Merge tag 'fscache-rewrite-20220111' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs

Pull fscache rewrite from David Howells:
 "This is a set of patches that rewrites the fscache driver and the
  cachefiles driver, significantly simplifying the code compared to
  what's upstream, removing the complex operation scheduling and object
  state machine in favour of something much smaller and simpler.

  The series is structured such that the first few patches disable
  fscache use by the network filesystems using it, remove the cachefiles
  driver entirely and as much of the fscache driver as can be got away
  with without causing build failures in the network filesystems.

  The patches after that recreate fscache and then cachefiles,
  attempting to add the pieces in a logical order. Finally, the
  filesystems are reenabled and then the very last patch changes the
  documentation.

  [!] Note: I have dropped the cifs patch for the moment, leaving local
      caching in cifs disabled. I've been having trouble getting that
      working. I think I have it done, but it needs more testing (there
      seem to be some test failures occurring with v5.16 also from
      xfstests), so I propose deferring that patch to the end of the
      merge window.

  WHY REWRITE?
  ============

  Fscache's operation scheduling API was intended to handle sequencing
  of cache operations, which were all required (where possible) to run
  asynchronously in parallel with the operations being done by the
  network filesystem, whilst allowing the cache to be brought online and
  offline and to interrupt service for invalidation.

  With the advent of the tmpfile capacity in the VFS, however, an
  opportunity arises to do invalidation much more simply, without having
  to wait for I/O that's actually in progress: Cachefiles can simply
  create a tmpfile, cut over the file pointer for the backing object
  attached to a cookie and abandon the in-progress I/O, dismissing it
  upon completion.

  Future work here would involve using Omar Sandoval's vfs_link() with
  AT_LINK_REPLACE[1] to allow an extant file to be displaced by a new
  hard link from a tmpfile as currently I have to unlink the old file
  first.

  These patches can also simplify the object state handling as I/O
  operations to the cache don't all have to be brought to a stop in
  order to invalidate a file. To that end, and with an eye on to writing
  a new backing cache model in the future, I've taken the opportunity to
  simplify the indexing structure.

  I've separated the index cookie concept from the file cookie concept
  by C type now. The former is now called a "volume cookie" (struct
  fscache_volume) and there is a container of file cookies. There are
  then just the two levels. All the index cookie levels are collapsed
  into a single volume cookie, and this has a single printable string as
  a key. For instance, an AFS volume would have a key of something like
  "afs,example.com,1000555", combining the filesystem name, cell name
  and volume ID. This is freeform, but must not have '/' chars in it.

  I've also eliminated all pointers back from fscache into the network
  filesystem. This required the duplication of a little bit of data in
  the cookie (cookie key, coherency data and file size), but it's not
  actually that much. This gets rid of problems with making sure we keep
  netfs data structures around so that the cache can access them.

  These patches mean that most of the code that was in the drivers
  before is simply gone and those drivers are now almost entirely new
  code. That being the case, there doesn't seem any particular reason to
  try and maintain bisectability across it. Further, there has to be a
  point in the middle where things are cut over as there's a single
  point everything has to go through (ie. /dev/cachefiles) and it can't
  be in use by two drivers at once.

  ISSUES YET OUTSTANDING
  ======================

  There are some issues still outstanding, unaddressed by this patchset,
  that will need fixing in future patchsets, but that don't stop this
  series from being usable:

  (1) The cachefiles driver needs to stop using the backing filesystem's
      metadata to store information about what parts of the cache are
      populated. This is not reliable with modern extent-based
      filesystems.

      Fixing this is deferred to a separate patchset as it involves
      negotiation with the network filesystem and the VM as to how much
      data to download to fulfil a read - which brings me on to (2)...

  (2) NFS (and CIFS with the dropped patch) do not take account of how
      the cache would like I/O to be structured to meet its granularity
      requirements. Previously, the cache used page granularity, which
      was fine as the network filesystems also dealt in page
      granularity, and the backing filesystem (ext4, xfs or whatever)
      did whatever it did out of sight. However, we now have folios to
      deal with and the cache will now have to store its own metadata to
      track its contents.

      The change I'm looking at making for cachefiles is to store
      content bitmaps in one or more xattrs and making a bit in the map
      correspond to something like a 256KiB block. However, the size of
      an xattr and the fact that they have to be read/updated in one go
      means that I'm looking at covering 1GiB of data per 512-byte map
      and storing each map in an xattr. Cachefiles has the potential to
      grow into a fully fledged filesystem of its very own if I'm not
      careful.

      However, I'm also looking at changing things even more radically
      and going to a different model of how the cache is arranged and
      managed - one that's more akin to the way, say, openafs does
      things - which brings me on to (3)...

  (3) The way cachefilesd does culling is very inefficient for large
      caches and it would be better to move it into the kernel if I can
      as cachefilesd has to keep asking the kernel if it can cull a
      file. Changing the way the backend works would allow this to be
      addressed.

  BITS THAT MAY BE CONTROVERSIAL
  ==============================

  There are some bits I've added that may be controversial:

  (1) I've provided a flag, S_KERNEL_FILE, that cachefiles uses to check
      if a files is already being used by some other kernel service
      (e.g. a duplicate cachefiles cache in the same directory) and
      reject it if it is. This isn't entirely necessary, but it helps
      prevent accidental data corruption.

      I don't want to use S_SWAPFILE as that has other effects, but
      quite possibly swapon() should set S_KERNEL_FILE too.

      Note that it doesn't prevent userspace from interfering, though
      perhaps it should. (I have made it prevent a marked directory from
      being rmdir-able).

  (2) Cachefiles wants to keep the backing file for a cookie open whilst
      we might need to write to it from network filesystem writeback.
      The problem is that the network filesystem unuses its cookie when
      its file is closed, and so we have nothing pinning the cachefiles
      file open and it will get closed automatically after a short time
      to avoid EMFILE/ENFILE problems.

      Reopening the cache file, however, is a problem if this is being
      done due to writeback triggered by exit(). Some filesystems will
      oops if we try to open a file in that context because they want to
      access current->fs or suchlike.

      To get around this, I added the following:

      (A) An inode flag, I_PINNING_FSCACHE_WB, to be set on a network
          filesystem inode to indicate that we have a usage count on the
          cookie caching that inode.

      (B) A flag in struct writeback_control, unpinned_fscache_wb, that
          is set when __writeback_single_inode() clears the last dirty
          page from i_pages - at which point it clears
          I_PINNING_FSCACHE_WB and sets this flag.

          This has to be done here so that clearing I_PINNING_FSCACHE_WB
          can be done atomically with the check of PAGECACHE_TAG_DIRTY
          that clears I_DIRTY_PAGES.

      (C) A function, fscache_set_page_dirty(), which if it is not set,
          sets I_PINNING_FSCACHE_WB and calls fscache_use_cookie() to
          pin the cache resources.

      (D) A function, fscache_unpin_writeback(), to be called by
          ->write_inode() to unuse the cookie.

      (E) A function, fscache_clear_inode_writeback(), to be called when
          the inode is evicted, before clear_inode() is called. This
          cleans up any lingering I_PINNING_FSCACHE_WB.

      The network filesystem can then use these tools to make sure that
      fscache_write_to_cache() can write locally modified data to the
      cache as well as to the server.

      For the future, I'm working on write helpers for netfs lib that
      should allow this facility to be removed by keeping track of the
      dirty regions separately - but that's incomplete at the moment and
      is also going to be affected by folios, one way or another, since
      it deals with pages"

Link: https://lore.kernel.org/all/510611.1641942444@warthog.procyon.org.uk/
Tested-by: Dominique Martinet <asmadeus@codewreck.org> # 9p
Tested-by: kafs-testing@auristor.com # afs
Tested-by: Jeff Layton <jlayton@kernel.org> # ceph
Tested-by: Dave Wysochanski <dwysocha@redhat.com> # nfs
Tested-by: Daire Byrne <daire@dneg.com> # nfs

* tag 'fscache-rewrite-20220111' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs: (67 commits)
  9p, afs, ceph, nfs: Use current_is_kswapd() rather than gfpflags_allow_blocking()
  fscache: Add a tracepoint for cookie use/unuse
  fscache: Rewrite documentation
  ceph: add fscache writeback support
  ceph: conversion to new fscache API
  nfs: Implement cache I/O by accessing the cache directly
  nfs: Convert to new fscache volume/cookie API
  9p: Copy local writes to the cache when writing to the server
  9p: Use fscache indexing rewrite and reenable caching
  afs: Skip truncation on the server of data we haven't written yet
  afs: Copy local writes to the cache when writing to the server
  afs: Convert afs to use the new fscache API
  fscache, cachefiles: Display stat of culling events
  fscache, cachefiles: Display stats of no-space events
  cachefiles: Allow cachefiles to actually function
  fscache, cachefiles: Store the volume coherency data
  cachefiles: Implement the I/O routines
  cachefiles: Implement cookie resize for truncate
  cachefiles: Implement begin and end I/O operation
  cachefiles: Implement backing file wrangling
  ...

16 files changed, 2172 insertions, 3050 deletions

diff --git a/fs/cachefiles/Kconfig b/fs/cachefiles/Kconfig
index 6827b40f7ddc..719faeeda168 100644
--- a/fs/cachefiles/Kconfig
+++ b/fs/cachefiles/Kconfig
@@ -19,3 +19,10 @@ config CACHEFILES_DEBUG
 	  caching on files module.  If this is set, the debugging output may be
 	  enabled by setting bits in /sys/modules/cachefiles/parameter/debug or
 	  by including a debugging specifier in /etc/cachefilesd.conf.
+
+config CACHEFILES_ERROR_INJECTION
+	bool "Provide error injection for cachefiles"
+	depends on CACHEFILES && SYSCTL
+	help
+	  This permits error injection to be enabled in cachefiles whilst a
+	  cache is in service.
diff --git a/fs/cachefiles/Makefile b/fs/cachefiles/Makefile
index 02fd17731769..16d811f1a2fa 100644
--- a/fs/cachefiles/Makefile
+++ b/fs/cachefiles/Makefile
@@ -4,15 +4,17 @@
 #
 
 cachefiles-y := \
-	bind.o \
+	cache.o \
 	daemon.o \
 	interface.o \
 	io.o \
 	key.o \
 	main.o \
 	namei.o \
-	rdwr.o \
 	security.o \
+	volume.o \
 	xattr.o
 
+cachefiles-$(CONFIG_CACHEFILES_ERROR_INJECTION) += error_inject.o
+
 obj-$(CONFIG_CACHEFILES) := cachefiles.o
diff --git a/fs/cachefiles/bind.c b/fs/cachefiles/bind.c
deleted file mode 100644
index 146291be6263..000000000000
--- a/fs/cachefiles/bind.c
+++ /dev/null
@@ -1,278 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-or-later
-/* Bind and unbind a cache from the filesystem backing it
- *
- * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
- * Written by David Howells (dhowells@redhat.com)
- */
-
-#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/ctype.h>
-#include <linux/xattr.h>
-#include "internal.h"
-
-static int cachefiles_daemon_add_cache(struct cachefiles_cache *caches);
-
-/*
- * bind a directory as a cache
- */
-int cachefiles_daemon_bind(struct cachefiles_cache *cache, char *args)
-{
-	_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 &&
-	       cache->brun_percent  < 100);
-
-	if (*args) {
-		pr_err("'bind' command doesn't take an argument\n");
-		return -EINVAL;
-	}
-
-	if (!cache->rootdirname) {
-		pr_err("No cache directory specified\n");
-		return -EINVAL;
-	}
-
-	/* don't permit already bound caches to be re-bound */
-	if (test_bit(CACHEFILES_READY, &cache->flags)) {
-		pr_err("Cache already bound\n");
-		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_daemon_add_cache(cache);
-}
-
-/*
- * add a cache
- */
-static int cachefiles_daemon_add_cache(struct cachefiles_cache *cache)
-{
-	struct cachefiles_object *fsdef;
-	struct path path;
-	struct kstatfs stats;
-	struct dentry *graveyard, *cachedir, *root;
-	const struct cred *saved_cred;
-	int ret;
-
-	_enter("");
-
-	/* we want to work under the module's security ID */
-	ret = cachefiles_get_security_ID(cache);
-	if (ret < 0)
-		return ret;
-
-	cachefiles_begin_secure(cache, &saved_cred);
-
-	/* allocate the root index object */
-	ret = -ENOMEM;
-
-	fsdef = kmem_cache_alloc(cachefiles_object_jar, GFP_KERNEL);
-	if (!fsdef)
-		goto error_root_object;
-
-	ASSERTCMP(fsdef->backer, ==, NULL);
-
-	atomic_set(&fsdef->usage, 1);
-	fsdef->type = FSCACHE_COOKIE_TYPE_INDEX;
-
-	/* look up the directory at the root of the cache */
-	ret = kern_path(cache->rootdirname, LOOKUP_DIRECTORY, &path);
-	if (ret < 0)
-		goto error_open_root;
-
-	cache->mnt = path.mnt;
-	root = path.dentry;
-
-	ret = -EINVAL;
-	if (is_idmapped_mnt(path.mnt)) {
-		pr_warn("File cache on idmapped mounts not supported");
-		goto error_unsupported;
-	}
-
-	/* check parameters */
-	ret = -EOPNOTSUPP;
-	if (d_is_negative(root) ||
-	    !d_backing_inode(root)->i_op->lookup ||
-	    !d_backing_inode(root)->i_op->mkdir ||
-	    !(d_backing_inode(root)->i_opflags & IOP_XATTR) ||
-	    !root->d_sb->s_op->statfs ||
-	    !root->d_sb->s_op->sync_fs)
-		goto error_unsupported;
-
-	ret = -EROFS;
-	if (sb_rdonly(root->d_sb))
-		goto error_unsupported;
-
-	/* determine the security of the on-disk cache as this governs
-	 * security ID of files we create */
-	ret = cachefiles_determine_cache_security(cache, root, &saved_cred);
-	if (ret < 0)
-		goto error_unsupported;
-
-	/* get the cache size and blocksize */
-	ret = vfs_statfs(&path, &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 - ilog2(stats.f_bsize);
-
-	_debug("blksize %u (shift %u)",
-	       cache->bsize, cache->bshift);
-
-	_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} 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");
-	if (IS_ERR(cachedir)) {
-		ret = PTR_ERR(cachedir);
-		goto error_unsupported;
-	}
-
-	fsdef->dentry = cachedir;
-	fsdef->fscache.cookie = NULL;
-
-	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,
-			   "%s",
-			   fsdef->dentry->d_sb->s_id);
-
-	fscache_object_init(&fsdef->fscache, &fscache_fsdef_index,
-			    &cache->cache);
-
-	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);
-
-	pr_info("File cache on %s registered\n", cache->cache.identifier);
-
-	/* check how much space the cache has */
-	cachefiles_has_space(cache, 0, 0);
-	cachefiles_end_secure(cache, saved_cred);
-	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:
-	cachefiles_end_secure(cache, saved_cred);
-	pr_err("Failed to register: %d\n", ret);
-	return ret;
-}
-
-/*
- * unbind a cache on fd release
- */
-void cachefiles_daemon_unbind(struct cachefiles_cache *cache)
-{
-	_enter("");
-
-	if (test_bit(CACHEFILES_READY, &cache->flags)) {
-		pr_info("File cache on %s unregistering\n",
-			cache->cache.identifier);
-
-		fscache_withdraw_cache(&cache->cache);
-	}
-
-	dput(cache->graveyard);
-	mntput(cache->mnt);
-
-	kfree(cache->rootdirname);
-	kfree(cache->secctx);
-	kfree(cache->tag);
-
-	_leave("");
-}
diff --git a/fs/cachefiles/cache.c b/fs/cachefiles/cache.c
new file mode 100644
index 000000000000..ce4d4785003c
--- /dev/null
+++ b/fs/cachefiles/cache.c
@@ -0,0 +1,378 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/* Manage high-level VFS aspects of a cache.
+ *
+ * Copyright (C) 2007, 2021 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ */
+
+#include <linux/slab.h>
+#include <linux/statfs.h>
+#include <linux/namei.h>
+#include "internal.h"
+
+/*
+ * Bring a cache online.
+ */
+int cachefiles_add_cache(struct cachefiles_cache *cache)
+{
+	struct fscache_cache *cache_cookie;
+	struct path path;
+	struct kstatfs stats;
+	struct dentry *graveyard, *cachedir, *root;
+	const struct cred *saved_cred;
+	int ret;
+
+	_enter("");
+
+	cache_cookie = fscache_acquire_cache(cache->tag);
+	if (IS_ERR(cache_cookie))
+		return PTR_ERR(cache_cookie);
+
+	/* we want to work under the module's security ID */
+	ret = cachefiles_get_security_ID(cache);
+	if (ret < 0)
+		goto error_getsec;
+
+	cachefiles_begin_secure(cache, &saved_cred);
+
+	/* look up the directory at the root of the cache */
+	ret = kern_path(cache->rootdirname, LOOKUP_DIRECTORY, &path);
+	if (ret < 0)
+		goto error_open_root;
+
+	cache->mnt = path.mnt;
+	root = path.dentry;
+
+	ret = -EINVAL;
+	if (is_idmapped_mnt(path.mnt)) {
+		pr_warn("File cache on idmapped mounts not supported");
+		goto error_unsupported;
+	}
+
+	/* check parameters */
+	ret = -EOPNOTSUPP;
+	if (d_is_negative(root) ||
+	    !d_backing_inode(root)->i_op->lookup ||
+	    !d_backing_inode(root)->i_op->mkdir ||
+	    !(d_backing_inode(root)->i_opflags & IOP_XATTR) ||
+	    !root->d_sb->s_op->statfs ||
+	    !root->d_sb->s_op->sync_fs ||
+	    root->d_sb->s_blocksize > PAGE_SIZE)
+		goto error_unsupported;
+
+	ret = -EROFS;
+	if (sb_rdonly(root->d_sb))
+		goto error_unsupported;
+
+	/* determine the security of the on-disk cache as this governs
+	 * security ID of files we create */
+	ret = cachefiles_determine_cache_security(cache, root, &saved_cred);
+	if (ret < 0)
+		goto error_unsupported;
+
+	/* get the cache size and blocksize */
+	ret = vfs_statfs(&path, &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 - ilog2(stats.f_bsize);
+
+	_debug("blksize %u (shift %u)",
+	       cache->bsize, cache->bshift);
+
+	_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} 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", NULL);
+	if (IS_ERR(cachedir)) {
+		ret = PTR_ERR(cachedir);
+		goto error_unsupported;
+	}
+
+	cache->store = cachedir;
+
+	/* get the graveyard directory */
+	graveyard = cachefiles_get_directory(cache, root, "graveyard", NULL);
+	if (IS_ERR(graveyard)) {
+		ret = PTR_ERR(graveyard);
+		goto error_unsupported;
+	}
+
+	cache->graveyard = graveyard;
+	cache->cache = cache_cookie;
+
+	ret = fscache_add_cache(cache_cookie, &cachefiles_cache_ops, cache);
+	if (ret < 0)
+		goto error_add_cache;
+
+	/* done */
+	set_bit(CACHEFILES_READY, &cache->flags);
+	dput(root);
+
+	pr_info("File cache on %s registered\n", cache_cookie->name);
+
+	/* check how much space the cache has */
+	cachefiles_has_space(cache, 0, 0, cachefiles_has_space_check);
+	cachefiles_end_secure(cache, saved_cred);
+	_leave(" = 0 [%px]", cache->cache);
+	return 0;
+
+error_add_cache:
+	cachefiles_put_directory(cache->graveyard);
+	cache->graveyard = NULL;
+error_unsupported:
+	cachefiles_put_directory(cache->store);
+	cache->store = NULL;
+	mntput(cache->mnt);
+	cache->mnt = NULL;
+	dput(root);
+error_open_root:
+	cachefiles_end_secure(cache, saved_cred);
+error_getsec:
+	fscache_relinquish_cache(cache_cookie);
+	cache->cache = NULL;
+	pr_err("Failed to register: %d\n", ret);
+	return ret;
+}
+
+/*
+ * 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 fnr, unsigned bnr,
+			 enum cachefiles_has_space_for reason)
+{
+	struct kstatfs stats;
+	u64 b_avail, b_writing;
+	int ret;
+
+	struct path path = {
+		.mnt	= cache->mnt,
+		.dentry	= cache->mnt->mnt_root,
+	};
+
+	//_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));
+
+	ret = vfs_statfs(&path, &stats);
+	if (ret < 0) {
+		trace_cachefiles_vfs_error(NULL, d_inode(path.dentry), ret,
+					   cachefiles_trace_statfs_error);
+		if (ret == -EIO)
+			cachefiles_io_error(cache, "statfs failed");
+		_leave(" = %d", ret);
+		return ret;
+	}
+
+	b_avail = stats.f_bavail >> cache->bshift;
+	b_writing = atomic_long_read(&cache->b_writing);
+	if (b_avail > b_writing)
+		b_avail -= b_writing;
+	else
+		b_avail = 0;
+
+	//_debug("avail %llu,%llu",
+	//       (unsigned long long)stats.f_ffree,
+	//       (unsigned long long)b_avail);
+
+	/* see if there is sufficient space */
+	if (stats.f_ffree > fnr)
+		stats.f_ffree -= fnr;
+	else
+		stats.f_ffree = 0;
+
+	if (b_avail > bnr)
+		b_avail -= bnr;
+	else
+		b_avail = 0;
+
+	ret = -ENOBUFS;
+	if (stats.f_ffree < cache->fstop ||
+	    b_avail < cache->bstop)
+		goto stop_and_begin_cull;
+
+	ret = 0;
+	if (stats.f_ffree < cache->fcull ||
+	    b_avail < cache->bcull)
+		goto begin_cull;
+
+	if (test_bit(CACHEFILES_CULLING, &cache->flags) &&
+	    stats.f_ffree >= cache->frun &&
+	    b_avail >= cache->brun &&
+	    test_and_clear_bit(CACHEFILES_CULLING, &cache->flags)
+	    ) {
+		_debug("cease culling");
+		cachefiles_state_changed(cache);
+	}
+
+	//_leave(" = 0");
+	return 0;
+
+stop_and_begin_cull:
+	switch (reason) {
+	case cachefiles_has_space_for_write:
+		fscache_count_no_write_space();
+		break;
+	case cachefiles_has_space_for_create:
+		fscache_count_no_create_space();
+		break;
+	default:
+		break;
+	}
+begin_cull:
+	if (!test_and_set_bit(CACHEFILES_CULLING, &cache->flags)) {
+		_debug("### CULL CACHE ###");
+		cachefiles_state_changed(cache);
+	}
+
+	_leave(" = %d", ret);
+	return ret;
+}
+
+/*
+ * Mark all the objects as being out of service and queue them all for cleanup.
+ */
+static void cachefiles_withdraw_objects(struct cachefiles_cache *cache)
+{
+	struct cachefiles_object *object;
+	unsigned int count = 0;
+
+	_enter("");
+
+	spin_lock(&cache->object_list_lock);
+
+	while (!list_empty(&cache->object_list)) {
+		object = list_first_entry(&cache->object_list,
+					  struct cachefiles_object, cache_link);
+		cachefiles_see_object(object, cachefiles_obj_see_withdrawal);
+		list_del_init(&object->cache_link);
+		fscache_withdraw_cookie(object->cookie);
+		count++;
+		if ((count & 63) == 0) {
+			spin_unlock(&cache->object_list_lock);
+			cond_resched();
+			spin_lock(&cache->object_list_lock);
+		}
+	}
+
+	spin_unlock(&cache->object_list_lock);
+	_leave(" [%u objs]", count);
+}
+
+/*
+ * Withdraw volumes.
+ */
+static void cachefiles_withdraw_volumes(struct cachefiles_cache *cache)
+{
+	_enter("");
+
+	for (;;) {
+		struct cachefiles_volume *volume = NULL;
+
+		spin_lock(&cache->object_list_lock);
+		if (!list_empty(&cache->volumes)) {
+			volume = list_first_entry(&cache->volumes,
+						  struct cachefiles_volume, cache_link);
+			list_del_init(&volume->cache_link);
+		}
+		spin_unlock(&cache->object_list_lock);
+		if (!volume)
+			break;
+
+		cachefiles_withdraw_volume(volume);
+	}
+
+	_leave("");
+}
+
+/*
+ * Sync a cache to backing disk.
+ */
+static void cachefiles_sync_cache(struct cachefiles_cache *cache)
+{
+	const struct cred *saved_cred;
+	int ret;
+
+	_enter("%s", cache->cache->name);
+
+	/* make sure all pages pinned by operations on behalf of the netfs are
+	 * written to disc */
+	cachefiles_begin_secure(cache, &saved_cred);
+	down_read(&cache->mnt->mnt_sb->s_umount);
+	ret = sync_filesystem(cache->mnt->mnt_sb);
+	up_read(&cache->mnt->mnt_sb->s_umount);
+	cachefiles_end_secure(cache, saved_cred);
+
+	if (ret == -EIO)
+		cachefiles_io_error(cache,
+				    "Attempt to sync backing fs superblock returned error %d",
+				    ret);
+}
+
+/*
+ * Withdraw cache objects.
+ */
+void cachefiles_withdraw_cache(struct cachefiles_cache *cache)
+{
+	struct fscache_cache *fscache = cache->cache;
+
+	pr_info("File cache on %s unregistering\n", fscache->name);
+
+	fscache_withdraw_cache(fscache);
+
+	/* we now have to destroy all the active objects pertaining to this
+	 * cache - which we do by passing them off to thread pool to be
+	 * disposed of */
+	cachefiles_withdraw_objects(cache);
+	fscache_wait_for_objects(fscache);
+
+	cachefiles_withdraw_volumes(cache);
+	cachefiles_sync_cache(cache);
+	cache->cache = NULL;
+	fscache_relinquish_cache(fscache);
+}
diff --git a/fs/cachefiles/daemon.c b/fs/cachefiles/daemon.c
index 752c1e43416f..40a792421fc1 100644
--- a/fs/cachefiles/daemon.c
+++ b/fs/cachefiles/daemon.c
@@ -1,7 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 /* Daemon interface
  *
- * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
+ * Copyright (C) 2007, 2021 Red Hat, Inc. All Rights Reserved.
  * Written by David Howells (dhowells@redhat.com)
  */
 
