The following system calls are used with this API:
With careful programming, an application can use inotify to efficiently monitor and cache the state of a set of filesystem objects. However, robust applications should allow for the fact that bugs in the monitoring logic or races of the kind described below may leave the cache inconsistent with the filesystem state. It is probably wise to to do some consistency checking, and rebuild the cache when inconsistencies are detected.
Each successful read(2) returns a buffer containing one or more of the following structures:
struct inotify_event {
int wd; /* Watch descriptor */
uint32_t mask; /* Mask describing event */
uint32_t cookie; /* Unique cookie associating related
events (for rename(2)) */
uint32_t len; /* Size of name field */
char name[]; /* Optional null-terminated name */
};
wd identifies the watch for which this event occurs. It is one of the watch descriptors returned by a previous call to inotify_add_watch(2).
mask contains bits that describe the event that occurred (see below).
cookie is a unique integer that connects related events. Currently this is used only for rename events, and allows the resulting pair of IN_MOVED_FROM and IN_MOVED_TO events to be connected by the application. For all other event types, cookie is set to 0.
The name field is present only when an event is returned for a file inside a watched directory; it identifies the file pathname relative to the watched directory. This pathname is null-terminated, and may include further null bytes ('\0') to align subsequent reads to a suitable address boundary.
The len field counts all of the bytes in name, including the null bytes; the length of each inotify_event structure is thus sizeof(struct inotify_event)+len.
The behavior when the buffer given to read(2) is too small to return information about the next event depends on the kernel version: in kernels before 2.6.21, read(2) returns 0; since kernel 2.6.21, read(2) fails with the error EINVAL. Specifying a buffer of size
sizeof(struct inotify_event) + NAME_MAX + 1
will be sufficient to read at least one event.
When monitoring a directory:
When events are generated for objects inside a watched directory, the name field in the returned inotify_event structure identifies the name of the file within the directory.
The IN_ALL_EVENTS macro is defined as a bit mask of all of the above events. This macro can be used as the mask argument when calling inotify_add_watch(2).
Two additional convenience macros are defined:
The following further bits can be specified in mask when calling inotify_add_watch(2):
The following bits may be set in the mask field returned by read(2):
Suppose an application is watching the directories dir1 and dir2, and the file dir1/myfile. The following examples show some events that may be generated.
Suppose that dir1/xx and dir2/yy are (the only) links to the same file, and an application is watching dir1, dir2, dir1/xx, and dir2/yy. Executing the following calls in the order given below will generate the following events:
Suppose an application is watching the directory dir and (the empty) directory dir/subdir. The following examples show some events that may be generated.
Since Linux 2.6.25, signal-driven I/O notification is available for inotify file descriptors; see the discussion of F_SETFL (for setting the O_ASYNC flag), F_SETOWN, and F_SETSIG in fcntl(2). The siginfo_t structure (described in sigaction(2)) that is passed to the signal handler has the following fields set: si_fd is set to the inotify file descriptor number; si_signo is set to the signal number; si_code is set to POLL_IN; and POLLIN is set in si_band.
If successive output inotify events produced on the inotify file descriptor are identical (same wd, mask, cookie, and name), then they are coalesced into a single event if the older event has not yet been read (but see BUGS). This reduces the amount of kernel memory required for the event queue, but also means that an application can't use inotify to reliably count file events.
The events returned by reading from an inotify file descriptor form an ordered queue. Thus, for example, it is guaranteed that when renaming from one directory to another, events will be produced in the correct order on the inotify file descriptor.
The FIONREAD ioctl(2) returns the number of bytes available to read from an inotify file descriptor.
Inotify 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. (Applications must fall back to polling the filesystem to catch such events.) Furthermore, various pseudo-filesystems such as /proc, /sys, and /dev/pts are not monitorable with inotify.
The inotify API does not report file accesses and modifications that may occur because of mmap(2), msync(2), and munmap(2).
The inotify API identifies affected files by filename. However, by the time an application processes an inotify event, the filename may already have been deleted or renamed.
The inotify API identifies events via watch descriptors. It is the application's responsibility to cache a mapping (if one is needed) between watch descriptors and pathnames. Be aware that directory renamings may affect multiple cached pathnames.
Inotify monitoring of directories is not recursive: to monitor subdirectories under a directory, additional watches must be created. This can take a significant amount time for large directory trees.
