File linux-2.6-cachefiles.patch of Package kernel

Overview Repositories Revisions Requests Users Attributes Meta

File linux-2.6-cachefiles.patch of Package kernel

Date: Thu, 27 Jul 2006 19:53:48 +0100
From: David Howells <dhowells@redhat.com>
Subject: [RHEL5 PATCH 28/30] FS-Cache: CacheFiles: A cache that backs onto a mounted filesystem

Add a cache backend that permits a mounted filesystem to be used as a backing
store for the cache.

CacheFiles uses a userspace daemon to do some of the cache management - such as
reaping stale nodes and culling.  This is called cachefilesd and lives in
/sbin.  The source for the daemon can be downloaded from:

http://people.redhat.com/~dhowells/cachefs/cachefilesd.c

And an example configuration from:

http://people.redhat.com/~dhowells/cachefs/cachefilesd.conf

The filesystem and data integrity of the cache are only as good as those of the
filesystem providing the backing services.  Note that CacheFiles does not
attempt to journal anything since the journalling interfaces of the various
filesystems are very specific in nature.

CacheFiles creates a proc-file - "/proc/fs/cachefiles" - that is used for
communication with the daemon.  Only one thing may have this open at once, and
whilst it is open, a cache is at least partially in existence.  The daemon
opens this and sends commands down it to control the cache.

CacheFiles is currently limited to a single cache.

CacheFiles attempts to maintain at least a certain percentage of free space on
the filesystem, shrinking the cache by culling the objects it contains to make
space if necessary - see the "Cache Culling" section.  This means it can be
placed on the same medium as a live set of data, and will expand to make use of
spare space and automatically contract when the set of data requires more
space.

Further changes [try #11] that have been made:

(*) Make the calls to the statfs() superblock op supply a dentry not a
     vfsmount.

(*) CONFIG_CACHEFILES_DEBUG permits _enter(), _debug() and _exit() to be
     enabled dynamically.

(*) debugging macros are checked by gcc for printf format compliance even when
     completely disabled.

============
REQUIREMENTS
============

The use of CacheFiles and its daemon requires the following features to be
available in the system and in the cache filesystem:

- dnotify.

- extended attributes (xattrs).

- openat() and friends.

- bmap() support on files in the filesystem (FIBMAP ioctl).

- The use of bmap() to detect a partial page at the end of the file.

It is strongly recommended that the "dir_index" option is enabled on Ext3
filesystems being used as a cache.

=============
CONFIGURATION
=============

The cache is configured by a script in /etc/cachefilesd.conf.  These commands
set up cache ready for use.  The following script commands are available:

(*) brun <N>%
 (*) bcull <N>%
 (*) bstop <N>%

Configure the culling limits.  Optional.  See the section on culling
	The defaults are 7%, 5% and 1% respectively.

(*) dir <path>

Specify the directory containing the root of the cache.  Mandatory.

(*) tag <name>

Specify a tag to FS-Cache to use in distinguishing multiple caches.
	Optional.  The default is "CacheFiles".

(*) debug <mask>

Specify a numeric bitmask to control debugging in the kernel module.
	Optional.  The default is zero (all off).

==================
STARTING THE CACHE
==================

The cache is started by running the daemon.  The daemon opens the cache proc
file, configures the cache and tells it to begin caching.  At that point the
cache binds to fscache and the cache becomes live.

The daemon is run as follows:

/sbin/cachefilesd [-d]* [-s] [-n] [-f <configfile>]

The flags are:

(*) -d

Increase the debugging level.  This can be specified multiple times and
	is cumulative with itself.

(*) -s

Send messages to stderr instead of syslog.

(*) -n

Don't daemonise and go into background.

(*) -f <configfile>

Use an alternative configuration file rather than the default one.

===============
THINGS TO AVOID
===============

Do not mount other things within the cache as this will cause problems.  The
kernel module contains its own very cut-down path walking facility that ignores
mountpoints, but the daemon can't avoid them.

Do not create, rename or unlink files and directories in the cache whilst the
cache is active, as this may cause the state to become uncertain.

Renaming files in the cache might make objects appear to be other objects (the
filename is part of the lookup key).

Do not change or remove the extended attributes attached to cache files by the
cache as this will cause the cache state management to get confused.

Do not create files or directories in the cache, lest the cache get confused or
serve incorrect data.

Do not chmod files in the cache.  The module creates things with minimal
permissions to prevent random users being able to access them directly.

=============
CACHE CULLING
=============

The cache may need culling occasionally to make space.  This involves
discarding objects from the cache that have been used less recently than
anything else.  Culling is based on the access time of data objects.  Empty
directories are culled if not in use.

Cache culling is done on the basis of the percentage of blocks available in the
underlying filesystem.  There are three "limits":

(*) brun

If the amount of available space in the cache rises above this limit, then
     culling is turned off.

(*) bcull

If the amount of available space in the cache falls below this limit, then
     culling is started.

(*) bstop

If the amount of available space in the cache falls below this limit, then
     no further allocation of disk space is permitted until culling has raised
     the amount above this limit again.

These must be configured thusly:

0 <= bstop < bcull < brun < 100

Note that these are percentages of available space, and do _not_ appear as 100
minus the percentage displayed by the "df" program.

The userspace daemon scans the cache to build up a table of cullable objects.
These are then culled in least recently used order.  A new scan of the cache is
started as soon as space is made in the table.  Objects will be skipped if
their atimes have changed or if the kernel module says it is still using them.

===============
CACHE STRUCTURE
===============

The CacheFiles module will create two directories in the directory it was
given:

(*) cache/

(*) graveyard/

The active cache objects all reside in the first directory.  The CacheFiles
kernel module moves any retired or culled objects that it can't simply unlink
to the graveyard from which the daemon will actually delete them.

The daemon uses dnotify to monitor the graveyard directory, and will delete
anything that appears therein.

The module represents index objects as directories with the filename "I..." or
"J...".  Note that the "cache/" directory is itself a special index.

Data objects are represented as files if they have no children, or directories
if they do.  Their filenames all begin "D..." or "E...".  If represented as a
directory, data objects will have a file in the directory called "data" that
actually holds the data.

Special objects are similar to data objects, except their filenames begin
"S..." or "T...".

If an object has children, then it will be represented as a directory.
Immediately in the representative directory are a collection of directories
named for hash values of the child object keys with an '@' prepended.  Into
this directory, if possible, will be placed the representations of the child
objects:

INDEX     INDEX      INDEX                             DATA FILES
	========= ========== ================================= ================
	cache/@4a/I03nfs/@30/Ji000000000000000--fHg8hi8400
	cache/@4a/I03nfs/@30/Ji000000000000000--fHg8hi8400/@75/Es0g000w...DB1ry
	cache/@4a/I03nfs/@30/Ji000000000000000--fHg8hi8400/@75/Es0g000w...N22ry
	cache/@4a/I03nfs/@30/Ji000000000000000--fHg8hi8400/@75/Es0g000w...FP1ry

If the key is so long that it exceeds NAME_MAX with the decorations added on to
it, then it will be cut into pieces, the first few of which will be used to
make a nest of directories, and the last one of which will be the objects
inside the last directory.  The names of the intermediate directories will have
'+' prepended:

J1223/@23/+xy...z/+kl...m/Epqr

Note that keys are raw data, and not only may they exceed NAME_MAX in size,
they may also contain things like '/' and NUL characters, and so they may not
be suitable for turning directly into a filename.

To handle this, CacheFiles will use a suitably printable filename directly and
"base-64" encode ones that aren't directly suitable.  The two versions of
object filenames indicate the encoding:

OBJECT TYPE	PRINTABLE	ENCODED
	===============	===============	===============
	Index		"I..."		"J..."
	Data		"D..."		"E..."
	Special		"S..."		"T..."

Intermediate directories are always "@" or "+" as appropriate.

Each object in the cache has an extended attribute label that holds the object
type ID (required to distinguish special objects) and the auxiliary data from
the netfs.  The latter is used to detect stale objects in the cache and update
or retire them.

Note that CacheFiles will erase from the cache any file it doesn't recognise or
any file of an incorrect type (such as a FIFO file or a device file).

This documentation is added by the patch to:

Documentation/filesystems/caching/cachefiles.txt

Signed-Off-By: David Howells <dhowells@redhat.com>
---

diff --git a/Documentation/filesystems/caching/cachefiles.txt b/Documentation/filesystems/caching/cachefiles.txt
new file mode 100644
index 0000000..37b6385
--- /dev/null
+++ b/Documentation/filesystems/caching/cachefiles.txt
@@ -0,0 +1,281 @@
+	       ===============================================
+	       CacheFiles: CACHE ON ALREADY MOUNTED FILESYSTEM
+	       ===============================================
+
+Contents:
+
+ (*) Overview.
+
+ (*) Requirements.
+
+ (*) Configuration.
+
+ (*) Starting the cache.
+
+ (*) Things to avoid.
+
+
+========
+OVERVIEW
+========
+
+CacheFiles is a caching backend that's meant to use as a cache a directory on
+an already mounted filesystem of a local type (such as Ext3).
+
+CacheFiles uses a userspace daemon to do some of the cache management - such as
+reaping stale nodes and culling.  This is called cachefilesd and lives in
+/sbin.
+
+The filesystem and data integrity of the cache are only as good as those of the
+filesystem providing the backing services.  Note that CacheFiles does not
+attempt to journal anything since the journalling interfaces of the various
+filesystems are very specific in nature.
+
+CacheFiles creates a proc-file - "/proc/fs/cachefiles" - that is used for
+communication with the daemon.  Only one thing may have this open at once, and
+whilst it is open, a cache is at least partially in existence.  The daemon
+opens this and sends commands down it to control the cache.
+
+CacheFiles is currently limited to a single cache.
+
+CacheFiles attempts to maintain at least a certain percentage of free space on
+the filesystem, shrinking the cache by culling the objects it contains to make
+space if necessary - see the "Cache Culling" section.  This means it can be
+placed on the same medium as a live set of data, and will expand to make use of
+spare space and automatically contract when the set of data requires more
+space.
+
+
+============
+REQUIREMENTS
+============
+
+The use of CacheFiles and its daemon requires the following features to be
+available in the system and in the cache filesystem:
+
+	- dnotify.
+
+	- extended attributes (xattrs).
+
+	- openat() and friends.
+
+	- bmap() support on files in the filesystem (FIBMAP ioctl).
+
+	- The use of bmap() to detect a partial page at the end of the file.
+
+It is strongly recommended that the "dir_index" option is enabled on Ext3
+filesystems being used as a cache.
+
+
+=============
+CONFIGURATION
+=============
+
+The cache is configured by a script in /etc/cachefilesd.conf.  These commands
+set up cache ready for use.  The following script commands are available:
+
+ (*) brun <N>%
+ (*) bcull <N>%
+ (*) bstop <N>%
+
+	Configure the culling limits.  Optional.  See the section on culling
+	The defaults are 7%, 5% and 1% respectively.
+
+ (*) dir <path>
+
+	Specify the directory containing the root of the cache.  Mandatory.
+
+ (*) tag <name>
+
+	Specify a tag to FS-Cache to use in distinguishing multiple caches.
+	Optional.  The default is "CacheFiles".
+
+ (*) debug <mask>
+
+	Specify a numeric bitmask to control debugging in the kernel module.
+	Optional.  The default is zero (all off).  The following values can be
+	OR'd into the mask to collect various information:
+
+		1	Turn on trace of function entry (_enter() macros)
+		2	Turn on trace of function exit (_leave() macros)
+		4	Turn on trace of internal debug points (_debug())
+
+	This mask can also be set through /proc/sys/fs/cachefiles/debug.
+
+
+==================
+STARTING THE CACHE
+==================
+
+The cache is started by running the daemon.  The daemon opens the cache proc
+file, configures the cache and tells it to begin caching.  At that point the
+cache binds to fscache and the cache becomes live.
+
+The daemon is run as follows:
+
+	/sbin/cachefilesd [-d]* [-s] [-n] [-f <configfile>]
+
+The flags are:
+
+ (*) -d
+
+	Increase the debugging level.  This can be specified multiple times and
+	is cumulative with itself.
+
+ (*) -s
+
+	Send messages to stderr instead of syslog.
+
+ (*) -n
+
+	Don't daemonise and go into background.
+
+ (*) -f <configfile>
+
+	Use an alternative configuration file rather than the default one.
+
+
+===============
+THINGS TO AVOID
+===============
+
+Do not mount other things within the cache as this will cause problems.  The
+kernel module contains its own very cut-down path walking facility that ignores
+mountpoints, but the daemon can't avoid them.
+
+Do not create, rename or unlink files and directories in the cache whilst the
+cache is active, as this may cause the state to become uncertain.
+
+Renaming files in the cache might make objects appear to be other objects (the
+filename is part of the lookup key).
+
+Do not change or remove the extended attributes attached to cache files by the
+cache as this will cause the cache state management to get confused.
+
+Do not create files or directories in the cache, lest the cache get confused or
+serve incorrect data.
+
+Do not chmod files in the cache.  The module creates things with minimal
+permissions to prevent random users being able to access them directly.
+
+
+=============
+CACHE CULLING
+=============
+
+The cache may need culling occasionally to make space.  This involves
+discarding objects from the cache that have been used less recently than
+anything else.  Culling is based on the access time of data objects.  Empty
+directories are culled if not in use.
+
+Cache culling is done on the basis of the percentage of blocks available in the
+underlying filesystem.  There are three "limits":
+
+ (*) brun
+
+     If the amount of available space in the cache rises above this limit, then
+     culling is turned off.
+
+ (*) bcull
+
+     If the amount of available space in the cache falls below this limit, then
+     culling is started.
+
+ (*) bstop
+
+     If the amount of available space in the cache falls below this limit, then
+     no further allocation of disk space is permitted until culling has raised
+     the amount above this limit again.
+
+These must be configured thusly:
+
+	0 <= bstop < bcull < brun < 100
+
+Note that these are percentages of available space, and do _not_ appear as 100
+minus the percentage displayed by the "df" program.
+
+The userspace daemon scans the cache to build up a table of cullable objects.
+These are then culled in least recently used order.  A new scan of the cache is
+started as soon as space is made in the table.  Objects will be skipped if
+their atimes have changed or if the kernel module says it is still using them.
+
+
+===============
+CACHE STRUCTURE
+===============
+
+The CacheFiles module will create two directories in the directory it was
+given:
+
+ (*) cache/
+
+ (*) graveyard/
+
+The active cache objects all reside in the first directory.  The CacheFiles
+kernel module moves any retired or culled objects that it can't simply unlink
+to the graveyard from which the daemon will actually delete them.
+
+The daemon uses dnotify to monitor the graveyard directory, and will delete
+anything that appears therein.
+
+
+The module represents index objects as directories with the filename "I..." or
+"J...".  Note that the "cache/" directory is itself a special index.
+
+Data objects are represented as files if they have no children, or directories
+if they do.  Their filenames all begin "D..." or "E...".  If represented as a
+directory, data objects will have a file in the directory called "data" that
+actually holds the data.
+
+Special objects are similar to data objects, except their filenames begin
+"S..." or "T...".
+
+
+If an object has children, then it will be represented as a directory.
+Immediately in the representative directory are a collection of directories
+named for hash values of the child object keys with an '@' prepended.  Into
+this directory, if possible, will be placed the representations of the child
+objects:
+
+	INDEX     INDEX      INDEX                             DATA FILES
+	========= ========== ================================= ================
+	cache/@4a/I03nfs/@30/Ji000000000000000--fHg8hi8400
+	cache/@4a/I03nfs/@30/Ji000000000000000--fHg8hi8400/@75/Es0g000w...DB1ry
+	cache/@4a/I03nfs/@30/Ji000000000000000--fHg8hi8400/@75/Es0g000w...N22ry
+	cache/@4a/I03nfs/@30/Ji000000000000000--fHg8hi8400/@75/Es0g000w...FP1ry
+
+
+If the key is so long that it exceeds NAME_MAX with the decorations added on to
+it, then it will be cut into pieces, the first few of which will be used to
+make a nest of directories, and the last one of which will be the objects
+inside the last directory.  The names of the intermediate directories will have
+'+' prepended:
+
+	J1223/@23/+xy...z/+kl...m/Epqr
+
+
+Note that keys are raw data, and not only may they exceed NAME_MAX in size,
+they may also contain things like '/' and NUL characters, and so they may not
+be suitable for turning directly into a filename.
+
+To handle this, CacheFiles will use a suitably printable filename directly and
+"base-64" encode ones that aren't directly suitable.  The two versions of
+object filenames indicate the encoding:
+
+	OBJECT TYPE	PRINTABLE	ENCODED
+	===============	===============	===============
+	Index		"I..."		"J..."
+	Data		"D..."		"E..."
+	Special		"S..."		"T..."
+
+Intermediate directories are always "@" or "+" as appropriate.
+
+
+Each object in the cache has an extended attribute label that holds the object
+type ID (required to distinguish special objects) and the auxiliary data from
+the netfs.  The latter is used to detect stale objects in the cache and update
+or retire them.
+
+
+Note that CacheFiles will erase from the cache any file it doesn't recognise or
+any file of an incorrect type (such as a FIFO file or a device file).
diff --git a/fs/Kconfig b/fs/Kconfig
index 36d0051..afec7e1 100644
--- a/fs/Kconfig
+++ b/fs/Kconfig
@@ -543,6 +543,26 @@ config FSCACHE
 
 	  See Documentation/filesystems/caching/fscache.txt for more information.
 