@@ -41,6 +41,8 @@ static int cachefiles_daemon_dir(struct cachefiles_cache *, char *);
 static int cachefiles_daemon_inuse(struct cachefiles_cache *, char *);
 static int cachefiles_daemon_secctx(struct cachefiles_cache *, char *);
 static int cachefiles_daemon_tag(struct cachefiles_cache *, char *);
+static int cachefiles_daemon_bind(struct cachefiles_cache *, char *);
+static void cachefiles_daemon_unbind(struct cachefiles_cache *);
 
 static unsigned long cachefiles_open;
 
@@ -78,7 +80,7 @@ static const struct cachefiles_daemon_cmd cachefiles_daemon_cmds[] = {
 
 
 /*
- * do various checks
+ * Prepare a cache for caching.
  */
 static int cachefiles_daemon_open(struct inode *inode, struct file *file)
 {
@@ -102,9 +104,10 @@ static int cachefiles_daemon_open(struct inode *inode, struct file *file)
 	}
 
 	mutex_init(&cache->daemon_mutex);
-	cache->active_nodes = RB_ROOT;
-	rwlock_init(&cache->active_lock);
 	init_waitqueue_head(&cache->daemon_pollwq);
+	INIT_LIST_HEAD(&cache->volumes);
+	INIT_LIST_HEAD(&cache->object_list);
+	spin_lock_init(&cache->object_list_lock);
 
 	/* set default caching limits
 	 * - limit at 1% free space and/or free files
@@ -124,7 +127,7 @@ static int cachefiles_daemon_open(struct inode *inode, struct file *file)
 }
 
 /*
- * release a cache
+ * Release a cache.
  */
 static int cachefiles_daemon_release(struct inode *inode, struct file *file)
 {
@@ -138,8 +141,6 @@ static int cachefiles_daemon_release(struct inode *inode, struct file *file)
 
 	cachefiles_daemon_unbind(cache);
 
-	ASSERT(!cache->active_nodes.rb_node);
-
 	/* clean up the control file interface */
 	cache->cachefilesd = NULL;
 	file->private_data = NULL;
@@ -152,7 +153,7 @@ static int cachefiles_daemon_release(struct inode *inode, struct file *file)
 }
 
 /*
- * read the cache state
+ * Read the cache state.
  */
 static ssize_t cachefiles_daemon_read(struct file *file, char __user *_buffer,
 				      size_t buflen, loff_t *pos)
@@ -169,7 +170,7 @@ static ssize_t cachefiles_daemon_read(struct file *file, char __user *_buffer,
 		return 0;
 
 	/* check how much space the cache has */
-	cachefiles_has_space(cache, 0, 0);
+	cachefiles_has_space(cache, 0, 0, cachefiles_has_space_check);
 
 	/* summarise */
 	f_released = atomic_xchg(&cache->f_released, 0);
@@ -206,7 +207,7 @@ static ssize_t cachefiles_daemon_read(struct file *file, char __user *_buffer,
 }
 
 /*
- * command the cache
+ * Take a command from cachefilesd, parse it and act on it.
  */
 static ssize_t cachefiles_daemon_write(struct file *file,
 				       const char __user *_data,
@@ -225,7 +226,7 @@ static ssize_t cachefiles_daemon_write(struct file *file,
 	if (test_bit(CACHEFILES_DEAD, &cache->flags))
 		return -EIO;
 
-	if (datalen < 0 || datalen > PAGE_SIZE - 1)
+	if (datalen > PAGE_SIZE - 1)
 		return -EOPNOTSUPP;
 
 	/* drag the command string into the kernel so we can parse it */
@@ -284,7 +285,7 @@ found_command:
 }
 
 /*
- * poll for culling state
+ * Poll for culling state
  * - use EPOLLOUT to indicate culling state
  */
 static __poll_t cachefiles_daemon_poll(struct file *file,
@@ -306,7 +307,7 @@ static __poll_t cachefiles_daemon_poll(struct file *file,
 }
 
 /*
- * give a range error for cache space constraints
+ * Give a range error for cache space constraints
  * - can be tail-called
  */
 static int cachefiles_daemon_range_error(struct cachefiles_cache *cache,
@@ -318,7 +319,7 @@ static int cachefiles_daemon_range_error(struct cachefiles_cache *cache,
 }
 
 /*
- * set the percentage of files at which to stop culling
+ * Set the percentage of files at which to stop culling
  * - command: "frun <N>%"
  */
 static int cachefiles_daemon_frun(struct cachefiles_cache *cache, char *args)
@@ -342,7 +343,7 @@ static int cachefiles_daemon_frun(struct cachefiles_cache *cache, char *args)
 }
 
 /*
- * set the percentage of files at which to start culling
+ * Set the percentage of files at which to start culling
  * - command: "fcull <N>%"
  */
 static int cachefiles_daemon_fcull(struct cachefiles_cache *cache, char *args)
@@ -366,7 +367,7 @@ static int cachefiles_daemon_fcull(struct cachefiles_cache *cache, char *args)
 }
 
 /*
- * set the percentage of files at which to stop allocating
+ * Set the percentage of files at which to stop allocating
  * - command: "fstop <N>%"
  */
 static int cachefiles_daemon_fstop(struct cachefiles_cache *cache, char *args)
@@ -382,7 +383,7 @@ static int cachefiles_daemon_fstop(struct cachefiles_cache *cache, char *args)
 	if (args[0] != '%' || args[1] != '\0')
 		return -EINVAL;
 
-	if (fstop < 0 || fstop >= cache->fcull_percent)
+	if (fstop >= cache->fcull_percent)
 		return cachefiles_daemon_range_error(cache, args);
 
 	cache->fstop_percent = fstop;
@@ -390,7 +391,7 @@ static int cachefiles_daemon_fstop(struct cachefiles_cache *cache, char *args)
 }
 
 /*
- * set the percentage of blocks at which to stop culling
+ * Set the percentage of blocks at which to stop culling
  * - command: "brun <N>%"
  */
 static int cachefiles_daemon_brun(struct cachefiles_cache *cache, char *args)
@@ -414,7 +415,7 @@ static int cachefiles_daemon_brun(struct cachefiles_cache *cache, char *args)
 }
 
 /*
- * set the percentage of blocks at which to start culling
+ * Set the percentage of blocks at which to start culling
  * - command: "bcull <N>%"
  */
 static int cachefiles_daemon_bcull(struct cachefiles_cache *cache, char *args)
@@ -438,7 +439,7 @@ static int cachefiles_daemon_bcull(struct cachefiles_cache *cache, char *args)
 }
 
 /*
- * set the percentage of blocks at which to stop allocating
+ * Set the percentage of blocks at which to stop allocating
  * - command: "bstop <N>%"
  */
 static int cachefiles_daemon_bstop(struct cachefiles_cache *cache, char *args)
@@ -454,7 +455,7 @@ static int cachefiles_daemon_bstop(struct cachefiles_cache *cache, char *args)
 	if (args[0] != '%' || args[1] != '\0')
 		return -EINVAL;
 
-	if (bstop < 0 || bstop >= cache->bcull_percent)
+	if (bstop >= cache->bcull_percent)
 		return cachefiles_daemon_range_error(cache, args);
 
 	cache->bstop_percent = bstop;
@@ -462,7 +463,7 @@ static int cachefiles_daemon_bstop(struct cachefiles_cache *cache, char *args)
 }
 
 /*
- * set the cache directory
+ * Set the cache directory
  * - command: "dir <name>"
  */
 static int cachefiles_daemon_dir(struct cachefiles_cache *cache, char *args)
@@ -490,7 +491,7 @@ static int cachefiles_daemon_dir(struct cachefiles_cache *cache, char *args)
 }
 
 /*
- * set the cache security context
+ * Set the cache security context
  * - command: "secctx <ctx>"
  */
 static int cachefiles_daemon_secctx(struct cachefiles_cache *cache, char *args)
@@ -518,7 +519,7 @@ static int cachefiles_daemon_secctx(struct cachefiles_cache *cache, char *args)
 }
 
 /*
- * set the cache tag
+ * Set the cache tag
  * - command: "tag <name>"
  */
 static int cachefiles_daemon_tag(struct cachefiles_cache *cache, char *args)
@@ -544,7 +545,7 @@ static int cachefiles_daemon_tag(struct cachefiles_cache *cache, char *args)
 }
 
 /*
- * request a node in the cache be culled from the current working directory
+ * Request a node in the cache be culled from the current working directory
  * - command: "cull <name>"
  */
 static int cachefiles_daemon_cull(struct cachefiles_cache *cache, char *args)
@@ -568,7 +569,6 @@ static int cachefiles_daemon_cull(struct cachefiles_cache *cache, char *args)
 		return -EIO;
 	}
 
-	/* extract the directory dentry from the cwd */
 	get_fs_pwd(current->fs, &path);
 
 	if (!d_can_lookup(path.dentry))
@@ -593,7 +593,7 @@ inval:
 }
 
 /*
- * set debugging mode
+ * Set debugging mode
  * - command: "debug <mask>"
  */
 static int cachefiles_daemon_debug(struct cachefiles_cache *cache, char *args)
@@ -616,7 +616,7 @@ inval:
 }
 
 /*
- * find out whether an object in the current working directory is in use or not
+ * Find out whether an object in the current working directory is in use or not
  * - command: "inuse <name>"
  */
 static int cachefiles_daemon_inuse(struct cachefiles_cache *cache, char *args)
@@ -640,7 +640,6 @@ static int cachefiles_daemon_inuse(struct cachefiles_cache *cache, char *args)
 		return -EIO;
 	}
 
-	/* extract the directory dentry from the cwd */
 	get_fs_pwd(current->fs, &path);
 
 	if (!d_can_lookup(path.dentry))
@@ -665,84 +664,65 @@ inval:
 }
 
 /*
- * see if we have space for a number of pages and/or a number of files in the
- * cache
+ * Bind a directory as a cache
  */
-int cachefiles_has_space(struct cachefiles_cache *cache,
-			 unsigned fnr, unsigned bnr)
+static int cachefiles_daemon_bind(struct cachefiles_cache *cache, char *args)
 {
-	struct kstatfs stats;
-	struct path path = {
-		.mnt	= cache->mnt,
-		.dentry	= cache->mnt->mnt_root,
-	};
-	int ret;
+	_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);
+
+	if (cache->fstop_percent >= cache->fcull_percent ||
+	    cache->fcull_percent >= cache->frun_percent ||
+	    cache->frun_percent  >= 100)
+		return -ERANGE;
+
+	if (cache->bstop_percent >= cache->bcull_percent ||
+	    cache->bcull_percent >= cache->brun_percent ||
+	    cache->brun_percent  >= 100)
+		return -ERANGE;
 
-	//_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));
-
-	ret = vfs_statfs(&path, &stats);
-	if (ret < 0) {
-		if (ret == -EIO)
-			cachefiles_io_error(cache, "statfs failed");
-		_leave(" = %d", ret);
-		return ret;
+	if (*args) {
+		pr_err("'bind' command doesn't take an argument\n");
+		return -EINVAL;
 	}
 
-	stats.f_bavail >>= cache->bshift;
-
-	//_debug("avail %llu,%llu",
-	//       (unsigned long long) stats.f_ffree,
-	//       (unsigned long long) stats.f_bavail);
-
-	/* see if there is sufficient space */
-	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_ffree < cache->fstop ||
-	    stats.f_bavail < cache->bstop)
-		goto begin_cull;
-
-	ret = 0;
-	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 &&
-	    test_and_clear_bit(CACHEFILES_CULLING, &cache->flags)
-	    ) {
-		_debug("cease culling");
-		cachefiles_state_changed(cache);
+	if (!cache->rootdirname) {
+		pr_err("No cache directory specified\n");
+		return -EINVAL;
 	}
 
-	//_leave(" = 0");
-	return 0;
-
-begin_cull:
-	if (!test_and_set_bit(CACHEFILES_CULLING, &cache->flags)) {
-		_debug("### CULL CACHE ###");
-		cachefiles_state_changed(cache);
+	/* Don't permit already bound caches to be re-bound */
+	if (test_bit(CACHEFILES_READY, &cache->flags)) {
+		pr_err("Cache already bound\n");
+		return -EBUSY;
 	}
 
