kernel_optimize_test/Documentation/filesystems/inotify.txt

				   inotify
	    a powerful yet simple file change notification system



Document started 15 Mar 2005 by Robert Love <rml@novell.com>


(i) User Interface

Inotify is controlled by a set of three system calls and normal file I/O on a
returned file descriptor.

First step in using inotify is to initialise an inotify instance:

	int fd = inotify_init ();

Each instance is associated with a unique, ordered queue.

Change events are managed by "watches".  A watch is an (object,mask) pair where
the object is a file or directory and the mask is a bit mask of one or more
inotify events that the application wishes to receive.  See <linux/inotify.h>
for valid events.  A watch is referenced by a watch descriptor, or wd.

Watches are added via a path to the file.

Watches on a directory will return events on any files inside of the directory.

Adding a watch is simple:

	int wd = inotify_add_watch (fd, path, mask);

Where "fd" is the return value from inotify_init(), path is the path to the
object to watch, and mask is the watch mask (see <linux/inotify.h>).

You can update an existing watch in the same manner, by passing in a new mask.

An existing watch is removed via

	int ret = inotify_rm_watch (fd, wd);

Events are provided in the form of an inotify_event structure that is read(2)
from a given inotify instance.  The filename is of dynamic length and follows
the struct. It is of size len.  The filename is padded with null bytes to
ensure proper alignment.  This padding is reflected in len.

You can slurp multiple events by passing a large buffer, for example

	size_t len = read (fd, buf, BUF_LEN);

Where "buf" is a pointer to an array of "inotify_event" structures at least
BUF_LEN bytes in size.  The above example will return as many events as are
available and fit in BUF_LEN.

Each inotify instance fd is also select()- and poll()-able.

You can find the size of the current event queue via the standard FIONREAD
ioctl on the fd returned by inotify_init().

All watches are destroyed and cleaned up on close.


(ii)

Prototypes:

	int inotify_init (void);
	int inotify_add_watch (int fd, const char *path, __u32 mask);
	int inotify_rm_watch (int fd, __u32 mask);


(iii) Internal Kernel Implementation

Each inotify instance is associated with an inotify_device structure.

Each watch is associated with an inotify_watch structure.  Watches are chained
off of each associated device and each associated inode.

See fs/inotify.c for the locking and lifetime rules.


(iv) Rationale

Q: What is the design decision behind not tying the watch to the open fd of
   the watched object?

A: Watches are associated with an open inotify device, not an open file.
   This solves the primary problem with dnotify: keeping the file open pins
   the file and thus, worse, pins the mount.  Dnotify is therefore infeasible
   for use on a desktop system with removable media as the media cannot be
   unmounted.  Watching a file should not require that it be open.

Q: What is the design decision behind using an-fd-per-instance as opposed to
   an fd-per-watch?

A: An fd-per-watch quickly consumes more file descriptors than are allowed,
   more fd's than are feasible to manage, and more fd's than are optimally
   select()-able.  Yes, root can bump the per-process fd limit and yes, users
   can use epoll, but requiring both is a silly and extraneous requirement.
   A watch consumes less memory than an open file, separating the number
   spaces is thus sensible.  The current design is what user-space developers
   want: Users initialize inotify, once, and add n watches, requiring but one
   fd and no twiddling with fd limits.  Initializing an inotify instance two
   thousand times is silly.  If we can implement user-space's preferences 
   cleanly--and we can, the idr layer makes stuff like this trivial--then we 
   should.

   There are other good arguments.  With a single fd, there is a single
   item to block on, which is mapped to a single queue of events.  The single
   fd returns all watch events and also any potential out-of-band data.  If
   every fd was a separate watch,

   - There would be no way to get event ordering.  Events on file foo and
     file bar would pop poll() on both fd's, but there would be no way to tell
     which happened first.  A single queue trivially gives you ordering.  Such
     ordering is crucial to existing applications such as Beagle.  Imagine
     "mv a b ; mv b a" events without ordering.

   - We'd have to maintain n fd's and n internal queues with state,
     versus just one.  It is a lot messier in the kernel.  A single, linear
     queue is the data structure that makes sense.

   - User-space developers prefer the current API.  The Beagle guys, for
     example, love it.  Trust me, I asked.  It is not a surprise: Who'd want
     to manage and block on 1000 fd's via select?

   - No way to get out of band data.

   - 1024 is still too low.  ;-)

   When you talk about designing a file change notification system that
   scales to 1000s of directories, juggling 1000s of fd's just does not seem
   the right interface.  It is too heavy.

   Additionally, it _is_ possible to  more than one instance  and
   juggle more than one queue and thus more than one associated fd.  There
   need not be a one-fd-per-process mapping; it is one-fd-per-queue and a
   process can easily want more than one queue.

Q: Why the system call approach?

A: The poor user-space interface is the second biggest problem with dnotify.
   Signals are a terrible, terrible interface for file notification.  Or for
   anything, for that matter.  The ideal solution, from all perspectives, is a
   file descriptor-based one that allows basic file I/O and poll/select.
   Obtaining the fd and managing the watches could have been done either via a
   device file or a family of new system calls.  We decided to implement a
   family of system calls because that is the preffered approach for new kernel
   interfaces.  The only real difference was whether we wanted to use open(2)
   and ioctl(2) or a couple of new system calls.  System calls beat ioctls.