+config CACHEFILES
+	tristate "Filesystem caching on files"
+	depends on FSCACHE
+	help
+	  This permits use of a mounted filesystem as a cache for other
+	  filesystems - primarily networking filesystems - thus allowing fast
+	  local disk to enhance the speed of slower devices.
+
+	  See Documentation/filesystems/caching/cachefiles.txt for more
+	  information.
+
+config CACHEFILES_DEBUG
+	bool "Debug CacheFiles"
+	depends on CACHEFILES
+	help
+	  This permits debugging to be dynamically enabled in the filesystem
+	  caching on files module.  If this is set, the debugging output may be
+	  enabled by setting bits in /proc/sys/fs/cachefiles/debug or by
+	  including a debugging specifier in /etc/cachefilesd.conf.
+
 endmenu
 
 menu "CD-ROM/DVD Filesystems"
diff --git a/fs/Makefile b/fs/Makefile
index 0b0e827..e5efb4b 100644
--- a/fs/Makefile
+++ b/fs/Makefile
@@ -101,5 +101,6 @@ obj-$(CONFIG_AFS_FS)		+= afs/
 obj-$(CONFIG_BEFS_FS)		+= befs/
 obj-$(CONFIG_HOSTFS)		+= hostfs/
 obj-$(CONFIG_HPPFS)		+= hppfs/
+obj-$(CONFIG_CACHEFILES)	+= cachefiles/
 obj-$(CONFIG_DEBUG_FS)		+= debugfs/
 obj-$(CONFIG_OCFS2_FS)		+= ocfs2/
diff --git a/fs/cachefiles/Makefile b/fs/cachefiles/Makefile
new file mode 100644
index 0000000..c1522d9
--- /dev/null
+++ b/fs/cachefiles/Makefile
@@ -0,0 +1,18 @@
+#
+# Makefile for caching in a mounted filesystem
+#
+
+cachefiles-objs := \
+	cf-bind.o \
+	cf-interface.o \
+	cf-key.o \
+	cf-main.o \
+	cf-namei.o \
+	cf-proc.o \
+	cf-xattr.o
+
+ifeq ($(CONFIG_SYSCTL),y)
+cachefiles-objs += cf-sysctl.o
+endif
+
+obj-$(CONFIG_CACHEFILES) := cachefiles.o
diff --git a/fs/cachefiles/cf-bind.c b/fs/cachefiles/cf-bind.c
new file mode 100644
index 0000000..d8cb4c3
--- /dev/null
+++ b/fs/cachefiles/cf-bind.c
@@ -0,0 +1,279 @@
+/* cf-bind.c: bind and unbind a cache from the filesystem backing it
+ *
+ * Copyright (C) 2006 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/sched.h>
+#include <linux/completion.h>
+#include <linux/slab.h>
+#include <linux/fs.h>
+#include <linux/file.h>
+#include <linux/namei.h>
+#include <linux/mount.h>
+#include <linux/namespace.h>
+#include <linux/statfs.h>
+#include <linux/proc_fs.h>
+#include <linux/ctype.h>
+#include "internal.h"
+
+static int cachefiles_proc_add_cache(struct cachefiles_cache *cache,
+				     struct vfsmount *mnt);
+
+/*****************************************************************************/
+/*
+ * bind a directory as a cache
+ */
+int cachefiles_proc_bind(struct cachefiles_cache *cache, char *args)
+{
+	_enter("{%u,%u,%u},%s",
+	       cache->brun_percent,
+	       cache->bcull_percent,
+	       cache->bstop_percent,
+	       args);
+
+	/* start by checking things over */
+	ASSERT(cache->bstop_percent >= 0 &&
+	       cache->bstop_percent < cache->bcull_percent &&
+	       cache->bcull_percent < cache->brun_percent &&
+	       cache->brun_percent  < 100);
+
+	if (*args) {
+		kerror("'bind' command doesn't take an argument");
+		return -EINVAL;
+	}
+
+	if (!cache->rootdirname) {
+		kerror("No cache directory specified");
+		return -EINVAL;
+	}
+
+	/* don't permit already bound caches to be re-bound */
+	if (test_bit(CACHEFILES_READY, &cache->flags)) {
+		kerror("Cache already bound");
+		return -EBUSY;
+	}
+
+	/* make sure we have copies of the tag and dirname strings */
+	if (!cache->tag) {
+		/* the tag string is released by the fops->release()
+		 * function, so we don't release it on error here */
+		cache->tag = kstrdup("CacheFiles", GFP_KERNEL);
+		if (!cache->tag)
+			return -ENOMEM;
+	}
+
+	/* add the cache */
+	return cachefiles_proc_add_cache(cache, NULL);
+}
+
+/*****************************************************************************/
+/*
+ * add a cache
+ */
+static int cachefiles_proc_add_cache(struct cachefiles_cache *cache,
+				     struct vfsmount *mnt)
+{
+	struct cachefiles_object *fsdef;
+	struct nameidata nd;
+	struct kstatfs stats;
+	struct dentry *graveyard, *cachedir, *root;
+	int ret;
+
+	_enter("");
+
+	/* allocate the root index object */
+	ret = -ENOMEM;
+
+	fsdef = kmem_cache_alloc(cachefiles_object_jar, SLAB_KERNEL);
+	if (!fsdef)
+		goto error_root_object;
+
+	atomic_set(&fsdef->usage, 1);
+	atomic_set(&fsdef->fscache_usage, 1);
+	fsdef->type = FSCACHE_COOKIE_TYPE_INDEX;
+
+	_debug("- fsdef %p", fsdef);
+
+	/* look up the directory at the root of the cache */
+	memset(&nd, 0, sizeof(nd));
+
+	ret = path_lookup(cache->rootdirname, LOOKUP_DIRECTORY, &nd);
+	if (ret < 0)
+		goto error_open_root;
+
+	/* bind to the special mountpoint we've prepared */
+	if (mnt) {
+		atomic_inc(&nd.mnt->mnt_sb->s_active);
+		mnt->mnt_sb = nd.mnt->mnt_sb;
+		mnt->mnt_flags = nd.mnt->mnt_flags;
+		mnt->mnt_flags |= MNT_NOSUID | MNT_NOEXEC | MNT_NODEV;
+		mnt->mnt_root = dget(nd.dentry);
+		mnt->mnt_mountpoint = mnt->mnt_root;
+
+		/* copy the name, but ignore kstrdup() failing ENOMEM - we'll
+		 * just end up with an devicenameless mountpoint */
+		mnt->mnt_devname = kstrdup(nd.mnt->mnt_devname, GFP_KERNEL);
+		path_release(&nd);
+
+		cache->mnt = mntget(mnt);
+		root = dget(mnt->mnt_root);
+	} else {
+		cache->mnt = nd.mnt;
+		root = nd.dentry;
+
+		nd.mnt = NULL;
+		nd.dentry = NULL;
+		path_release(&nd);
+	}
+
+	/* check parameters */
+	ret = -EOPNOTSUPP;
+	if (!root->d_inode ||
+	    !root->d_inode->i_op ||
+	    !root->d_inode->i_op->lookup ||
+	    !root->d_inode->i_op->mkdir ||
+	    !root->d_inode->i_op->setxattr ||
+	    !root->d_inode->i_op->getxattr ||
+	    !root->d_sb ||
+	    !root->d_sb->s_op ||
+	    !root->d_sb->s_op->statfs ||
+	    !root->d_sb->s_op->sync_fs)
+		goto error_unsupported;
+
+	ret = -EROFS;
+	if (root->d_sb->s_flags & MS_RDONLY)
+		goto error_unsupported;
+
+	/* get the cache size and blocksize */
+	ret = root->d_sb->s_op->statfs(root, &stats);
+	if (ret < 0)
+		goto error_unsupported;
+
+	ret = -ERANGE;
+	if (stats.f_bsize <= 0)
+		goto error_unsupported;
+
+	ret = -EOPNOTSUPP;
+	if (stats.f_bsize > PAGE_SIZE)
+		goto error_unsupported;
+
+	cache->bsize = stats.f_bsize;
+	cache->bshift = 0;
+	if (stats.f_bsize < PAGE_SIZE)
+		cache->bshift = PAGE_SHIFT - long_log2(stats.f_bsize);
+
+	_debug("blksize %u (shift %u)",
+	       cache->bsize, cache->bshift);
+
+	_debug("size %llu, avail %llu", stats.f_blocks, stats.f_bavail);
+
+	/* set up caching limits */
+	stats.f_blocks >>= cache->bshift;
+	do_div(stats.f_blocks, 100);
+	cache->bstop = stats.f_blocks * cache->bstop_percent;
+	cache->bcull = stats.f_blocks * cache->bcull_percent;
+	cache->brun  = stats.f_blocks * cache->brun_percent;
+
+	_debug("limits {%llu,%llu,%llu}",
+	       cache->brun,
+	       cache->bcull,
+	       cache->bstop);
+
+	/* get the cache directory and check its type */
+	cachedir = cachefiles_get_directory(cache, root, "cache");
+	if (IS_ERR(cachedir)) {
+		ret = PTR_ERR(cachedir);
+		goto error_unsupported;
+	}
+
+	fsdef->dentry = cachedir;
+
+	ret = cachefiles_check_object_type(fsdef);
+	if (ret < 0)
+		goto error_unsupported;
+
+	/* get the graveyard directory */
+	graveyard = cachefiles_get_directory(cache, root, "graveyard");
+	if (IS_ERR(graveyard)) {
+		ret = PTR_ERR(graveyard);
+		goto error_unsupported;
+	}
+
+	cache->graveyard = graveyard;
+
+	/* publish the cache */
+	fscache_init_cache(&cache->cache,
+			   &cachefiles_cache_ops,
+			   "%02x:%02x",
+			   MAJOR(fsdef->dentry->d_sb->s_dev),
+			   MINOR(fsdef->dentry->d_sb->s_dev)
+			   );
+
+	ret = fscache_add_cache(&cache->cache, &fsdef->fscache, cache->tag);
+	if (ret < 0)
+		goto error_add_cache;
+
+	/* done */
+	set_bit(CACHEFILES_READY, &cache->flags);
+	dput(root);
+
+	printk(KERN_INFO "CacheFiles:"
+	       " File cache on %s registered\n",
+	       cache->cache.identifier);
+
+	/* check how much space the cache has */
+	cachefiles_has_space(cache, 0);
+
+	return 0;
+
+error_add_cache:
+	dput(cache->graveyard);
+	cache->graveyard = NULL;
+error_unsupported:
+	mntput(cache->mnt);
+	cache->mnt = NULL;
+	dput(fsdef->dentry);
+	fsdef->dentry = NULL;
+	dput(root);
+error_open_root:
+	kmem_cache_free(cachefiles_object_jar, fsdef);
+error_root_object:
+	kerror("Failed to register: %d", ret);
+	return ret;
+}
+
+/*****************************************************************************/
+/*
+ * unbind a cache on fd release
+ */
+void cachefiles_proc_unbind(struct cachefiles_cache *cache)
+{
+	_enter("");
+
+	if (test_bit(CACHEFILES_READY, &cache->flags)) {
+		printk(KERN_INFO "CacheFiles:"
+		       " File cache on %s unregistering\n",
+		       cache->cache.identifier);
+
+		fscache_withdraw_cache(&cache->cache);
+	}
+
+	if (cache->cache.fsdef)
+		cache->cache.ops->put_object(cache->cache.fsdef);
+
+	dput(cache->graveyard);
+	mntput(cache->mnt);
+
+	kfree(cache->rootdirname);
+	kfree(cache->tag);
+
+	_leave("");
+}
diff --git a/fs/cachefiles/cf-interface.c b/fs/cachefiles/cf-interface.c
new file mode 100644
index 0000000..b5ca30f
--- /dev/null
+++ b/fs/cachefiles/cf-interface.c
@@ -0,0 +1,1299 @@
+/* cf-interface.c: CacheFiles to FS-Cache interface
+ *
+ * Copyright (C) 2006 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include <linux/file.h>
+#include <linux/mount.h>
+#include <linux/statfs.h>
+#include <linux/buffer_head.h>
+#include "internal.h"
+
+#define list_to_page(head) (list_entry((head)->prev, struct page, lru))
+#define log2(n) ffz(~(n))
+
+/*****************************************************************************/
+/*
+ * look up the nominated node in this cache, creating it if necessary
+ */
+static struct fscache_object *cachefiles_lookup_object(
+	struct fscache_cache *_cache,
+	struct fscache_object *_parent,
+	struct fscache_cookie *cookie)
+{
+	struct cachefiles_object *parent, *object;
+	struct cachefiles_cache *cache;
+	struct cachefiles_xattr *auxdata;
+	unsigned keylen, auxlen;
+	void *buffer;
+	char *key;
+	int ret;
+
+	ASSERT(_parent);
+
+	cache = container_of(_cache, struct cachefiles_cache, cache);
+	parent = container_of(_parent, struct cachefiles_object, fscache);
+
+	_enter("{%s},%p,%p", cache->cache.identifier, parent, cookie);
+
+	/* create a new object record and a temporary leaf image */
+	object = kmem_cache_alloc(cachefiles_object_jar, SLAB_KERNEL);
+	if (!object)
+		goto nomem_object;
+
+	atomic_set(&object->usage, 1);
+	atomic_set(&object->fscache_usage, 1);
+
+	fscache_object_init(&object->fscache);
+	object->fscache.cookie = cookie;
+	object->fscache.cache = parent->fscache.cache;
+
+	object->type = cookie->def->type;
+
+	/* get hold of the raw key
+	 * - stick the length on the front and leave space on the back for the
+	 *   encoder
+	 */
+	buffer = kmalloc((2 + 512) + 3, GFP_KERNEL);
+	if (!buffer)
+		goto nomem_buffer;
+
+	keylen = cookie->def->get_key(cookie->netfs_data, buffer + 2, 512);
+	ASSERTCMP(keylen, <, 512);
+
+	*(uint16_t *)buffer = keylen;
+	((char *)buffer)[keylen + 2] = 0;
+	((char *)buffer)[keylen + 3] = 0;
+	((char *)buffer)[keylen + 4] = 0;
+
+	/* turn the raw key into something that can work with as a filename */
+	key = cachefiles_cook_key(buffer, keylen + 2, object->type);
+	if (!key)
+		goto nomem_key;
+
+	/* get hold of the auxiliary data and prepend the object type */
+	auxdata = buffer;
+	auxlen = 0;
+	if (cookie->def->get_aux) {
+		auxlen = cookie->def->get_aux(cookie->netfs_data,
+					      auxdata->data, 511);
+		ASSERTCMP(auxlen, <, 511);
+	}
+
+	auxdata->len = auxlen + 1;
+	auxdata->type = cookie->def->type;
+
+	/* look up the key, creating any missing bits */
+	ret = cachefiles_walk_to_object(parent, object, key, auxdata);
+	if (ret < 0)
+		goto lookup_failed;
+
+	kfree(buffer);
+	kfree(key);
+	_leave(" = %p", &object->fscache);
+	return &object->fscache;
+
+lookup_failed:
+	kmem_cache_free(cachefiles_object_jar, object);
+	kfree(buffer);
+	kfree(key);
+	_leave(" = %d", ret);
+	return ERR_PTR(ret);
+
+nomem_key:
+	kfree(buffer);
+nomem_buffer:
+	kmem_cache_free(cachefiles_object_jar, object);
+nomem_object:
+	_leave(" = -ENOMEM");
+	return ERR_PTR(-ENOMEM);
+
+}
+
+/*****************************************************************************/
+/*
+ * increment the usage count on an inode object (may fail if unmounting)
+ */
+static struct fscache_object *cachefiles_grab_object(struct fscache_object *_object)
+{
+	struct cachefiles_object *object;
+
+	_enter("%p", _object);
+
+	object = container_of(_object, struct cachefiles_object, fscache);
+
+#ifdef CACHEFILES_DEBUG_SLAB
+	ASSERT((atomic_read(&object->fscache_usage) & 0xffff0000) != 0x6b6b0000);
+#endif
+
+	atomic_inc(&object->fscache_usage);
+	return &object->fscache;
+
+}
+
+/*****************************************************************************/
+/*
+ * lock the semaphore on an object object
+ */
+static void cachefiles_lock_object(struct fscache_object *_object)
+{
+	struct cachefiles_object *object;
+
+	_enter("%p", _object);
+
+	object = container_of(_object, struct cachefiles_object, fscache);
+
+#ifdef CACHEFILES_DEBUG_SLAB
+	ASSERT((atomic_read(&object->fscache_usage) & 0xffff0000) != 0x6b6b0000);
+#endif
+
+	down_write(&object->sem);
+
+}
+
+/*****************************************************************************/
+/*
+ * unlock the semaphore on an object object
+ */
+static void cachefiles_unlock_object(struct fscache_object *_object)
+{
+	struct cachefiles_object *object;
+
+	_enter("%p", _object);
+
+	object = container_of(_object, struct cachefiles_object, fscache);
+	up_write(&object->sem);
+
+}
+
+/*****************************************************************************/
+/*
+ * update the auxilliary data for an object object on disk
+ */
+static void cachefiles_update_object(struct fscache_object *_object)
+{
+	struct cachefiles_object *object;
+	struct cachefiles_cache *cache;
+
+	_enter("%p", _object);
+
+	object = container_of(_object, struct cachefiles_object, fscache);
+	cache = container_of(object->fscache.cache, struct cachefiles_cache, cache);
+
+	//cachefiles_tree_update_object(super, object);
+
+}
+
+/*****************************************************************************/
+/*
+ * dispose of a reference to an object object
+ */
+static void cachefiles_put_object(struct fscache_object *_object)
+{
+	struct cachefiles_object *object;
+	struct cachefiles_cache *cache;
+
+	ASSERT(_object);
+
+	object = container_of(_object, struct cachefiles_object, fscache);
+	_enter("%p{%d}", object, atomic_read(&object->usage));
+
+	ASSERT(object);
+
+	cache = container_of(object->fscache.cache,
+			     struct cachefiles_cache, cache);
+
+#ifdef CACHEFILES_DEBUG_SLAB
+	ASSERT((atomic_read(&object->fscache_usage) & 0xffff0000) != 0x6b6b0000);
+#endif
+
+	if (!atomic_dec_and_test(&object->fscache_usage))
+		return;
+
+	_debug("- kill object %p", object);
+
+	/* delete retired objects */
+	if (test_bit(FSCACHE_OBJECT_RECYCLING, &object->fscache.flags) &&
+	    _object != cache->cache.fsdef
+	    ) {
+		_debug("- retire object %p", object);
+		cachefiles_delete_object(cache, object);
+	}
+
+	/* close the filesystem stuff attached to the object */
+	if (object->backer != object->dentry) {
+		dput(object->backer);
+		object->backer = NULL;
+	}
+
+	/* note that an object is now inactive */
+	write_lock(&cache->active_lock);
+	rb_erase(&object->active_node, &cache->active_nodes);
+	write_unlock(&cache->active_lock);
+
+	dput(object->dentry);
+	object->dentry = NULL;
+
+	/* then dispose of the object */
+	kmem_cache_free(cachefiles_object_jar, object);
+
+	_leave("");
+
+}
+
+/*****************************************************************************/
+/*
+ * sync a cache
+ */
+static void cachefiles_sync_cache(struct fscache_cache *_cache)
+{
+	struct cachefiles_cache *cache;
+	int ret;
+
+	_enter("%p", _cache);
+
+	cache = container_of(_cache, struct cachefiles_cache, cache);
+
+	/* make sure all pages pinned by operations on behalf of the netfs are
+	 * written to disc */
+	ret = fsync_super(cache->mnt->mnt_sb);
+	if (ret == -EIO)
+		cachefiles_io_error(cache,
+				    "Attempt to sync backing fs superblock"
+				    " returned error %d",
+				    ret);
+
+}
+
+/*****************************************************************************/
+/*
+ * set the data size on an object
+ */
+static int cachefiles_set_i_size(struct fscache_object *_object, loff_t i_size)
+{
+	struct cachefiles_object *object;
+	struct iattr newattrs;
+	int ret;
+
+	_enter("%p,%llu", _object, i_size);
+
+	object = container_of(_object, struct cachefiles_object, fscache);
+
+	if (i_size == object->i_size)
+		return 0;
+
+	if (!object->backer)
+		return -ENOBUFS;
+
+	ASSERT(S_ISREG(object->backer->d_inode->i_mode));
+
+	newattrs.ia_size = i_size;
+	newattrs.ia_valid = ATTR_SIZE;
+
+	mutex_lock(&object->backer->d_inode->i_mutex);
+	ret = notify_change(object->backer, &newattrs);
+	mutex_unlock(&object->backer->d_inode->i_mutex);
+
+	if (ret == -EIO) {
+		cachefiles_io_error_obj(object, "Size set failed");
+		ret = -ENOBUFS;
+	}
+
+	_leave(" = %d", ret);
+	return ret;
+
+}
+
+/*****************************************************************************/
+/*
+ * see if we have space for a number of pages in the cache
+ */
+int cachefiles_has_space(struct cachefiles_cache *cache, unsigned nr)
+{
+	struct kstatfs stats;
+	int ret;
+
+	_enter("{%llu,%llu,%llu},%d",
+	       cache->brun, cache->bcull, cache->bstop,  nr);
+
+	/* find out how many pages of blockdev are available */
+	memset(&stats, 0, sizeof(stats));
+
+	ret = cache->mnt->mnt_sb->s_op->statfs(cache->mnt->mnt_root, &stats);
+	if (ret < 0) {
+		if (ret == -EIO)
+			cachefiles_io_error(cache, "statfs failed");
+		return ret;
+	}
+
+	stats.f_bavail >>= cache->bshift;
+
+	_debug("avail %llu", stats.f_bavail);
+
+	/* see if there is sufficient space */
+	stats.f_bavail -= nr;
+
+	ret = -ENOBUFS;
+	if (stats.f_bavail < cache->bstop)
+		goto begin_cull;
+
+	ret = 0;
+	if (stats.f_bavail < cache->bcull)
+		goto begin_cull;
+
+	if (test_bit(CACHEFILES_CULLING, &cache->flags) &&
+	    stats.f_bavail >= cache->brun
+	    ) {
+		if (test_and_clear_bit(CACHEFILES_CULLING, &cache->flags)) {
+			_debug("cease culling");
+			send_sigurg(&cache->cachefilesd->f_owner);
+		}
+	}
+
+	_leave(" = 0");
+	return 0;
+
+begin_cull:
+	if (!test_and_set_bit(CACHEFILES_CULLING, &cache->flags)) {
+		_debug("### CULL CACHE ###");
+		send_sigurg(&cache->cachefilesd->f_owner);
+	}
+
+	_leave(" = %d", ret);
+	return ret;
+
+}
+
+/*****************************************************************************/
+/*
+ * waiting reading backing files
+ */
+static int cachefiles_read_waiter(wait_queue_t *wait, unsigned mode,
+				  int sync, void *_key)
+{
+	struct cachefiles_one_read *monitor =
+		container_of(wait, struct cachefiles_one_read, monitor);
+	struct wait_bit_key *key = _key;
+	struct page *page = wait->private;
+
+	ASSERT(key);
+
+	_enter("{%lu},%u,%d,{%p,%u}",
+	       monitor->netfs_page->index, mode, sync,
+	       key->flags, key->bit_nr);
+
+	if (key->flags != &page->flags ||
+	    key->bit_nr != PG_locked)
+		return 0;
+
+	_debug("--- monitor %p %lx ---", page, page->flags);
+
+	if (!PageUptodate(page) && !PageError(page))
+		dump_stack();
+
+	/* remove from the waitqueue */
+	list_del(&wait->task_list);
+
+	/* move onto the action list and queue for keventd */
+	ASSERT(monitor->object);
+
+	spin_lock(&monitor->object->work_lock);
+	list_move(&monitor->obj_link, &monitor->object->read_list);
+	spin_unlock(&monitor->object->work_lock);
+
+	schedule_work(&monitor->object->read_work);
+
+	return 0;
+
+}
+
+/*****************************************************************************/
+/*
+ * let keventd drive the copying of pages
+ */
+void cachefiles_read_copier_work(void *_object)
+{
+	struct cachefiles_one_read *monitor;
+	struct cachefiles_object *object = _object;
+	struct fscache_cookie *cookie = object->fscache.cookie;
+	struct pagevec pagevec;
+	int error, max;
+
+	_enter("{ino=%lu}", object->backer->d_inode->i_ino);
+
+	pagevec_init(&pagevec, 0);
+
+	max = 8;
+	spin_lock_irq(&object->work_lock);
+
+	while (!list_empty(&object->read_list)) {