-	_leave(" = %d", ret);
-	return ret;
+	return cachefiles_add_cache(cache);
+}
+
+/*
+ * Unbind a cache.
+ */
+static void cachefiles_daemon_unbind(struct cachefiles_cache *cache)
+{
+	_enter("");
+
+	if (test_bit(CACHEFILES_READY, &cache->flags))
+		cachefiles_withdraw_cache(cache);
+
+	cachefiles_put_directory(cache->graveyard);
+	cachefiles_put_directory(cache->store);
+	mntput(cache->mnt);
+
+	kfree(cache->rootdirname);
+	kfree(cache->secctx);
+	kfree(cache->tag);
+
+	_leave("");
 }
diff --git a/fs/cachefiles/error_inject.c b/fs/cachefiles/error_inject.c
new file mode 100644
index 000000000000..58f8aec964e4
--- /dev/null
+++ b/fs/cachefiles/error_inject.c
@@ -0,0 +1,46 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/* Error injection handling.
+ *
+ * Copyright (C) 2021 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ */
+
+#include <linux/sysctl.h>
+#include "internal.h"
+
+unsigned int cachefiles_error_injection_state;
+
+static struct ctl_table_header *cachefiles_sysctl;
+static struct ctl_table cachefiles_sysctls[] = {
+	{
+		.procname	= "error_injection",
+		.data		= &cachefiles_error_injection_state,
+		.maxlen		= sizeof(unsigned int),
+		.mode		= 0644,
+		.proc_handler	= proc_douintvec,
+	},
+	{}
+};
+
+static struct ctl_table cachefiles_sysctls_root[] = {
+	{
+		.procname	= "cachefiles",
+		.mode		= 0555,
+		.child		= cachefiles_sysctls,
+	},
+	{}
+};
+
+int __init cachefiles_register_error_injection(void)
+{
+	cachefiles_sysctl = register_sysctl_table(cachefiles_sysctls_root);
+	if (!cachefiles_sysctl)
+		return -ENOMEM;
+	return 0;
+
+}
+
+void cachefiles_unregister_error_injection(void)
+{
+	unregister_sysctl_table(cachefiles_sysctl);
+}
diff --git a/fs/cachefiles/interface.c b/fs/cachefiles/interface.c
index da28ac1fa225..51c968cd00a6 100644
--- a/fs/cachefiles/interface.c
+++ b/fs/cachefiles/interface.c
@@ -1,572 +1,445 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 /* FS-Cache interface to CacheFiles
  *
- * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
+ * Copyright (C) 2021 Red Hat, Inc. All Rights Reserved.
  * Written by David Howells (dhowells@redhat.com)
  */
 
 #include <linux/slab.h>
 #include <linux/mount.h>
+#include <linux/xattr.h>
+#include <linux/file.h>
+#include <linux/falloc.h>
+#include <trace/events/fscache.h>
 #include "internal.h"
 
-struct cachefiles_lookup_data {
-	struct cachefiles_xattr	*auxdata;	/* auxiliary data */
-	char			*key;		/* key path */
-};
-
-static int cachefiles_attr_changed(struct fscache_object *_object);
+static atomic_t cachefiles_object_debug_id;
 
 /*
- * allocate an object record for a cookie lookup and prepare the lookup data
+ * Allocate a cache object record.
  */
-static struct fscache_object *cachefiles_alloc_object(
-	struct fscache_cache *_cache,
-	struct fscache_cookie *cookie)
+static
+struct cachefiles_object *cachefiles_alloc_object(struct fscache_cookie *cookie)
 {
-	struct cachefiles_lookup_data *lookup_data;
+	struct fscache_volume *vcookie = cookie->volume;
+	struct cachefiles_volume *volume = vcookie->cache_priv;
 	struct cachefiles_object *object;
-	struct cachefiles_cache *cache;
-	struct cachefiles_xattr *auxdata;
-	unsigned keylen, auxlen;
-	void *buffer, *p;
-	char *key;
 
-	cache = container_of(_cache, struct cachefiles_cache, cache);
+	_enter("{%s},%x,", vcookie->key, cookie->debug_id);
 
-	_enter("{%s},%x,", cache->cache.identifier, cookie->debug_id);
-
-	lookup_data = kmalloc(sizeof(*lookup_data), cachefiles_gfp);
-	if (!lookup_data)
-		goto nomem_lookup_data;
-
-	/* create a new object record and a temporary leaf image */
-	object = kmem_cache_alloc(cachefiles_object_jar, cachefiles_gfp);
+	object = kmem_cache_zalloc(cachefiles_object_jar, GFP_KERNEL);
 	if (!object)
-		goto nomem_object;
-
-	ASSERTCMP(object->backer, ==, NULL);
+		return NULL;
 
-	BUG_ON(test_bit(CACHEFILES_OBJECT_ACTIVE, &object->flags));
-	atomic_set(&object->usage, 1);
+	refcount_set(&object->ref, 1);
 
-	fscache_object_init(&object->fscache, cookie, &cache->cache);
+	spin_lock_init(&object->lock);
+	INIT_LIST_HEAD(&object->cache_link);
+	object->volume = volume;
+	object->debug_id = atomic_inc_return(&cachefiles_object_debug_id);
+	object->cookie = fscache_get_cookie(cookie, fscache_cookie_get_attach_object);
 
-	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, cachefiles_gfp);
-	if (!buffer)
-		goto nomem_buffer;
-
-	keylen = cookie->key_len;
-	if (keylen <= sizeof(cookie->inline_key))
-		p = cookie->inline_key;
-	else
-		p = cookie->key;
-	memcpy(buffer + 2, p, keylen);
-
-	*(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 = cookie->aux_len;
-	if (auxlen) {
-		if (auxlen <= sizeof(cookie->inline_aux))
-			p = cookie->inline_aux;
-		else
-			p = cookie->aux;
-		memcpy(auxdata->data, p, auxlen);
-	}
-
-	auxdata->len = auxlen + 1;
-	auxdata->type = cookie->type;
-
-	lookup_data->auxdata = auxdata;
-	lookup_data->key = key;
-	object->lookup_data = lookup_data;
-
-	_leave(" = %x [%p]", object->fscache.debug_id, lookup_data);
-	return &object->fscache;
-
-nomem_key:
-	kfree(buffer);
-nomem_buffer:
-	BUG_ON(test_bit(CACHEFILES_OBJECT_ACTIVE, &object->flags));
-	kmem_cache_free(cachefiles_object_jar, object);
-	fscache_object_destroyed(&cache->cache);
-nomem_object:
-	kfree(lookup_data);
-nomem_lookup_data:
-	_leave(" = -ENOMEM");
-	return ERR_PTR(-ENOMEM);
+	fscache_count_object(vcookie->cache);
+	trace_cachefiles_ref(object->debug_id, cookie->debug_id, 1,
+			     cachefiles_obj_new);
+	return object;
 }
 
 /*
- * attempt to look up the nominated node in this cache
- * - return -ETIMEDOUT to be scheduled again
+ * Note that an object has been seen.
  */
-static int cachefiles_lookup_object(struct fscache_object *_object)
+void cachefiles_see_object(struct cachefiles_object *object,
+			   enum cachefiles_obj_ref_trace why)
 {
-	struct cachefiles_lookup_data *lookup_data;
-	struct cachefiles_object *parent, *object;
-	struct cachefiles_cache *cache;
-	const struct cred *saved_cred;
-	int ret;
-
-	_enter("{OBJ%x}", _object->debug_id);
-
-	cache = container_of(_object->cache, struct cachefiles_cache, cache);
-	parent = container_of(_object->parent,
-			      struct cachefiles_object, fscache);
-	object = container_of(_object, struct cachefiles_object, fscache);
-	lookup_data = object->lookup_data;
-
-	ASSERTCMP(lookup_data, !=, NULL);
-
-	/* look up the key, creating any missing bits */
-	cachefiles_begin_secure(cache, &saved_cred);
-	ret = cachefiles_walk_to_object(parent, object,
-					lookup_data->key,
-					lookup_data->auxdata);
-	cachefiles_end_secure(cache, saved_cred);
-
-	/* polish off by setting the attributes of non-index files */
-	if (ret == 0 &&
-	    object->fscache.cookie->def->type != FSCACHE_COOKIE_TYPE_INDEX)
-		cachefiles_attr_changed(&object->fscache);
-
-	if (ret < 0 && ret != -ETIMEDOUT) {
-		if (ret != -ENOBUFS)
-			pr_warn("Lookup failed error %d\n", ret);
-		fscache_object_lookup_error(&object->fscache);
-	}
-
-	_leave(" [%d]", ret);
-	return ret;
+	trace_cachefiles_ref(object->debug_id, object->cookie->debug_id,
+			     refcount_read(&object->ref), why);
 }
 
 /*
- * indication of lookup completion
+ * Increment the usage count on an object;
  */
-static void cachefiles_lookup_complete(struct fscache_object *_object)
+struct cachefiles_object *cachefiles_grab_object(struct cachefiles_object *object,
+						 enum cachefiles_obj_ref_trace why)
 {
-	struct cachefiles_object *object;
-
-	object = container_of(_object, struct cachefiles_object, fscache);
-
-	_enter("{OBJ%x,%p}", object->fscache.debug_id, object->lookup_data);
+	int r;
 
-	if (object->lookup_data) {
-		kfree(object->lookup_data->key);
-		kfree(object->lookup_data->auxdata);
-		kfree(object->lookup_data);
-		object->lookup_data = NULL;
-	}
+	__refcount_inc(&object->ref, &r);
+	trace_cachefiles_ref(object->debug_id, object->cookie->debug_id, r, why);
+	return object;
 }
 
 /*
- * increment the usage count on an inode object (may fail if unmounting)
+ * dispose of a reference to an object
  */
-static
-struct fscache_object *cachefiles_grab_object(struct fscache_object *_object,
-					      enum fscache_obj_ref_trace why)
+void cachefiles_put_object(struct cachefiles_object *object,
+			   enum cachefiles_obj_ref_trace why)
 {
-	struct cachefiles_object *object =
-		container_of(_object, struct cachefiles_object, fscache);
-	int u;
+	unsigned int object_debug_id = object->debug_id;
+	unsigned int cookie_debug_id = object->cookie->debug_id;
+	struct fscache_cache *cache;
+	bool done;
+	int r;
+
+	done = __refcount_dec_and_test(&object->ref, &r);
+	trace_cachefiles_ref(object_debug_id, cookie_debug_id, r, why);
+	if (done) {
+		_debug("- kill object OBJ%x", object_debug_id);
+
+		ASSERTCMP(object->file, ==, NULL);
 
-	_enter("{OBJ%x,%d}", _object->debug_id, atomic_read(&object->usage));
+		kfree(object->d_name);
 
-#ifdef CACHEFILES_DEBUG_SLAB
-	ASSERT((atomic_read(&object->usage) & 0xffff0000) != 0x6b6b0000);
-#endif
+		cache = object->volume->cache->cache;
+		fscache_put_cookie(object->cookie, fscache_cookie_put_object);
+		object->cookie = NULL;
+		kmem_cache_free(cachefiles_object_jar, object);
+		fscache_uncount_object(cache);
+	}
 
-	u = atomic_inc_return(&object->usage);
-	trace_cachefiles_ref(object, _object->cookie,
-			     (enum cachefiles_obj_ref_trace)why, u);
-	return &object->fscache;
+	_leave("");
 }
 
 /*
- * update the auxiliary data for an object object on disk
+ * Adjust the size of a cache file if necessary to match the DIO size.  We keep
+ * the EOF marker a multiple of DIO blocks so that we don't fall back to doing
+ * non-DIO for a partial block straddling the EOF, but we also have to be
+ * careful of someone expanding the file and accidentally accreting the
+ * padding.
  */
-static void cachefiles_update_object(struct fscache_object *_object)
+static int cachefiles_adjust_size(struct cachefiles_object *object)
 {
-	struct cachefiles_object *object;
-	struct cachefiles_xattr *auxdata;
-	struct cachefiles_cache *cache;
-	struct fscache_cookie *cookie;
-	const struct cred *saved_cred;
-	const void *aux;
-	unsigned auxlen;
+	struct iattr newattrs;
+	struct file *file = object->file;
+	uint64_t ni_size;
+	loff_t oi_size;
+	int ret;
 
-	_enter("{OBJ%x}", _object->debug_id);
+	ni_size = object->cookie->object_size;
+	ni_size = round_up(ni_size, CACHEFILES_DIO_BLOCK_SIZE);
 
-	object = container_of(_object, struct cachefiles_object, fscache);
-	cache = container_of(object->fscache.cache, struct cachefiles_cache,
-			     cache);
+	_enter("{OBJ%x},[%llu]",
+	       object->debug_id, (unsigned long long) ni_size);
 
-	if (!fscache_use_cookie(_object)) {
-		_leave(" [relinq]");
-		return;
-	}
+	if (!file)
+		return -ENOBUFS;
 
-	cookie = object->fscache.cookie;
-	auxlen = cookie->aux_len;
+	oi_size = i_size_read(file_inode(file));
+	if (oi_size == ni_size)
+		return 0;
 
-	if (!auxlen) {
-		fscache_unuse_cookie(_object);
-		_leave(" [no aux]");
-		return;
-	}
+	inode_lock(file_inode(file));
 
-	auxdata = kmalloc(2 + auxlen + 3, cachefiles_gfp);
-	if (!auxdata) {
-		fscache_unuse_cookie(_object);
-		_leave(" [nomem]");
-		return;
+	/* if there's an extension to a partial page at the end of the backing
+	 * file, we need to discard the partial page so that we pick up new
+	 * data after it */
+	if (oi_size & ~PAGE_MASK && ni_size > oi_size) {
+		_debug("discard tail %llx", oi_size);
+		newattrs.ia_valid = ATTR_SIZE;
+		newattrs.ia_size = oi_size & PAGE_MASK;
+		ret = cachefiles_inject_remove_error();
+		if (ret == 0)
+			ret = notify_change(&init_user_ns, file->f_path.dentry,
+					    &newattrs, NULL);
+		if (ret < 0)
+			goto truncate_failed;
 	}
 
-	aux = (auxlen <= sizeof(cookie->inline_aux)) ?
-		cookie->inline_aux : cookie->aux;
+	newattrs.ia_valid = ATTR_SIZE;
+	newattrs.ia_size = ni_size;
+	ret = cachefiles_inject_write_error();
+	if (ret == 0)
+		ret = notify_change(&init_user_ns, file->f_path.dentry,
+				    &newattrs, NULL);
 
-	memcpy(auxdata->data, aux, auxlen);
-	fscache_unuse_cookie(_object);
+truncate_failed:
+	inode_unlock(file_inode(file));
 
-	auxdata->len = auxlen + 1;
-	auxdata->type = cookie->type;
+	if (ret < 0)
+		trace_cachefiles_io_error(NULL, file_inode(file), ret,
+					  cachefiles_trace_notify_change_error);
+	if (ret == -EIO) {
+		cachefiles_io_error_obj(object, "Size set failed");
+		ret = -ENOBUFS;
+	}
 
-	cachefiles_begin_secure(cache, &saved_cred);
-	cachefiles_update_object_xattr(object, auxdata);
-	cachefiles_end_secure(cache, saved_cred);
-	kfree(auxdata);
-	_leave("");
+	_leave(" = %d", ret);
+	return ret;
 }
 
 /*
- * discard the resources pinned by an object and effect retirement if
- * requested
+ * Attempt to look up the nominated node in this cache
  */
-static void cachefiles_drop_object(struct fscache_object *_object)
+static bool cachefiles_lookup_cookie(struct fscache_cookie *cookie)
 {
 	struct cachefiles_object *object;
-	struct cachefiles_cache *cache;
+	struct cachefiles_cache *cache = cookie->volume->cache->cache_priv;
 	const struct cred *saved_cred;
-	struct inode *inode;
-	blkcnt_t i_blocks = 0;
+	bool success;
 
-	ASSERT(_object);
+	object = cachefiles_alloc_object(cookie);
+	if (!object)
+		goto fail;
 
-	object = container_of(_object, struct cachefiles_object, fscache);
+	_enter("{OBJ%x}", object->debug_id);
 
-	_enter("{OBJ%x,%d}",
-	       object->fscache.debug_id, atomic_read(&object->usage));
+	if (!cachefiles_cook_key(object))
+		goto fail_put;
 
-	cache = container_of(object->fscache.cache,
-			     struct cachefiles_cache, cache);
+	cookie->cache_priv = object;
 
-#ifdef CACHEFILES_DEBUG_SLAB
-	ASSERT((atomic_read(&object->usage) & 0xffff0000) != 0x6b6b0000);
-#endif
+	cachefiles_begin_secure(cache, &saved_cred);
 
-	/* We need to tidy the object up if we did in fact manage to open it.
-	 * It's possible for us to get here before the object is fully
-	 * initialised if the parent goes away or the object gets retired
-	 * before we set it up.
-	 */
-	if (object->dentry) {
-		/* delete retired objects */
-		if (test_bit(FSCACHE_OBJECT_RETIRED, &object->fscache.flags) &&
-		    _object != cache->cache.fsdef
-		    ) {
-			_debug("- retire object OBJ%x", object->fscache.debug_id);
-			inode = d_backing_inode(object->dentry);
-			if (inode)
-				i_blocks = inode->i_blocks;
-
-			cachefiles_begin_secure(cache, &saved_cred);
-			cachefiles_delete_object(cache, object);
-			cachefiles_end_secure(cache, saved_cred);
-		}
+	success = cachefiles_look_up_object(object);
+	if (!success)
+		goto fail_withdraw;
 
-		/* close the filesystem stuff attached to the object */
-		if (object->backer != object->dentry)
-			dput(object->backer);
-		object->backer = NULL;
-	}
+	cachefiles_see_object(object, cachefiles_obj_see_lookup_cookie);
+
+	spin_lock(&cache->object_list_lock);
+	list_add(&object->cache_link, &cache->object_list);
+	spin_unlock(&cache->object_list_lock);
+	cachefiles_adjust_size(object);
 
-	/* note that the object is now inactive */
-	if (test_bit(CACHEFILES_OBJECT_ACTIVE, &object->flags))
-		cachefiles_mark_object_inactive(cache, object, i_blocks);
+	cachefiles_end_secure(cache, saved_cred);
+	_leave(" = t");
+	return true;
 
-	dput(object->dentry);
-	object->dentry = NULL;
+fail_withdraw:
+	cachefiles_end_secure(cache, saved_cred);
+	cachefiles_see_object(object, cachefiles_obj_see_lookup_failed);
+	fscache_caching_failed(cookie);
+	_debug("failed c=%08x o=%08x", cookie->debug_id, object->debug_id);
+	/* The caller holds an access count on the cookie, so we need them to
+	 * drop it before we can withdraw the object.
+	 */
+	return false;
 
-	_leave("");
+fail_put:
+	cachefiles_put_object(object, cachefiles_obj_put_alloc_fail);
+fail:
+	return false;
 }
 
