kernel_optimize_test/Documentation/filesystems/relayfs.txt

relayfs - a high-speed data relay filesystem
============================================

relayfs is a filesystem designed to provide an efficient mechanism for
tools and facilities to relay large and potentially sustained streams
of data from kernel space to user space.

The main abstraction of relayfs is the 'channel'.  A channel consists
of a set of per-cpu kernel buffers each represented by a file in the
relayfs filesystem.  Kernel clients write into a channel using
efficient write functions which automatically log to the current cpu's
channel buffer.  User space applications mmap() the per-cpu files and
retrieve the data as it becomes available.

The format of the data logged into the channel buffers is completely
up to the relayfs client; relayfs does however provide hooks which
allow clients to impose some structure on the buffer data.  Nor does
relayfs implement any form of data filtering - this also is left to
the client.  The purpose is to keep relayfs as simple as possible.

This document provides an overview of the relayfs API.  The details of
the function parameters are documented along with the functions in the
filesystem code - please see that for details.

Semantics
=========

Each relayfs channel has one buffer per CPU, each buffer has one or
more sub-buffers. Messages are written to the first sub-buffer until
it is too full to contain a new message, in which case it it is
written to the next (if available).  Messages are never split across
sub-buffers.  At this point, userspace can be notified so it empties
the first sub-buffer, while the kernel continues writing to the next.

When notified that a sub-buffer is full, the kernel knows how many
bytes of it are padding i.e. unused.  Userspace can use this knowledge
to copy only valid data.

After copying it, userspace can notify the kernel that a sub-buffer
has been consumed.

relayfs can operate in a mode where it will overwrite data not yet
collected by userspace, and not wait for it to consume it.

relayfs itself does not provide for communication of such data between
userspace and kernel, allowing the kernel side to remain simple and not
impose a single interface on userspace. It does provide a separate
helper though, described below.

klog, relay-app & librelay
==========================

relayfs itself is ready to use, but to make things easier, two
additional systems are provided.  klog is a simple wrapper to make
writing formatted text or raw data to a channel simpler, regardless of
whether a channel to write into exists or not, or whether relayfs is
compiled into the kernel or is configured as a module.  relay-app is
the kernel counterpart of userspace librelay.c, combined these two
files provide glue to easily stream data to disk, without having to
bother with housekeeping.  klog and relay-app can be used together,
with klog providing high-level logging functions to the kernel and
relay-app taking care of kernel-user control and disk-logging chores.

It is possible to use relayfs without relay-app & librelay, but you'll
have to implement communication between userspace and kernel, allowing
both to convey the state of buffers (full, empty, amount of padding).

klog, relay-app and librelay can be found in the relay-apps tarball on
http://relayfs.sourceforge.net

The relayfs user space API
==========================

relayfs implements basic file operations for user space access to
relayfs channel buffer data.  Here are the file operations that are
available and some comments regarding their behavior:

open()	 enables user to open an _existing_ buffer.

mmap()	 results in channel buffer being mapped into the caller's
	 memory space. Note that you can't do a partial mmap - you must
	 map the entire file, which is NRBUF * SUBBUFSIZE.

read()	 read the contents of a channel buffer.  The bytes read are
	 'consumed' by the reader i.e. they won't be available again
	 to subsequent reads.  If the channel is being used in
	 no-overwrite mode (the default), it can be read at any time
	 even if there's an active kernel writer.  If the channel is
	 being used in overwrite mode and there are active channel
	 writers, results may be unpredictable - users should make
	 sure that all logging to the channel has ended before using
	 read() with overwrite mode.

poll()	 POLLIN/POLLRDNORM/POLLERR supported.  User applications are
	 notified when sub-buffer boundaries are crossed.

close() decrements the channel buffer's refcount.  When the refcount
	reaches 0 i.e. when no process or kernel client has the buffer
	open, the channel buffer is freed.


In order for a user application to make use of relayfs files, the
relayfs filesystem must be mounted.  For example,

	mount -t relayfs relayfs /mnt/relay

NOTE:	relayfs doesn't need to be mounted for kernel clients to create
	or use channels - it only needs to be mounted when user space
	applications need access to the buffer data.