+		monitor = list_entry(object->read_list.next,
+				     struct cachefiles_one_read, obj_link);
+		list_del(&monitor->obj_link);
+
+		spin_unlock_irq(&object->work_lock);
+
+		_debug("- copy {%lu}", monitor->back_page->index);
+
+		error = -EIO;
+		if (PageUptodate(monitor->back_page)) {
+			copy_highpage(monitor->netfs_page, monitor->back_page);
+
+			pagevec_add(&pagevec, monitor->netfs_page);
+			cookie->def->mark_pages_cached(
+				cookie->netfs_data,
+				monitor->netfs_page->mapping,
+				&pagevec);
+			pagevec_reinit(&pagevec);
+
+			error = 0;
+		}
+
+		if (error)
+			cachefiles_io_error_obj(
+				object,
+				"readpage failed on backing file %lx",
+				(unsigned long) monitor->back_page->flags);
+
+		page_cache_release(monitor->back_page);
+
+		monitor->end_io_func(monitor->netfs_page,
+				     monitor->context,
+				     error);
+
+		page_cache_release(monitor->netfs_page);
+		fscache_put_context(cookie, monitor->context);
+		kfree(monitor);
+
+		/* let keventd have some air occasionally */
+		max--;
+		if (max < 0 || need_resched()) {
+			if (!list_empty(&object->read_list))
+				schedule_work(&object->read_work);
+			_leave(" [maxed out]");
+			return;
+		}
+
+		spin_lock_irq(&object->work_lock);
+	}
+
+	spin_unlock_irq(&object->work_lock);
+
+	_leave("");
+
+}
+
+/*****************************************************************************/
+/*
+ * read the corresponding page to the given set from the backing file
+ * - an uncertain page is simply discarded, to be tried again another time
+ */
+static int cachefiles_read_backing_file_one(struct cachefiles_object *object,
+					    fscache_rw_complete_t end_io_func,
+					    void *context,
+					    struct page *netpage,
+					    struct pagevec *lru_pvec)
+{
+	struct cachefiles_one_read *monitor;
+	struct address_space *bmapping;
+	struct page *newpage, *backpage;
+	int ret;
+
+	_enter("");
+
+	ASSERTCMP(pagevec_count(lru_pvec), ==, 0);
+	pagevec_reinit(lru_pvec);
+
+	_debug("read back %p{%lu,%d}",
+	       netpage, netpage->index, page_count(netpage));
+
+	monitor = kzalloc(sizeof(*monitor), GFP_KERNEL);
+	if (!monitor)
+		goto nomem;
+
+	monitor->netfs_page = netpage;
+	monitor->object = object;
+	monitor->end_io_func = end_io_func;
+	monitor->context = fscache_get_context(object->fscache.cookie,
+					       context);
+
+	init_waitqueue_func_entry(&monitor->monitor, cachefiles_read_waiter);
+
+	/* attempt to get hold of the backing page */
+	bmapping = object->backer->d_inode->i_mapping;
+	newpage = NULL;
+
+	for (;;) {
+		backpage = find_get_page(bmapping, netpage->index);
+		if (backpage)
+			goto backing_page_already_present;
+
+		if (!newpage) {
+			newpage = page_cache_alloc_cold(bmapping);
+			if (!newpage)
+				goto nomem_monitor;
+		}
+
+		ret = add_to_page_cache(newpage, bmapping,
+					netpage->index, GFP_KERNEL);
+		if (ret == 0)
+			goto installed_new_backing_page;
+		if (ret != -EEXIST)
+			goto nomem_page;
+	}
+
+	/* we've installed a new backing page, so now we need to add it
+	 * to the LRU list and start it reading */
+installed_new_backing_page:
+	_debug("- new %p", newpage);
+
+	backpage = newpage;
+	newpage = NULL;
+
+	page_cache_get(backpage);
+	pagevec_add(lru_pvec, backpage);
+	__pagevec_lru_add(lru_pvec);
+
+	ret = bmapping->a_ops->readpage(NULL, backpage);
+	if (ret < 0)
+		goto read_error;
+
+	/* set the monitor to transfer the data across */
+monitor_backing_page:
+	_debug("- monitor add");
+
+	/* install the monitor */
+	page_cache_get(monitor->netfs_page);
+	page_cache_get(backpage);
+	monitor->back_page = backpage;
+
+	spin_lock_irq(&object->work_lock);
+	list_add_tail(&monitor->obj_link, &object->read_pend_list);
+	spin_unlock_irq(&object->work_lock);
+
+	monitor->monitor.private = backpage;
+	install_page_waitqueue_monitor(backpage, &monitor->monitor);
+	monitor = NULL;
+
+	/* but the page may have been read before the monitor was
+	 * installed, so the monitor may miss the event - so we have to
+	 * ensure that we do get one in such a case */
+	if (!TestSetPageLocked(backpage))
+		unlock_page(backpage);
+	goto success;
+
+	/* if the backing page is already present, it can be in one of
+	 * three states: read in progress, read failed or read okay */
+backing_page_already_present:
+	_debug("- present");
+
+	if (newpage) {
+		page_cache_release(newpage);
+		newpage = NULL;
+	}
+
+	if (PageError(backpage))
+		goto io_error;
+
+	if (PageUptodate(backpage))
+		goto backing_page_already_uptodate;
+
+	goto monitor_backing_page;
+
+	/* the backing page is already up to date, attach the netfs
+	 * page to the pagecache and LRU and copy the data across */
+backing_page_already_uptodate:
+	_debug("- uptodate");
+
+	copy_highpage(netpage, backpage);
+	end_io_func(netpage, context, 0);
+
+success:
+	_debug("success");
+	ret = 0;
+
+out:
+	if (backpage)
+		page_cache_release(backpage);
+	if (monitor) {
+		fscache_put_context(object->fscache.cookie, monitor->context);
+		kfree(monitor);
+	}
+
+	_leave(" = %d", ret);
+	return ret;
+
+read_error:
+	_debug("read error %d", ret);
+	if (ret == -ENOMEM)
+		goto out;
+io_error:
+	cachefiles_io_error_obj(object, "page read error on backing file");
+	ret = -EIO;
+	goto out;
+
+nomem_page:
+	page_cache_release(newpage);
+nomem_monitor:
+	fscache_put_context(object->fscache.cookie, monitor->context);
+	kfree(monitor);
+nomem:
+	_leave(" = -ENOMEM");
+	return -ENOMEM;
+
+}
+
+/*****************************************************************************/
+/*
+ * read a page from the cache or allocate a block in which to store it
+ * - cache withdrawal is prevented by the caller
+ * - returns -EINTR if interrupted
+ * - returns -ENOMEM if ran out of memory
+ * - returns -ENOBUFS if no buffers can be made available
+ * - returns -ENOBUFS if page is beyond EOF
+ * - if the page is backed by a block in the cache:
+ *   - a read will be started which will call the callback on completion
+ *   - 0 will be returned
+ * - else if the page is unbacked:
+ *   - the metadata will be retained
+ *   - -ENODATA will be returned
+ */
+static int cachefiles_read_or_alloc_page(struct fscache_object *_object,
+					 struct page *page,
+					 fscache_rw_complete_t end_io_func,
+					 void *context,
+					 unsigned long gfp)
+{
+	struct cachefiles_object *object;
+	struct cachefiles_cache *cache;
+	struct fscache_cookie *cookie;
+	struct pagevec pagevec;
+	struct inode *inode;
+	sector_t block0, block;
+	unsigned shift;
+	int ret;
+
+	object = container_of(_object, struct cachefiles_object, fscache);
+	cache = container_of(object->fscache.cache, struct cachefiles_cache, cache);
+
+	_enter("{%p},{%lx},,,", object, page->index);
+
+	if (!object->backer)
+		return -ENOBUFS;
+
+	inode = object->backer->d_inode;
+	ASSERT(S_ISREG(inode->i_mode));
+	ASSERT(inode->i_mapping->a_ops->bmap);
+	ASSERT(inode->i_mapping->a_ops->readpages);
+
+	/* calculate the shift required to use bmap */
+	if (inode->i_sb->s_blocksize > PAGE_SIZE)
+		return -ENOBUFS;
+
+	shift = PAGE_SHIFT - inode->i_sb->s_blocksize_bits;
+
+	cookie = object->fscache.cookie;
+
+	pagevec_init(&pagevec, 0);
+
+	/* we assume the absence or presence of the first block is a good
+	 * enough indication for the page as a whole
+	 * - TODO: don't use bmap() for this as it is _not_ actually good
+	 *   enough for this as it doesn't indicate errors, but it's all we've
+	 *   got for the moment
+	 */
+	block0 = page->index;
+	block0 <<= shift;
+
+	block = inode->i_mapping->a_ops->bmap(inode->i_mapping, block0);
+	_debug("%llx -> %llx", block0, block);
+
+	if (block) {
+		/* submit the apparently valid page to the backing fs to be
+		 * read from disk */
+		ret = cachefiles_read_backing_file_one(object,
+						       end_io_func,
+						       context,
+						       page,
+						       &pagevec);
+		ret = 0;
+	} else if (cachefiles_has_space(cache, 1) == 0) {
+		/* there's space in the cache we can use */
+		pagevec_add(&pagevec, page);
+		cookie->def->mark_pages_cached(cookie->netfs_data,
+					       page->mapping, &pagevec);
+		ret = -ENODATA;
+	} else {
+		ret = -ENOBUFS;
+	}
+
+	_leave(" = %d", ret);
+	return ret;
+
+}
+
+/*****************************************************************************/
+/*
+ * read the corresponding pages to the given set from the backing file
+ * - any uncertain pages are simply discarded, to be tried again another time
+ */
+static int cachefiles_read_backing_file(struct cachefiles_object *object,
+					fscache_rw_complete_t end_io_func,
+					void *context,
+					struct address_space *mapping,
+					struct list_head *list,
+					struct pagevec *lru_pvec)
+{
+	struct cachefiles_one_read *monitor = NULL;
+	struct address_space *bmapping = object->backer->d_inode->i_mapping;
+	struct page *newpage = NULL, *netpage, *_n, *backpage = NULL;
+	int ret = 0;
+
+	_enter("");
+
+	ASSERTCMP(pagevec_count(lru_pvec), ==, 0);
+	pagevec_reinit(lru_pvec);
+
+	list_for_each_entry_safe(netpage, _n, list, lru) {
+		list_del(&netpage->lru);
+
+		_debug("read back %p{%lu,%d}",
+		       netpage, netpage->index, page_count(netpage));
+
+		if (!monitor) {
+			monitor = kzalloc(sizeof(*monitor), GFP_KERNEL);
+			if (!monitor)
+				goto nomem;
+
+			monitor->object = object;
+			monitor->end_io_func = end_io_func;
+			monitor->context = fscache_get_context(
+				object->fscache.cookie, context);
+
+			init_waitqueue_func_entry(&monitor->monitor,
+						  cachefiles_read_waiter);
+		}
+
+		for (;;) {
+			backpage = find_get_page(bmapping, netpage->index);
+			if (backpage)
+				goto backing_page_already_present;
+
+			if (!newpage) {
+				newpage = page_cache_alloc_cold(bmapping);
+				if (!newpage)
+					goto nomem;
+			}
+
+			ret = add_to_page_cache(newpage, bmapping,
+						netpage->index, GFP_KERNEL);
+			if (ret == 0)
+				goto installed_new_backing_page;
+			if (ret != -EEXIST)
+				goto nomem;
+		}
+
+		/* we've installed a new backing page, so now we need to add it
+		 * to the LRU list and start it reading */
+	installed_new_backing_page:
+		_debug("- new %p", newpage);
+
+		backpage = newpage;
+		newpage = NULL;
+
+		page_cache_get(backpage);
+		if (!pagevec_add(lru_pvec, backpage))
+			__pagevec_lru_add(lru_pvec);
+
+	reread_backing_page:
+		ret = bmapping->a_ops->readpage(NULL, backpage);
+		if (ret < 0)
+			goto read_error;
+
+		/* add the netfs page to the pagecache and LRU, and set the
+		 * monitor to transfer the data across */
+	monitor_backing_page:
+		_debug("- monitor add");
+
+		ret = add_to_page_cache(netpage, mapping, netpage->index,
+					GFP_KERNEL);
+		if (ret < 0) {
+			if (ret == -EEXIST) {
+				page_cache_release(netpage);
+				continue;
+			}
+			goto nomem;
+		}
+
+		page_cache_get(netpage);
+		if (!pagevec_add(lru_pvec, netpage))
+			__pagevec_lru_add(lru_pvec);
+
+		/* install a monitor */
+		page_cache_get(netpage);
+		monitor->netfs_page = netpage;
+
+		page_cache_get(backpage);
+		monitor->back_page = backpage;
+
+		spin_lock_irq(&object->work_lock);
+		list_add_tail(&monitor->obj_link, &object->read_pend_list);
+		spin_unlock_irq(&object->work_lock);
+
+		monitor->monitor.private = backpage;
+		install_page_waitqueue_monitor(backpage, &monitor->monitor);
+		monitor = NULL;
+
+		/* but the page may have been read before the monitor was
+		 * installed, so the monitor may miss the event - so we have to
+		 * ensure that we do get one in such a case */
+		if (!TestSetPageLocked(backpage)) {
+			_debug("2unlock %p", backpage);
+			unlock_page(backpage);
+		}
+
+		page_cache_release(backpage);
+		backpage = NULL;
+
+		page_cache_release(netpage);
+		netpage = NULL;
+		continue;
+
+		/* if the backing page is already present, it can be in one of
+		 * three states: read in progress, read failed or read okay */
+	backing_page_already_present:
+		_debug("- present %p", backpage);
+
+		if (PageError(backpage))
+			goto io_error;
+
+		if (PageUptodate(backpage))
+			goto backing_page_already_uptodate;
+
+		_debug("- not ready %p{%lx}", backpage, backpage->flags);
+
+		if (TestSetPageLocked(backpage))
+			goto monitor_backing_page;
+
+		if (PageError(backpage)) {
+			unlock_page(backpage);
+			goto io_error;
+		}
+
+		if (PageUptodate(backpage))
+			goto backing_page_already_uptodate_unlock;
+
+		/* we've locked a page that's neither up to date nor erroneous,
+		 * so we need to attempt to read it again */
+		goto reread_backing_page;
+
+		/* the backing page is already up to date, attach the netfs
+		 * page to the pagecache and LRU and copy the data across */
+	backing_page_already_uptodate_unlock:
+		unlock_page(backpage);
+	backing_page_already_uptodate:
+		_debug("- uptodate");
+
+		ret = add_to_page_cache(netpage, mapping, netpage->index,
+					GFP_KERNEL);
+		if (ret < 0) {
+			if (ret == -EEXIST) {
+				page_cache_release(netpage);
+				continue;
+			}
+			goto nomem;
+		}
+
+		copy_highpage(netpage, backpage);
+
+		page_cache_release(backpage);
+		backpage = NULL;
+
+		page_cache_get(netpage);
+		if (!pagevec_add(lru_pvec, netpage))
+			__pagevec_lru_add(lru_pvec);
+
+		end_io_func(netpage, context, 0);
+
+		page_cache_release(netpage);
+		netpage = NULL;
+		continue;
+	}
+
+	netpage = NULL;
+
+	_debug("out");
+
+out:
+	/* tidy up */
+	pagevec_lru_add(lru_pvec);
+
+	if (newpage)
+		page_cache_release(newpage);
+	if (netpage)
+		page_cache_release(netpage);
+	if (backpage)
+		page_cache_release(backpage);
+	if (monitor) {
+		fscache_put_context(object->fscache.cookie, monitor->context);
+		kfree(monitor);
+	}
+
+	list_for_each_entry_safe(netpage, _n, list, lru) {
+		list_del(&netpage->lru);
+		page_cache_release(netpage);
+	}
+
+	_leave(" = %d", ret);
+	return ret;
+
+nomem:
+	_debug("nomem");
+	ret = -ENOMEM;
+	goto out;
+
+read_error:
+	_debug("read error %d", ret);
+	if (ret == -ENOMEM)
+		goto out;
+io_error:
+	cachefiles_io_error_obj(object, "page read error on backing file");
+	ret = -EIO;
+	goto out;
+
+}
+
+/*****************************************************************************/
+/*
+ * read a list of pages from the cache or allocate blocks in which to store
+ * them
+ */
+static int cachefiles_read_or_alloc_pages(struct fscache_object *_object,
+					  struct address_space *mapping,
+					  struct list_head *pages,
+					  unsigned *nr_pages,
+					  fscache_rw_complete_t end_io_func,
+					  void *context,
+					  unsigned long gfp)
+{
+	struct cachefiles_object *object;
+	struct cachefiles_cache *cache;
+	struct fscache_cookie *cookie;
+	struct list_head backpages;
+	struct pagevec pagevec;
+	struct inode *inode;
+	struct page *page, *_n;
+	unsigned shift, nrbackpages;
+	int ret, ret2, space;
+
+	object = container_of(_object, struct cachefiles_object, fscache);
+	cache = container_of(object->fscache.cache, struct cachefiles_cache, cache);
+
+	_enter("{%p},,%d,,", object, *nr_pages);
+
+	if (!object->backer)
+		return -ENOBUFS;
+
+	space = 1;
+	if (cachefiles_has_space(cache, *nr_pages) < 0)
+		space = 0;
+
+	inode = object->backer->d_inode;
+	ASSERT(S_ISREG(inode->i_mode));
+	ASSERT(inode->i_mapping->a_ops->bmap);
+	ASSERT(inode->i_mapping->a_ops->readpages);
+
+	/* calculate the shift required to use bmap */
+	if (inode->i_sb->s_blocksize > PAGE_SIZE)
+		return -ENOBUFS;
+
+	shift = PAGE_SHIFT - inode->i_sb->s_blocksize_bits;
+
+	pagevec_init(&pagevec, 0);
+
+	cookie = object->fscache.cookie;
+
+	INIT_LIST_HEAD(&backpages);
+	nrbackpages = 0;
+
+	ret = space ? -ENODATA : -ENOBUFS;
+	list_for_each_entry_safe(page, _n, pages, lru) {
+		sector_t block0, block;
+
+		/* we assume the absence or presence of the first block is a
+		 * good enough indication for the page as a whole
+		 * - TODO: don't use bmap() for this as it is _not_ actually
+		 *   good enough for this as it doesn't indicate errors, but
+		 *   it's all we've got for the moment
+		 */
+		block0 = page->index;
+		block0 <<= shift;
+
+		block = inode->i_mapping->a_ops->bmap(inode->i_mapping,
+						      block0);
+		_debug("%llx -> %llx", block0, block);
+
+		if (block) {
+			/* we have data - add it to the list to give to the
+			 * backing fs */
+			list_move(&page->lru, &backpages);
+			(*nr_pages)--;
+			nrbackpages++;
+		} else if (space && pagevec_add(&pagevec, page) == 0) {
+			cookie->def->mark_pages_cached(cookie->netfs_data,
+						       mapping, &pagevec);
+			pagevec_reinit(&pagevec);
+			ret = -ENODATA;
+		}
+	}
+
+	if (pagevec_count(&pagevec) > 0) {
+		cookie->def->mark_pages_cached(cookie->netfs_data,
+					       mapping, &pagevec);
+		pagevec_reinit(&pagevec);
+	}
+
+	if (list_empty(pages))
+		ret = 0;
+
+	/* submit the apparently valid pages to the backing fs to be read from disk */
+	if (nrbackpages > 0) {
+		ret2 = cachefiles_read_backing_file(object,
+						    end_io_func,
+						    context,
+						    mapping,
+						    &backpages,
+						    &pagevec);
+
+		ASSERTCMP(pagevec_count(&pagevec), ==, 0);
+
+		if (ret2 == -ENOMEM || ret2 == -EINTR)
+			ret = ret2;
+	}
+
+	_leave(" = %d [nr=%u%s]",
+	       ret, *nr_pages, list_empty(pages) ? " empty" : "");
+	return ret;
+
+}
+
+/*****************************************************************************/
+/*
+ * read a page from the cache or allocate a block in which to store it
+ * - cache withdrawal is prevented by the caller
+ * - returns -EINTR if interrupted
+ * - returns -ENOMEM if ran out of memory
+ * - returns -ENOBUFS if no buffers can be made available
+ * - returns -ENOBUFS if page is beyond EOF
+ * - otherwise:
+ *   - the metadata will be retained
+ *   - 0 will be returned
+ */
+static int cachefiles_allocate_page(struct fscache_object *_object,
+				    struct page *page,
+				    unsigned long gfp)
+{
+	struct cachefiles_object *object;
+	struct cachefiles_cache *cache;
+
+	object = container_of(_object, struct cachefiles_object, fscache);
+	cache = container_of(object->fscache.cache,
+			     struct cachefiles_cache, cache);
+
+	_enter("%p,{%lx},,,", object, page->index);
+
+	return cachefiles_has_space(cache, 1);
+
+}
+
+/*****************************************************************************/
+/*
+ * page storer
+ */
+void cachefiles_write_work(void *_object)
+{
+	struct cachefiles_one_write *writer;
+	struct cachefiles_object *object = _object;
+	int ret, max;
+
+	_enter("%p", object);
+
+	ASSERT(!irqs_disabled());
+
+	spin_lock_irq(&object->work_lock);
+	max = 8;
+
+	while (!list_empty(&object->write_list)) {
+		writer = list_entry(object->write_list.next,
+				    struct cachefiles_one_write, obj_link);
+		list_del(&writer->obj_link);
+
+		spin_unlock_irq(&object->work_lock);
+
+		_debug("- store {%lu}", writer->netfs_page->index);
+
+		ret = generic_file_buffered_write_one_kernel_page(
+			object->backer->d_inode->i_mapping,
+			writer->netfs_page->index,
+			writer->netfs_page);
+
+		if (ret == -ENOSPC) {
+			ret = -ENOBUFS;
+		} else if (ret == -EIO) {
+			cachefiles_io_error_obj(object,
+						"write page to backing file"
+						" failed");
+			ret = -ENOBUFS;
+		}
+
+		_debug("- callback");
+		writer->end_io_func(writer->netfs_page,
+				    writer->context,
+				    ret);
+
+		_debug("- put net");
+		page_cache_release(writer->netfs_page);
+		fscache_put_context(object->fscache.cookie, writer->context);
+		kfree(writer);
+
+		/* let keventd have some air occasionally */
+		max--;
+		if (max < 0 || need_resched()) {
+			if (!list_empty(&object->write_list))
+				schedule_work(&object->write_work);
+			_leave(" [maxed out]");
+			return;
+		}
+
+		_debug("- next");
+		spin_lock_irq(&object->work_lock);
+	}
+
+	spin_unlock_irq(&object->work_lock);
+	_leave("");
+
+}
+
+/*****************************************************************************/
+/*
+ * request a page be stored in the cache
+ * - cache withdrawal is prevented by the caller
+ * - this request may be ignored if there's no cache block available, in which
+ *   case -ENOBUFS will be returned
+ * - if the op is in progress, 0 will be returned
+ */
+static int cachefiles_write_page(struct fscache_object *_object,