 /*
- * dispose of a reference to an object
+ * Shorten the backing object to discard any dirty data and free up
+ * any unused granules.
  */
-void cachefiles_put_object(struct fscache_object *_object,
-			   enum fscache_obj_ref_trace why)
+static bool cachefiles_shorten_object(struct cachefiles_object *object,
+				      struct file *file, loff_t new_size)
 {
-	struct cachefiles_object *object;
-	struct fscache_cache *cache;
-	int u;
-
-	ASSERT(_object);
-
-	object = container_of(_object, struct cachefiles_object, fscache);
-
-	_enter("{OBJ%x,%d}",
-	       object->fscache.debug_id, atomic_read(&object->usage));
-
-#ifdef CACHEFILES_DEBUG_SLAB
-	ASSERT((atomic_read(&object->usage) & 0xffff0000) != 0x6b6b0000);
-#endif
-
-	ASSERTIFCMP(object->fscache.parent,
-		    object->fscache.parent->n_children, >, 0);
-
-	u = atomic_dec_return(&object->usage);
-	trace_cachefiles_ref(object, _object->cookie,
-			     (enum cachefiles_obj_ref_trace)why, u);
-	ASSERTCMP(u, !=, -1);
-	if (u == 0) {
-		_debug("- kill object OBJ%x", object->fscache.debug_id);
+	struct cachefiles_cache *cache = object->volume->cache;
+	struct inode *inode = file_inode(file);
+	loff_t i_size, dio_size;
+	int ret;
 
-		ASSERT(!test_bit(CACHEFILES_OBJECT_ACTIVE, &object->flags));
-		ASSERTCMP(object->fscache.parent, ==, NULL);
-		ASSERTCMP(object->backer, ==, NULL);
-		ASSERTCMP(object->dentry, ==, NULL);
-		ASSERTCMP(object->fscache.n_ops, ==, 0);
-		ASSERTCMP(object->fscache.n_children, ==, 0);
+	dio_size = round_up(new_size, CACHEFILES_DIO_BLOCK_SIZE);
+	i_size = i_size_read(inode);
+
+	trace_cachefiles_trunc(object, inode, i_size, dio_size,
+			       cachefiles_trunc_shrink);
+	ret = cachefiles_inject_remove_error();
+	if (ret == 0)
+		ret = vfs_truncate(&file->f_path, dio_size);
+	if (ret < 0) {
+		trace_cachefiles_io_error(object, file_inode(file), ret,
+					  cachefiles_trace_trunc_error);
+		cachefiles_io_error_obj(object, "Trunc-to-size failed %d", ret);
+		cachefiles_remove_object_xattr(cache, object, file->f_path.dentry);
+		return false;
+	}
 
-		if (object->lookup_data) {
-			kfree(object->lookup_data->key);
-			kfree(object->lookup_data->auxdata);
-			kfree(object->lookup_data);
-			object->lookup_data = NULL;
+	if (new_size < dio_size) {
+		trace_cachefiles_trunc(object, inode, dio_size, new_size,
+				       cachefiles_trunc_dio_adjust);
+		ret = cachefiles_inject_write_error();
+		if (ret == 0)
+			ret = vfs_fallocate(file, FALLOC_FL_ZERO_RANGE,
+					    new_size, dio_size);
+		if (ret < 0) {
+			trace_cachefiles_io_error(object, file_inode(file), ret,
+						  cachefiles_trace_fallocate_error);
+			cachefiles_io_error_obj(object, "Trunc-to-dio-size failed %d", ret);
+			cachefiles_remove_object_xattr(cache, object, file->f_path.dentry);
+			return false;
 		}
-
-		cache = object->fscache.cache;
-		fscache_object_destroy(&object->fscache);
-		kmem_cache_free(cachefiles_object_jar, object);
-		fscache_object_destroyed(cache);
 	}
 
-	_leave("");
+	return true;
 }
 
 /*
- * sync a cache
+ * Resize the backing object.
  */
-static void cachefiles_sync_cache(struct fscache_cache *_cache)
+static void cachefiles_resize_cookie(struct netfs_cache_resources *cres,
+				     loff_t new_size)
 {
-	struct cachefiles_cache *cache;
+	struct cachefiles_object *object = cachefiles_cres_object(cres);
+	struct cachefiles_cache *cache = object->volume->cache;
+	struct fscache_cookie *cookie = object->cookie;
 	const struct cred *saved_cred;
-	int ret;
+	struct file *file = cachefiles_cres_file(cres);
+	loff_t old_size = cookie->object_size;
 
-	_enter("%s", _cache->tag->name);
+	_enter("%llu->%llu", old_size, new_size);
 
-	cache = container_of(_cache, struct cachefiles_cache, cache);
-
-	/* make sure all pages pinned by operations on behalf of the netfs are
-	 * written to disc */
-	cachefiles_begin_secure(cache, &saved_cred);
-	down_read(&cache->mnt->mnt_sb->s_umount);
-	ret = sync_filesystem(cache->mnt->mnt_sb);
-	up_read(&cache->mnt->mnt_sb->s_umount);
-	cachefiles_end_secure(cache, saved_cred);
+	if (new_size < old_size) {
+		cachefiles_begin_secure(cache, &saved_cred);
+		cachefiles_shorten_object(object, file, new_size);
+		cachefiles_end_secure(cache, saved_cred);
+		object->cookie->object_size = new_size;
+		return;
+	}
 
-	if (ret == -EIO)
-		cachefiles_io_error(cache,
-				    "Attempt to sync backing fs superblock"
-				    " returned error %d",
-				    ret);
+	/* The file is being expanded.  We don't need to do anything
+	 * particularly.  cookie->initial_size doesn't change and so the point
+	 * at which we have to download before doesn't change.
+	 */
+	cookie->object_size = new_size;
 }
 
 /*
- * check if the backing cache is updated to FS-Cache
- * - called by FS-Cache when evaluates if need to invalidate the cache
+ * Commit changes to the object as we drop it.
  */
-static int cachefiles_check_consistency(struct fscache_operation *op)
+static void cachefiles_commit_object(struct cachefiles_object *object,
+				     struct cachefiles_cache *cache)
 {
-	struct cachefiles_object *object;
-	struct cachefiles_cache *cache;
-	const struct cred *saved_cred;
-	int ret;
+	bool update = false;
 
-	_enter("{OBJ%x}", op->object->debug_id);
+	if (test_and_clear_bit(FSCACHE_COOKIE_LOCAL_WRITE, &object->cookie->flags))
+		update = true;
+	if (test_and_clear_bit(FSCACHE_COOKIE_NEEDS_UPDATE, &object->cookie->flags))
+		update = true;
+	if (update)
+		cachefiles_set_object_xattr(object);
 
-	object = container_of(op->object, struct cachefiles_object, fscache);
-	cache = container_of(object->fscache.cache,
-			     struct cachefiles_cache, cache);
+	if (test_bit(CACHEFILES_OBJECT_USING_TMPFILE, &object->flags))
+		cachefiles_commit_tmpfile(cache, object);
+}
 
-	cachefiles_begin_secure(cache, &saved_cred);
-	ret = cachefiles_check_auxdata(object);
-	cachefiles_end_secure(cache, saved_cred);
+/*
+ * Finalise and object and close the VFS structs that we have.
+ */
+static void cachefiles_clean_up_object(struct cachefiles_object *object,
+				       struct cachefiles_cache *cache)
+{
+	if (test_bit(FSCACHE_COOKIE_RETIRED, &object->cookie->flags)) {
+		if (!test_bit(CACHEFILES_OBJECT_USING_TMPFILE, &object->flags)) {
+			cachefiles_see_object(object, cachefiles_obj_see_clean_delete);
+			_debug("- inval object OBJ%x", object->debug_id);
+			cachefiles_delete_object(object, FSCACHE_OBJECT_WAS_RETIRED);
+		} else {
+			cachefiles_see_object(object, cachefiles_obj_see_clean_drop_tmp);
+			_debug("- inval object OBJ%x tmpfile", object->debug_id);
+		}
+	} else {
+		cachefiles_see_object(object, cachefiles_obj_see_clean_commit);
+		cachefiles_commit_object(object, cache);
+	}
 
-	_leave(" = %d", ret);
-	return ret;
+	cachefiles_unmark_inode_in_use(object, object->file);
+	if (object->file) {
+		fput(object->file);
+		object->file = NULL;
+	}
 }
 
 /*
- * notification the attributes on an object have changed
- * - called with reads/writes excluded by FS-Cache
+ * Withdraw caching for a cookie.
  */
-static int cachefiles_attr_changed(struct fscache_object *_object)
+static void cachefiles_withdraw_cookie(struct fscache_cookie *cookie)
 {
-	struct cachefiles_object *object;
-	struct cachefiles_cache *cache;
+	struct cachefiles_object *object = cookie->cache_priv;
+	struct cachefiles_cache *cache = object->volume->cache;
 	const struct cred *saved_cred;
-	struct iattr newattrs;
-	uint64_t ni_size;
-	loff_t oi_size;
-	int ret;
-
-	ni_size = _object->store_limit_l;
-
-	_enter("{OBJ%x},[%llu]",
-	       _object->debug_id, (unsigned long long) ni_size);
-
-	object = container_of(_object, struct cachefiles_object, fscache);
-	cache = container_of(object->fscache.cache,
-			     struct cachefiles_cache, cache);
-
-	if (ni_size == object->i_size)
-		return 0;
-
-	if (!object->backer)
-		return -ENOBUFS;
 
-	ASSERT(d_is_reg(object->backer));
+	_enter("o=%x", object->debug_id);
+	cachefiles_see_object(object, cachefiles_obj_see_withdraw_cookie);
 
-	fscache_set_store_limit(&object->fscache, ni_size);
-
-	oi_size = i_size_read(d_backing_inode(object->backer));
-	if (oi_size == ni_size)
-		return 0;
-
-	cachefiles_begin_secure(cache, &saved_cred);
-	inode_lock(d_inode(object->backer));
-
-	/* if there's an extension to a partial page at the end of the backing
-	 * file, we need to discard the partial page so that we pick up new
-	 * data after it */
-	if (oi_size & ~PAGE_MASK && ni_size > oi_size) {
-		_debug("discard tail %llx", oi_size);
-		newattrs.ia_valid = ATTR_SIZE;
-		newattrs.ia_size = oi_size & PAGE_MASK;
-		ret = notify_change(&init_user_ns, object->backer, &newattrs, NULL);
-		if (ret < 0)
-			goto truncate_failed;
+	if (!list_empty(&object->cache_link)) {
+		spin_lock(&cache->object_list_lock);
+		cachefiles_see_object(object, cachefiles_obj_see_withdrawal);
+		list_del_init(&object->cache_link);
+		spin_unlock(&cache->object_list_lock);
 	}
 
-	newattrs.ia_valid = ATTR_SIZE;
-	newattrs.ia_size = ni_size;
-	ret = notify_change(&init_user_ns, object->backer, &newattrs, NULL);
-
-truncate_failed:
-	inode_unlock(d_inode(object->backer));
-	cachefiles_end_secure(cache, saved_cred);
-
-	if (ret == -EIO) {
-		fscache_set_store_limit(&object->fscache, 0);
-		cachefiles_io_error_obj(object, "Size set failed");
-		ret = -ENOBUFS;
+	if (object->file) {
+		cachefiles_begin_secure(cache, &saved_cred);
+		cachefiles_clean_up_object(object, cache);
+		cachefiles_end_secure(cache, saved_cred);
 	}
 
-	_leave(" = %d", ret);
-	return ret;
+	cookie->cache_priv = NULL;
+	cachefiles_put_object(object, cachefiles_obj_put_detach);
 }
 
 /*
- * Invalidate an object
+ * Invalidate the storage associated with a cookie.
  */
-static void cachefiles_invalidate_object(struct fscache_operation *op)
+static bool cachefiles_invalidate_cookie(struct fscache_cookie *cookie)
 {
-	struct cachefiles_object *object;
-	struct cachefiles_cache *cache;
-	const struct cred *saved_cred;
-	struct path path;
-	uint64_t ni_size;
-	int ret;
+	struct cachefiles_object *object = cookie->cache_priv;
+	struct file *new_file, *old_file;
+	bool old_tmpfile;
 
-	object = container_of(op->object, struct cachefiles_object, fscache);
-	cache = container_of(object->fscache.cache,
-			     struct cachefiles_cache, cache);
+	_enter("o=%x,[%llu]", object->debug_id, object->cookie->object_size);
 
-	ni_size = op->object->store_limit_l;
+	old_tmpfile = test_bit(CACHEFILES_OBJECT_USING_TMPFILE, &object->flags);
 
-	_enter("{OBJ%x},[%llu]",
-	       op->object->debug_id, (unsigned long long)ni_size);
+	if (!object->file) {
+		fscache_resume_after_invalidation(cookie);
+		_leave(" = t [light]");
+		return true;
+	}
 
-	if (object->backer) {
-		ASSERT(d_is_reg(object->backer));
+	new_file = cachefiles_create_tmpfile(object);
+	if (IS_ERR(new_file))
+		goto failed;
 
-		fscache_set_store_limit(&object->fscache, ni_size);
+	/* Substitute the VFS target */
+	_debug("sub");
+	spin_lock(&object->lock);
 
-		path.dentry = object->backer;
-		path.mnt = cache->mnt;
+	old_file = object->file;
+	object->file = new_file;
+	object->content_info = CACHEFILES_CONTENT_NO_DATA;
+	set_bit(CACHEFILES_OBJECT_USING_TMPFILE, &object->flags);
+	set_bit(FSCACHE_COOKIE_NEEDS_UPDATE, &object->cookie->flags);
 
-		cachefiles_begin_secure(cache, &saved_cred);
-		ret = vfs_truncate(&path, 0);
-		if (ret == 0)
-			ret = vfs_truncate(&path, ni_size);
-		cachefiles_end_secure(cache, saved_cred);
+	spin_unlock(&object->lock);
+	_debug("subbed");
+
+	/* Allow I/O to take place again */
+	fscache_resume_after_invalidation(cookie);
+
+	if (old_file) {
+		if (!old_tmpfile) {
+			struct cachefiles_volume *volume = object->volume;
+			struct dentry *fan = volume->fanout[(u8)cookie->key_hash];
 
-		if (ret != 0) {
-			fscache_set_store_limit(&object->fscache, 0);
-			if (ret == -EIO)
-				cachefiles_io_error_obj(object,
-							"Invalidate failed");
+			inode_lock_nested(d_inode(fan), I_MUTEX_PARENT);
+			cachefiles_bury_object(volume->cache, object, fan,
+					       old_file->f_path.dentry,
+					       FSCACHE_OBJECT_INVALIDATED);
 		}
+		fput(old_file);
 	}
 
-	fscache_op_complete(op, true);
-	_leave("");
-}
+	_leave(" = t");
+	return true;
 
-/*
- * dissociate a cache from all the pages it was backing
- */
-static void cachefiles_dissociate_pages(struct fscache_cache *cache)
-{
-	_enter("");
+failed:
+	_leave(" = f");
+	return false;
 }
 
 const struct fscache_cache_ops cachefiles_cache_ops = {
 	.name			= "cachefiles",
-	.alloc_object		= cachefiles_alloc_object,
-	.lookup_object		= cachefiles_lookup_object,
-	.lookup_complete	= cachefiles_lookup_complete,
-	.grab_object		= cachefiles_grab_object,
-	.update_object		= cachefiles_update_object,
-	.invalidate_object	= cachefiles_invalidate_object,
-	.drop_object		= cachefiles_drop_object,
-	.put_object		= cachefiles_put_object,
-	.sync_cache		= cachefiles_sync_cache,
-	.attr_changed		= cachefiles_attr_changed,
-	.read_or_alloc_page	= cachefiles_read_or_alloc_page,
-	.read_or_alloc_pages	= cachefiles_read_or_alloc_pages,
-	.allocate_page		= cachefiles_allocate_page,
-	.allocate_pages		= cachefiles_allocate_pages,
-	.write_page		= cachefiles_write_page,
-	.uncache_page		= cachefiles_uncache_page,
-	.dissociate_pages	= cachefiles_dissociate_pages,
-	.check_consistency	= cachefiles_check_consistency,
-	.begin_read_operation	= cachefiles_begin_read_operation,
+	.acquire_volume		= cachefiles_acquire_volume,
+	.free_volume		= cachefiles_free_volume,
+	.lookup_cookie		= cachefiles_lookup_cookie,
+	.withdraw_cookie	= cachefiles_withdraw_cookie,
+	.invalidate_cookie	= cachefiles_invalidate_cookie,
+	.begin_operation	= cachefiles_begin_operation,
+	.resize_cookie		= cachefiles_resize_cookie,
+	.prepare_to_write	= cachefiles_prepare_to_write,
 };
diff --git a/fs/cachefiles/internal.h b/fs/cachefiles/internal.h
index 0a511c36dab8..8dd54d9375b6 100644
--- a/fs/cachefiles/internal.h
+++ b/fs/cachefiles/internal.h
@@ -1,7 +1,7 @@
 /* SPDX-License-Identifier: GPL-2.0-or-later */
 /* General netfs cache on cache files internal defs
  *
- * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
+ * Copyright (C) 2021 Red Hat, Inc. All Rights Reserved.
  * Written by David Howells (dhowells@redhat.com)
  */
 
@@ -13,58 +13,72 @@
 
 
 #include <linux/fscache-cache.h>
-#include <linux/timer.h>
-#include <linux/wait_bit.h>
 #include <linux/cred.h>
-#include <linux/workqueue.h>
 #include <linux/security.h>
 
+#define CACHEFILES_DIO_BLOCK_SIZE 4096
+
 struct cachefiles_cache;
 struct cachefiles_object;
 
-extern unsigned cachefiles_debug;
-#define CACHEFILES_DEBUG_KENTER	1
-#define CACHEFILES_DEBUG_KLEAVE	2
-#define CACHEFILES_DEBUG_KDEBUG	4
+enum cachefiles_content {
+	/* These values are saved on disk */
+	CACHEFILES_CONTENT_NO_DATA	= 0, /* No content stored */
+	CACHEFILES_CONTENT_SINGLE	= 1, /* Content is monolithic, all is present */
+	CACHEFILES_CONTENT_ALL		= 2, /* Content is all present, no map */
+	CACHEFILES_CONTENT_BACKFS_MAP	= 3, /* Content is piecemeal, mapped through backing fs */
+	CACHEFILES_CONTENT_DIRTY	= 4, /* Content is dirty (only seen on disk) */
+	nr__cachefiles_content
+};
 
-#define cachefiles_gfp (__GFP_RECLAIM | __GFP_NORETRY | __GFP_NOMEMALLOC)
+/*
+ * Cached volume representation.
+ */
+struct cachefiles_volume {
+	struct cachefiles_cache		*cache;
+	struct list_head		cache_link;	/* Link in cache->volumes */
+	struct fscache_volume		*vcookie;	/* The netfs's representation */
+	struct dentry			*dentry;	/* The volume dentry */
+	struct dentry			*fanout[256];	/* Fanout subdirs */
+};
 