The relayfs kernel API
======================

Here's a summary of the API relayfs provides to in-kernel clients:


  channel management functions:

    relay_open(base_filename, parent, subbuf_size, n_subbufs,
               callbacks)
    relay_close(chan)
    relay_flush(chan)
    relay_reset(chan)
    relayfs_create_dir(name, parent)
    relayfs_remove_dir(dentry)
    relayfs_create_file(name, parent, mode, fops, data)
    relayfs_remove_file(dentry)

  channel management typically called on instigation of userspace:

    relay_subbufs_consumed(chan, cpu, subbufs_consumed)

  write functions:

    relay_write(chan, data, length)
    __relay_write(chan, data, length)
    relay_reserve(chan, length)

  callbacks:

    subbuf_start(buf, subbuf, prev_subbuf, prev_padding)
    buf_mapped(buf, filp)
    buf_unmapped(buf, filp)
    create_buf_file(filename, parent, mode, buf, is_global)
    remove_buf_file(dentry)

  helper functions:

    relay_buf_full(buf)
    subbuf_start_reserve(buf, length)


Creating a channel
------------------

relay_open() is used to create a channel, along with its per-cpu
channel buffers.  Each channel buffer will have an associated file
created for it in the relayfs filesystem, which can be opened and
mmapped from user space if desired.  The files are named
basename0...basenameN-1 where N is the number of online cpus, and by
default will be created in the root of the filesystem.  If you want a
directory structure to contain your relayfs files, you can create it
with relayfs_create_dir() and pass the parent directory to
relay_open().  Clients are responsible for cleaning up any directory
structure they create when the channel is closed - use
relayfs_remove_dir() for that.

The total size of each per-cpu buffer is calculated by multiplying the
number of sub-buffers by the sub-buffer size passed into relay_open().
The idea behind sub-buffers is that they're basically an extension of
double-buffering to N buffers, and they also allow applications to
easily implement random-access-on-buffer-boundary schemes, which can
be important for some high-volume applications.  The number and size
of sub-buffers is completely dependent on the application and even for
the same application, different conditions will warrant different
values for these parameters at different times.  Typically, the right
values to use are best decided after some experimentation; in general,
though, it's safe to assume that having only 1 sub-buffer is a bad
idea - you're guaranteed to either overwrite data or lose events
depending on the channel mode being used.

Channel 'modes'
---------------

relayfs channels can be used in either of two modes - 'overwrite' or
'no-overwrite'.  The mode is entirely determined by the implementation
of the subbuf_start() callback, as described below.  In 'overwrite'
mode, also known as 'flight recorder' mode, writes continuously cycle
around the buffer and will never fail, but will unconditionally
overwrite old data regardless of whether it's actually been consumed.
In no-overwrite mode, writes will fail i.e. data will be lost, if the
number of unconsumed sub-buffers equals the total number of
sub-buffers in the channel.  It should be clear that if there is no
consumer or if the consumer can't consume sub-buffers fast enought,
data will be lost in either case; the only difference is whether data
is lost from the beginning or the end of a buffer.

As explained above, a relayfs channel is made of up one or more
per-cpu channel buffers, each implemented as a circular buffer
subdivided into one or more sub-buffers.  Messages are written into
the current sub-buffer of the channel's current per-cpu buffer via the
write functions described below.  Whenever a message can't fit into
the current sub-buffer, because there's no room left for it, the
client is notified via the subbuf_start() callback that a switch to a
new sub-buffer is about to occur.  The client uses this callback to 1)
initialize the next sub-buffer if appropriate 2) finalize the previous
sub-buffer if appropriate and 3) return a boolean value indicating
whether or not to actually go ahead with the sub-buffer switch.