+				 struct page *page,
+				 fscache_rw_complete_t end_io_func,
+				 void *context,
+				 unsigned long gfp)
+{
+//	struct cachefiles_one_write *writer;
+	struct cachefiles_object *object;
+	int ret;
+
+	object = container_of(_object, struct cachefiles_object, fscache);
+
+	_enter("%p,%p{%lx},,,", object, page, page->index);
+
+	if (!object->backer)
+		return -ENOBUFS;
+
+	ASSERT(S_ISREG(object->backer->d_inode->i_mode));
+
+#if 0 // set to 1 for deferred writing
+	/* queue the operation for deferred processing by keventd */
+	writer = kzalloc(sizeof(*writer), GFP_KERNEL);
+	if (!writer)
+		return -ENOMEM;
+
+	page_cache_get(page);
+	writer->netfs_page = page;
+	writer->object = object;
+	writer->end_io_func = end_io_func;
+	writer->context = facache_get_context(object->fscache.cookie, context);
+
+	spin_lock_irq(&object->work_lock);
+	list_add_tail(&writer->obj_link, &object->write_list);
+	spin_unlock_irq(&object->work_lock);
+
+	schedule_work(&object->write_work);
+	ret = 0;
+
+#else
+	/* copy the page to ext3 and let it store it in its own time */
+	ret = generic_file_buffered_write_one_kernel_page(
+		object->backer->d_inode->i_mapping, page->index, page);
+
+	if (ret != 0) {
+		if (ret == -EIO)
+			cachefiles_io_error_obj(object,
+						"write page to backing file"
+						" failed");
+		ret = -ENOBUFS;
+	}
+
+	end_io_func(page, context, ret);
+#endif
+
+	_leave(" = %d", ret);
+	return ret;
+
+}
+
+/*****************************************************************************/
+/*
+ * detach a backing block from a page
+ * - cache withdrawal is prevented by the caller
+ */
+static void cachefiles_uncache_pages(struct fscache_object *_object,
+				     struct pagevec *pagevec)
+{
+	struct cachefiles_object *object;
+	struct cachefiles_cache *cache;
+
+	object = container_of(_object, struct cachefiles_object, fscache);
+	cache = container_of(object->fscache.cache,
+			     struct cachefiles_cache, cache);
+
+	_enter("%p,{%lu,%lx},,,",
+	       object, pagevec->nr, pagevec->pages[0]->index);
+
+}
+
+/*****************************************************************************/
+/*
+ * dissociate a cache from all the pages it was backing
+ */
+static void cachefiles_dissociate_pages(struct fscache_cache *cache)
+{
+	_enter("");
+
+}
+
+struct fscache_cache_ops cachefiles_cache_ops = {
+	.name			= "cachefiles",
+	.lookup_object		= cachefiles_lookup_object,
+	.grab_object		= cachefiles_grab_object,
+	.lock_object		= cachefiles_lock_object,
+	.unlock_object		= cachefiles_unlock_object,
+	.update_object		= cachefiles_update_object,
+	.put_object		= cachefiles_put_object,
+	.sync_cache		= cachefiles_sync_cache,
+	.set_i_size		= cachefiles_set_i_size,
+	.read_or_alloc_page	= cachefiles_read_or_alloc_page,
+	.read_or_alloc_pages	= cachefiles_read_or_alloc_pages,
+	.allocate_page		= cachefiles_allocate_page,
+	.write_page		= cachefiles_write_page,
+	.uncache_pages		= cachefiles_uncache_pages,
+	.dissociate_pages	= cachefiles_dissociate_pages,
+};
diff --git a/fs/cachefiles/cf-key.c b/fs/cachefiles/cf-key.c
new file mode 100644
index 0000000..cfae5a7
--- /dev/null
+++ b/fs/cachefiles/cf-key.c
@@ -0,0 +1,157 @@
+/* cf-key.c: Key to pathname encoder
+ *
+ * Copyright (C) 2006 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/slab.h>
+#include "internal.h"
+
+static const char cachefiles_charmap[64] =
+	"0123456789"			/* 0 - 9 */
+	"abcdefghijklmnopqrstuvwxyz"	/* 10 - 35 */
+	"ABCDEFGHIJKLMNOPQRSTUVWXYZ"	/* 36 - 61 */
+	"_-"				/* 62 - 63 */
+	;
+
+static const char cachefiles_filecharmap[256] = {
+	/* we skip space and tab and control chars */
+	[ 33 ... 46 ] = 1,		/* '!' -> '.' */
+	/* we skip '/' as it's significant to pathwalk */
+	[ 48 ... 127 ] = 1,		/* '0' -> '~' */
+};
+
+/*****************************************************************************/
+/*
+ * turn the raw key into something cooked
+ * - the raw key should include the length in the two bytes at the front
+ * - the key may be up to 514 bytes in length (including the length word)
+ *   - "base64" encode the strange keys, mapping 3 bytes of raw to four of
+ *     cooked
+ *   - need to cut the cooked key into 252 char lengths (189 raw bytes)
+ */
+char *cachefiles_cook_key(const u8 *raw, int keylen, uint8_t type)
+{
+	unsigned char csum, ch;
+	unsigned int acc;
+	char *key;
+	int loop, len, max, seg, mark, print;
+
+	_enter(",%d", keylen);
+
+	BUG_ON(keylen < 2 || keylen > 514);
+
+	csum = raw[0] + raw[1];
+	print = 1;
+	for (loop = 2; loop < keylen; loop++) {
+		ch = raw[loop];
+		csum += ch;
+		print &= cachefiles_filecharmap[ch];
+	}
+
+	if (print) {
+		/* if the path is usable ASCII, then we render it directly */
+		max = keylen - 2;
+		max += 2;	/* two base64'd length chars on the front */
+		max += 5;	/* @checksum/M */
+		max += 3 * 2;	/* maximum number of segment dividers (".../M")
+				 * is ((514 + 251) / 252) = 3
+				 */
+		max += 1;	/* NUL on end */
+	} else {
+		/* calculate the maximum length of the cooked key */
+		keylen = (keylen + 2) / 3;
+
+		max = keylen * 4;
+		max += 5;	/* @checksum/M */
+		max += 3 * 2;	/* maximum number of segment dividers (".../M")
+				 * is ((514 + 188) / 189) = 3
+				 */
+		max += 1;	/* NUL on end */
+	}
+
+	_debug("max: %d", max);
+
+	key = kmalloc(max, GFP_KERNEL);
+	if (!key)
+		return NULL;
+
+	len = 0;
+
+	/* build the cooked key */
+	sprintf(key, "@%02x/+", (unsigned) csum);
+	len = 5;
+	mark = len - 1;
+
+	if (print) {
+		acc = *(uint16_t *) raw;
+		raw += 2;
+
+		key[len + 1] = cachefiles_charmap[acc & 63];
+		acc >>= 6;
+		key[len] = cachefiles_charmap[acc & 63];
+		len += 2;
+
+		seg = 250;
+		for (loop = keylen; loop > 0; loop--) {
+			if (seg <= 0) {
+				key[len++] = '/';
+				mark = len;
+				key[len++] = '+';
+				seg = 252;
+			}
+
+			key[len++] = *raw++;
+			ASSERT(len < max);
+		}
+
+		switch (type) {
+		case FSCACHE_COOKIE_TYPE_INDEX:		type = 'I';	break;
+		case FSCACHE_COOKIE_TYPE_DATAFILE:	type = 'D';	break;
+		default:				type = 'S';	break;
+		}
+	} else {
+		seg = 252;
+		for (loop = keylen; loop > 0; loop--) {
+			if (seg <= 0) {
+				key[len++] = '/';
+				mark = len;
+				key[len++] = '+';
+				seg = 252;
+			}
+
+			acc = *raw++;
+			acc |= *raw++ << 8;
+			acc |= *raw++ << 16;
+
+			_debug("acc: %06x", acc);
+
+			key[len++] = cachefiles_charmap[acc & 63];
+			acc >>= 6;
+			key[len++] = cachefiles_charmap[acc & 63];
+			acc >>= 6;
+			key[len++] = cachefiles_charmap[acc & 63];
+			acc >>= 6;
+			key[len++] = cachefiles_charmap[acc & 63];
+
+			ASSERT(len < max);
+		}
+
+		switch (type) {
+		case FSCACHE_COOKIE_TYPE_INDEX:		type = 'J';	break;
+		case FSCACHE_COOKIE_TYPE_DATAFILE:	type = 'E';	break;
+		default:				type = 'T';	break;
+		}
+	}
+
+	key[mark] = type;
+	key[len] = 0;
+
+	_leave(" = %p %d:[%s]", key, len, key);
+	return key;
+}
diff --git a/fs/cachefiles/cf-main.c b/fs/cachefiles/cf-main.c
new file mode 100644
index 0000000..da8d020
--- /dev/null
+++ b/fs/cachefiles/cf-main.c
@@ -0,0 +1,129 @@
+/* cf-main.c: network filesystem caching backend to use cache files on a
+ *            premounted filesystem
+ *
+ * Copyright (C) 2006 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/sched.h>
+#include <linux/completion.h>
+#include <linux/slab.h>
+#include <linux/fs.h>
+#include <linux/file.h>
+#include <linux/namei.h>
+#include <linux/mount.h>
+#include <linux/statfs.h>
+#include <linux/proc_fs.h>
+#include <linux/sysctl.h>
+#include "internal.h"
+
+unsigned long cachefiles_debug;
+
+static int cachefiles_init(void);
+static void cachefiles_exit(void);
+
+fs_initcall(cachefiles_init);
+module_exit(cachefiles_exit);
+
+MODULE_DESCRIPTION("Mounted-filesystem based cache");
+MODULE_AUTHOR("Red Hat, Inc.");
+MODULE_LICENSE("GPL");
+
+kmem_cache_t *cachefiles_object_jar;
+
+static void cachefiles_object_init_once(void *_object, kmem_cache_t *cachep,
+					unsigned long flags)
+{
+	struct cachefiles_object *object = _object;
+
+	switch (flags & (SLAB_CTOR_VERIFY | SLAB_CTOR_CONSTRUCTOR)) {
+	case SLAB_CTOR_CONSTRUCTOR:
+		memset(object, 0, sizeof(*object));
+		fscache_object_init(&object->fscache);
+		init_rwsem(&object->sem);
+		spin_lock_init(&object->work_lock);
+		INIT_LIST_HEAD(&object->read_list);
+		INIT_LIST_HEAD(&object->read_pend_list);
+		INIT_WORK(&object->read_work, &cachefiles_read_copier_work,
+			  object);
+		INIT_LIST_HEAD(&object->write_list);
+		INIT_WORK(&object->write_work, &cachefiles_write_work, object);
+		break;
+
+	default:
+		break;
+	}
+}
+
+/*****************************************************************************/
+/*
+ * initialise the fs caching module
+ */
+static int __init cachefiles_init(void)
+{
+	struct proc_dir_entry *pde;
+	int ret;
+
+	/* create a proc entry to use as a handle for the userspace daemon */
+	ret = -ENOMEM;
+
+	pde = create_proc_entry("cachefiles", 0600, proc_root_fs);
+	if (!pde) {
+		kerror("Unable to create /proc/fs/cachefiles");
+		goto error_proc;
+	}
+
+	if (cachefiles_sysctl_init() < 0) {
+		kerror("Unable to create sysctl parameters");
+		goto error_object_jar;
+	}
+
+	pde->owner = THIS_MODULE;
+	pde->proc_fops = &cachefiles_proc_fops;
+	cachefiles_proc = pde;
+
+	/* create an object jar */
+	cachefiles_object_jar =
+		kmem_cache_create("cachefiles_object_jar",
+				  sizeof(struct cachefiles_object),
+				  0,
+				  SLAB_HWCACHE_ALIGN,
+				  cachefiles_object_init_once,
+				  NULL);
+	if (!cachefiles_object_jar) {
+		printk(KERN_NOTICE
+		       "CacheFiles: Failed to allocate an object jar\n");
+		goto error_sysctl;
+	}
+
+	printk(KERN_INFO "CacheFiles: Loaded\n");
+	return 0;
+
+error_sysctl:
+	cachefiles_sysctl_cleanup();
+error_object_jar:
+	remove_proc_entry("cachefiles", proc_root_fs);
+error_proc:
+	kerror("failed to register: %d", ret);
+	return ret;
+}
+
+/*****************************************************************************/
+/*
+ * clean up on module removal
+ */
+static void __exit cachefiles_exit(void)
+{
+	printk(KERN_INFO "CacheFiles: Unloading\n");
+
+	kmem_cache_destroy(cachefiles_object_jar);
+	remove_proc_entry("cachefiles", proc_root_fs);
+	cachefiles_sysctl_cleanup();
+}
diff --git a/fs/cachefiles/cf-namei.c b/fs/cachefiles/cf-namei.c
new file mode 100644
index 0000000..20db88a
--- /dev/null
+++ b/fs/cachefiles/cf-namei.c
@@ -0,0 +1,825 @@
+/* cf-namei.c: CacheFiles path walking and related routines
+ *
+ * Copyright (C) 2006 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/file.h>
+#include <linux/fs.h>
+#include <linux/fsnotify.h>
+#include <linux/quotaops.h>
+#include <linux/xattr.h>
+#include <linux/mount.h>
+#include <linux/namei.h>
+#include "internal.h"
+
+/*****************************************************************************/
+/*
+ * record the fact that an object is now active
+ */
+static void cachefiles_mark_object_active(struct cachefiles_cache *cache,
+					  struct cachefiles_object *object)
+{
+	struct cachefiles_object *xobject;
+	struct rb_node **_p, *_parent = NULL;
+	struct dentry *dentry;
+
+	write_lock(&cache->active_lock);
+
+	dentry = object->dentry;
+	_p = &cache->active_nodes.rb_node;
+	while (*_p) {
+		_parent = *_p;
+		xobject = rb_entry(_parent,
+				   struct cachefiles_object, active_node);
+
+		if (xobject->dentry > dentry)
+			_p = &(*_p)->rb_left;
+		else if (xobject->dentry < dentry)
+			_p = &(*_p)->rb_right;
+		else
+			BUG(); /* uh oh... this dentry shouldn't be here */
+	}
+
+	rb_link_node(&object->active_node, _parent, _p);
+	rb_insert_color(&object->active_node, &cache->active_nodes);
+
+	write_unlock(&cache->active_lock);
+}
+
+/*****************************************************************************/
+/*
+ * delete an object representation from the cache
+ * - file backed objects are unlinked
+ * - directory backed objects are stuffed into the graveyard for userspace to
+ *   delete
+ * - unlocks the directory mutex
+ */
+static int cachefiles_bury_object(struct cachefiles_cache *cache,
+				  struct dentry *dir,
+				  struct dentry *rep)
+{
+	struct dentry *grave, *alt, *trap;
+	struct qstr name;
+	const char *old_name;
+	char nbuffer[8 + 8 + 1];
+	int ret;
+
+	_enter(",'%*.*s','%*.*s'",
+	       dir->d_name.len, dir->d_name.len, dir->d_name.name,
+	       rep->d_name.len, rep->d_name.len, rep->d_name.name);
+
+	/* non-directories can just be unlinked */
+	if (!S_ISDIR(rep->d_inode->i_mode)) {
+		_debug("unlink stale object");
+		ret = dir->d_inode->i_op->unlink(dir->d_inode, rep);
+
+		mutex_unlock(&dir->d_inode->i_mutex);
+
+		if (ret == 0) {
+			_debug("d_delete");
+			d_delete(rep);
+		} else if (ret == -EIO) {
+			cachefiles_io_error(cache, "Unlink failed");
+		}
+
+		_leave(" = %d", ret);
+		return ret;
+	}
+
+	/* directories have to be moved to the graveyard */
+	_debug("move stale object to graveyard");
+	mutex_unlock(&dir->d_inode->i_mutex);
+
+try_again:
+	/* first step is to make up a grave dentry in the graveyard */
+	sprintf(nbuffer, "%08x%08x",
+		(uint32_t) xtime.tv_sec,
+		(uint32_t) atomic_inc_return(&cache->gravecounter));
+
+	name.name = nbuffer;
+	name.len = strlen(name.name);
+
+	/* hash the name */
+	name.hash = full_name_hash(name.name, name.len);
+
+	if (dir->d_op && dir->d_op->d_hash) {
+		ret = dir->d_op->d_hash(dir, &name);
+		if (ret < 0) {
+			if (ret == -EIO)
+				cachefiles_io_error(cache, "Hash failed");
+
+			_leave(" = %d", ret);
+			return ret;
+		}
+	}
+
+	/* do the multiway lock magic */
+	trap = lock_rename(cache->graveyard, dir);
+
+	/* do some checks before getting the grave dentry */
+	if (rep->d_parent != dir) {
+		/* the entry was probably culled when we dropped the parent dir
+		 * lock */
+		unlock_rename(cache->graveyard, dir);
+		_leave(" = 0 [culled?]");
+		return 0;
+	}
+
+	if (!S_ISDIR(cache->graveyard->d_inode->i_mode)) {
+		unlock_rename(cache->graveyard, dir);
+		cachefiles_io_error(cache, "Graveyard no longer a directory");
+		return -EIO;
+	}
+
+	if (trap == rep) {
+		unlock_rename(cache->graveyard, dir);
+		cachefiles_io_error(cache, "May not make directory loop");
+		return -EIO;
+	}
+
+	if (d_mountpoint(rep)) {
+		unlock_rename(cache->graveyard, dir);
+		cachefiles_io_error(cache, "Mountpoint in cache");
+		return -EIO;
+	}
+
+	/* see if there's a dentry already there for this name */
+	grave = d_lookup(cache->graveyard, &name);
+	if (!grave) {
+		_debug("not found");
+
+		grave = d_alloc(cache->graveyard, &name);
+		if (!grave) {
+			unlock_rename(cache->graveyard, dir);
+			_leave(" = -ENOMEM");
+			return -ENOMEM;
+		}
+
+		alt = cache->graveyard->d_inode->i_op->lookup(
+			cache->graveyard->d_inode, grave, NULL);
+		if (IS_ERR(alt)) {
+			unlock_rename(cache->graveyard, dir);
+			dput(grave);
+
+			if (PTR_ERR(alt) == -ENOMEM) {
+				_leave(" = -ENOMEM");
+				return -ENOMEM;
+			}
+
+			cachefiles_io_error(cache, "Lookup error %ld",
+					    PTR_ERR(alt));
+			return -EIO;
+		}
+
+		if (alt) {
+			dput(grave);
+			grave = alt;
+		}
+	}
+
+	if (grave->d_inode) {
+		unlock_rename(cache->graveyard, dir);
+		dput(grave);
+		grave = NULL;
+		cond_resched();
+		goto try_again;
+	}
+
+	if (d_mountpoint(grave)) {
+		unlock_rename(cache->graveyard, dir);
+		dput(grave);
+		cachefiles_io_error(cache, "Mountpoint in graveyard");
+		return -EIO;
+	}
+
+	/* target should not be an ancestor of source */
+	if (trap == grave) {
+		unlock_rename(cache->graveyard, dir);
+		dput(grave);
+		cachefiles_io_error(cache, "May not make directory loop");
+		return -EIO;
+	}
+
+	/* attempt the rename */
+	DQUOT_INIT(dir->d_inode);
+	DQUOT_INIT(cache->graveyard->d_inode);
+
+	old_name = fsnotify_oldname_init(rep->d_name.name);
+
+	ret = dir->d_inode->i_op->rename(dir->d_inode, rep,
+					 cache->graveyard->d_inode, grave);
+
+	if (ret == 0) {
+		d_move(rep, grave);
+		fsnotify_move(dir->d_inode, cache->graveyard->d_inode,
+			      old_name, rep->d_name.name, 1,
+			      grave->d_inode, rep->d_inode);
+	} else if (ret != -ENOMEM) {
+		cachefiles_io_error(cache, "Rename failed with error %d", ret);
+	}
+
+	fsnotify_oldname_free(old_name);
+
+	unlock_rename(cache->graveyard, dir);
+	dput(grave);
+	_leave(" = 0");
+	return 0;
+}
+
+/*****************************************************************************/
+/*
+ * delete an object representation from the cache
+ */
+int cachefiles_delete_object(struct cachefiles_cache *cache,
+			     struct cachefiles_object *object)
+{
+	struct dentry *dir;
+	int ret;
+
+	_enter(",{%p}", object->dentry);
+
+	ASSERT(object->dentry);
+	ASSERT(object->dentry->d_inode);
+	ASSERT(object->dentry->d_parent);
+
+	dir = dget_parent(object->dentry);
+
+	mutex_lock(&dir->d_inode->i_mutex);
+	ret = cachefiles_bury_object(cache, dir, object->dentry);
+
+	dput(dir);
+	_leave(" = %d", ret);
+	return ret;
+}
+
+/*****************************************************************************/
+/*
+ * walk from the parent object to the child object through the backing
+ * filesystem, creating directories as we go
+ */
+int cachefiles_walk_to_object(struct cachefiles_object *parent,
+			      struct cachefiles_object *object,
+			      char *key,
+			      struct cachefiles_xattr *auxdata)
+{
+	struct cachefiles_cache *cache;
+	struct dentry *dir, *next = NULL, *new;
+	struct qstr name;
+	uid_t fsuid;
+	gid_t fsgid;
+	int ret;
+
+	_enter("{%p}", parent->dentry);
+
+	cache = container_of(parent->fscache.cache,
+			     struct cachefiles_cache, cache);
+
+	ASSERT(parent->dentry);
+	ASSERT(parent->dentry->d_inode);
+
+	if (!(S_ISDIR(parent->dentry->d_inode->i_mode))) {
+		// TODO: convert file to dir
+		_leave("looking up in none directory");
+		return -ENOBUFS;
+	}
+
+	fsuid = current->fsuid;
+	fsgid = current->fsgid;
+	current->fsuid = 0;
+	current->fsgid = 0;
+
+	dir = dget(parent->dentry);
+
+advance:
+	/* attempt to transit the first directory component */
+	name.name = key;
+	key = strchr(key, '/');
+	if (key) {
+		name.len = key - (char *) name.name;
+		*key++ = 0;
+	} else {
+		name.len = strlen(name.name);
+	}
+
+	/* hash the name */
+	name.hash = full_name_hash(name.name, name.len);
+
+	if (dir->d_op && dir->d_op->d_hash) {
+		ret = dir->d_op->d_hash(dir, &name);
+		if (ret < 0) {
+			cachefiles_io_error(cache, "Hash failed");
+			goto error_out2;
+		}
+	}
+
+lookup_again:
+	/* search the current directory for the element name */
+	_debug("lookup '%s' %x", name.name, name.hash);
+
+	mutex_lock(&dir->d_inode->i_mutex);
+
+	next = d_lookup(dir, &name);
+	if (!next) {
+		_debug("not found");
+
+		new = d_alloc(dir, &name);
+		if (!new)
+			goto nomem_d_alloc;
+
+		ASSERT(dir->d_inode->i_op);
+		ASSERT(dir->d_inode->i_op->lookup);
+
+		next = dir->d_inode->i_op->lookup(dir->d_inode, new, NULL);
+		if (IS_ERR(next))
+			goto lookup_error;
+
+		if (!next)
+			next = new;
+		else
+			dput(new);
+
+		if (next->d_inode) {
+			ret = -EPERM;
+			if (!next->d_inode->i_op ||
+			    !next->d_inode->i_op->setxattr ||
+			    !next->d_inode->i_op->getxattr ||
+			    !next->d_inode->i_op->removexattr)
+				goto error;
+
+			if (key && (!next->d_inode->i_op->lookup ||
+				    !next->d_inode->i_op->mkdir ||
+				    !next->d_inode->i_op->create ||
+				    !next->d_inode->i_op->rename ||
+				    !next->d_inode->i_op->rmdir ||
+				    !next->d_inode->i_op->unlink))
+				goto error;
+		}
+	}
+
+	_debug("next -> %p %s", next, next->d_inode ? "positive" : "negative");
+
+	if (!key)
+		object->new = !next->d_inode;
+
+	/* we need to create the object if it's negative */
+	if (key || object->type == FSCACHE_COOKIE_TYPE_INDEX) {
+		/* index objects and intervening tree levels must be subdirs */
+		if (!next->d_inode) {
+			DQUOT_INIT(dir->d_inode);
+			ret = dir->d_inode->i_op->mkdir(dir->d_inode, next, 0);
+			if (ret < 0)
+				goto create_error;
+
+			ASSERT(next->d_inode);
+
+			fsnotify_mkdir(dir->d_inode, next);
+
+			_debug("mkdir -> %p{%p{ino=%lu}}",
+			       next, next->d_inode, next->d_inode->i_ino);
+
+		} else if (!S_ISDIR(next->d_inode->i_mode)) {
+			kerror("inode %lu is not a directory",
+			       next->d_inode->i_ino);