 /*
- * node records
+ * Backing file state.
  */
 struct cachefiles_object {
-	struct fscache_object		fscache;	/* fscache handle */
-	struct cachefiles_lookup_data	*lookup_data;	/* cached lookup data */
-	struct dentry			*dentry;	/* the file/dir representing this object */
-	struct dentry			*backer;	/* backing file */
-	loff_t				i_size;		/* object size */
+	struct fscache_cookie		*cookie;	/* Netfs data storage object cookie */
+	struct cachefiles_volume	*volume;	/* Cache volume that holds this object */
+	struct list_head		cache_link;	/* Link in cache->*_list */
+	struct file			*file;		/* The file representing this object */
+	char				*d_name;	/* Backing file name */
+	int				debug_id;
+	spinlock_t			lock;
+	refcount_t			ref;
+	u8				d_name_len;	/* Length of filename */
+	enum cachefiles_content		content_info:8;	/* Info about content presence */
 	unsigned long			flags;
-#define CACHEFILES_OBJECT_ACTIVE	0		/* T if marked active */
-	atomic_t			usage;		/* object usage count */
-	uint8_t				type;		/* object type */
-	uint8_t				new;		/* T if object new */
-	spinlock_t			work_lock;
-	struct rb_node			active_node;	/* link in active tree (dentry is key) */
+#define CACHEFILES_OBJECT_USING_TMPFILE	0		/* Have an unlinked tmpfile */
 };
 
-extern struct kmem_cache *cachefiles_object_jar;
-
 /*
  * Cache files cache definition
  */
 struct cachefiles_cache {
-	struct fscache_cache		cache;		/* FS-Cache record */
+	struct fscache_cache		*cache;		/* Cache cookie */
 	struct vfsmount			*mnt;		/* mountpoint holding the cache */
+	struct dentry			*store;		/* Directory into which live objects go */
 	struct dentry			*graveyard;	/* directory into which dead objects go */
 	struct file			*cachefilesd;	/* manager daemon handle */
+	struct list_head		volumes;	/* List of volume objects */
+	struct list_head		object_list;	/* List of active objects */
+	spinlock_t			object_list_lock; /* Lock for volumes and object_list */
 	const struct cred		*cache_cred;	/* security override for accessing cache */
 	struct mutex			daemon_mutex;	/* command serialisation mutex */
 	wait_queue_head_t		daemon_pollwq;	/* poll waitqueue for daemon */
-	struct rb_root			active_nodes;	/* active nodes (can't be culled) */
-	rwlock_t			active_lock;	/* lock for active_nodes */
 	atomic_t			gravecounter;	/* graveyard uniquifier */
 	atomic_t			f_released;	/* number of objects released lately */
 	atomic_long_t			b_released;	/* number of blocks released lately */
+	atomic_long_t			b_writing;	/* Number of blocks being written */
 	unsigned			frun_percent;	/* when to stop culling (% files) */
 	unsigned			fcull_percent;	/* when to start culling (% files) */
 	unsigned			fstop_percent;	/* when to stop allocating (% files) */
@@ -89,38 +103,19 @@ struct cachefiles_cache {
 	char				*tag;		/* cache binding tag */
 };
 
-/*
- * backing file read tracking
- */
-struct cachefiles_one_read {
-	wait_queue_entry_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 fscache_retrieval	*op;		/* retrieval op covering this */
-	struct list_head		op_link;	/* link in op's todo list */
-};
-
-/*
- * 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;
-};
+#include <trace/events/cachefiles.h>
 
-/*
- * auxiliary data xattr buffer
- */
-struct cachefiles_xattr {
-	uint16_t			len;
-	uint8_t				type;
-	uint8_t				data[];
-};
+static inline
+struct file *cachefiles_cres_file(struct netfs_cache_resources *cres)
+{
+	return cres->cache_priv2;
+}
 
-#include <trace/events/cachefiles.h>
+static inline
+struct cachefiles_object *cachefiles_cres_object(struct netfs_cache_resources *cres)
+{
+	return fscache_cres_cookie(cres)->cache_priv;
+}
 
 /*
  * note change of state for daemon
@@ -132,74 +127,118 @@ static inline void cachefiles_state_changed(struct cachefiles_cache *cache)
 }
 
 /*
- * bind.c
+ * cache.c
  */
-extern int cachefiles_daemon_bind(struct cachefiles_cache *cache, char *args);
-extern void cachefiles_daemon_unbind(struct cachefiles_cache *cache);
+extern int cachefiles_add_cache(struct cachefiles_cache *cache);
+extern void cachefiles_withdraw_cache(struct cachefiles_cache *cache);
+
+enum cachefiles_has_space_for {
+	cachefiles_has_space_check,
+	cachefiles_has_space_for_write,
+	cachefiles_has_space_for_create,
+};
+extern int cachefiles_has_space(struct cachefiles_cache *cache,
+				unsigned fnr, unsigned bnr,
+				enum cachefiles_has_space_for reason);
 
 /*
  * daemon.c
  */
 extern const struct file_operations cachefiles_daemon_fops;
 
-extern int cachefiles_has_space(struct cachefiles_cache *cache,
-				unsigned fnr, unsigned bnr);
+/*
+ * error_inject.c
+ */
+#ifdef CONFIG_CACHEFILES_ERROR_INJECTION
+extern unsigned int cachefiles_error_injection_state;
+extern int cachefiles_register_error_injection(void);
+extern void cachefiles_unregister_error_injection(void);
+
+#else
+#define cachefiles_error_injection_state 0
+
+static inline int cachefiles_register_error_injection(void)
+{
+	return 0;
+}
+
+static inline void cachefiles_unregister_error_injection(void)
+{
+}
+#endif
+
+
+static inline int cachefiles_inject_read_error(void)
+{
+	return cachefiles_error_injection_state & 2 ? -EIO : 0;
+}
+
+static inline int cachefiles_inject_write_error(void)
+{
+	return cachefiles_error_injection_state & 2 ? -EIO :
+		cachefiles_error_injection_state & 1 ? -ENOSPC :
+		0;
+}
+
+static inline int cachefiles_inject_remove_error(void)
+{
+	return cachefiles_error_injection_state & 2 ? -EIO : 0;
+}
 
 /*
  * interface.c
  */
 extern const struct fscache_cache_ops cachefiles_cache_ops;
+extern void cachefiles_see_object(struct cachefiles_object *object,
+				  enum cachefiles_obj_ref_trace why);
+extern struct cachefiles_object *cachefiles_grab_object(struct cachefiles_object *object,
+							enum cachefiles_obj_ref_trace why);
+extern void cachefiles_put_object(struct cachefiles_object *object,
+				  enum cachefiles_obj_ref_trace why);
 
-void cachefiles_put_object(struct fscache_object *_object,
-			   enum fscache_obj_ref_trace why);
+/*
+ * io.c
+ */
+extern bool cachefiles_begin_operation(struct netfs_cache_resources *cres,
+				       enum fscache_want_state want_state);
 
 /*
  * key.c
  */
-extern char *cachefiles_cook_key(const u8 *raw, int keylen, uint8_t type);
+extern bool cachefiles_cook_key(struct cachefiles_object *object);
+
+/*
+ * main.c
+ */
+extern struct kmem_cache *cachefiles_object_jar;
 
 /*
  * namei.c
  */
-extern void cachefiles_mark_object_inactive(struct cachefiles_cache *cache,
-					    struct cachefiles_object *object,
-					    blkcnt_t i_blocks);
-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,
-				     const char *key,
-				     struct cachefiles_xattr *auxdata);
+extern void cachefiles_unmark_inode_in_use(struct cachefiles_object *object,
+					   struct file *file);
+extern int cachefiles_bury_object(struct cachefiles_cache *cache,
+				  struct cachefiles_object *object,
+				  struct dentry *dir,
+				  struct dentry *rep,
+				  enum fscache_why_object_killed why);
+extern int cachefiles_delete_object(struct cachefiles_object *object,
+				    enum fscache_why_object_killed why);
+extern bool cachefiles_look_up_object(struct cachefiles_object *object);
 extern struct dentry *cachefiles_get_directory(struct cachefiles_cache *cache,
 					       struct dentry *dir,
-					       const char *name);
+					       const char *name,
+					       bool *_is_new);
+extern void cachefiles_put_directory(struct dentry *dir);
 
 extern int cachefiles_cull(struct cachefiles_cache *cache, struct dentry *dir,
 			   char *filename);
 
 extern int cachefiles_check_in_use(struct cachefiles_cache *cache,
 				   struct dentry *dir, char *filename);
-
-/*
- * rdwr.c
- */
-extern int cachefiles_read_or_alloc_page(struct fscache_retrieval *,
-					 struct page *, gfp_t);
-extern int cachefiles_read_or_alloc_pages(struct fscache_retrieval *,
-					  struct list_head *, unsigned *,
-					  gfp_t);
-extern int cachefiles_allocate_page(struct fscache_retrieval *, struct page *,
-				    gfp_t);
-extern int cachefiles_allocate_pages(struct fscache_retrieval *,
-				     struct list_head *, unsigned *, gfp_t);
-extern int cachefiles_write_page(struct fscache_storage *, struct page *);
-extern void cachefiles_uncache_page(struct fscache_object *, struct page *);
-
-/*
- * rdwr2.c
- */
-extern int cachefiles_begin_read_operation(struct netfs_read_request *,
-					   struct fscache_retrieval *);
+extern struct file *cachefiles_create_tmpfile(struct cachefiles_object *object);
+extern bool cachefiles_commit_tmpfile(struct cachefiles_cache *cache,
+				      struct cachefiles_object *object);
 
 /*
  * security.c
@@ -222,28 +261,32 @@ static inline void cachefiles_end_secure(struct cachefiles_cache *cache,
 }
 
 /*
+ * volume.c
+ */
+void cachefiles_acquire_volume(struct fscache_volume *volume);
+void cachefiles_free_volume(struct fscache_volume *volume);
+void cachefiles_withdraw_volume(struct cachefiles_volume *volume);
+
+/*
  * 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_update_object_xattr(struct cachefiles_object *object,
-					  struct cachefiles_xattr *auxdata);
-extern int cachefiles_check_auxdata(struct cachefiles_object *object);
-extern int cachefiles_check_object_xattr(struct cachefiles_object *object,
-					 struct cachefiles_xattr *auxdata);
+extern int cachefiles_set_object_xattr(struct cachefiles_object *object);
+extern int cachefiles_check_auxdata(struct cachefiles_object *object,
+				    struct file *file);
 extern int cachefiles_remove_object_xattr(struct cachefiles_cache *cache,
+					  struct cachefiles_object *object,
 					  struct dentry *dentry);
-
+extern void cachefiles_prepare_to_write(struct fscache_cookie *cookie);
+extern bool cachefiles_set_volume_xattr(struct cachefiles_volume *volume);
+extern int cachefiles_check_volume_xattr(struct cachefiles_volume *volume);
 
 /*
- * error handling
+ * Error handling
  */
-
 #define cachefiles_io_error(___cache, FMT, ...)		\
 do {							\
 	pr_err("I/O Error: " FMT"\n", ##__VA_ARGS__);	\
-	fscache_io_error(&(___cache)->cache);		\
+	fscache_io_error((___cache)->cache);		\
 	set_bit(CACHEFILES_DEAD, &(___cache)->flags);	\
 } while (0)
 
@@ -251,15 +294,20 @@ do {							\
 do {									\
 	struct cachefiles_cache *___cache;				\
 									\
-	___cache = container_of((object)->fscache.cache,		\
-				struct cachefiles_cache, cache);	\
-	cachefiles_io_error(___cache, FMT, ##__VA_ARGS__);		\
+	___cache = (object)->volume->cache;				\
+	cachefiles_io_error(___cache, FMT " [o=%08x]", ##__VA_ARGS__,	\
+			    (object)->debug_id);			\
 } while (0)
 
 
 /*
- * debug tracing
+ * Debug tracing
  */
+extern unsigned cachefiles_debug;
+#define CACHEFILES_DEBUG_KENTER	1
+#define CACHEFILES_DEBUG_KLEAVE	2
+#define CACHEFILES_DEBUG_KDEBUG	4
+
 #define dbgprintk(FMT, ...) \
 	printk(KERN_DEBUG "[%-6.6s] "FMT"\n", current->comm, ##__VA_ARGS__)
 
diff --git a/fs/cachefiles/io.c b/fs/cachefiles/io.c
index effe37ef8629..60b1eac2ce78 100644
--- a/fs/cachefiles/io.c
+++ b/fs/cachefiles/io.c
@@ -9,8 +9,9 @@
 #include <linux/slab.h>
 #include <linux/file.h>
 #include <linux/uio.h>
+#include <linux/falloc.h>
 #include <linux/sched/mm.h>
-#include <linux/netfs.h>
+#include <trace/events/fscache.h>
 #include "internal.h"
 
 struct cachefiles_kiocb {
@@ -21,14 +22,18 @@ struct cachefiles_kiocb {
 		size_t		skipped;
 		size_t		len;
 	};
+	struct cachefiles_object *object;
 	netfs_io_terminated_t	term_func;
 	void			*term_func_priv;
 	bool			was_async;
+	unsigned int		inval_counter;	/* Copy of cookie->inval_counter */
+	u64			b_writing;
 };
 
 static inline void cachefiles_put_kiocb(struct cachefiles_kiocb *ki)
 {
 	if (refcount_dec_and_test(&ki->ki_refcnt)) {
+		cachefiles_put_object(ki->object, cachefiles_obj_put_ioreq);
 		fput(ki->iocb.ki_filp);
 		kfree(ki);
 	}
@@ -40,12 +45,22 @@ static inline void cachefiles_put_kiocb(struct cachefiles_kiocb *ki)
 static void cachefiles_read_complete(struct kiocb *iocb, long ret)
 {
 	struct cachefiles_kiocb *ki = container_of(iocb, struct cachefiles_kiocb, iocb);
+	struct inode *inode = file_inode(ki->iocb.ki_filp);
 
 	_enter("%ld", ret);
 
+	if (ret < 0)
+		trace_cachefiles_io_error(ki->object, inode, ret,
+					  cachefiles_trace_read_error);
+
 	if (ki->term_func) {
-		if (ret >= 0)
-			ret += ki->skipped;
+		if (ret >= 0) {
+			if (ki->object->cookie->inval_counter == ki->inval_counter)
+				ki->skipped += ret;
+			else
+				ret = -ESTALE;
+		}
+
 		ki->term_func(ki->term_func_priv, ret, ki->was_async);
 	}
 
@@ -58,16 +73,24 @@ static void cachefiles_read_complete(struct kiocb *iocb, long ret)
 static int cachefiles_read(struct netfs_cache_resources *cres,
 			   loff_t start_pos,
 			   struct iov_iter *iter,
-			   bool seek_data,
+			   enum netfs_read_from_hole read_hole,
 			   netfs_io_terminated_t term_func,
 			   void *term_func_priv)
 {
+	struct cachefiles_object *object;
 	struct cachefiles_kiocb *ki;
-	struct file *file = cres->cache_priv2;
+	struct file *file;
 	unsigned int old_nofs;
 	ssize_t ret = -ENOBUFS;
 	size_t len = iov_iter_count(iter), skipped = 0;
 
+	if (!fscache_wait_for_operation(cres, FSCACHE_WANT_READ))
+		goto presubmission_error;
+
+	fscache_count_read();
+	object = cachefiles_cres_object(cres);
+	file = cachefiles_cres_file(cres);
+
 	_enter("%pD,%li,%llx,%zx/%llx",
 	       file, file_inode(file)->i_ino, start_pos, len,
 	       i_size_read(file_inode(file)));
@@ -75,10 +98,12 @@ static int cachefiles_read(struct netfs_cache_resources *cres,
 	/* If the caller asked us to seek for data before doing the read, then
 	 * we should do that now.  If we find a gap, we fill it with zeros.
 	 */
-	if (seek_data) {
+	if (read_hole != NETFS_READ_HOLE_IGNORE) {
 		loff_t off = start_pos, off2;
 
-		off2 = vfs_llseek(file, off, SEEK_DATA);
+		off2 = cachefiles_inject_read_error();
+		if (off2 == 0)
+			off2 = vfs_llseek(file, off, SEEK_DATA);
 		if (off2 < 0 && off2 >= (loff_t)-MAX_ERRNO && off2 != -ENXIO) {
 			skipped = 0;
 			ret = off2;
@@ -90,6 +115,10 @@ static int cachefiles_read(struct netfs_cache_resources *cres,
 			 * in the region, so clear the rest of the buffer and
 			 * return success.
 			 */
+			ret = -ENODATA;
+			if (read_hole == NETFS_READ_HOLE_FAIL)
+				goto presubmission_error;
+
 			iov_iter_zero(len, iter);
 			skipped = len;
 			ret = 0;
@@ -100,7 +129,7 @@ static int cachefiles_read(struct netfs_cache_resources *cres,
 		iov_iter_zero(skipped, iter);
 	}
 
-	ret = -ENOBUFS;
+	ret = -ENOMEM;
 	ki = kzalloc(sizeof(struct cachefiles_kiocb), GFP_KERNEL);
 	if (!ki)
 		goto presubmission_error;
@@ -112,6 +141,8 @@ static int cachefiles_read(struct netfs_cache_resources *cres,
 	ki->iocb.ki_hint	= ki_hint_validate(file_write_hint(file));
 	ki->iocb.ki_ioprio	= get_current_ioprio();
 	ki->skipped		= skipped;
+	ki->object		= object;
+	ki->inval_counter	= cres->inval_counter;
 	ki->term_func		= term_func;
 	ki->term_func_priv	= term_func_priv;
 	ki->was_async		= true;
@@ -120,9 +151,13 @@ static int cachefiles_read(struct netfs_cache_resources *cres,
 		ki->iocb.ki_complete = cachefiles_read_complete;
 
 	get_file(ki->iocb.ki_filp);
+	cachefiles_grab_object(object, cachefiles_obj_get_ioreq);
 
+	trace_cachefiles_read(object, file_inode(file), ki->iocb.ki_pos, len - skipped);
 	old_nofs = memalloc_nofs_save();
-	ret = vfs_iocb_iter_read(file, &ki->iocb, iter);
+	ret = cachefiles_inject_read_error();
+	if (ret == 0)
+		ret = vfs_iocb_iter_read(file, &ki->iocb, iter);
 	memalloc_nofs_restore(old_nofs);
 	switch (ret) {
 	case -EIOCBQUEUED:
@@ -162,6 +197,7 @@ presubmission_error:
 static void cachefiles_write_complete(struct kiocb *iocb, long ret)
 {
 	struct cachefiles_kiocb *ki = container_of(iocb, struct cachefiles_kiocb, iocb);
+	struct cachefiles_object *object = ki->object;
 	struct inode *inode = file_inode(ki->iocb.ki_filp);
 
 	_enter("%ld", ret);
@@ -170,9 +206,14 @@ static void cachefiles_write_complete(struct kiocb *iocb, long ret)
 	__sb_writers_acquired(inode->i_sb, SB_FREEZE_WRITE);
 	__sb_end_write(inode->i_sb, SB_FREEZE_WRITE);
 
+	if (ret < 0)
+		trace_cachefiles_io_error(object, inode, ret,
+					  cachefiles_trace_write_error);
+
+	atomic_long_sub(ki->b_writing, &object->volume->cache->b_writing);
+	set_bit(FSCACHE_COOKIE_HAVE_DATA, &object->cookie->flags);
 	if (ki->term_func)
 		ki->term_func(ki->term_func_priv, ret, ki->was_async);
-
 	cachefiles_put_kiocb(ki);
 }
 