To implement 'no-overwrite' mode, the userspace client would provide
an implementation of the subbuf_start() callback something like the
following:

static int subbuf_start(struct rchan_buf *buf,
                        void *subbuf,
			void *prev_subbuf,
			unsigned int prev_padding)
{
	if (prev_subbuf)
		*((unsigned *)prev_subbuf) = prev_padding;

	if (relay_buf_full(buf))
		return 0;

	subbuf_start_reserve(buf, sizeof(unsigned int));

	return 1;
}

If the current buffer is full i.e. all sub-buffers remain unconsumed,
the callback returns 0 to indicate that the buffer switch should not
occur yet i.e. until the consumer has had a chance to read the current
set of ready sub-buffers.  For the relay_buf_full() function to make
sense, the consumer is reponsible for notifying relayfs when
sub-buffers have been consumed via relay_subbufs_consumed().  Any
subsequent attempts to write into the buffer will again invoke the
subbuf_start() callback with the same parameters; only when the
consumer has consumed one or more of the ready sub-buffers will
relay_buf_full() return 0, in which case the buffer switch can
continue.

The implementation of the subbuf_start() callback for 'overwrite' mode
would be very similar:

static int subbuf_start(struct rchan_buf *buf,
                        void *subbuf,
			void *prev_subbuf,
			unsigned int prev_padding)
{
	if (prev_subbuf)
		*((unsigned *)prev_subbuf) = prev_padding;

	subbuf_start_reserve(buf, sizeof(unsigned int));

	return 1;
}

In this case, the relay_buf_full() check is meaningless and the
callback always returns 1, causing the buffer switch to occur
unconditionally.  It's also meaningless for the client to use the
relay_subbufs_consumed() function in this mode, as it's never
consulted.

The default subbuf_start() implementation, used if the client doesn't
define any callbacks, or doesn't define the subbuf_start() callback,
implements the simplest possible 'no-overwrite' mode i.e. it does
nothing but return 0.

Header information can be reserved at the beginning of each sub-buffer
by calling the subbuf_start_reserve() helper function from within the
subbuf_start() callback.  This reserved area can be used to store
whatever information the client wants.  In the example above, room is
reserved in each sub-buffer to store the padding count for that
sub-buffer.  This is filled in for the previous sub-buffer in the
subbuf_start() implementation; the padding value for the previous
sub-buffer is passed into the subbuf_start() callback along with a
pointer to the previous sub-buffer, since the padding value isn't
known until a sub-buffer is filled.  The subbuf_start() callback is
also called for the first sub-buffer when the channel is opened, to
give the client a chance to reserve space in it.  In this case the
previous sub-buffer pointer passed into the callback will be NULL, so
the client should check the value of the prev_subbuf pointer before
writing into the previous sub-buffer.

Writing to a channel
--------------------

kernel clients write data into the current cpu's channel buffer using
relay_write() or __relay_write().  relay_write() is the main logging
function - it uses local_irqsave() to protect the buffer and should be
used if you might be logging from interrupt context.  If you know
you'll never be logging from interrupt context, you can use
__relay_write(), which only disables preemption.  These functions
don't return a value, so you can't determine whether or not they
failed - the assumption is that you wouldn't want to check a return
value in the fast logging path anyway, and that they'll always succeed
unless the buffer is full and no-overwrite mode is being used, in
which case you can detect a failed write in the subbuf_start()
callback by calling the relay_buf_full() helper function.

relay_reserve() is used to reserve a slot in a channel buffer which
can be written to later.  This would typically be used in applications
that need to write directly into a channel buffer without having to
stage data in a temporary buffer beforehand.  Because the actual write
may not happen immediately after the slot is reserved, applications
using relay_reserve() can keep a count of the number of bytes actually
written, either in space reserved in the sub-buffers themselves or as
a separate array.  See the 'reserve' example in the relay-apps tarball
at http://relayfs.sourceforge.net for an example of how this can be
done.  Because the write is under control of the client and is
separated from the reserve, relay_reserve() doesn't protect the buffer
at all - it's up to the client to provide the appropriate
synchronization when using relay_reserve().

Closing a channel
-----------------

The client calls relay_close() when it's finished using the channel.
The channel and its associated buffers are destroyed when there are no
longer any references to any of the channel buffers.  relay_flush()
forces a sub-buffer switch on all the channel buffers, and can be used
to finalize and process the last sub-buffers before the channel is
closed.