+			ret = -ENOBUFS;
+			goto error;
+		}
+
+	} else {
+		/* non-index objects start out life as files */
+		if (!next->d_inode) {
+			DQUOT_INIT(dir->d_inode);
+			ret = dir->d_inode->i_op->create(dir->d_inode, next,
+							 S_IFREG, NULL);
+			if (ret < 0)
+				goto create_error;
+
+			ASSERT(next->d_inode);
+
+			fsnotify_create(dir->d_inode, next);
+
+			_debug("create -> %p{%p{ino=%lu}}",
+			       next, next->d_inode, next->d_inode->i_ino);
+
+		} else if (!S_ISDIR(next->d_inode->i_mode) &&
+			   !S_ISREG(next->d_inode->i_mode)
+			   ) {
+			kerror("inode %lu is not a file or directory",
+			       next->d_inode->i_ino);
+			ret = -ENOBUFS;
+			goto error;
+		}
+	}
+
+	/* process the next component */
+	if (key) {
+		_debug("advance");
+		mutex_unlock(&dir->d_inode->i_mutex);
+		dput(dir);
+		dir = next;
+		next = NULL;
+		goto advance;
+	}
+
+	/* we've found the object we were looking for */
+	object->dentry = next;
+
+	/* if we've found that the terminal object exists, then we need to
+	 * check its attributes and delete it if it's out of date */
+	if (!object->new) {
+		_debug("validate '%*.*s'",
+		       next->d_name.len, next->d_name.len, next->d_name.name);
+
+		ret = cachefiles_check_object_xattr(object, auxdata);
+		if (ret == -ESTALE) {
+			/* delete the object (the deleter drops the directory
+			 * mutex) */
+			object->dentry = NULL;
+
+			ret = cachefiles_bury_object(cache, dir, next);
+			dput(next);
+			next = NULL;
+
+			if (ret < 0)
+				goto delete_error;
+
+			_debug("redo lookup");
+			goto lookup_again;
+		}
+	}
+
+	/* note that we're now using this object */
+	cachefiles_mark_object_active(cache, object);
+
+	mutex_unlock(&dir->d_inode->i_mutex);
+	dput(dir);
+	dir = NULL;
+
+	if (object->new) {
+		/* attach data to a newly constructed terminal object */
+		ret = cachefiles_set_object_xattr(object, auxdata);
+		if (ret < 0)
+			goto check_error;
+	} else {
+		/* always update the atime on an object we've just looked up
+		 * (this is used to keep track of culling, and atimes are only
+		 * updated by read, write and readdir but not lookup or
+		 * open) */
+		touch_atime(cache->mnt, next);
+	}
+
+	/* open a file interface onto a data file */
+	if (object->type != FSCACHE_COOKIE_TYPE_INDEX) {
+		if (S_ISREG(object->dentry->d_inode->i_mode)) {
+			const struct address_space_operations *aops;
+
+			ret = -EPERM;
+			aops = object->dentry->d_inode->i_mapping->a_ops;
+			if (!aops->bmap ||
+			    !aops->prepare_write ||
+			    !aops->commit_write)
+				goto check_error;
+
+			object->backer = object->dentry;
+		} else {
+			BUG(); // TODO: open file in data-class subdir
+		}
+	}
+
+	current->fsuid = fsuid;
+	current->fsgid = fsgid;
+	object->new = 0;
+
+	_leave(" = 0 [%lu]", object->dentry->d_inode->i_ino);
+	return 0;
+
+create_error:
+	if (ret == -EIO)
+		cachefiles_io_error(cache, "create/mkdir failed");
+	goto error;
+
+check_error:
+	write_lock(&cache->active_lock);
+	rb_erase(&object->active_node, &cache->active_nodes);
+	write_unlock(&cache->active_lock);
+
+	dput(object->dentry);
+	object->dentry = NULL;
+	goto error_out;
+
+delete_error:
+	_debug("delete error %d", ret);
+	goto error_out2;
+
+lookup_error:
+	_debug("lookup error %ld", PTR_ERR(next));
+	dput(new);
+	ret = PTR_ERR(next);
+	if (ret == -EIO)
+		cachefiles_io_error(cache, "Lookup failed");
+	next = NULL;
+	goto error;
+
+nomem_d_alloc:
+	ret = -ENOMEM;
+error:
+	mutex_unlock(&dir->d_inode->i_mutex);
+	dput(next);
+error_out2:
+	dput(dir);
+error_out:
+	current->fsuid = fsuid;
+	current->fsgid = fsgid;
+
+	_leave(" = ret");
+	return ret;
+}
+
+/*****************************************************************************/
+/*
+ * get a subdirectory
+ */
+struct dentry *cachefiles_get_directory(struct cachefiles_cache *cache,
+					struct dentry *dir,
+					const char *dirname)
+{
+	struct dentry *subdir, *new;
+	struct qstr name;
+	uid_t fsuid;
+	gid_t fsgid;
+	int ret;
+
+	_enter("");
+
+	/* set up the name */
+	name.name = dirname;
+	name.len = strlen(dirname);
+	name.hash = full_name_hash(name.name, name.len);
+
+	if (dir->d_op && dir->d_op->d_hash) {
+		ret = dir->d_op->d_hash(dir, &name);
+		if (ret < 0) {
+			if (ret == -EIO)
+				kerror("Hash failed");
+			_leave(" = %d", ret);
+			return ERR_PTR(ret);
+		}
+	}
+
+	/* search the current directory for the element name */
+	_debug("lookup '%s' %x", name.name, name.hash);
+
+	fsuid = current->fsuid;
+	fsgid = current->fsgid;
+	current->fsuid = 0;
+	current->fsgid = 0;
+
+	mutex_lock(&dir->d_inode->i_mutex);
+
+	subdir = d_lookup(dir, &name);
+	if (!subdir) {
+		_debug("not found");
+
+		new = d_alloc(dir, &name);
+		if (!new)
+			goto nomem_d_alloc;
+
+		subdir = dir->d_inode->i_op->lookup(dir->d_inode, new, NULL);
+		if (IS_ERR(subdir))
+			goto lookup_error;
+
+		if (!subdir)
+			subdir = new;
+		else
+			dput(new);
+	}
+
+	_debug("subdir -> %p %s",
+	       subdir, subdir->d_inode ? "positive" : "negative");
+
+	/* we need to create the subdir if it doesn't exist yet */
+	if (!subdir->d_inode) {
+		DQUOT_INIT(dir->d_inode);
+		ret = dir->d_inode->i_op->mkdir(dir->d_inode, subdir, 0700);
+		if (ret < 0)
+			goto mkdir_error;
+
+		ASSERT(subdir->d_inode);
+
+		fsnotify_mkdir(dir->d_inode, subdir);
+
+		_debug("mkdir -> %p{%p{ino=%lu}}",
+		       subdir,
+		       subdir->d_inode,
+		       subdir->d_inode->i_ino);
+	}
+
+	mutex_unlock(&dir->d_inode->i_mutex);
+
+	current->fsuid = fsuid;
+	current->fsgid = fsgid;
+
+	/* we need to make sure the subdir is a directory */
+	ASSERT(subdir->d_inode);
+
+	if (!S_ISDIR(subdir->d_inode->i_mode)) {
+		kerror("%s is not a directory", dirname);
+		ret = -EIO;
+		goto check_error;
+	}
+
+	ret = -EPERM;
+	if (!subdir->d_inode->i_op ||
+	    !subdir->d_inode->i_op->setxattr ||
+	    !subdir->d_inode->i_op->getxattr ||
+	    !subdir->d_inode->i_op->lookup ||
+	    !subdir->d_inode->i_op->mkdir ||
+	    !subdir->d_inode->i_op->create ||
+	    !subdir->d_inode->i_op->rename ||
+	    !subdir->d_inode->i_op->rmdir ||
+	    !subdir->d_inode->i_op->unlink)
+		goto check_error;
+
+	_leave(" = [%lu]", subdir->d_inode->i_ino);
+	return subdir;
+
+check_error:
+	dput(subdir);
+	_leave(" = %d [check]", ret);
+	return ERR_PTR(ret);
+
+mkdir_error:
+	mutex_unlock(&dir->d_inode->i_mutex);
+	kerror("mkdir %s failed with error %d", dirname, ret);
+	goto error_out;
+
+lookup_error:
+	mutex_unlock(&dir->d_inode->i_mutex);
+	dput(new);
+	ret = PTR_ERR(subdir);
+	kerror("Lookup %s failed with error %d", dirname, ret);
+	goto error_out;
+
+nomem_d_alloc:
+	mutex_unlock(&dir->d_inode->i_mutex);
+	ret = -ENOMEM;
+	goto error_out;
+
+error_out:
+	current->fsuid = fsuid;
+	current->fsgid = fsgid;
+	_leave(" = %d", ret);
+	return ERR_PTR(ret);
+}
+
+/*****************************************************************************/
+/*
+ * cull an object if it's not in use
+ * - called only by cache manager daemon
+ */
+int cachefiles_cull(struct cachefiles_cache *cache, struct dentry *dir,
+		    char *filename)
+{
+	struct cachefiles_object *object;
+	struct rb_node *_n;
+	struct dentry *victim, *new;
+	struct qstr name;
+	int ret;
+
+	_enter(",%*.*s/,%s",
+	       dir->d_name.len, dir->d_name.len, dir->d_name.name, filename);
+
+	/* set up the name */
+	name.name = filename;
+	name.len = strlen(filename);
+	name.hash = full_name_hash(name.name, name.len);
+
+	if (dir->d_op && dir->d_op->d_hash) {
+		ret = dir->d_op->d_hash(dir, &name);
+		if (ret < 0) {
+			if (ret == -EIO)
+				cachefiles_io_error(cache, "Hash failed");
+			_leave(" = %d", ret);
+			return ret;
+		}
+	}
+
+	/* look up the victim */
+	mutex_lock(&dir->d_inode->i_mutex);
+
+	victim = d_lookup(dir, &name);
+	if (!victim) {
+		_debug("not found");
+
+		new = d_alloc(dir, &name);
+		if (!new)
+			goto nomem_d_alloc;
+
+		victim = dir->d_inode->i_op->lookup(dir->d_inode, new, NULL);
+		if (IS_ERR(victim))
+			goto lookup_error;
+
+		if (!victim)
+			victim = new;
+		else
+			dput(new);
+	}
+
+	_debug("victim -> %p %s",
+	       victim, victim->d_inode ? "positive" : "negative");
+
+	/* if the object is no longer there then we probably retired the object
+	 * at the netfs's request whilst the cull was in progress
+	 */
+	if (!victim->d_inode) {
+		mutex_unlock(&dir->d_inode->i_mutex);
+		dput(victim);
+		_leave(" = -ENOENT [absent]");
+		return -ENOENT;
+	}
+
+	/* check to see if we're using this object */
+	read_lock(&cache->active_lock);
+
+	_n = cache->active_nodes.rb_node;
+
+	while (_n) {
+		object = rb_entry(_n, struct cachefiles_object, active_node);
+
+		if (object->dentry > victim)
+			_n = _n->rb_left;
+		else if (object->dentry < victim)
+			_n = _n->rb_right;
+		else
+			goto object_in_use;
+	}
+
+	read_unlock(&cache->active_lock);
+
+	/* okay... the victim is not being used so we can cull it
+	 * - start by marking it as stale
+	 */
+	_debug("victim is cullable");
+
+	ret = cachefiles_remove_object_xattr(cache, victim);
+	if (ret < 0)
+		goto error_unlock;
+
+	/*  actually remove the victim (drops the dir mutex) */
+	_debug("bury");
+
+	ret = cachefiles_bury_object(cache, dir, victim);
+	if (ret < 0)
+		goto error;
+
+	dput(victim);
+	_leave(" = 0");
+	return 0;
+
+
+object_in_use:
+	read_unlock(&cache->active_lock);
+	mutex_unlock(&dir->d_inode->i_mutex);
+	dput(victim);
+	_leave(" = -EBUSY [in use]");
+	return -EBUSY;
+
+nomem_d_alloc:
+	mutex_unlock(&dir->d_inode->i_mutex);
+	_leave(" = -ENOMEM");
+	return -ENOMEM;
+
+lookup_error:
+	mutex_unlock(&dir->d_inode->i_mutex);
+	dput(new);
+	ret = PTR_ERR(victim);
+	if (ret == -EIO)
+		cachefiles_io_error(cache, "Lookup failed");
+	goto choose_error;
+
+error_unlock:
+	mutex_unlock(&dir->d_inode->i_mutex);
+error:
+	dput(victim);
+choose_error:
+	if (ret == -ENOENT) {
+		/* file or dir now absent - probably retired by netfs */
+		_leave(" = -ESTALE [absent]");
+		return -ESTALE;
+	}
+
+	if (ret != -ENOMEM) {
+		kerror("Internal error: %d", ret);
+		ret = -EIO;
+	}
+
+	_leave(" = %d", ret);
+	return ret;
+}
diff --git a/fs/cachefiles/cf-proc.c b/fs/cachefiles/cf-proc.c
new file mode 100644
index 0000000..c14320e
--- /dev/null
+++ b/fs/cachefiles/cf-proc.c
@@ -0,0 +1,498 @@
+/* cf-proc.c: /proc/fs/cachefiles interface
+ *
+ * Copyright (C) 2006 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/sched.h>
+#include <linux/completion.h>
+#include <linux/slab.h>
+#include <linux/fs.h>
+#include <linux/file.h>
+#include <linux/namei.h>
+#include <linux/mount.h>
+#include <linux/namespace.h>
+#include <linux/statfs.h>
+#include <linux/proc_fs.h>
+#include <linux/ctype.h>
+#include "internal.h"
+
+static int cachefiles_proc_open(struct inode *, struct file *);
+static int cachefiles_proc_release(struct inode *, struct file *);
+static ssize_t cachefiles_proc_read(struct file *, char __user *, size_t, loff_t *);
+static ssize_t cachefiles_proc_write(struct file *, const char __user *, size_t, loff_t *);
+static int cachefiles_proc_brun(struct cachefiles_cache *cache, char *args);
+static int cachefiles_proc_bcull(struct cachefiles_cache *cache, char *args);
+static int cachefiles_proc_bstop(struct cachefiles_cache *cache, char *args);
+static int cachefiles_proc_cull(struct cachefiles_cache *cache, char *args);
+static int cachefiles_proc_debug(struct cachefiles_cache *cache, char *args);
+static int cachefiles_proc_dir(struct cachefiles_cache *cache, char *args);
+static int cachefiles_proc_tag(struct cachefiles_cache *cache, char *args);
+
+struct proc_dir_entry *cachefiles_proc;
+
+static unsigned long cachefiles_open;
+
+struct file_operations cachefiles_proc_fops = {
+	.open		= cachefiles_proc_open,
+	.release	= cachefiles_proc_release,
+	.read		= cachefiles_proc_read,
+	.write		= cachefiles_proc_write,
+};
+
+struct cachefiles_proc_cmd {
+	char name[8];
+	int (*handler)(struct cachefiles_cache *cache, char *args);
+};
+
+static const struct cachefiles_proc_cmd cachefiles_proc_cmds[] = {
+	{ "bind",	cachefiles_proc_bind	},
+	{ "brun",	cachefiles_proc_brun	},
+	{ "bcull",	cachefiles_proc_bcull	},
+	{ "bstop",	cachefiles_proc_bstop	},
+	{ "cull",	cachefiles_proc_cull	},
+	{ "debug",	cachefiles_proc_debug	},
+	{ "dir",	cachefiles_proc_dir	},
+	{ "tag",	cachefiles_proc_tag	},
+	{ "",		NULL			}
+};
+
+
+/*****************************************************************************/
+/*
+ * do various checks
+ */
+static int cachefiles_proc_open(struct inode *inode, struct file *file)
+{
+	struct cachefiles_cache *cache;
+
+	_enter("");
+
+	/* only the superuser may do this */
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	/* /proc/fs/cachefiles may only be open once at a time */
+	if (xchg(&cachefiles_open, 1) == 1)
+		return -EBUSY;
+
+	/* allocate a cache record */
+	cache = kzalloc(sizeof(struct cachefiles_cache), GFP_KERNEL);
+	if (!cache) {
+		cachefiles_open = 0;
+		return -ENOMEM;
+	}
+
+	cache->active_nodes = RB_ROOT;
+	rwlock_init(&cache->active_lock);
+
+	/* set default caching limits
+	 * - limit at 1% free space
+	 * - cull below 5% free space
+	 * - cease culling above 7% free space
+	 */
+	cache->brun_percent = 7;
+	cache->bcull_percent = 5;
+	cache->bstop_percent = 1;
+
+	file->private_data = cache;
+	cache->cachefilesd = file;
+	return 0;
+}
+
+/*****************************************************************************/
+/*
+ * release a cache
+ */
+static int cachefiles_proc_release(struct inode *inode, struct file *file)
+{
+	struct cachefiles_cache *cache = file->private_data;
+
+	_enter("");
+
+	ASSERT(cache);
+
+	set_bit(CACHEFILES_DEAD, &cache->flags);
+
+	cachefiles_proc_unbind(cache);
+
+	ASSERT(!cache->active_nodes.rb_node);
+
+	/* clean up the control file interface */
+	cache->cachefilesd = NULL;
+	file->private_data = NULL;
+	cachefiles_open = 0;
+
+	kfree(cache);
+
+	_leave("");
+	return 0;
+}
+
+/*****************************************************************************/
+/*
+ * read the cache state
+ */
+static ssize_t cachefiles_proc_read(struct file *file, char __user *_buffer,
+				    size_t buflen, loff_t *pos)
+{
+	struct cachefiles_cache *cache = file->private_data;
+	char buffer[256];
+	int n;
+
+	_enter(",,%zu,", buflen);
+
+	if (!test_bit(CACHEFILES_READY, &cache->flags))
+		return 0;
+
+	/* check how much space the cache has */
+	cachefiles_has_space(cache, 0);
+
+	/* summarise */
+	n = snprintf(buffer, sizeof(buffer),
+		     "cull=%c"
+		     " brun=%llx"
+		     " bcull=%llx"
+		     " bstop=%llx",
+		     test_bit(CACHEFILES_CULLING, &cache->flags) ? '1' : '0',
+		     cache->brun,
+		     cache->bcull,
+		     cache->bstop
+		     );
+
+	if (n > buflen)
+		return -EMSGSIZE;
+
+	if (copy_to_user(_buffer, buffer, n) != 0)
+		return -EFAULT;
+
+	return n;
+}
+
+/*****************************************************************************/
+/*
+ * command the cache
+ */
+static ssize_t cachefiles_proc_write(struct file *file,
+				     const char __user *_data, size_t datalen,
+				     loff_t *pos)
+{
+	const struct cachefiles_proc_cmd *cmd;
+	struct cachefiles_cache *cache = file->private_data;
+	ssize_t ret;
+	char *data, *args, *cp;
+
+	_enter(",,%zu,", datalen);
+
+	ASSERT(cache);
+
+	if (test_bit(CACHEFILES_DEAD, &cache->flags))
+		return -EIO;
+
+	if (datalen < 0 || datalen > PAGE_SIZE - 1)
+		return -EOPNOTSUPP;
+
+	/* drag the command string into the kernel so we can parse it */
+	data = kmalloc(datalen + 1, GFP_KERNEL);
+	if (!data)
+		return -ENOMEM;
+
+	ret = -EFAULT;
+	if (copy_from_user(data, _data, datalen) != 0)
+		goto error;
+
+	data[datalen] = '\0';
+
+	ret = -EINVAL;
+	if (memchr(data, '\0', datalen))
+		goto error;
+
+	/* strip any newline */
+	cp = memchr(data, '\n', datalen);
+	if (cp) {
+		if (cp == data)
+			goto error;
+
+		*cp = '\0';
+	}
+
+	/* parse the command */
+	ret = -EOPNOTSUPP;
+
+	for (args = data; *args; args++)
+		if (isspace(*args))
+			break;
+	if (*args) {
+		if (args == data)
+			goto error;
+		*args = '\0';
+		for (args++; isspace(*args); args++)
+			continue;
+	}
+
+	/* run the appropriate command handler */
+	for (cmd = cachefiles_proc_cmds; cmd->name[0]; cmd++)
+		if (strcmp(cmd->name, data) == 0)
+			goto found_command;
+
+error:
+	kfree(data);
+	_leave(" = %d", ret);
+	return ret;
+
+found_command:
+	mutex_lock(&file->f_dentry->d_inode->i_mutex);
+
+	ret = -EIO;
+	if (!test_bit(CACHEFILES_DEAD, &cache->flags))
+		ret = cmd->handler(cache, args);
+
+	mutex_unlock(&file->f_dentry->d_inode->i_mutex);
+
+	if (ret == 0)
+		ret = datalen;
+	goto error;
+}
+
+/*****************************************************************************/
+/*
+ * give a range error for cache space constraints
+ * - can be tail-called
+ */
+static int cachefiles_proc_range_error(struct cachefiles_cache *cache, char *args)
+{
+	kerror("Free space limits must be in range"
+	       " 0%%<=bstop<bcull<brun<100%%");
+
+	return -EINVAL;
+}
+
+/*****************************************************************************/
+/*
+ * set the percentage of blocks at which to stop culling
+ * - command: "brun <N>%"
+ */
+static int cachefiles_proc_brun(struct cachefiles_cache *cache, char *args)
+{
+	unsigned long brun;
+
+	_enter(",%s", args);
+
+	if (!*args)
+		return -EINVAL;
+
+	brun = simple_strtoul(args, &args, 10);
+	if (args[0] != '%' || args[1] != '\0')
+		return -EINVAL;
+
+	if (brun <= cache->bcull_percent || brun >= 100)
+		return cachefiles_proc_range_error(cache, args);
+
+	cache->brun_percent = brun;
+	return 0;
+}
+
+/*****************************************************************************/
+/*
+ * set the percentage of blocks at which to start culling
+ * - command: "bcull <N>%"
+ */
+static int cachefiles_proc_bcull(struct cachefiles_cache *cache, char *args)
+{
+	unsigned long bcull;
+
+	_enter(",%s", args);
+
+	if (!*args)
+		return -EINVAL;
+
+	bcull = simple_strtoul(args, &args, 10);
+	if (args[0] != '%' || args[1] != '\0')
+		return -EINVAL;
+
+	if (bcull <= cache->bstop_percent || bcull >= cache->brun_percent)
+		return cachefiles_proc_range_error(cache, args);
+
+	cache->bcull_percent = bcull;
+	return 0;
+}
+
+/*****************************************************************************/
+/*
+ * set the percentage of blocks at which to stop allocating
+ * - command: "bstop <N>%"
+ */
+static int cachefiles_proc_bstop(struct cachefiles_cache *cache, char *args)
+{
+	unsigned long bstop;
+
+	_enter(",%s", args);
+
+	if (!*args)
+		return -EINVAL;
+
+	bstop = simple_strtoul(args, &args, 10);
+	if (args[0] != '%' || args[1] != '\0')
+		return -EINVAL;
+
+	if (bstop < 0 || bstop >= cache->bcull_percent)
+		return cachefiles_proc_range_error(cache, args);
+
+	cache->bstop_percent = bstop;
+	return 0;
+}
+
+/*****************************************************************************/
+/*
+ * set the cache directory
+ * - command: "dir <name>"
+ */
+static int cachefiles_proc_dir(struct cachefiles_cache *cache, char *args)
+{
+	char *dir;
+
+	_enter(",%s", args);
+
+	if (!*args) {
+		kerror("Empty directory specified");
+		return -EINVAL;
+	}
+
+	if (cache->rootdirname) {
+		kerror("Second cache directory specified");
+		return -EEXIST;
+	}
+
+	dir = kstrdup(args, GFP_KERNEL);
+	if (!dir)
+		return -ENOMEM;
+
+	cache->rootdirname = dir;
+	return 0;
+}
+
+/*****************************************************************************/
+/*
+ * set the cache tag
+ * - command: "tag <name>"
+ */
+static int cachefiles_proc_tag(struct cachefiles_cache *cache, char *args)
+{
+	char *tag;
+
+	_enter(",%s", args);
+
+	if (!*args) {
+		kerror("Empty tag specified");
+		return -EINVAL;
+	}
+
+	if (cache->tag)
+		return -EEXIST;
+
+	tag = kstrdup(args, GFP_KERNEL);
+	if (!tag)
+		return -ENOMEM;
+
+	cache->tag = tag;
+	return 0;
+}
+
+/*****************************************************************************/
+/*
+ * request a node in the cache be culled
+ * - command: "cull <dirfd> <name>"
+ */
+static int cachefiles_proc_cull(struct cachefiles_cache *cache, char *args)
+{
+	struct dentry *dir;
+	struct file *dirfile;