@@ -185,17 +226,27 @@ static int cachefiles_write(struct netfs_cache_resources *cres,
 			    netfs_io_terminated_t term_func,
 			    void *term_func_priv)
 {
+	struct cachefiles_object *object;
+	struct cachefiles_cache *cache;
 	struct cachefiles_kiocb *ki;
 	struct inode *inode;
-	struct file *file = cres->cache_priv2;
+	struct file *file;
 	unsigned int old_nofs;
 	ssize_t ret = -ENOBUFS;
 	size_t len = iov_iter_count(iter);
 
+	if (!fscache_wait_for_operation(cres, FSCACHE_WANT_WRITE))
+		goto presubmission_error;
+	fscache_count_write();
+	object = cachefiles_cres_object(cres);
+	cache = object->volume->cache;
+	file = cachefiles_cres_file(cres);
+
 	_enter("%pD,%li,%llx,%zx/%llx",
 	       file, file_inode(file)->i_ino, start_pos, len,
 	       i_size_read(file_inode(file)));
 
+	ret = -ENOMEM;
 	ki = kzalloc(sizeof(struct cachefiles_kiocb), GFP_KERNEL);
 	if (!ki)
 		goto presubmission_error;
@@ -206,14 +257,18 @@ static int cachefiles_write(struct netfs_cache_resources *cres,
 	ki->iocb.ki_flags	= IOCB_DIRECT | IOCB_WRITE;
 	ki->iocb.ki_hint	= ki_hint_validate(file_write_hint(file));
 	ki->iocb.ki_ioprio	= get_current_ioprio();
+	ki->object		= object;
+	ki->inval_counter	= cres->inval_counter;
 	ki->start		= start_pos;
 	ki->len			= len;
 	ki->term_func		= term_func;
 	ki->term_func_priv	= term_func_priv;
 	ki->was_async		= true;
+	ki->b_writing		= (len + (1 << cache->bshift)) >> cache->bshift;
 
 	if (ki->term_func)
 		ki->iocb.ki_complete = cachefiles_write_complete;
+	atomic_long_add(ki->b_writing, &cache->b_writing);
 
 	/* Open-code file_start_write here to grab freeze protection, which
 	 * will be released by another thread in aio_complete_rw().  Fool
@@ -225,9 +280,13 @@ static int cachefiles_write(struct netfs_cache_resources *cres,
 	__sb_writers_release(inode->i_sb, SB_FREEZE_WRITE);
 
 	get_file(ki->iocb.ki_filp);
+	cachefiles_grab_object(object, cachefiles_obj_get_ioreq);
 
+	trace_cachefiles_write(object, inode, ki->iocb.ki_pos, len);
 	old_nofs = memalloc_nofs_save();
-	ret = vfs_iocb_iter_write(file, &ki->iocb, iter);
+	ret = cachefiles_inject_write_error();
+	if (ret == 0)
+		ret = vfs_iocb_iter_write(file, &ki->iocb, iter);
 	memalloc_nofs_restore(old_nofs);
 	switch (ret) {
 	case -EIOCBQUEUED:
@@ -257,8 +316,8 @@ in_progress:
 
 presubmission_error:
 	if (term_func)
-		term_func(term_func_priv, -ENOMEM, false);
-	return -ENOMEM;
+		term_func(term_func_priv, ret, false);
+	return ret;
 }
 
 /*
@@ -268,47 +327,82 @@ presubmission_error:
 static enum netfs_read_source cachefiles_prepare_read(struct netfs_read_subrequest *subreq,
 						      loff_t i_size)
 {
-	struct fscache_retrieval *op = subreq->rreq->cache_resources.cache_priv;
+	enum cachefiles_prepare_read_trace why;
+	struct netfs_read_request *rreq = subreq->rreq;
+	struct netfs_cache_resources *cres = &rreq->cache_resources;
 	struct cachefiles_object *object;
 	struct cachefiles_cache *cache;
+	struct fscache_cookie *cookie = fscache_cres_cookie(cres);
 	const struct cred *saved_cred;
-	struct file *file = subreq->rreq->cache_resources.cache_priv2;
+	struct file *file = cachefiles_cres_file(cres);
+	enum netfs_read_source ret = NETFS_DOWNLOAD_FROM_SERVER;
 	loff_t off, to;
+	ino_t ino = file ? file_inode(file)->i_ino : 0;
 
 	_enter("%zx @%llx/%llx", subreq->len, subreq->start, i_size);
 
-	object = container_of(op->op.object,
-			      struct cachefiles_object, fscache);
-	cache = container_of(object->fscache.cache,
-			     struct cachefiles_cache, cache);
+	if (subreq->start >= i_size) {
+		ret = NETFS_FILL_WITH_ZEROES;
+		why = cachefiles_trace_read_after_eof;
+		goto out_no_object;
+	}
 
-	if (!file)
-		goto cache_fail_nosec;
+	if (test_bit(FSCACHE_COOKIE_NO_DATA_TO_READ, &cookie->flags)) {
+		__set_bit(NETFS_SREQ_WRITE_TO_CACHE, &subreq->flags);
+		why = cachefiles_trace_read_no_data;
+		goto out_no_object;
+	}
 
-	if (subreq->start >= i_size)
-		return NETFS_FILL_WITH_ZEROES;
+	/* The object and the file may be being created in the background. */
+	if (!file) {
+		why = cachefiles_trace_read_no_file;
+		if (!fscache_wait_for_operation(cres, FSCACHE_WANT_READ))
+			goto out_no_object;
+		file = cachefiles_cres_file(cres);
+		if (!file)
+			goto out_no_object;
+		ino = file_inode(file)->i_ino;
+	}
 
+	object = cachefiles_cres_object(cres);
+	cache = object->volume->cache;
 	cachefiles_begin_secure(cache, &saved_cred);
 
-	off = vfs_llseek(file, subreq->start, SEEK_DATA);
+	off = cachefiles_inject_read_error();
+	if (off == 0)
+		off = vfs_llseek(file, subreq->start, SEEK_DATA);
 	if (off < 0 && off >= (loff_t)-MAX_ERRNO) {
-		if (off == (loff_t)-ENXIO)
+		if (off == (loff_t)-ENXIO) {
+			why = cachefiles_trace_read_seek_nxio;
 			goto download_and_store;
-		goto cache_fail;
+		}
+		trace_cachefiles_io_error(object, file_inode(file), off,
+					  cachefiles_trace_seek_error);
+		why = cachefiles_trace_read_seek_error;
+		goto out;
 	}
 
-	if (off >= subreq->start + subreq->len)
+	if (off >= subreq->start + subreq->len) {
+		why = cachefiles_trace_read_found_hole;
 		goto download_and_store;
+	}
 
 	if (off > subreq->start) {
 		off = round_up(off, cache->bsize);
 		subreq->len = off - subreq->start;
+		why = cachefiles_trace_read_found_part;
 		goto download_and_store;
 	}
 
-	to = vfs_llseek(file, subreq->start, SEEK_HOLE);
-	if (to < 0 && to >= (loff_t)-MAX_ERRNO)
-		goto cache_fail;
+	to = cachefiles_inject_read_error();
+	if (to == 0)
+		to = vfs_llseek(file, subreq->start, SEEK_HOLE);
+	if (to < 0 && to >= (loff_t)-MAX_ERRNO) {
+		trace_cachefiles_io_error(object, file_inode(file), to,
+					  cachefiles_trace_seek_error);
+		why = cachefiles_trace_read_seek_error;
+		goto out;
+	}
 
 	if (to < subreq->start + subreq->len) {
 		if (subreq->start + subreq->len >= i_size)
@@ -318,32 +412,119 @@ static enum netfs_read_source cachefiles_prepare_read(struct netfs_read_subreque
 		subreq->len = to - subreq->start;
 	}
 
-	cachefiles_end_secure(cache, saved_cred);
-	return NETFS_READ_FROM_CACHE;
+	why = cachefiles_trace_read_have_data;
+	ret = NETFS_READ_FROM_CACHE;
+	goto out;
 
 download_and_store:
-	if (cachefiles_has_space(cache, 0, (subreq->len + PAGE_SIZE - 1) / PAGE_SIZE) == 0)
-		__set_bit(NETFS_SREQ_WRITE_TO_CACHE, &subreq->flags);
-cache_fail:
+	__set_bit(NETFS_SREQ_WRITE_TO_CACHE, &subreq->flags);
+out:
 	cachefiles_end_secure(cache, saved_cred);
-cache_fail_nosec:
-	return NETFS_DOWNLOAD_FROM_SERVER;
+out_no_object:
+	trace_cachefiles_prep_read(subreq, ret, why, ino);
+	return ret;
 }
 
 /*
  * Prepare for a write to occur.
  */
-static int cachefiles_prepare_write(struct netfs_cache_resources *cres,
-				    loff_t *_start, size_t *_len, loff_t i_size)
+static int __cachefiles_prepare_write(struct netfs_cache_resources *cres,
+				      loff_t *_start, size_t *_len, loff_t i_size,
+				      bool no_space_allocated_yet)
 {
-	loff_t start = *_start;
+	struct cachefiles_object *object = cachefiles_cres_object(cres);
+	struct cachefiles_cache *cache = object->volume->cache;
+	struct file *file = cachefiles_cres_file(cres);
+	loff_t start = *_start, pos;
 	size_t len = *_len, down;
+	int ret;
 
 	/* Round to DIO size */
 	down = start - round_down(start, PAGE_SIZE);
 	*_start = start - down;
 	*_len = round_up(down + len, PAGE_SIZE);
-	return 0;
+
+	/* We need to work out whether there's sufficient disk space to perform
+	 * the write - but we can skip that check if we have space already
+	 * allocated.
+	 */
+	if (no_space_allocated_yet)
+		goto check_space;
+
+	pos = cachefiles_inject_read_error();
+	if (pos == 0)
+		pos = vfs_llseek(file, *_start, SEEK_DATA);
+	if (pos < 0 && pos >= (loff_t)-MAX_ERRNO) {
+		if (pos == -ENXIO)
+			goto check_space; /* Unallocated tail */
+		trace_cachefiles_io_error(object, file_inode(file), pos,
+					  cachefiles_trace_seek_error);
+		return pos;
+	}
+	if ((u64)pos >= (u64)*_start + *_len)
+		goto check_space; /* Unallocated region */
+
+	/* We have a block that's at least partially filled - if we're low on
+	 * space, we need to see if it's fully allocated.  If it's not, we may
+	 * want to cull it.
+	 */
+	if (cachefiles_has_space(cache, 0, *_len / PAGE_SIZE,
+				 cachefiles_has_space_check) == 0)
+		return 0; /* Enough space to simply overwrite the whole block */
+
+	pos = cachefiles_inject_read_error();
+	if (pos == 0)
+		pos = vfs_llseek(file, *_start, SEEK_HOLE);
+	if (pos < 0 && pos >= (loff_t)-MAX_ERRNO) {
+		trace_cachefiles_io_error(object, file_inode(file), pos,
+					  cachefiles_trace_seek_error);
+		return pos;
+	}
+	if ((u64)pos >= (u64)*_start + *_len)
+		return 0; /* Fully allocated */
+
+	/* Partially allocated, but insufficient space: cull. */
+	fscache_count_no_write_space();
+	ret = cachefiles_inject_remove_error();
+	if (ret == 0)
+		ret = vfs_fallocate(file, FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE,
+				    *_start, *_len);
+	if (ret < 0) {
+		trace_cachefiles_io_error(object, file_inode(file), ret,
+					  cachefiles_trace_fallocate_error);
+		cachefiles_io_error_obj(object,
+					"CacheFiles: fallocate failed (%d)\n", ret);
+		ret = -EIO;
+	}
+
+	return ret;
+
+check_space:
+	return cachefiles_has_space(cache, 0, *_len / PAGE_SIZE,
+				    cachefiles_has_space_for_write);
+}
+
+static int cachefiles_prepare_write(struct netfs_cache_resources *cres,
+				    loff_t *_start, size_t *_len, loff_t i_size,
+				    bool no_space_allocated_yet)
+{
+	struct cachefiles_object *object = cachefiles_cres_object(cres);
+	struct cachefiles_cache *cache = object->volume->cache;
+	const struct cred *saved_cred;
+	int ret;
+
+	if (!cachefiles_cres_file(cres)) {
+		if (!fscache_wait_for_operation(cres, FSCACHE_WANT_WRITE))
+			return -ENOBUFS;
+		if (!cachefiles_cres_file(cres))
+			return -ENOBUFS;
+	}
+
+	cachefiles_begin_secure(cache, &saved_cred);
+	ret = __cachefiles_prepare_write(cres, _start, _len, i_size,
+					 no_space_allocated_yet);
+	cachefiles_end_secure(cache, saved_cred);
+	return ret;
 }
 
 /*
@@ -351,19 +532,11 @@ static int cachefiles_prepare_write(struct netfs_cache_resources *cres,
  */
 static void cachefiles_end_operation(struct netfs_cache_resources *cres)
 {
-	struct fscache_retrieval *op = cres->cache_priv;
-	struct file *file = cres->cache_priv2;
-
-	_enter("");
+	struct file *file = cachefiles_cres_file(cres);
 
 	if (file)
 		fput(file);
-	if (op) {
-		fscache_op_complete(&op->op, false);
-		fscache_put_retrieval(op);
-	}
-
-	_leave("");
+	fscache_end_cookie_access(fscache_cres_cookie(cres), fscache_access_io_end);
 }
 
 static const struct netfs_cache_ops cachefiles_netfs_cache_ops = {
@@ -377,44 +550,25 @@ static const struct netfs_cache_ops cachefiles_netfs_cache_ops = {
 /*
  * Open the cache file when beginning a cache operation.
  */
-int cachefiles_begin_read_operation(struct netfs_read_request *rreq,
-				    struct fscache_retrieval *op)
+bool cachefiles_begin_operation(struct netfs_cache_resources *cres,
+				enum fscache_want_state want_state)
 {
-	struct cachefiles_object *object;
-	struct cachefiles_cache *cache;
-	struct path path;
-	struct file *file;
+	struct cachefiles_object *object = cachefiles_cres_object(cres);
+
+	if (!cachefiles_cres_file(cres)) {
+		cres->ops = &cachefiles_netfs_cache_ops;
+		if (object->file) {
+			spin_lock(&object->lock);
+			if (!cres->cache_priv2 && object->file)
+				cres->cache_priv2 = get_file(object->file);
+			spin_unlock(&object->lock);
+		}
+	}
 
-	_enter("");
-
-	object = container_of(op->op.object,
-			      struct cachefiles_object, fscache);
-	cache = container_of(object->fscache.cache,
-			     struct cachefiles_cache, cache);
-
-	path.mnt = cache->mnt;
-	path.dentry = object->backer;
-	file = open_with_fake_path(&path, O_RDWR | O_LARGEFILE | O_DIRECT,
-				   d_inode(object->backer), cache->cache_cred);
-	if (IS_ERR(file))
-		return PTR_ERR(file);
-	if (!S_ISREG(file_inode(file)->i_mode))
-		goto error_file;
-	if (unlikely(!file->f_op->read_iter) ||
-	    unlikely(!file->f_op->write_iter)) {
-		pr_notice("Cache does not support read_iter and write_iter\n");
-		goto error_file;
+	if (!cachefiles_cres_file(cres) && want_state != FSCACHE_WANT_PARAMS) {
+		pr_err("failed to get cres->file\n");
+		return false;
 	}
 
-	fscache_get_retrieval(op);
-	rreq->cache_resources.cache_priv = op;
-	rreq->cache_resources.cache_priv2 = file;
-	rreq->cache_resources.ops = &cachefiles_netfs_cache_ops;
-	rreq->cache_resources.debug_id = object->fscache.debug_id;
-	_leave("");
-	return 0;
-
-error_file:
-	fput(file);
-	return -EIO;
+	return true;
 }
diff --git a/fs/cachefiles/key.c b/fs/cachefiles/key.c
index 7f94efc97e23..bf935e25bdbe 100644
--- a/fs/cachefiles/key.c
+++ b/fs/cachefiles/key.c
@@ -1,7 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 /* Key to pathname encoder
  *
- * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
+ * Copyright (C) 2021 Red Hat, Inc. All Rights Reserved.
  * Written by David Howells (dhowells@redhat.com)
  */
 
@@ -22,134 +22,117 @@ static const char cachefiles_filecharmap[256] = {
 	[48 ... 127] = 1,		/* '0' -> '~' */
 };
 
+static inline unsigned int how_many_hex_digits(unsigned int x)
+{
+	return x ? round_up(ilog2(x) + 1, 4) / 4 : 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)
+ * - the key may be up to NAME_MAX 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)
+bool cachefiles_cook_key(struct cachefiles_object *object)
 {
-	unsigned char csum, ch;
-	unsigned int acc;
-	char *key;
-	int loop, len, max, seg, mark, print;
+	const u8 *key = fscache_get_key(object->cookie), *kend;
+	unsigned char ch;
+	unsigned int acc, i, n, nle, nbe, keylen = object->cookie->key_len;
+	unsigned int b64len, len, print, pad;
+	char *name, sep;
 
-	_enter(",%d", keylen);
+	_enter(",%u,%*phN", keylen, keylen, key);
 
-	BUG_ON(keylen < 2 || keylen > 514);
+	BUG_ON(keylen > NAME_MAX - 3);
 
-	csum = raw[0] + raw[1];
 	print = 1;
-	for (loop = 2; loop < keylen; loop++) {
-		ch = raw[loop];
-		csum += ch;
+	for (i = 0; i < keylen; i++) {
+		ch = key[i];
 		print &= cachefiles_filecharmap[ch];
 	}
 
+	/* If the path is usable ASCII, then we render it directly */
 	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 */
+		len = 1 + keylen;
+		name = kmalloc(len + 1, GFP_KERNEL);
+		if (!name)
+			return false;
+
+		name[0] = 'D'; /* Data object type, string encoding */
+		memcpy(name + 1, key, keylen);
+		goto success;
 	}
 
-	max += 1;	/* 2nd NUL on end */
-
-	_debug("max: %d", max);
-
-	key = kmalloc(max, cachefiles_gfp);
-	if (!key)
-		return NULL;
-
-	len = 0;
-
-	/* build the cooked key */
-	sprintf(key, "@%02x%c+", (unsigned) csum, 0);
-	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++] = '\0';
-				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++] = '\0';
-				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);
-		}
+	/* See if it makes sense to encode it as "hex,hex,hex" for each 32-bit
+	 * chunk.  We rely on the key having been padded out to a whole number
+	 * of 32-bit words.
+	 */
+	n = round_up(keylen, 4);
+	nbe = nle = 0;
+	for (i = 0; i < n; i += 4) {
+		u32 be = be32_to_cpu(*(__be32 *)(key + i));
+		u32 le = le32_to_cpu(*(__le32 *)(key + i));
+
+		nbe += 1 + how_many_hex_digits(be);
+		nle += 1 + how_many_hex_digits(le);
+	}
 
