_EVENT_INFO_TYPE_OLD_DFID_NAME, and
214 FAN_EVENT_INFO_TYPE_NEW_DFID_NAME. The additional record in‐
215 cludes a file handle that identifies the filesystem child object
216 that the directory entry is referring to.
217 FAN_REPORT_DFID_NAME_TARGET
219 This is a synonym for (FAN_REPORT_DFID_NAME|FAN_RE‐
220 PORT_FID|FAN_REPORT_TARGET_FID).
221 FAN_REPORT_PIDFD (since Linux 5.15)
223 Events for fanotify groups initialized with this flag will con‐
224 tain an additional information record alongside the generic fan‐
225 otify_event_metadata structure. This information record will be
226 of type FAN_EVENT_INFO_TYPE_PIDFD and will contain a pidfd for
227 the process that was responsible for generating an event. A
228 pidfd returned in this information record object is no different
229 to the pidfd that is returned when calling pidfd_open(2). Usage
230 of this information record are for applications that may be in‐
231 terested in reliably determining whether the process responsible
232 for generating an event has been recycled or terminated. The
233 use of the FAN_REPORT_TID flag along with FAN_REPORT_PIDFD is
234 currently not supported and attempting to do so will result in
235 the error EINVAL being returned. This limitation is currently
236 imposed by the pidfd API as it currently only supports the cre‐
237 ation of pidfds for thread-group leaders. Creating pidfds for
238 non-thread-group leaders may be supported at some point in the
239 future, so this restriction may eventually be lifted. For more
240 details on information records, see fanotify(7).
241 The event_f_flags argument defines the file status flags that will be
243 set on the open file descriptions that are created for fanotify events.
244 For details of these flags, see the description of the flags values in
245 open(2). event_f_flags includes a multi-bit field for the access mode.
246 This field can take the following values:
247 O_RDONLY
249 This value allows only read access.
250 O_WRONLY
252 This value allows only write access.
253 O_RDWR This value allows read and write access.
255 Additional bits can be set in event_f_flags. The most useful values
257 are:
258 O_LARGEFILE
260 Enable support for files exceeding 2 GB. Failing to set this
261 flag will result in an EOVERFLOW error when trying to open a
262 large file which is monitored by an fanotify group on a 32-bit
263 system.
264 O_CLOEXEC (since Linux 3.18)
266 Enable the close-on-exec flag for the file descriptor. See the
267 description of the O_CLOEXEC flag in open(2) for reasons why
268 this may be useful.
269 The following are also allowable: O_APPEND, O_DSYNC, O_NOATIME, O_NON‐
271 BLOCK, and O_SYNC. Specifying any other flag in event_f_flags yields
272 the error EINVAL (but see BUGS).
273
275 On success, fanotify_init() returns a new file descriptor. On error,
276 -1 is returned, and errno is set to indicate the error.
277
279 EINVAL An invalid value was passed in flags or event_f_flags.
280 FAN_ALL_INIT_FLAGS (deprecated since Linux 4.20) defines all al‐
281 lowable bits for flags.
282 EMFILE The number of fanotify groups for this user exceeds the limit.
284 See fanotify(7) for details about this limit.
285 EMFILE The per-process limit on the number of open file descriptors has
287 been reached.
288 ENOMEM The allocation of memory for the notification group failed.
290 ENOSYS This kernel does not implement fanotify_init(). The fanotify
292 API is available only if the kernel was configured with CON‐
293 FIG_FANOTIFY.
294 EPERM The operation is not permitted because the caller lacks a re‐
296 quired capability.
297
299 Prior to Linux 5.13, calling fanotify_init() required the CAP_SYS_ADMIN
300 capability. Since Linux 5.13, users may call fanotify_init() without
301 the CAP_SYS_ADMIN capability to create and initialize an fanotify group
302 with limited functionality.
303 The limitations imposed on an event listener created by a user without
305 the
306 CAP_SYS_ADMIN capability are as follows:
307 • The user cannot request for an unlimited event queue by using
309 FAN_UNLIMITED_QUEUE.
310 • The user cannot request for an unlimited number of marks by
312 using FAN_UNLIMITED_MARKS.
313 • The user cannot request to use either notification classes
315 FAN_CLASS_CONTENT or FAN_CLASS_PRE_CONTENT. This means that
316 user cannot request permission events.
317 • The user is required to create a group that identifies
319 filesystem objects by file handles, for example, by providing
320 the FAN_REPORT_FID flag.
321 • The user is limited to only mark inodes. The ability to mark
323 a mount or filesystem via fanotify_mark() through the use of
324 FAN_MARK_MOUNT or FAN_MARK_FILESYSTEM is not permitted.
325 • The event object in the event queue is limited in terms of
327 the information that is made available to the unprivileged
328 user. A user will also not receive the pid that generated
329 the event, unless the listening process itself generated the
330 event.
331
333 Linux.
334
336 Linux 2.6.37.
337
339 The following bug was present before Linux 3.18:
340 • The O_CLOEXEC is ignored when passed in event_f_flags.
342 The following bug was present before Linux 3.14:
344 • The event_f_flags argument is not checked for invalid flags. Flags
346 that are intended only for internal use, such as FMODE_EXEC, can be
347 set, and will consequently be set for the file descriptors returned
348 when reading from the fanotify file descriptor.
349
351 fanotify_mark(2), fanotify(7)
352Linux man-pages 6.04 2023-03-30 fanotify_init(2)