+	int dirfd, fput_needed, ret;
+
+	_enter(",%s", args);
+
+	dirfd = simple_strtoul(args, &args, 0);
+
+	if (!args || !isspace(*args))
+		goto inval;
+
+	while (isspace(*args))
+		args++;
+
+	if (!*args)
+		goto inval;
+
+	if (strchr(args, '/'))
+		goto inval;
+
+	if (!test_bit(CACHEFILES_READY, &cache->flags)) {
+		kerror("cull applied to unready cache");
+		return -EIO;
+	}
+
+	if (test_bit(CACHEFILES_DEAD, &cache->flags)) {
+		kerror("cull applied to dead cache");
+		return -EIO;
+	}
+
+	/* extract the directory dentry from the fd */
+	dirfile = fget_light(dirfd, &fput_needed);
+	if (!dirfile) {
+		kerror("cull dirfd not open");
+		return -EBADF;
+	}
+
+	dir = dget(dirfile->f_dentry);
+	fput_light(dirfile, fput_needed);
+	dirfile = NULL;
+
+	if (!S_ISDIR(dir->d_inode->i_mode))
+		goto notdir;
+
+	ret = cachefiles_cull(cache, dir, args);
+
+	dput(dir);
+	_leave(" = %d", ret);
+	return ret;
+
+notdir:
+	dput(dir);
+	kerror("cull command requires dirfd to be a directory");
+	return -ENOTDIR;
+
+inval:
+	kerror("cull command requires dirfd and filename");
+	return -EINVAL;
+}
+
+/*****************************************************************************/
+/*
+ * set debugging mode
+ * - command: "debug <mask>"
+ */
+static int cachefiles_proc_debug(struct cachefiles_cache *cache, char *args)
+{
+	unsigned long mask;
+
+	_enter(",%s", args);
+
+	mask = simple_strtoul(args, &args, 0);
+
+	if (!args || !isspace(*args))
+		goto inval;
+
+	cachefiles_debug = mask;
+	_leave(" = 0");
+	return 0;
+
+inval:
+	kerror("debug command requires mask");
+	return -EINVAL;
+}
diff --git a/fs/cachefiles/cf-sysctl.c b/fs/cachefiles/cf-sysctl.c
new file mode 100644
index 0000000..4d6fc04
--- /dev/null
+++ b/fs/cachefiles/cf-sysctl.c
@@ -0,0 +1,69 @@
+/* cf-sysctl.c: Control parameters
+ *
+ * Copyright (C) 2006 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/sysctl.h>
+#include "internal.h"
+
+static struct ctl_table_header *cachefiles_sysctl;
+
+/*
+ * Something that isn't CTL_ANY, CTL_NONE or a value that may clash.
+ * Use the same values as fs/nfs/sysctl.c
+ */
+#define CTL_UNNUMBERED -2
+
+static ctl_table cachefiles_sysctl_table[] = {
+        {
+		.ctl_name	= CTL_UNNUMBERED,
+		.procname	= "debug",
+		.data		= &cachefiles_debug,
+		.maxlen		= sizeof(unsigned long),
+		.mode		= 0644,
+		.proc_handler	= &proc_doulongvec_minmax
+	},
+	{ .ctl_name = 0 }
+};
+
+static ctl_table cachefiles_sysctl_dir[] = {
+	{
+		.ctl_name	= CTL_UNNUMBERED,
+		.procname	= "cachefiles",
+		.maxlen		= 0,
+		.mode		= 0555,
+		.child		= cachefiles_sysctl_table
+	},
+	{ .ctl_name = 0 }
+};
+
+static ctl_table cachefiles_sysctl_root[] = {
+	{
+		.ctl_name = CTL_FS,
+		.procname = "fs",
+		.mode = 0555,
+		.child = cachefiles_sysctl_dir,
+	},
+	{ .ctl_name = 0 }
+};
+
+int __init cachefiles_sysctl_init(void)
+{
+	cachefiles_sysctl = register_sysctl_table(cachefiles_sysctl_root, 0);
+	if (!cachefiles_sysctl)
+		return -ENOMEM;
+	return 0;
+}
+
+void __exit cachefiles_sysctl_cleanup(void)
+{
+	unregister_sysctl_table(cachefiles_sysctl);
+}
diff --git a/fs/cachefiles/cf-xattr.c b/fs/cachefiles/cf-xattr.c
new file mode 100644
index 0000000..8952bf0
--- /dev/null
+++ b/fs/cachefiles/cf-xattr.c
@@ -0,0 +1,295 @@
+/* cf-xattr.c: CacheFiles extended attribute management
+ *
+ * Copyright (C) 2006 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/file.h>
+#include <linux/fs.h>
+#include <linux/fsnotify.h>
+#include <linux/quotaops.h>
+#include <linux/xattr.h>
+#include "internal.h"
+
+static const char cachefiles_xattr_cache[] = XATTR_USER_PREFIX "CacheFiles.cache";
+
+/*****************************************************************************/
+/*
+ * check the type label on an object
+ * - done using xattrs
+ */
+int cachefiles_check_object_type(struct cachefiles_object *object)
+{
+	struct dentry *dentry = object->dentry;
+	char type[3], xtype[3];
+	int ret;
+
+	ASSERT(dentry);
+	ASSERT(dentry->d_inode);
+	ASSERT(dentry->d_inode->i_op);
+	ASSERT(dentry->d_inode->i_op->setxattr);
+	ASSERT(dentry->d_inode->i_op->getxattr);
+
+	if (!object->fscache.cookie)
+		strcpy(type, "C3");
+	else
+		snprintf(type, 3, "%02x", object->fscache.cookie->def->type);
+
+	_enter("%p{%s}", object, type);
+
+	mutex_lock(&dentry->d_inode->i_mutex);
+
+	/* attempt to install a type label directly */
+	ret = dentry->d_inode->i_op->setxattr(dentry, cachefiles_xattr_cache,
+					      type, 2, XATTR_CREATE);
+	if (ret == 0) {
+		_debug("SET");
+		fsnotify_xattr(dentry);
+		mutex_unlock(&dentry->d_inode->i_mutex);
+		goto error;
+	}
+
+	if (ret != -EEXIST) {
+		kerror("Can't set xattr on %*.*s [%lu] (err %d)",
+		       dentry->d_name.len, dentry->d_name.len,
+		       dentry->d_name.name, dentry->d_inode->i_ino,
+		       -ret);
+		goto error;
+	}
+
+	/* read the current type label */
+	ret = dentry->d_inode->i_op->getxattr(dentry, cachefiles_xattr_cache,
+					      xtype, 3);
+	if (ret < 0) {
+		if (ret == -ERANGE)
+			goto bad_type_length;
+
+		kerror("Can't read xattr on %*.*s [%lu] (err %d)",
+		       dentry->d_name.len, dentry->d_name.len,
+		       dentry->d_name.name, dentry->d_inode->i_ino,
+		       -ret);
+		goto error;
+	}
+
+	/* check the type is what we're expecting */
+	if (ret != 2)
+		goto bad_type_length;
+
+	if (xtype[0] != type[0] || xtype[1] != type[1])
+		goto bad_type;
+
+	ret = 0;
+
+error:
+	mutex_unlock(&dentry->d_inode->i_mutex);
+	_leave(" = %d", ret);
+	return ret;
+
+bad_type_length:
+	kerror("Cache object %lu type xattr length incorrect",
+	       dentry->d_inode->i_ino);
+	ret = -EIO;
+	goto error;
+
+bad_type:
+	xtype[2] = 0;
+	kerror("Cache object %*.*s [%lu] type %s not %s",
+	       dentry->d_name.len, dentry->d_name.len,
+	       dentry->d_name.name, dentry->d_inode->i_ino,
+	       xtype, type);
+	ret = -EIO;
+	goto error;
+}
+
+/*****************************************************************************/
+/*
+ * set the state xattr on a cache file
+ */
+int cachefiles_set_object_xattr(struct cachefiles_object *object,
+				struct cachefiles_xattr *auxdata)
+{
+	struct dentry *dentry = object->dentry;
+	int ret;
+
+	ASSERT(object->fscache.cookie);
+	ASSERT(dentry);
+	ASSERT(dentry->d_inode->i_op->setxattr);
+
+	_enter("%p,#%d", object, auxdata->len);
+
+	/* attempt to install the cache metadata directly */
+	mutex_lock(&dentry->d_inode->i_mutex);
+
+	_debug("SET %s #%u",
+	       object->fscache.cookie->def->name, auxdata->len);
+
+	ret = dentry->d_inode->i_op->setxattr(dentry, cachefiles_xattr_cache,
+					      &auxdata->type, auxdata->len,
+					      XATTR_CREATE);
+	if (ret == 0)
+		fsnotify_xattr(dentry);
+	else if (ret != -ENOMEM)
+		cachefiles_io_error_obj(object,
+					"Failed to set xattr with error %d",
+					ret);
+
+	mutex_unlock(&dentry->d_inode->i_mutex);
+	_leave(" = %d", ret);
+	return ret;
+}
+
+/*****************************************************************************/
+/*
+ * check the state xattr on a cache file
+ * - return -ESTALE if the object should be deleted
+ */
+int cachefiles_check_object_xattr(struct cachefiles_object *object,
+				  struct cachefiles_xattr *auxdata)
+{
+	struct cachefiles_xattr *auxbuf;
+	struct dentry *dentry = object->dentry;
+	int ret;
+
+	_enter("%p,#%d", object, auxdata->len);
+
+	ASSERT(dentry);
+	ASSERT(dentry->d_inode);
+	ASSERT(dentry->d_inode->i_op->setxattr);
+	ASSERT(dentry->d_inode->i_op->getxattr);
+
+	auxbuf = kmalloc(sizeof(struct cachefiles_xattr) + 512, GFP_KERNEL);
+	if (!auxbuf) {
+		_leave(" = -ENOMEM");
+		return -ENOMEM;
+	}
+
+	mutex_lock(&dentry->d_inode->i_mutex);
+
+	/* read the current type label */
+	ret = dentry->d_inode->i_op->getxattr(dentry, cachefiles_xattr_cache,
+					      &auxbuf->type, 512 + 1);
+	if (ret < 0) {
+		if (ret == -ENODATA)
+			goto stale; /* no attribute - power went off
+				     * mid-cull? */
+
+		if (ret == -ERANGE)
+			goto bad_type_length;
+
+		cachefiles_io_error_obj(object,
+					"can't read xattr on %lu (err %d)",
+					dentry->d_inode->i_ino, -ret);
+		goto error;
+	}
+
+	/* check the on-disk object */
+	if (ret < 1)
+		goto bad_type_length;
+
+	if (auxbuf->type != auxdata->type)
+		goto stale;
+
+	auxbuf->len = ret;
+
+	/* consult the netfs */
+	if (object->fscache.cookie->def->check_aux) {
+		fscache_checkaux_t result;
+		unsigned int dlen;
+
+		dlen = auxbuf->len - 1;
+
+		_debug("checkaux %s #%u",
+		       object->fscache.cookie->def->name, dlen);
+
+		result = object->fscache.cookie->def->check_aux(
+			object->fscache.cookie->netfs_data,
+			&auxbuf->data, dlen);
+
+		switch (result) {
+			/* entry okay as is */
+		case FSCACHE_CHECKAUX_OKAY:
+			goto okay;
+
+			/* entry requires update */
+		case FSCACHE_CHECKAUX_NEEDS_UPDATE:
+			break;
+
+			/* entry requires deletion */
+		case FSCACHE_CHECKAUX_OBSOLETE:
+			goto stale;
+
+		default:
+			BUG();
+		}
+
+		/* update the current label */
+		ret = dentry->d_inode->i_op->setxattr(dentry,
+						      cachefiles_xattr_cache,
+						      &auxdata->type,
+						      auxdata->len,
+						      XATTR_REPLACE);
+		if (ret < 0) {
+			cachefiles_io_error_obj(object,
+						"Can't update xattr on %lu"
+						" (error %d)",
+						dentry->d_inode->i_ino, -ret);
+			goto error;
+		}
+	}
+
+okay:
+	ret = 0;
+
+error:
+	mutex_unlock(&dentry->d_inode->i_mutex);
+	kfree(auxbuf);
+	_leave(" = %d", ret);
+	return ret;
+
+bad_type_length:
+	kerror("Cache object %lu xattr length incorrect",
+	       dentry->d_inode->i_ino);
+	ret = -EIO;
+	goto error;
+
+stale:
+	ret = -ESTALE;
+	goto error;
+}
+
+/*****************************************************************************/
+/*
+ * remove the object's xattr to mark it stale
+ */
+int cachefiles_remove_object_xattr(struct cachefiles_cache *cache,
+				   struct dentry *dentry)
+{
+	int ret;
+
+	mutex_lock(&dentry->d_inode->i_mutex);
+
+	ret = dentry->d_inode->i_op->removexattr(dentry,
+						 cachefiles_xattr_cache);
+
+	mutex_unlock(&dentry->d_inode->i_mutex);
+
+	if (ret < 0) {
+		if (ret == -ENOENT || ret == -ENODATA)
+			ret = 0;
+		else if (ret != -ENOMEM)
+			cachefiles_io_error(cache,
+					    "Can't remove xattr from %lu"
+					    " (error %d)",
+					    dentry->d_inode->i_ino, -ret);
+	}
+
+	_leave(" = %d", ret);
+	return ret;
+}
diff --git a/fs/cachefiles/internal.h b/fs/cachefiles/internal.h
new file mode 100644
index 0000000..edda6e7
--- /dev/null
+++ b/fs/cachefiles/internal.h
@@ -0,0 +1,308 @@
+/* internal.h: general netfs cache on cache files internal defs
+ *
+ * Copyright (C) 2006 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ *
+ *
+ * CacheFiles layout:
+ *
+ *	/..../CacheDir/
+ *		index
+ *		0/
+ *		1/
+ *		2/
+ *		  index
+ *		  0/
+ *		  1/
+ *		  2/
+ *		    index
+ *		    0
+ *		    1
+ *		    2
+ */
+
+#include <linux/fscache-cache.h>
+#include <linux/timer.h>
+#include <linux/wait.h>
+#include <linux/workqueue.h>
+
+struct cachefiles_cache;
+struct cachefiles_object;
+
+extern unsigned long cachefiles_debug;
+#define CACHEFILES_DEBUG_KENTER	1
+#define CACHEFILES_DEBUG_KLEAVE	2
+#define CACHEFILES_DEBUG_KDEBUG	4
+
+extern struct fscache_cache_ops cachefiles_cache_ops;
+extern struct proc_dir_entry *cachefiles_proc;
+extern struct file_operations cachefiles_proc_fops;
+
+/*****************************************************************************/
+/*
+ * node records
+ */
+struct cachefiles_object {
+	struct fscache_object		fscache;	/* fscache handle */
+	struct dentry			*dentry;	/* the file/dir representing this object */
+	struct dentry			*backer;	/* backing file */
+	loff_t				i_size;		/* object size */
+	atomic_t			usage;		/* basic object usage count */
+	atomic_t			fscache_usage;	/* FSDEF object usage count */
+	uint8_t				type;		/* object type */
+	uint8_t				new;		/* T if object new */
+	spinlock_t			work_lock;
+	struct rw_semaphore		sem;
+	struct work_struct		read_work;	/* read page copier */
+	struct list_head		read_list;	/* pages to copy */
+	struct list_head		read_pend_list;	/* pages to pending read from backer */
+	struct work_struct		write_work;	/* page writer */
+	struct list_head		write_list;	/* pages to store */
+	struct rb_node			active_node;	/* link in active tree (dentry is key) */
+};
+
+extern kmem_cache_t *cachefiles_object_jar;
+
+/*****************************************************************************/
+/*
+ * Cache files cache definition
+ */
+struct cachefiles_cache {
+	struct fscache_cache		cache;		/* FS-Cache record */
+	struct vfsmount			*mnt;		/* mountpoint holding the cache */
+	struct dentry			*graveyard;	/* directory into which dead objects go */
+	struct file			*cachefilesd;	/* manager daemon handle */
+	struct rb_root			active_nodes;	/* active nodes (can't be culled) */
+	rwlock_t			active_lock;	/* lock for active_nodes */
+	atomic_t			gravecounter;	/* graveyard uniquifier */
+	unsigned			brun_percent;	/* when to stop culling (%) */
+	unsigned			bcull_percent;	/* when to start culling (%) */
+	unsigned			bstop_percent;	/* when to stop allocating (%) */
+	unsigned			bsize;		/* cache's block size */
+	unsigned			bshift;		/* min(log2 (PAGE_SIZE / bsize), 0) */
+	sector_t			brun;		/* when to stop culling */
+	sector_t			bcull;		/* when to start culling */
+	sector_t			bstop;		/* when to stop allocating */
+	unsigned long			flags;
+#define CACHEFILES_READY		0	/* T if cache prepared */
+#define CACHEFILES_DEAD			1	/* T if cache dead */
+#define CACHEFILES_CULLING		2	/* T if cull engaged */
+	char				*rootdirname;	/* name of cache root directory */
+	char				*tag;		/* cache binding tag */
+};
+
+/*****************************************************************************/
+/*
+ * backing file read tracking
+ */
+struct cachefiles_one_read {
+	wait_queue_t			monitor;	/* link into monitored waitqueue */
+	struct page			*back_page;	/* backing file page we're waiting for */
+	struct page			*netfs_page;	/* netfs page we're going to fill */
+	struct cachefiles_object	*object;
+	struct list_head		obj_link;	/* link in object's lists */
+	fscache_rw_complete_t		end_io_func;
+	void				*context;
+};
+
+/*****************************************************************************/
+/*
+ * backing file write tracking
+ */
+struct cachefiles_one_write {
+	struct page			*netfs_page;	/* netfs page to copy */
+	struct cachefiles_object	*object;
+	struct list_head		obj_link;	/* link in object's lists */
+	fscache_rw_complete_t		end_io_func;
+	void				*context;
+};
+
+/*****************************************************************************/
+/*
+ * auxiliary data xattr buffer
+ */
+struct cachefiles_xattr {
+	uint16_t			len;
+	uint8_t				type;
+	uint8_t				data[];
+};
+
+
+/* cf-bind.c */
+extern int cachefiles_proc_bind(struct cachefiles_cache *cache, char *args);
+extern void cachefiles_proc_unbind(struct cachefiles_cache *cache);
+
+/* cf-interface.c */
+extern void cachefiles_read_copier_work(void *_object);
+extern void cachefiles_write_work(void *_object);
+extern int cachefiles_has_space(struct cachefiles_cache *cache, unsigned nr);
+
+/* cf-key.c */
+extern char *cachefiles_cook_key(const u8 *raw, int keylen, uint8_t type);
+
+/* cf-namei.c */
+extern int cachefiles_delete_object(struct cachefiles_cache *cache,
+				    struct cachefiles_object *object);
+extern int cachefiles_walk_to_object(struct cachefiles_object *parent,
+				     struct cachefiles_object *object,
+				     char *key,
+				     struct cachefiles_xattr *auxdata);
+extern struct dentry *cachefiles_get_directory(struct cachefiles_cache *cache,
+					       struct dentry *dir,
+					       const char *name);
+
+extern int cachefiles_cull(struct cachefiles_cache *cache, struct dentry *dir,
+			   char *filename);
+
+/* cf-sysctl.c */
+extern int __init cachefiles_sysctl_init(void);
+extern void __exit cachefiles_sysctl_cleanup(void);
+
+/* cf-xattr.c */
+extern int cachefiles_check_object_type(struct cachefiles_object *object);
+extern int cachefiles_set_object_xattr(struct cachefiles_object *object,
+				       struct cachefiles_xattr *auxdata);
+extern int cachefiles_check_object_xattr(struct cachefiles_object *object,
+					 struct cachefiles_xattr *auxdata);
+extern int cachefiles_remove_object_xattr(struct cachefiles_cache *cache,
+					  struct dentry *dentry);
+
+
+/*****************************************************************************/
+/*
+ * error handling
+ */
+#define kerror(FMT,...) printk(KERN_ERR "CacheFiles: "FMT"\n" ,##__VA_ARGS__);
+
+#define cachefiles_io_error(___cache, FMT, ...)		\
+do {							\
+	kerror("I/O Error: " FMT ,##__VA_ARGS__);	\
+	fscache_io_error(&(___cache)->cache);		\
+	set_bit(CACHEFILES_DEAD, &(___cache)->flags);	\
+} while(0)
+
+#define cachefiles_io_error_obj(object, FMT, ...)			\
+do {									\
+	struct cachefiles_cache *___cache;				\
+									\
+	___cache = container_of((object)->fscache.cache,		\
+				struct cachefiles_cache, cache);	\
+	cachefiles_io_error(___cache, FMT ,##__VA_ARGS__);		\
+} while(0)
+
+
+/*****************************************************************************/
+/*
+ * debug tracing
+ */
+#define dbgprintk(FMT,...) \
+	printk("[%-6.6s] "FMT"\n",current->comm ,##__VA_ARGS__)
+
+/* make sure we maintain the format strings, even when debugging is disabled */
+static inline void _dbprintk(const char *fmt, ...)
+	__attribute__((format(printf,1,2)));
+static inline void _dbprintk(const char *fmt, ...)
+{
+}
+
+#define kenter(FMT,...)	dbgprintk("==> %s("FMT")",__FUNCTION__ ,##__VA_ARGS__)
+#define kleave(FMT,...)	dbgprintk("<== %s()"FMT"",__FUNCTION__ ,##__VA_ARGS__)
+#define kdebug(FMT,...)	dbgprintk(FMT ,##__VA_ARGS__)
+
+
+#if defined(__KDEBUG)
+#define _enter(FMT,...)	kenter(FMT,##__VA_ARGS__)
+#define _leave(FMT,...)	kleave(FMT,##__VA_ARGS__)
+#define _debug(FMT,...)	kdebug(FMT,##__VA_ARGS__)
+
+#elif defined(CONFIG_CACHEFILES_DEBUG)
+#define _enter(FMT,...)					\
+do {							\
+	if (cachefiles_debug & CACHEFILES_DEBUG_KENTER)	\
+		kenter(FMT,##__VA_ARGS__);		\
+} while (0)
+
+#define _leave(FMT,...)					\
+do {							\
+	if (cachefiles_debug & CACHEFILES_DEBUG_KLEAVE)	\
+		kleave(FMT,##__VA_ARGS__);		\
+} while (0)
+
+#define _debug(FMT,...)					\
+do {							\
+	if (cachefiles_debug & CACHEFILES_DEBUG_KDEBUG)	\
+		kdebug(FMT,##__VA_ARGS__);		\
+} while (0)
+
+#else
+#define _enter(FMT,...)	_dbprintk("==> %s("FMT")",__FUNCTION__ ,##__VA_ARGS__)
+#define _leave(FMT,...)	_dbprintk("<== %s()"FMT"",__FUNCTION__ ,##__VA_ARGS__)
+#define _debug(FMT,...)	_dbprintk(FMT ,##__VA_ARGS__)
+#endif
+
+#if 1 // defined(__KDEBUGALL)
+
+#define ASSERT(X)							\
+do {									\
+	if (unlikely(!(X))) {						\
+		printk(KERN_ERR "\n");					\
+		printk(KERN_ERR "CacheFiles: Assertion failed\n");	\
+		BUG();							\
+	}								\
+} while(0)
+
+#define ASSERTCMP(X, OP, Y)						\
+do {									\
+	if (unlikely(!((X) OP (Y)))) {					\
+		printk(KERN_ERR "\n");					\
+		printk(KERN_ERR "CacheFiles: Assertion failed\n");	\
+		printk(KERN_ERR "%lx " #OP " %lx is false\n",		\
+		       (unsigned long)(X), (unsigned long)(Y));		\
+		BUG();							\
+	}								\
+} while(0)
+
+#define ASSERTIF(C, X)							\
+do {									\
+	if (unlikely((C) && !(X))) {					\
+		printk(KERN_ERR "\n");					\
+		printk(KERN_ERR "CacheFiles: Assertion failed\n");	\
+		BUG();							\
+	}								\
+} while(0)
+
+#define ASSERTIFCMP(C, X, OP, Y)					\
+do {									\
+	if (unlikely((C) && !((X) OP (Y)))) {				\
+		printk(KERN_ERR "\n");					\
+		printk(KERN_ERR "CacheFiles: Assertion failed\n");	\
+		printk(KERN_ERR "%lx " #OP " %lx is false\n",		\
+		       (unsigned long)(X), (unsigned long)(Y));		\
+		BUG();							\
+	}								\
+} while(0)
+
+#else
+
+#define ASSERT(X)				\
+do {						\
+} while(0)
+
+#define ASSERTCMP(X, OP, Y)			\
+do {						\
+} while(0)
+
+#define ASSERTIF(C, X)				\
+do {						\
+} while(0)
+
+#define ASSERTIFCMP(C, X, OP, Y)		\
+do {						\
+} while(0)
+
+#endif
Date: Mon, 16 Oct 2006 12:53:28 +0100
From: David Howells <dhowells@redhat.com>
Subject: [RHEL5 PATCH] CacheFiles: Fix double unlock bug