-		switch (type) {
-		case FSCACHE_COOKIE_TYPE_INDEX:		type = 'J';	break;
-		case FSCACHE_COOKIE_TYPE_DATAFILE:	type = 'E';	break;
-		default:				type = 'T';	break;
+	b64len = DIV_ROUND_UP(keylen, 3);
+	pad = b64len * 3 - keylen;
+	b64len = 2 + b64len * 4; /* Length if we base64-encode it */
+	_debug("len=%u nbe=%u nle=%u b64=%u", keylen, nbe, nle, b64len);
+	if (nbe < b64len || nle < b64len) {
+		unsigned int nlen = min(nbe, nle) + 1;
+		name = kmalloc(nlen, GFP_KERNEL);
+		if (!name)
+			return false;
+		sep = (nbe <= nle) ? 'S' : 'T'; /* Encoding indicator */
+		len = 0;
+		for (i = 0; i < n; i += 4) {
+			u32 x;
+			if (nbe <= nle)
+				x = be32_to_cpu(*(__be32 *)(key + i));
+			else
+				x = le32_to_cpu(*(__le32 *)(key + i));
+			name[len++] = sep;
+			if (x != 0)
+				len += snprintf(name + len, nlen - len, "%x", x);
+			sep = ',';
 		}
+		goto success;
 	}
 
-	key[mark] = type;
-	key[len++] = 0;
-	key[len] = 0;
+	/* We need to base64-encode it */
+	name = kmalloc(b64len + 1, GFP_KERNEL);
+	if (!name)
+		return false;
+
+	name[0] = 'E';
+	name[1] = '0' + pad;
+	len = 2;
+	kend = key + keylen;
+	do {
+		acc  = *key++;
+		if (key < kend) {
+			acc |= *key++ << 8;
+			if (key < kend)
+				acc |= *key++ << 16;
+		}
 
-	_leave(" = %s %d", key, len);
-	return key;
+		name[len++] = cachefiles_charmap[acc & 63];
+		acc >>= 6;
+		name[len++] = cachefiles_charmap[acc & 63];
+		acc >>= 6;
+		name[len++] = cachefiles_charmap[acc & 63];
+		acc >>= 6;
+		name[len++] = cachefiles_charmap[acc & 63];
+	} while (key < kend);
+
+success:
+	name[len] = 0;
+	object->d_name = name;
+	object->d_name_len = len;
+	_leave(" = %s", object->d_name);
+	return true;
 }
diff --git a/fs/cachefiles/main.c b/fs/cachefiles/main.c
index 9c8d34c49b12..3f369c6f816d 100644
--- a/fs/cachefiles/main.c
+++ b/fs/cachefiles/main.c
@@ -2,7 +2,7 @@
 /* Network filesystem caching backend to use cache files on a premounted
  * filesystem
  *
- * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
+ * Copyright (C) 2021 Red Hat, Inc. All Rights Reserved.
  * Written by David Howells (dhowells@redhat.com)
  */
 
@@ -18,6 +18,8 @@
 #include <linux/statfs.h>
 #include <linux/sysctl.h>
 #include <linux/miscdevice.h>
+#include <linux/netfs.h>
+#include <trace/events/netfs.h>
 #define CREATE_TRACE_POINTS
 #include "internal.h"
 
@@ -37,14 +39,6 @@ static struct miscdevice cachefiles_dev = {
 	.fops	= &cachefiles_daemon_fops,
 };
 
-static void cachefiles_object_init_once(void *_object)
-{
-	struct cachefiles_object *object = _object;
-
-	memset(object, 0, sizeof(*object));
-	spin_lock_init(&object->work_lock);
-}
-
 /*
  * initialise the fs caching module
  */
@@ -52,6 +46,9 @@ static int __init cachefiles_init(void)
 {
 	int ret;
 
+	ret = cachefiles_register_error_injection();
+	if (ret < 0)
+		goto error_einj;
 	ret = misc_register(&cachefiles_dev);
 	if (ret < 0)
 		goto error_dev;
@@ -61,9 +58,7 @@ static int __init cachefiles_init(void)
 	cachefiles_object_jar =
 		kmem_cache_create("cachefiles_object_jar",
 				  sizeof(struct cachefiles_object),
-				  0,
-				  SLAB_HWCACHE_ALIGN,
-				  cachefiles_object_init_once);
+				  0, SLAB_HWCACHE_ALIGN, NULL);
 	if (!cachefiles_object_jar) {
 		pr_notice("Failed to allocate an object jar\n");
 		goto error_object_jar;
@@ -75,6 +70,8 @@ static int __init cachefiles_init(void)
 error_object_jar:
 	misc_deregister(&cachefiles_dev);
 error_dev:
+	cachefiles_unregister_error_injection();
+error_einj:
 	pr_err("failed to register: %d\n", ret);
 	return ret;
 }
@@ -90,6 +87,7 @@ static void __exit cachefiles_exit(void)
 
 	kmem_cache_destroy(cachefiles_object_jar);
 	misc_deregister(&cachefiles_dev);
+	cachefiles_unregister_error_injection();
 }
 
 module_exit(cachefiles_exit);
diff --git a/fs/cachefiles/namei.c b/fs/cachefiles/namei.c
index a9aca5ab5970..9bd692870617 100644
--- a/fs/cachefiles/namei.c
+++ b/fs/cachefiles/namei.c
@@ -1,295 +1,268 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 /* CacheFiles path walking and related routines
  *
- * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
+ * Copyright (C) 2021 Red Hat, Inc. All Rights Reserved.
  * Written by David Howells (dhowells@redhat.com)
  */
 
-#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 <linux/security.h>
-#include <linux/slab.h>
 #include "internal.h"
 
-#define CACHEFILES_KEYBUF_SIZE 512
-
 /*
- * dump debugging info about an object
+ * Mark the backing file as being a cache file if it's not already in use.  The
+ * mark tells the culling request command that it's not allowed to cull the
+ * file or directory.  The caller must hold the inode lock.
  */
-static noinline
-void __cachefiles_printk_object(struct cachefiles_object *object,
-				const char *prefix)
+static bool __cachefiles_mark_inode_in_use(struct cachefiles_object *object,
+					   struct dentry *dentry)
 {
-	struct fscache_cookie *cookie;
-	const u8 *k;
-	unsigned loop;
-
-	pr_err("%sobject: OBJ%x\n", prefix, object->fscache.debug_id);
-	pr_err("%sobjstate=%s fl=%lx wbusy=%x ev=%lx[%lx]\n",
-	       prefix, object->fscache.state->name,
-	       object->fscache.flags, work_busy(&object->fscache.work),
-	       object->fscache.events, object->fscache.event_mask);
-	pr_err("%sops=%u inp=%u exc=%u\n",
-	       prefix, object->fscache.n_ops, object->fscache.n_in_progress,
-	       object->fscache.n_exclusive);
-	pr_err("%sparent=%x\n",
-	       prefix, object->fscache.parent ? object->fscache.parent->debug_id : 0);
-
-	spin_lock(&object->fscache.lock);
-	cookie = object->fscache.cookie;
-	if (cookie) {
-		pr_err("%scookie=%x [pr=%x nd=%p fl=%lx]\n",
-		       prefix,
-		       cookie->debug_id,
-		       cookie->parent ? cookie->parent->debug_id : 0,
-		       cookie->netfs_data,
-		       cookie->flags);
-		pr_err("%skey=[%u] '", prefix, cookie->key_len);
-		k = (cookie->key_len <= sizeof(cookie->inline_key)) ?
-			cookie->inline_key : cookie->key;
-		for (loop = 0; loop < cookie->key_len; loop++)
-			pr_cont("%02x", k[loop]);
-		pr_cont("'\n");
+	struct inode *inode = d_backing_inode(dentry);
+	bool can_use = false;
+
+	if (!(inode->i_flags & S_KERNEL_FILE)) {
+		inode->i_flags |= S_KERNEL_FILE;
+		trace_cachefiles_mark_active(object, inode);
+		can_use = true;
 	} else {
-		pr_err("%scookie=NULL\n", prefix);
+		pr_notice("cachefiles: Inode already in use: %pd\n", dentry);
 	}
-	spin_unlock(&object->fscache.lock);
+
+	return can_use;
 }
 
-/*
- * dump debugging info about a pair of objects
- */
-static noinline void cachefiles_printk_object(struct cachefiles_object *object,
-					      struct cachefiles_object *xobject)
+static bool cachefiles_mark_inode_in_use(struct cachefiles_object *object,
+					 struct dentry *dentry)
 {
-	if (object)
-		__cachefiles_printk_object(object, "");
-	if (xobject)
-		__cachefiles_printk_object(xobject, "x");
+	struct inode *inode = d_backing_inode(dentry);
+	bool can_use;
+
+	inode_lock(inode);
+	can_use = __cachefiles_mark_inode_in_use(object, dentry);
+	inode_unlock(inode);
+	return can_use;
 }
 
 /*
- * mark the owner of a dentry, if there is one, to indicate that that dentry
- * has been preemptively deleted
- * - the caller must hold the i_mutex on the dentry's parent as required to
- *   call vfs_unlink(), vfs_rmdir() or vfs_rename()
+ * Unmark a backing inode.  The caller must hold the inode lock.
  */
-static void cachefiles_mark_object_buried(struct cachefiles_cache *cache,
-					  struct dentry *dentry,
-					  enum fscache_why_object_killed why)
+static void __cachefiles_unmark_inode_in_use(struct cachefiles_object *object,
+					     struct dentry *dentry)
 {
-	struct cachefiles_object *object;
-	struct rb_node *p;
-
-	_enter(",'%pd'", dentry);
+	struct inode *inode = d_backing_inode(dentry);
 
-	write_lock(&cache->active_lock);
+	inode->i_flags &= ~S_KERNEL_FILE;
+	trace_cachefiles_mark_inactive(object, inode);
+}
 
-	p = cache->active_nodes.rb_node;
-	while (p) {
-		object = rb_entry(p, struct cachefiles_object, active_node);
-		if (object->dentry > dentry)
-			p = p->rb_left;
-		else if (object->dentry < dentry)
-			p = p->rb_right;
-		else
-			goto found_dentry;
+/*
+ * Unmark a backing inode and tell cachefilesd that there's something that can
+ * be culled.
+ */
+void cachefiles_unmark_inode_in_use(struct cachefiles_object *object,
+				    struct file *file)
+{
+	struct cachefiles_cache *cache = object->volume->cache;
+	struct inode *inode = file_inode(file);
+
+	if (inode) {
+		inode_lock(inode);
+		__cachefiles_unmark_inode_in_use(object, file->f_path.dentry);
+		inode_unlock(inode);
+
+		if (!test_bit(CACHEFILES_OBJECT_USING_TMPFILE, &object->flags)) {
+			atomic_long_add(inode->i_blocks, &cache->b_released);
+			if (atomic_inc_return(&cache->f_released))
+				cachefiles_state_changed(cache);
+		}
 	}
+}
 
-	write_unlock(&cache->active_lock);
-	trace_cachefiles_mark_buried(NULL, dentry, why);
-	_leave(" [no owner]");
-	return;
+/*
+ * get a subdirectory
+ */
+struct dentry *cachefiles_get_directory(struct cachefiles_cache *cache,
+					struct dentry *dir,
+					const char *dirname,
+					bool *_is_new)
+{
+	struct dentry *subdir;
+	struct path path;
+	int ret;
 
-	/* found the dentry for  */
-found_dentry:
-	kdebug("preemptive burial: OBJ%x [%s] %pd",
-	       object->fscache.debug_id,
-	       object->fscache.state->name,
-	       dentry);
+	_enter(",,%s", dirname);
 
-	trace_cachefiles_mark_buried(object, dentry, why);
+	/* search the current directory for the element name */
+	inode_lock_nested(d_inode(dir), I_MUTEX_PARENT);
 
-	if (fscache_object_is_live(&object->fscache)) {
-		pr_err("\n");
-		pr_err("Error: Can't preemptively bury live object\n");
-		cachefiles_printk_object(object, NULL);
-	} else {
-		if (why != FSCACHE_OBJECT_IS_STALE)
-			fscache_object_mark_killed(&object->fscache, why);
+retry:
+	ret = cachefiles_inject_read_error();
+	if (ret == 0)
+		subdir = lookup_one_len(dirname, dir, strlen(dirname));
+	else
+		subdir = ERR_PTR(ret);
+	if (IS_ERR(subdir)) {
+		trace_cachefiles_vfs_error(NULL, d_backing_inode(dir),
+					   PTR_ERR(subdir),
+					   cachefiles_trace_lookup_error);
+		if (PTR_ERR(subdir) == -ENOMEM)
+			goto nomem_d_alloc;
+		goto lookup_error;
 	}
 
-	write_unlock(&cache->active_lock);
-	_leave(" [owner marked]");
-}
+	_debug("subdir -> %pd %s",
+	       subdir, d_backing_inode(subdir) ? "positive" : "negative");
 
-/*
- * record the fact that an object is now active
- */
-static int 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;
+	/* we need to create the subdir if it doesn't exist yet */
+	if (d_is_negative(subdir)) {
+		ret = cachefiles_has_space(cache, 1, 0,
+					   cachefiles_has_space_for_create);
+		if (ret < 0)
+			goto mkdir_error;
 
-	_enter(",%x", object->fscache.debug_id);
+		_debug("attempt mkdir");
 
-try_again:
-	write_lock(&cache->active_lock);
+		path.mnt = cache->mnt;
+		path.dentry = dir;
+		ret = security_path_mkdir(&path, subdir, 0700);
+		if (ret < 0)
+			goto mkdir_error;
+		ret = cachefiles_inject_write_error();
+		if (ret == 0)
+			ret = vfs_mkdir(&init_user_ns, d_inode(dir), subdir, 0700);
+		if (ret < 0) {
+			trace_cachefiles_vfs_error(NULL, d_inode(dir), ret,
+						   cachefiles_trace_mkdir_error);
+			goto mkdir_error;
+		}
 
-	dentry = object->dentry;
-	trace_cachefiles_mark_active(object, dentry);
+		if (unlikely(d_unhashed(subdir))) {
+			cachefiles_put_directory(subdir);
+			goto retry;
+		}
+		ASSERT(d_backing_inode(subdir));
 
-	if (test_and_set_bit(CACHEFILES_OBJECT_ACTIVE, &object->flags)) {
-		pr_err("Error: Object already active\n");
-		cachefiles_printk_object(object, NULL);
-		BUG();
+		_debug("mkdir -> %pd{ino=%lu}",
+		       subdir, d_backing_inode(subdir)->i_ino);
+		if (_is_new)
+			*_is_new = true;
 	}
 
-	_p = &cache->active_nodes.rb_node;
-	while (*_p) {
-		_parent = *_p;
-		xobject = rb_entry(_parent,
-				   struct cachefiles_object, active_node);
+	/* Tell rmdir() it's not allowed to delete the subdir */
+	inode_lock(d_inode(subdir));
+	inode_unlock(d_inode(dir));
 
-		ASSERT(xobject != object);
+	if (!__cachefiles_mark_inode_in_use(NULL, subdir))
+		goto mark_error;
 
-		if (xobject->dentry > dentry)
-			_p = &(*_p)->rb_left;
-		else if (xobject->dentry < dentry)
-			_p = &(*_p)->rb_right;
-		else
-			goto wait_for_old_object;
-	}
+	inode_unlock(d_inode(subdir));
 
-	rb_link_node(&object->active_node, _parent, _p);
-	rb_insert_color(&object->active_node, &cache->active_nodes);
+	/* we need to make sure the subdir is a directory */
+	ASSERT(d_backing_inode(subdir));
 
-	write_unlock(&cache->active_lock);
-	_leave(" = 0");
-	return 0;
+	if (!d_can_lookup(subdir)) {
+		pr_err("%s is not a directory\n", dirname);
+		ret = -EIO;
+		goto check_error;
+	}
 
-	/* an old object from a previous incarnation is hogging the slot - we
-	 * need to wait for it to be destroyed */
-wait_for_old_object:
-	trace_cachefiles_wait_active(object, dentry, xobject);
-	clear_bit(CACHEFILES_OBJECT_ACTIVE, &object->flags);
-
-	if (fscache_object_is_live(&xobject->fscache)) {
-		pr_err("\n");
-		pr_err("Error: Unexpected object collision\n");
-		cachefiles_printk_object(object, xobject);
-	}
-	atomic_inc(&xobject->usage);
-	write_unlock(&cache->active_lock);
-
-	if (test_bit(CACHEFILES_OBJECT_ACTIVE, &xobject->flags)) {
-		wait_queue_head_t *wq;
-
-		signed long timeout = 60 * HZ;
-		wait_queue_entry_t wait;
-		bool requeue;
-
-		/* if the object we're waiting for is queued for processing,
-		 * then just put ourselves on the queue behind it */
-		if (work_pending(&xobject->fscache.work)) {
-			_debug("queue OBJ%x behind OBJ%x immediately",
-			       object->fscache.debug_id,
-			       xobject->fscache.debug_id);
-			goto requeue;
-		}
+	ret = -EPERM;
+	if (!(d_backing_inode(subdir)->i_opflags & IOP_XATTR) ||
+	    !d_backing_inode(subdir)->i_op->lookup ||
+	    !d_backing_inode(subdir)->i_op->mkdir ||
+	    !d_backing_inode(subdir)->i_op->rename ||
+	    !d_backing_inode(subdir)->i_op->rmdir ||
+	    !d_backing_inode(subdir)->i_op->unlink)
+		goto check_error;
 
-		/* otherwise we sleep until either the object we're waiting for
-		 * is done, or the fscache_object is congested */
-		wq = bit_waitqueue(&xobject->flags, CACHEFILES_OBJECT_ACTIVE);
-		init_wait(&wait);
-		requeue = false;
-		do {
-			prepare_to_wait(wq, &wait, TASK_UNINTERRUPTIBLE);
-			if (!test_bit(CACHEFILES_OBJECT_ACTIVE, &xobject->flags))
-				break;
-
-			requeue = fscache_object_sleep_till_congested(&timeout);
-		} while (timeout > 0 && !requeue);
-		finish_wait(wq, &wait);
-
-		if (requeue &&
-		    test_bit(CACHEFILES_OBJECT_ACTIVE, &xobject->flags)) {
-			_debug("queue OBJ%x behind OBJ%x after wait",
-			       object->fscache.debug_id,
-			       xobject->fscache.debug_id);
-			goto requeue;
-		}
+	_leave(" = [%lu]", d_backing_inode(subdir)->i_ino);
+	return subdir;
 
-		if (timeout <= 0) {
-			pr_err("\n");
-			pr_err("Error: Overlong wait for old active object to go away\n");
-			cachefiles_printk_object(object, xobject);
-			goto requeue;
-		}
-	}
+check_error:
+	cachefiles_put_directory(subdir);
+	_leave(" = %d [check]", ret);
+	return ERR_PTR(ret);
 
-	ASSERT(!test_bit(CACHEFILES_OBJECT_ACTIVE, &xobject->flags));
+mark_error:
+	inode_unlock(d_inode(subdir));
+	dput(subdir);
+	return ERR_PTR(-EBUSY);
 