Creating non-relay files
------------------------

relay_open() automatically creates files in the relayfs filesystem to
represent the per-cpu kernel buffers; it's often useful for
applications to be able to create their own files alongside the relay
files in the relayfs filesystem as well e.g. 'control' files much like
those created in /proc or debugfs for similar purposes, used to
communicate control information between the kernel and user sides of a
relayfs application.  For this purpose the relayfs_create_file() and
relayfs_remove_file() API functions exist.  For relayfs_create_file(),
the caller passes in a set of user-defined file operations to be used
for the file and an optional void * to a user-specified data item,
which will be accessible via inode->u.generic_ip (see the relay-apps
tarball for examples).  The file_operations are a required parameter
to relayfs_create_file() and thus the semantics of these files are
completely defined by the caller.

See the relay-apps tarball at http://relayfs.sourceforge.net for
examples of how these non-relay files are meant to be used.

Creating relay files in other filesystems
-----------------------------------------

By default of course, relay_open() creates relay files in the relayfs
filesystem.  Because relay_file_operations is exported, however, it's
also possible to create and use relay files in other pseudo-filesytems
such as debugfs.

For this purpose, two callback functions are provided,
create_buf_file() and remove_buf_file().  create_buf_file() is called
once for each per-cpu buffer from relay_open() to allow the client to
create a file to be used to represent the corresponding buffer; if
this callback is not defined, the default implementation will create
and return a file in the relayfs filesystem to represent the buffer.
The callback should return the dentry of the file created to represent
the relay buffer.  Note that the parent directory passed to
relay_open() (and passed along to the callback), if specified, must
exist in the same filesystem the new relay file is created in.  If
create_buf_file() is defined, remove_buf_file() must also be defined;
it's responsible for deleting the file(s) created in create_buf_file()
and is called during relay_close().

The create_buf_file() implementation can also be defined in such a way
as to allow the creation of a single 'global' buffer instead of the
default per-cpu set.  This can be useful for applications interested
mainly in seeing the relative ordering of system-wide events without
the need to bother with saving explicit timestamps for the purpose of
merging/sorting per-cpu files in a postprocessing step.

To have relay_open() create a global buffer, the create_buf_file()
implementation should set the value of the is_global outparam to a
non-zero value in addition to creating the file that will be used to
represent the single buffer.  In the case of a global buffer,
create_buf_file() and remove_buf_file() will be called only once.  The
normal channel-writing functions e.g. relay_write() can still be used
- writes from any cpu will transparently end up in the global buffer -
but since it is a global buffer, callers should make sure they use the
proper locking for such a buffer, either by wrapping writes in a
spinlock, or by copying a write function from relayfs_fs.h and
creating a local version that internally does the proper locking.

See the 'exported-relayfile' examples in the relay-apps tarball for
examples of creating and using relay files in debugfs.

Misc
----

Some applications may want to keep a channel around and re-use it
rather than open and close a new channel for each use.  relay_reset()
can be used for this purpose - it resets a channel to its initial
state without reallocating channel buffer memory or destroying
existing mappings.  It should however only be called when it's safe to
do so i.e. when the channel isn't currently being written to.

Finally, there are a couple of utility callbacks that can be used for
different purposes.  buf_mapped() is called whenever a channel buffer
is mmapped from user space and buf_unmapped() is called when it's
unmapped.  The client can use this notification to trigger actions
within the kernel application, such as enabling/disabling logging to
the channel.


Resources
=========

For news, example code, mailing list, etc. see the relayfs homepage:

    http://relayfs.sourceforge.net


Credits
=======

The ideas and specs for relayfs came about as a result of discussions
on tracing involving the following:

Michel Dagenais		<michel.dagenais@polymtl.ca>
Richard Moore		<richardj_moore@uk.ibm.com>
Bob Wisniewski		<bob@watson.ibm.com>
Karim Yaghmour		<karim@opersys.com>
Tom Zanussi		<zanussi@us.ibm.com>

Also thanks to Hubertus Franke for a lot of useful suggestions and bug
reports.