If monitoring an entire directory subtree, and a new subdirectory is created in that tree or an existing directory is renamed into that tree, be aware that by the time you create a watch for the new subdirectory, new files (and subdirectories) may already exist inside the subdirectory. Therefore, you may want to scan the contents of the subdirectory immediately after adding the watch (and, if desired, recursively add watches for any subdirectories that it contains).
Note that the event queue can overflow. In this case, events are lost. Robust applications should handle the possibility of lost events gracefully. For example, it may be necessary to rebuild part or all of the application cache. (One simple, but possibly expensive, approach is to close the inotify file descriptor, empty the cache, create a new inotify file descriptor, and then re-create watches and cache entries for the objects to be monitored.)
These two events are usually consecutive in the event stream available when reading from the inotify file descriptor. However, this is not guaranteed. If multiple processes are triggering events for monitored objects, then (on rare occasions) an arbitrary number of other events may appear between the IN_MOVED_FROM and IN_MOVED_TO events. Furthermore, it is not guaranteed that the event pair is atomically inserted into the queue: there may be a brief interval where the IN_MOVED_FROM has appeared, but the IN_MOVED_TO has not.
Matching up the IN_MOVED_FROM and IN_MOVED_TO event pair generated by rename(2) is thus inherently racy. (Don't forget that if an object is renamed outside of a monitored directory, there may not even be an IN_MOVED_TO event.) Heuristic approaches (e.g., assume the events are always consecutive) can be used to ensure a match in most cases, but will inevitably miss some cases, causing the application to perceive the IN_MOVED_FROM and IN_MOVED_TO events as being unrelated. If watch descriptors are destroyed and re-created as a result, then those watch descriptors will be inconsistent with the watch descriptors in any pending events. (Re-creating the inotify file descriptor and rebuilding the cache may be useful to deal with this scenario.)
Applications should also allow for the possibility that the IN_MOVED_FROM event was the last event that could fit in the buffer returned by the current call to read(2), and the accompanying IN_MOVED_TO event might be fetched only on the next read(2), which should be done with a (small) timeout to allow for the fact that insertion of the IN_MOVED_FROM-IN_MOVED_TO event pair is not atomic, and also the possibility that there may not be any IN_MOVED_TO event.
As originally designed and implemented, the IN_ONESHOT flag did not cause an IN_IGNORED event to be generated when the watch was dropped after one event. However, as an unintended effect of other changes, since Linux 2.6.36, an IN_IGNORED event is generated in this case.
Before kernel 2.6.25, the kernel code that was intended to coalesce successive identical events (i.e., the two most recent events could potentially be coalesced if the older had not yet been read) instead checked if the most recent event could be coalesced with the oldest unread event.
When a watch descriptor is removed by calling inotify_rm_watch(2) (or because a watch file is deleted or the filesystem that contains it is unmounted), any pending unread events for that watch descriptor remain available to read. As watch descriptors are subsequently allocated with inotify_add_watch(2), the kernel cycles through the range of possible watch descriptors (0 to INT_MAX) incrementally. When allocating a free watch descriptor, no check is made to see whether that watch descriptor number has any pending unread events in the inotify queue. Thus, it can happen that a watch descriptor is reallocated even when pending unread events exist for a previous incarnation of that watch descriptor number, with the result that the application might then read those events and interpret them as belonging to the file associated with the newly recycled watch descriptor. In practice, the likelihood of hitting this bug may be extremely low, since it requires that an application cycle through INT_MAX watch descriptors, release a watch descriptor while leaving unread events for that watch descriptor in the queue, and then recycle that watch descriptor. For this reason, and because there have been no reports of the bug occurring in real-world applications, as of Linux 3.15, no kernel changes have yet been made to eliminate this possible bug.
The following output was recorded while editing the file /home/user/temp/foo and listing directory /tmp. Before the file and the directory were opened, IN_OPEN events occurred. After the file was closed, an IN_CLOSE_WRITE event occurred. After the directory was closed, an IN_CLOSE_NOWRITE event occurred. Execution of the program ended when the user pressed the ENTER key.