Fix a bug whereby cachefiles_check_object_type() doubly unlocks a mutex under
some circumstances [BZ 210701].

The first line after the "error:" label is another unlock of the same mutex,
so also unlocking just before going there is incorrect.

David

diff --git a/fs/cachefiles/cf-xattr.c b/fs/cachefiles/cf-xattr.c
index c03f699..e232dd6 100644
--- a/fs/cachefiles/cf-xattr.c
+++ b/fs/cachefiles/cf-xattr.c
@@ -51,7 +51,6 @@ int cachefiles_check_object_type(struct 
 	if (ret == 0) {
 		_debug("SET");
 		fsnotify_xattr(dentry);
-		mutex_unlock(&dentry->d_inode->i_mutex);
 		goto error;
 	}

Date: Tue, 17 Oct 2006 14:31:48 -0400
From: Peter Martuccelli <peterm@redhat.com>
Subject: [RHEL5 PATCH] Annotate the lock of i_mutex on /proc/fs/fscache

Hello,

Posting a RHEL5 Beta2 blocker for/from David as he is having 
problems getting this posted to rhkernel-list.

Regards,

Peter

<David's post start here>

Annotate the lock of i_mutex when handling a write to /proc/fs/fscache
so that the lockdep facility can see that it's okay to hold both this
lock and the lock on inodes in the backing files at the same time [BZ
204615].

Without this patch, the lockdep facility will see two inode i_mutex
locks being held at the same time - something which could cause a
deadlock under other circumstances - and generates a warning about it.

Ingo: Can you confirm that (a) this is the right way to handle things,
and (b) it's okay to use subclass 1.

David

Index: latest/fs/cachefiles/cf-proc.c
===================================================================
--- latest.orig/fs/cachefiles/cf-proc.c
+++ latest/fs/cachefiles/cf-proc.c
@@ -248,7 +248,7 @@ error:
 	return ret;
 
 found_command:
-	mutex_lock(&file->f_dentry->d_inode->i_mutex);
+	mutex_lock_nested(&file->f_dentry->d_inode->i_mutex, 1);
 
 	ret = -EIO;
 	if (!test_bit(CACHEFILES_DEAD, &cache->flags))
Date: Wed, 30 Aug 2006 18:04:26 -0400
From: Steve Dickson <SteveD@redhat.com>
Subject: Re: [RHEL5/FC6 PATCH] CacheFiles: Inode count maintenance

David Howells wrote:
>Maintain a certain amount of free inode/file space in addition to free disk
>space.
>
>The problem is that although CacheFiles makes sure that it leaves a certain
>amount of disk space free, and will shrink the cache if necessary, it will
>happily eat all the possible inodes on the backing filesystem and will
>otherwise sit there doing nothing, despite there being lots of free space 
>and
>no free inodes.
>
>I've tested this on my testbox using the same procedure as for a previous
>patch:
>
>	Set up a very small cache, for example with:
>
>		dd if=/dev/zero of=/root/fakedisc bs=10240 count=1024
>		mke2fs -j /root/fakedisc 
>		tune2fs -o user_xattr /root/fakedisc 
>		mount -o loop /root/fakedisc /var/fscache/
>		cachefilesd
>
>	And then running a "find" over a few thousand NFS files on an NFS 
>	share
>	that's mounted with "-o fsc".  This will eventually run the cache out
>	of free inodes, and ENOSPC will be returned by Ext3 preventing 
>	further
>	creation.
>
>As long as there is sufficient free disk space (which there should be), the
>cachefiles facility will just sit there rejecting all new requests without
>initiating culling.  Once this patch is applied, it will actively begin 
>culling
>to reduce the number of inodes it is consuming.
>
>Note that the documentation in the cachefilesd package also needs updating 
>to
>reflect the fact that there are now three new configuration options.
Attached is the patch what will apply to and was tested on a
RHEL5 2.6.17-1.2519.4.18 kernel (verses an rawhide and upstream kernel).

steved.

--- linux-2.6.17.i686/Documentation/filesystems/caching/cachefiles.txt.inode	2006-08-30 15:04:15.737846000 -0400
+++ linux-2.6.17.i686/Documentation/filesystems/caching/cachefiles.txt	2006-08-30 15:05:04.171612000 -0400
@@ -77,9 +77,15 @@ set up cache ready for use.  The followi
  (*) brun <N>%
  (*) bcull <N>%
  (*) bstop <N>%
+ (*) frun <N>%
+ (*) fcull <N>%
+ (*) fstop <N>%
 
 	Configure the culling limits.  Optional.  See the section on culling
-	The defaults are 7%, 5% and 1% respectively.
+	The defaults are 7% (run), 5% (cull) and 1% (stop) respectively.
+
+	The commands beginning with a 'b' are file space (block) limits, those
+	beginning with an 'f' are file count limits.
 
  (*) dir <path>
 
@@ -168,31 +174,37 @@ discarding objects from the cache that h
 anything else.  Culling is based on the access time of data objects.  Empty
 directories are culled if not in use.
 
-Cache culling is done on the basis of the percentage of blocks available in the
-underlying filesystem.  There are three "limits":
+Cache culling is done on the basis of the percentage of blocks and the
+percentage of files available in the underlying filesystem.  There are six
+"limits":
 
  (*) brun
+ (*) frun
 
-     If the amount of available space in the cache rises above this limit, then
-     culling is turned off.
+     If the amount of free space and the number of available files in the cache
+     rises above both these limits, then culling is turned off.
 
  (*) bcull
+ (*) fcull
 
-     If the amount of available space in the cache falls below this limit, then
-     culling is started.
+     If the amount of available space or the number of available files in the
+     cache falls below either of these limits, then culling is started.
 
  (*) bstop
+ (*) fstop
 
-     If the amount of available space in the cache falls below this limit, then
-     no further allocation of disk space is permitted until culling has raised
-     the amount above this limit again.
+     If the amount of available space or the number of available files in the
+     cache falls below either of these limits, then no further allocation of
+     disk space or files is permitted until culling has raised things above
+     these limits again.
 
 These must be configured thusly:
 
 	0 <= bstop < bcull < brun < 100
+	0 <= fstop < fcull < frun < 100
 
-Note that these are percentages of available space, and do _not_ appear as 100
-minus the percentage displayed by the "df" program.
+Note that these are percentages of available space and available files, and do
+_not_ appear as 100 minus the percentage displayed by the "df" program.
 
 The userspace daemon scans the cache to build up a table of cullable objects.
 These are then culled in least recently used order.  A new scan of the cache is
--- linux-2.6.17.i686/fs/cachefiles/cf-bind.c.inode	2006-08-30 15:00:18.549787000 -0400
+++ linux-2.6.17.i686/fs/cachefiles/cf-bind.c	2006-08-30 15:05:04.176612000 -0400
@@ -33,13 +33,21 @@ static int cachefiles_proc_add_cache(str
  */
 int cachefiles_proc_bind(struct cachefiles_cache *cache, char *args)
 {
-	_enter("{%u,%u,%u},%s",
+	_enter("{%u,%u,%u,%u,%u,%u},%s",
+	       cache->frun_percent,
+	       cache->fcull_percent,
+	       cache->fstop_percent,
 	       cache->brun_percent,
 	       cache->bcull_percent,
 	       cache->bstop_percent,
 	       args);
 
 	/* start by checking things over */
+	ASSERT(cache->fstop_percent >= 0 &&
+	       cache->fstop_percent < cache->fcull_percent &&
+	       cache->fcull_percent < cache->frun_percent &&
+	       cache->frun_percent  < 100);
+
 	ASSERT(cache->bstop_percent >= 0 &&
 	       cache->bstop_percent < cache->bcull_percent &&
 	       cache->bcull_percent < cache->brun_percent &&
@@ -173,19 +181,31 @@ static int cachefiles_proc_add_cache(str
 	_debug("blksize %u (shift %u)",
 	       cache->bsize, cache->bshift);
 
-	_debug("size %llu, avail %llu", stats.f_blocks, stats.f_bavail);
+	_debug("size %llu, avail %llu",
+	       (unsigned long long) stats.f_blocks,
+	       (unsigned long long) stats.f_bavail);
 
 	/* set up caching limits */
+	do_div(stats.f_files, 100);
+	cache->fstop = stats.f_files * cache->fstop_percent;
+	cache->fcull = stats.f_files * cache->fcull_percent;
+	cache->frun  = stats.f_files * cache->frun_percent;
+
+	_debug("limits {%llu,%llu,%llu} files",
+	       (unsigned long long) cache->frun,
+	       (unsigned long long) cache->fcull,
+	       (unsigned long long) cache->fstop);
+
 	stats.f_blocks >>= cache->bshift;
 	do_div(stats.f_blocks, 100);
 	cache->bstop = stats.f_blocks * cache->bstop_percent;
 	cache->bcull = stats.f_blocks * cache->bcull_percent;
 	cache->brun  = stats.f_blocks * cache->brun_percent;
 
-	_debug("limits {%llu,%llu,%llu}",
-	       cache->brun,
-	       cache->bcull,
-	       cache->bstop);
+	_debug("limits {%llu,%llu,%llu} blocks",
+	       (unsigned long long) cache->brun,
+	       (unsigned long long) cache->bcull,
+	       (unsigned long long) cache->bstop);
 
 	/* get the cache directory and check its type */
 	cachedir = cachefiles_get_directory(cache, root, "cache");
@@ -230,7 +250,7 @@ static int cachefiles_proc_add_cache(str
 	       cache->cache.identifier);
 
 	/* check how much space the cache has */
-	cachefiles_has_space(cache, 0);
+	cachefiles_has_space(cache, 0, 0);
 
 	return 0;
 
--- linux-2.6.17.i686/fs/cachefiles/cf-interface.c.inode	2006-08-30 15:00:18.565770000 -0400
+++ linux-2.6.17.i686/fs/cachefiles/cf-interface.c	2006-08-30 15:05:04.186612000 -0400
@@ -315,15 +315,23 @@ static int cachefiles_set_i_size(struct 
 
 /*****************************************************************************/
 /*
- * see if we have space for a number of pages in the cache
+ * see if we have space for a number of pages and/or a number of files in the
+ * cache
  */
-int cachefiles_has_space(struct cachefiles_cache *cache, unsigned nr)
+int cachefiles_has_space(struct cachefiles_cache *cache,
+			 unsigned fnr, unsigned bnr)
 {
 	struct kstatfs stats;
 	int ret;
 
-	_enter("{%llu,%llu,%llu},%d",
-	       cache->brun, cache->bcull, cache->bstop,  nr);
+	_enter("{%llu,%llu,%llu,%llu,%llu,%llu},%u,%u",
+	       (unsigned long long) cache->frun,
+	       (unsigned long long) cache->fcull,
+	       (unsigned long long) cache->fstop,
+	       (unsigned long long) cache->brun,
+	       (unsigned long long) cache->bcull,
+	       (unsigned long long) cache->bstop,
+	       fnr, bnr);
 
 	/* find out how many pages of blockdev are available */
 	memset(&stats, 0, sizeof(stats));
@@ -337,20 +345,33 @@ int cachefiles_has_space(struct cachefil
 
 	stats.f_bavail >>= cache->bshift;
 
-	_debug("avail %llu", stats.f_bavail);
+	_debug("avail %llu,%llu",
+	       (unsigned long long) stats.f_ffree,
+	       (unsigned long long) stats.f_bavail);
 
 	/* see if there is sufficient space */
-	stats.f_bavail -= nr;
+	if (stats.f_ffree > fnr)
+		stats.f_ffree -= fnr;
+	else
+		stats.f_ffree = 0;
+
+	if (stats.f_bavail > bnr)
+		stats.f_bavail -= bnr;
+	else
+		stats.f_bavail = 0;
 
 	ret = -ENOBUFS;
-	if (stats.f_bavail < cache->bstop)
+	if (stats.f_ffree < cache->fstop ||
+	    stats.f_bavail < cache->bstop)
 		goto begin_cull;
 
 	ret = 0;
-	if (stats.f_bavail < cache->bcull)
+	if (stats.f_ffree < cache->fcull ||
+	    stats.f_bavail < cache->bcull)
 		goto begin_cull;
 
 	if (test_bit(CACHEFILES_CULLING, &cache->flags) &&
+	    stats.f_ffree >= cache->frun &&
 	    stats.f_bavail >= cache->brun
 	    ) {
 		if (test_and_clear_bit(CACHEFILES_CULLING, &cache->flags)) {
@@ -715,7 +736,9 @@ static int cachefiles_read_or_alloc_page
 	block0 <<= shift;
 
 	block = inode->i_mapping->a_ops->bmap(inode->i_mapping, block0);
-	_debug("%llx -> %llx", block0, block);
+	_debug("%llx -> %llx",
+	       (unsigned long long) block0,
+	       (unsigned long long) block);
 
 	if (block) {
 		/* submit the apparently valid page to the backing fs to be
@@ -726,7 +749,7 @@ static int cachefiles_read_or_alloc_page
 						       page,
 						       &pagevec);
 		ret = 0;
-	} else if (cachefiles_has_space(cache, 1) == 0) {
+	} else if (cachefiles_has_space(cache, 0, 1) == 0) {
 		/* there's space in the cache we can use */
 		pagevec_add(&pagevec, page);
 		cookie->def->mark_pages_cached(cookie->netfs_data,
@@ -1004,7 +1027,7 @@ static int cachefiles_read_or_alloc_page
 		return -ENOBUFS;
 
 	space = 1;
-	if (cachefiles_has_space(cache, *nr_pages) < 0)
+	if (cachefiles_has_space(cache, 0, *nr_pages) < 0)
 		space = 0;
 
 	inode = object->backer->d_inode;
@@ -1040,7 +1063,9 @@ static int cachefiles_read_or_alloc_page
 
 		block = inode->i_mapping->a_ops->bmap(inode->i_mapping,
 						      block0);
-		_debug("%llx -> %llx", block0, block);
+		_debug("%llx -> %llx",
+		       (unsigned long long) block0,
+		       (unsigned long long) block);
 
 		if (block) {
 			/* we have data - add it to the list to give to the
@@ -1111,7 +1136,7 @@ static int cachefiles_allocate_page(stru
 
 	_enter("%p,{%lx},,,", object, page->index);
 
-	return cachefiles_has_space(cache, 1);
+	return cachefiles_has_space(cache, 0, 1);
 
 }
 
--- linux-2.6.17.i686/fs/cachefiles/cf-proc.c.inode	2006-08-30 15:00:18.596735000 -0400
+++ linux-2.6.17.i686/fs/cachefiles/cf-proc.c	2006-08-30 15:05:04.201616000 -0400
@@ -28,6 +28,9 @@ static int cachefiles_proc_open(struct i
 static int cachefiles_proc_release(struct inode *, struct file *);
 static ssize_t cachefiles_proc_read(struct file *, char __user *, size_t, loff_t *);
 static ssize_t cachefiles_proc_write(struct file *, const char __user *, size_t, loff_t *);
+static int cachefiles_proc_frun(struct cachefiles_cache *cache, char *args);
+static int cachefiles_proc_fcull(struct cachefiles_cache *cache, char *args);
+static int cachefiles_proc_fstop(struct cachefiles_cache *cache, char *args);
 static int cachefiles_proc_brun(struct cachefiles_cache *cache, char *args);
 static int cachefiles_proc_bcull(struct cachefiles_cache *cache, char *args);
 static int cachefiles_proc_bstop(struct cachefiles_cache *cache, char *args);
@@ -60,6 +63,9 @@ static const struct cachefiles_proc_cmd 
 	{ "cull",	cachefiles_proc_cull	},
 	{ "debug",	cachefiles_proc_debug	},
 	{ "dir",	cachefiles_proc_dir	},
+	{ "frun",	cachefiles_proc_frun	},
+	{ "fcull",	cachefiles_proc_fcull	},
+	{ "fstop",	cachefiles_proc_fstop	},
 	{ "tag",	cachefiles_proc_tag	},
 	{ "",		NULL			}
 };
@@ -94,10 +100,13 @@ static int cachefiles_proc_open(struct i
 	rwlock_init(&cache->active_lock);
 
 	/* set default caching limits
-	 * - limit at 1% free space
-	 * - cull below 5% free space
-	 * - cease culling above 7% free space
+	 * - limit at 1% free space and/or free files
+	 * - cull below 5% free space and/or free files
+	 * - cease culling above 7% free space and/or free files
 	 */
+	cache->frun_percent = 7;
+	cache->fcull_percent = 5;
+	cache->fstop_percent = 1;
 	cache->brun_percent = 7;
 	cache->bcull_percent = 5;
 	cache->bstop_percent = 1;
@@ -153,18 +162,24 @@ static ssize_t cachefiles_proc_read(stru
 		return 0;
 
 	/* check how much space the cache has */
-	cachefiles_has_space(cache, 0);
+	cachefiles_has_space(cache, 0, 0);
 
 	/* summarise */
 	n = snprintf(buffer, sizeof(buffer),
 		     "cull=%c"
+		     " frun=%llx"
+		     " fcull=%llx"
+		     " fstop=%llx"
 		     " brun=%llx"
 		     " bcull=%llx"
 		     " bstop=%llx",
 		     test_bit(CACHEFILES_CULLING, &cache->flags) ? '1' : '0',
-		     cache->brun,
-		     cache->bcull,
-		     cache->bstop
+		     (unsigned long long) cache->frun,
+		     (unsigned long long) cache->fcull,
+		     (unsigned long long) cache->fstop,
+		     (unsigned long long) cache->brun,
+		     (unsigned long long) cache->bcull,
+		     (unsigned long long) cache->bstop
 		     );
 
 	if (n > buflen)
@@ -269,13 +284,88 @@ found_command:
 static int cachefiles_proc_range_error(struct cachefiles_cache *cache, char *args)
 {
 	kerror("Free space limits must be in range"
-	       " 0%%<=bstop<bcull<brun<100%%");
+	       " 0%%<=stop<cull<run<100%%");
 
 	return -EINVAL;
 }
 
 /*****************************************************************************/
 /*
+ * set the percentage of files at which to stop culling
+ * - command: "frun <N>%"
+ */
+static int cachefiles_proc_frun(struct cachefiles_cache *cache, char *args)
+{
+	unsigned long frun;
+
+	_enter(",%s", args);
+
+	if (!*args)
+		return -EINVAL;
+
+	frun = simple_strtoul(args, &args, 10);
+	if (args[0] != '%' || args[1] != '\0')
+		return -EINVAL;
+
+	if (frun <= cache->fcull_percent || frun >= 100)
+		return cachefiles_proc_range_error(cache, args);
+
+	cache->frun_percent = frun;
+	return 0;
+}
+
+/*****************************************************************************/
+/*
+ * set the percentage of files at which to start culling
+ * - command: "fcull <N>%"
+ */
+static int cachefiles_proc_fcull(struct cachefiles_cache *cache, char *args)
+{
+	unsigned long fcull;
+
+	_enter(",%s", args);
+
+	if (!*args)
+		return -EINVAL;
+
+	fcull = simple_strtoul(args, &args, 10);
+	if (args[0] != '%' || args[1] != '\0')
+		return -EINVAL;
+
+	if (fcull <= cache->fstop_percent || fcull >= cache->frun_percent)
+		return cachefiles_proc_range_error(cache, args);
+
+	cache->fcull_percent = fcull;
+	return 0;
+}
+
+/*****************************************************************************/
+/*
+ * set the percentage of files at which to stop allocating
+ * - command: "fstop <N>%"
+ */
+static int cachefiles_proc_fstop(struct cachefiles_cache *cache, char *args)
+{
+	unsigned long fstop;
+
+	_enter(",%s", args);
+
+	if (!*args)
+		return -EINVAL;
+
+	fstop = simple_strtoul(args, &args, 10);
+	if (args[0] != '%' || args[1] != '\0')
+		return -EINVAL;
+
+	if (fstop < 0 || fstop >= cache->fcull_percent)
+		return cachefiles_proc_range_error(cache, args);
+
+	cache->fstop_percent = fstop;
+	return 0;
+}
+
+/*****************************************************************************/
+/*
  * set the percentage of blocks at which to stop culling
  * - command: "brun <N>%"
  */
--- linux-2.6.17.i686/fs/cachefiles/cf-namei.c.inode	2006-08-30 15:04:28.793964000 -0400
+++ linux-2.6.17.i686/fs/cachefiles/cf-namei.c	2006-08-30 15:05:04.193612000 -0400
@@ -373,6 +373,10 @@ lookup_again:
 	if (key || object->type == FSCACHE_COOKIE_TYPE_INDEX) {
 		/* index objects and intervening tree levels must be subdirs */
 		if (!next->d_inode) {
+			ret = cachefiles_has_space(cache, 1, 0);
+			if (ret < 0)
+				goto create_error;
+
 			DQUOT_INIT(dir->d_inode);
 			ret = dir->d_inode->i_op->mkdir(dir->d_inode, next, 0);
 			if (ret < 0)
@@ -395,6 +399,10 @@ lookup_again:
 	} else {
 		/* non-index objects start out life as files */
 		if (!next->d_inode) {
+			ret = cachefiles_has_space(cache, 1, 0);
+			if (ret < 0)
+				goto create_error;
+
 			DQUOT_INIT(dir->d_inode);
 			ret = dir->d_inode->i_op->create(dir->d_inode, next,
 							 S_IFREG, NULL);
@@ -611,6 +619,10 @@ struct dentry *cachefiles_get_directory(
 
 	/* we need to create the subdir if it doesn't exist yet */
 	if (!subdir->d_inode) {
+		ret = cachefiles_has_space(cache, 1, 0);
+		if (ret < 0)
+			goto mkdir_error;
+
 		DQUOT_INIT(dir->d_inode);
 		ret = dir->d_inode->i_op->mkdir(dir->d_inode, subdir, 0700);
 		if (ret < 0)
--- linux-2.6.17.i686/fs/cachefiles/internal.h.inode	2006-08-30 15:04:45.662097000 -0400
+++ linux-2.6.17.i686/fs/cachefiles/internal.h	2006-08-30 15:05:04.207612000 -0400
@@ -80,11 +80,17 @@ struct cachefiles_cache {
 	struct rb_root			active_nodes;	/* active nodes (can't be culled) */
 	rwlock_t			active_lock;	/* lock for active_nodes */
 	atomic_t			gravecounter;	/* graveyard uniquifier */
-	unsigned			brun_percent;	/* when to stop culling (%) */
-	unsigned			bcull_percent;	/* when to start culling (%) */
-	unsigned			bstop_percent;	/* when to stop allocating (%) */
+	unsigned			frun_percent;	/* when to stop culling (% files) */
+	unsigned			fcull_percent;	/* when to start culling (% files) */
+	unsigned			fstop_percent;	/* when to stop allocating (% files) */
+	unsigned			brun_percent;	/* when to stop culling (% blocks) */
+	unsigned			bcull_percent;	/* when to start culling (% blocks) */
+	unsigned			bstop_percent;	/* when to stop allocating (% blocks) */
 	unsigned			bsize;		/* cache's block size */
 	unsigned			bshift;		/* min(log2 (PAGE_SIZE / bsize), 0) */
+	uint64_t			frun;		/* when to stop culling */
+	uint64_t			fcull;		/* when to start culling */
+	uint64_t			fstop;		/* when to stop allocating */
 	sector_t			brun;		/* when to stop culling */
 	sector_t			bcull;		/* when to start culling */
 	sector_t			bstop;		/* when to stop allocating */
@@ -140,7 +146,8 @@ extern void cachefiles_proc_unbind(struc
 /* cf-interface.c */
 extern void cachefiles_read_copier_work(void *_object);
 extern void cachefiles_write_work(void *_object);
-extern int cachefiles_has_space(struct cachefiles_cache *cache, unsigned nr);
+extern int cachefiles_has_space(struct cachefiles_cache *cache,
+				unsigned fnr, unsigned bnr);
 
 /* cf-key.c */
 extern char *cachefiles_cook_key(const u8 *raw, int keylen, uint8_t type);

Places

File linux-2.6-cachefiles.patch of Package kernel

Places