$ ./a.out /tmp /home/user/temp Press enter key to terminate. Listening for events. IN_OPEN: /home/user/temp/foo [file] IN_CLOSE_WRITE: /home/user/temp/foo [file] IN_OPEN: /tmp/ [directory] IN_CLOSE_NOWRITE: /tmp/ [directory] Listening for events stopped.
#include <errno.h>
#include <poll.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/inotify.h>
#include <unistd.h>
/* Read all available inotify events from the file descriptor 'fd'.
wd is the table of watch descriptors for the directories in argv.
argc is the length of wd and argv.
argv is the list of watched directories.
Entry 0 of wd and argv is unused. */
static void
handle_events(int fd, int *wd, int argc, char* argv[])
{
/* Some systems cannot read integer variables if they are not
properly aligned. On other systems, incorrect alignment may
decrease performance. Hence, the buffer used for reading from
the inotify file descriptor should have the same alignment as
struct inotify_event. */
char buf[4096]
__attribute__ ((aligned(__alignof__(struct inotify_event))));
const struct inotify_event *event;
int i;
ssize_t len;
char *ptr;
/* Loop while events can be read from inotify file descriptor. */
for (;;) {
/* Read some events. */
len = read(fd, buf, sizeof buf);
if (len == -1 && errno != EAGAIN) {
perror("read");
exit(EXIT_FAILURE);
}
/* If the nonblocking read() found no events to read, then
it returns -1 with errno set to EAGAIN. In that case,
we exit the loop. */
if (len <= 0)
break;
/* Loop over all events in the buffer */
for (ptr = buf; ptr < buf + len;
ptr += sizeof(struct inotify_event) + event->len) {
event = (const struct inotify_event *) ptr;
/* Print event type */
if (event->mask & IN_OPEN)
printf("IN_OPEN: ");
if (event->mask & IN_CLOSE_NOWRITE)
printf("IN_CLOSE_NOWRITE: ");
if (event->mask & IN_CLOSE_WRITE)
printf("IN_CLOSE_WRITE: ");
/* Print the name of the watched directory */
for (i = 1; i < argc; ++i) {
if (wd[i] == event->wd) {
printf("%s/", argv[i]);
break;
}
}
/* Print the name of the file */
if (event->len)
printf("%s", event->name);
/* Print type of filesystem object */
if (event->mask & IN_ISDIR)
printf(" [directory]\n");
else
printf(" [file]\n");
}
}
}
int
main(int argc, char* argv[])
{
char buf;
int fd, i, poll_num;
int *wd;
nfds_t nfds;
struct pollfd fds[2];
if (argc < 2) {
printf("Usage: %s PATH [PATH ...]\n", argv[0]);
exit(EXIT_FAILURE);
}
printf("Press ENTER key to terminate.\n");
/* Create the file descriptor for accessing the inotify API */
fd = inotify_init1(IN_NONBLOCK);
if (fd == -1) {
perror("inotify_init1");
exit(EXIT_FAILURE);
}
/* Allocate memory for watch descriptors */
wd = calloc(argc, sizeof(int));
if (wd == NULL) {
perror("calloc");
exit(EXIT_FAILURE);
}
/* Mark directories for events
- file was opened
- file was closed */
for (i = 1; i < argc; i++) {
wd[i] = inotify_add_watch(fd, argv[i],
IN_OPEN | IN_CLOSE);
if (wd[i] == -1) {
fprintf(stderr, "Cannot watch '%s'\n", argv[i]);
perror("inotify_add_watch");
exit(EXIT_FAILURE);
}
}
/* Prepare for polling */
nfds = 2;
/* Console input */
fds[0].fd = STDIN_FILENO;
fds[0].events = POLLIN;
/* Inotify input */
fds[1].fd = fd;
fds[1].events = POLLIN;
/* Wait for events and/or terminal input */
printf("Listening for events.\n");
while (1) {
poll_num = poll(fds, nfds, -1);
if (poll_num == -1) {
if (errno == EINTR)
continue;
perror("poll");
exit(EXIT_FAILURE);
}
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')
continue;
break;
}
if (fds[1].revents & POLLIN) {
/* Inotify events are available */
handle_events(fd, wd, argc, argv);
}
}
}
printf("Listening for events stopped.\n");
/* Close inotify file descriptor */
close(fd);
free(wd);
exit(EXIT_SUCCESS);
}
Documentation/filesystems/inotify.txt in the Linux kernel source tree