-	cache->cache.ops->put_object(&xobject->fscache,
-		(enum fscache_obj_ref_trace)cachefiles_obj_put_wait_retry);
-	goto try_again;
+mkdir_error:
+	inode_unlock(d_inode(dir));
+	dput(subdir);
+	pr_err("mkdir %s failed with error %d\n", dirname, ret);
+	return ERR_PTR(ret);
+
+lookup_error:
+	inode_unlock(d_inode(dir));
+	ret = PTR_ERR(subdir);
+	pr_err("Lookup %s failed with error %d\n", dirname, ret);
+	return ERR_PTR(ret);
 
-requeue:
-	cache->cache.ops->put_object(&xobject->fscache,
-		(enum fscache_obj_ref_trace)cachefiles_obj_put_wait_timeo);
-	_leave(" = -ETIMEDOUT");
-	return -ETIMEDOUT;
+nomem_d_alloc:
+	inode_unlock(d_inode(dir));
+	_leave(" = -ENOMEM");
+	return ERR_PTR(-ENOMEM);
 }
 
 /*
- * Mark an object as being inactive.
+ * Put a subdirectory.
  */
-void cachefiles_mark_object_inactive(struct cachefiles_cache *cache,
-				     struct cachefiles_object *object,
-				     blkcnt_t i_blocks)
+void cachefiles_put_directory(struct dentry *dir)
 {
-	struct dentry *dentry = object->dentry;
-	struct inode *inode = d_backing_inode(dentry);
-
-	trace_cachefiles_mark_inactive(object, dentry, inode);
+	if (dir) {
+		inode_lock(dir->d_inode);
+		__cachefiles_unmark_inode_in_use(NULL, dir);
+		inode_unlock(dir->d_inode);
+		dput(dir);
+	}
+}
 
-	write_lock(&cache->active_lock);
-	rb_erase(&object->active_node, &cache->active_nodes);
-	clear_bit(CACHEFILES_OBJECT_ACTIVE, &object->flags);
-	write_unlock(&cache->active_lock);
+/*
+ * Remove a regular file from the cache.
+ */
+static int cachefiles_unlink(struct cachefiles_cache *cache,
+			     struct cachefiles_object *object,
+			     struct dentry *dir, struct dentry *dentry,
+			     enum fscache_why_object_killed why)
+{
+	struct path path = {
+		.mnt	= cache->mnt,
+		.dentry	= dir,
+	};
+	int ret;
 
-	wake_up_bit(&object->flags, CACHEFILES_OBJECT_ACTIVE);
+	trace_cachefiles_unlink(object, dentry, why);
+	ret = security_path_unlink(&path, dentry);
+	if (ret < 0) {
+		cachefiles_io_error(cache, "Unlink security error");
+		return ret;
+	}
 
-	/* This object can now be culled, so we need to let the daemon know
-	 * that there is something it can remove if it needs to.
-	 */
-	atomic_long_add(i_blocks, &cache->b_released);
-	if (atomic_inc_return(&cache->f_released))
-		cachefiles_state_changed(cache);
+	ret = cachefiles_inject_remove_error();
+	if (ret == 0) {
+		ret = vfs_unlink(&init_user_ns, d_backing_inode(dir), dentry, NULL);
+		if (ret == -EIO)
+			cachefiles_io_error(cache, "Unlink failed");
+	}
+	if (ret != 0)
+		trace_cachefiles_vfs_error(object, d_backing_inode(dir), ret,
+					   cachefiles_trace_unlink_error);
+	return ret;
 }
 
 /*
- * delete an object representation from the cache
- * - file backed objects are unlinked
- * - directory backed objects are stuffed into the graveyard for userspace to
+ * 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 cachefiles_object *object,
-				  struct dentry *dir,
-				  struct dentry *rep,
-				  bool preemptive,
-				  enum fscache_why_object_killed why)
+int cachefiles_bury_object(struct cachefiles_cache *cache,
+			   struct cachefiles_object *object,
+			   struct dentry *dir,
+			   struct dentry *rep,
+			   enum fscache_why_object_killed why)
 {
 	struct dentry *grave, *trap;
 	struct path path, path_to_graveyard;
@@ -298,29 +271,21 @@ static int cachefiles_bury_object(struct cachefiles_cache *cache,
 
 	_enter(",'%pd','%pd'", dir, rep);
 
+	if (rep->d_parent != dir) {
+		inode_unlock(d_inode(dir));
+		_leave(" = -ESTALE");
+		return -ESTALE;
+	}
+
 	/* non-directories can just be unlinked */
 	if (!d_is_dir(rep)) {
-		_debug("unlink stale object");
-
-		path.mnt = cache->mnt;
-		path.dentry = dir;
-		ret = security_path_unlink(&path, rep);
-		if (ret < 0) {
-			cachefiles_io_error(cache, "Unlink security error");
-		} else {
-			trace_cachefiles_unlink(object, rep, why);
-			ret = vfs_unlink(&init_user_ns, d_inode(dir), rep,
-					 NULL);
-
-			if (preemptive)
-				cachefiles_mark_object_buried(cache, rep, why);
-		}
+		dget(rep); /* Stop the dentry being negated if it's only pinned
+			    * by a file struct.
+			    */
+		ret = cachefiles_unlink(cache, object, dir, rep, why);
+		dput(rep);
 
 		inode_unlock(d_inode(dir));
-
-		if (ret == -EIO)
-			cachefiles_io_error(cache, "Unlink failed");
-
 		_leave(" = %d", ret);
 		return ret;
 	}
@@ -368,14 +333,16 @@ try_again:
 	grave = lookup_one_len(nbuffer, cache->graveyard, strlen(nbuffer));
 	if (IS_ERR(grave)) {
 		unlock_rename(cache->graveyard, dir);
+		trace_cachefiles_vfs_error(object, d_inode(cache->graveyard),
+					   PTR_ERR(grave),
+					   cachefiles_trace_lookup_error);
 
 		if (PTR_ERR(grave) == -ENOMEM) {
 			_leave(" = -ENOMEM");
 			return -ENOMEM;
 		}
 
-		cachefiles_io_error(cache, "Lookup error %ld",
-				    PTR_ERR(grave));
+		cachefiles_io_error(cache, "Lookup error %ld", PTR_ERR(grave));
 		return -EIO;
 	}
 
@@ -420,15 +387,18 @@ try_again:
 			.new_dentry	= grave,
 		};
 		trace_cachefiles_rename(object, rep, grave, why);
-		ret = vfs_rename(&rd);
+		ret = cachefiles_inject_read_error();
+		if (ret == 0)
+			ret = vfs_rename(&rd);
+		if (ret != 0)
+			trace_cachefiles_vfs_error(object, d_inode(dir), ret,
+						   cachefiles_trace_rename_error);
 		if (ret != 0 && ret != -ENOMEM)
 			cachefiles_io_error(cache,
 					    "Rename failed with error %d", ret);
-
-		if (preemptive)
-			cachefiles_mark_object_buried(cache, rep, why);
 	}
 
+	__cachefiles_unmark_inode_in_use(object, rep);
 	unlock_rename(cache->graveyard, dir);
 	dput(grave);
 	_leave(" = 0");
@@ -436,493 +406,358 @@ try_again:
 }
 
 /*
- * delete an object representation from the cache
+ * Delete a cache file.
  */
-int cachefiles_delete_object(struct cachefiles_cache *cache,
-			     struct cachefiles_object *object)
+int cachefiles_delete_object(struct cachefiles_object *object,
+			     enum fscache_why_object_killed why)
 {
-	struct dentry *dir;
+	struct cachefiles_volume *volume = object->volume;
+	struct dentry *dentry = object->file->f_path.dentry;
+	struct dentry *fan = volume->fanout[(u8)object->cookie->key_hash];
 	int ret;
 
-	_enter(",OBJ%x{%pd}", object->fscache.debug_id, object->dentry);
-
-	ASSERT(object->dentry);
-	ASSERT(d_backing_inode(object->dentry));
-	ASSERT(object->dentry->d_parent);
+	_enter(",OBJ%x{%pD}", object->debug_id, object->file);
 
-	dir = dget_parent(object->dentry);
+	/* Stop the dentry being negated if it's only pinned by a file struct. */
+	dget(dentry);
 
-	inode_lock_nested(d_inode(dir), I_MUTEX_PARENT);
-
-	if (test_bit(FSCACHE_OBJECT_KILLED_BY_CACHE, &object->fscache.flags)) {
-		/* object allocation for the same key preemptively deleted this
-		 * object's file so that it could create its own file */
-		_debug("object preemptively buried");
-		inode_unlock(d_inode(dir));
-		ret = 0;
-	} else {
-		/* we need to check that our parent is _still_ our parent - it
-		 * may have been renamed */
-		if (dir == object->dentry->d_parent) {
-			ret = cachefiles_bury_object(cache, object, dir,
-						     object->dentry, false,
-						     FSCACHE_OBJECT_WAS_RETIRED);
-		} else {
-			/* it got moved, presumably by cachefilesd culling it,
-			 * so it's no longer in the key path and we can ignore
-			 * it */
-			inode_unlock(d_inode(dir));
-			ret = 0;
-		}
-	}
-
-	dput(dir);
-	_leave(" = %d", ret);
+	inode_lock_nested(d_backing_inode(fan), I_MUTEX_PARENT);
+	ret = cachefiles_unlink(volume->cache, object, fan, dentry, why);
+	inode_unlock(d_backing_inode(fan));
+	dput(dentry);
 	return ret;
 }
 
 /*
- * walk from the parent object to the child object through the backing
- * filesystem, creating directories as we go
+ * Create a temporary file and leave it unattached and un-xattr'd until the
+ * time comes to discard the object from memory.
  */
-int cachefiles_walk_to_object(struct cachefiles_object *parent,
-			      struct cachefiles_object *object,
-			      const char *key,
-			      struct cachefiles_xattr *auxdata)
+struct file *cachefiles_create_tmpfile(struct cachefiles_object *object)
 {
-	struct cachefiles_cache *cache;
-	struct dentry *dir, *next = NULL;
-	struct inode *inode;
+	struct cachefiles_volume *volume = object->volume;
+	struct cachefiles_cache *cache = volume->cache;
+	const struct cred *saved_cred;
+	struct dentry *fan = volume->fanout[(u8)object->cookie->key_hash];
+	struct file *file;
 	struct path path;
-	const char *name;
-	int ret, nlen;
-
-	_enter("OBJ%x{%pd},OBJ%x,%s,",
-	       parent->fscache.debug_id, parent->dentry,
-	       object->fscache.debug_id, key);
-
-	cache = container_of(parent->fscache.cache,
-			     struct cachefiles_cache, cache);
-	path.mnt = cache->mnt;
-
-	ASSERT(parent->dentry);
-	ASSERT(d_backing_inode(parent->dentry));
-
-	if (!(d_is_dir(parent->dentry))) {
-		// TODO: convert file to dir
-		_leave("looking up in none directory");
-		return -ENOBUFS;
-	}
-
-	dir = dget(parent->dentry);
-
-advance:
-	/* attempt to transit the first directory component */
-	name = key;
-	nlen = strlen(key);
-
-	/* key ends in a double NUL */
-	key = key + nlen + 1;
-	if (!*key)
-		key = NULL;
-
-lookup_again:
-	/* search the current directory for the element name */
-	_debug("lookup '%s'", name);
-
-	inode_lock_nested(d_inode(dir), I_MUTEX_PARENT);
+	uint64_t ni_size = object->cookie->object_size;
+	long ret;
 
-	next = lookup_one_len(name, dir, nlen);
-	if (IS_ERR(next)) {
-		trace_cachefiles_lookup(object, next, NULL);
-		goto lookup_error;
-	}
+	ni_size = round_up(ni_size, CACHEFILES_DIO_BLOCK_SIZE);
 
-	inode = d_backing_inode(next);
-	trace_cachefiles_lookup(object, next, inode);
-	_debug("next -> %pd %s", next, inode ? "positive" : "negative");
-
-	if (!key)
-		object->new = !inode;
-
-	/* if this element of the path doesn't exist, then the lookup phase
-	 * failed, and we can release any readers in the certain knowledge that
-	 * there's nothing for them to actually read */
-	if (d_is_negative(next))
-		fscache_object_lookup_negative(&object->fscache);
-
-	/* 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 (d_is_negative(next)) {
-			ret = cachefiles_has_space(cache, 1, 0);
-			if (ret < 0)
-				goto no_space_error;
-
-			path.dentry = dir;
-			ret = security_path_mkdir(&path, next, 0);
-			if (ret < 0)
-				goto create_error;
-			ret = vfs_mkdir(&init_user_ns, d_inode(dir), next, 0);
-			if (!key)
-				trace_cachefiles_mkdir(object, next, ret);
-			if (ret < 0)
-				goto create_error;
-
-			if (unlikely(d_unhashed(next))) {
-				dput(next);
-				inode_unlock(d_inode(dir));
-				goto lookup_again;
-			}
-			ASSERT(d_backing_inode(next));
-
-			_debug("mkdir -> %pd{ino=%lu}",
-			       next, d_backing_inode(next)->i_ino);
-
-		} else if (!d_can_lookup(next)) {
-			pr_err("inode %lu is not a directory\n",
-			       d_backing_inode(next)->i_ino);
-			ret = -ENOBUFS;
-			goto error;
-		}
+	cachefiles_begin_secure(cache, &saved_cred);
 
-	} else {
-		/* non-index objects start out life as files */
-		if (d_is_negative(next)) {
-			ret = cachefiles_has_space(cache, 1, 0);
-			if (ret < 0)
-				goto no_space_error;
-
-			path.dentry = dir;
-			ret = security_path_mknod(&path, next, S_IFREG, 0);
-			if (ret < 0)
-				goto create_error;
-			ret = vfs_create(&init_user_ns, d_inode(dir), next,
-					 S_IFREG, true);
-			trace_cachefiles_create(object, next, ret);
-			if (ret < 0)
-				goto create_error;
-
-			ASSERT(d_backing_inode(next));
-
-			_debug("create -> %pd{ino=%lu}",
-			       next, d_backing_inode(next)->i_ino);
-
-		} else if (!d_can_lookup(next) &&
-			   !d_is_reg(next)
-			   ) {
-			pr_err("inode %lu is not a file or directory\n",
-			       d_backing_inode(next)->i_ino);
-			ret = -ENOBUFS;
-			goto error;
+	path.mnt = cache->mnt;
+	ret = cachefiles_inject_write_error();
+	if (ret == 0)
+		path.dentry = vfs_tmpfile(&init_user_ns, fan, S_IFREG, O_RDWR);
+	else
+		path.dentry = ERR_PTR(ret);
+	if (IS_ERR(path.dentry)) {
+		trace_cachefiles_vfs_error(object, d_inode(fan), PTR_ERR(path.dentry),
+					   cachefiles_trace_tmpfile_error);
+		if (PTR_ERR(path.dentry) == -EIO)
+			cachefiles_io_error_obj(object, "Failed to create tmpfile");
+		file = ERR_CAST(path.dentry);
+		goto out;
+	}
+
+	trace_cachefiles_tmpfile(object, d_backing_inode(path.dentry));
+
+	if (!cachefiles_mark_inode_in_use(object, path.dentry)) {
+		file = ERR_PTR(-EBUSY);
+		goto out_dput;
+	}
+
+	if (ni_size > 0) {
+		trace_cachefiles_trunc(object, d_backing_inode(path.dentry), 0, ni_size,
+				       cachefiles_trunc_expand_tmpfile);
+		ret = cachefiles_inject_write_error();
+		if (ret == 0)
+			ret = vfs_truncate(&path, ni_size);
+		if (ret < 0) {
+			trace_cachefiles_vfs_error(
+				object, d_backing_inode(path.dentry), ret,
+				cachefiles_trace_trunc_error);
+			file = ERR_PTR(ret);
+			goto out_dput;
 		}
 	}
 
-	/* process the next component */
-	if (key) {
-		_debug("advance");
-		inode_unlock(d_inode(dir));
-		dput(dir);
-		dir = next;
-		next = NULL;
-		goto advance;
+	file = open_with_fake_path(&path, O_RDWR | O_LARGEFILE | O_DIRECT,
+				   d_backing_inode(path.dentry), cache->cache_cred);
+	if (IS_ERR(file)) {
+		trace_cachefiles_vfs_error(object, d_backing_inode(path.dentry),
+					   PTR_ERR(file),
+					   cachefiles_trace_open_error);
+		goto out_dput;
+	}
+	if (unlikely(!file->f_op->read_iter) ||
+	    unlikely(!file->f_op->write_iter)) {
+		fput(file);
+		pr_notice("Cache does not support read_iter and write_iter\n");
+		file = ERR_PTR(-EINVAL);
 	}
 
-	/* 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 '%pd'", next);
-
-		ret = cachefiles_check_object_xattr(object, auxdata);
-		if (ret == -ESTALE) {
-			/* delete the object (the deleter drops the directory
-			 * mutex) */
-			object->dentry = NULL;
+out_dput:
+	dput(path.dentry);
+out:
+	cachefiles_end_secure(cache, saved_cred);
+	return file;
+}
 
-			ret = cachefiles_bury_object(cache, object, dir, next,
-						     true,
-						     FSCACHE_OBJECT_IS_STALE);
-			dput(next);
-			next = NULL;
+/*
+ * Create a new file.
+ */
+static bool cachefiles_create_file(struct cachefiles_object *object)
+{
+	struct file *file;
+	int ret;
 
-			if (ret < 0)
-				goto delete_error;
+	ret = cachefiles_has_space(object->volume->cache, 1, 0,
+				   cachefiles_has_space_for_create);
+	if (ret < 0)
+		return false;
 
-			_debug("redo lookup");
-			fscache_object_retrying_stale(&object->fscache);
-			goto lookup_again;
-		}
-	}
+	file = cachefiles_create_tmpfile(object);
+	if (IS_ERR(file))
+		return false;
 
-	/* note that we're now using this object */
-	ret = cachefiles_mark_object_active(cache, object);
+	set_bit(FSCACHE_COOKIE_NEEDS_UPDATE, &object->cookie->flags);
+	set_bit(CACHEFILES_OBJECT_USING_TMPFILE, &object->flags);
+	_debug("create -> %pD{ino=%lu}", file, file_inode(file)->i_ino);
+	object->file = file;
+	return true;
+}
 
-	inode_unlock(d_inode(dir));
-	dput(dir);
-	dir = NULL;
+/*
+ * Open an existing file, checking its attributes and replacing it if it is
+ * stale.
+ */
+static bool cachefiles_open_file(struct cachefiles_object *object,
+				 struct dentry *dentry)
+{
+	struct cachefiles_cache *cache = object->volume->cache;
+	struct file *file;
+	struct path path;
+	int ret;
 
-	if (ret == -ETIMEDOUT)
-		goto mark_active_timed_out;
+	_enter("%pd", dentry);
 
-	_debug("=== OBTAINED_OBJECT ===");
+	if (!cachefiles_mark_inode_in_use(object, dentry))
+		return false;
 
-	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) */
-		path.dentry = next;
-		touch_atime(&path);
-	}
-
-	/* open a file interface onto a data file */
-	if (object->type != FSCACHE_COOKIE_TYPE_INDEX) {
-		if (d_is_reg(object->dentry)) {
-			const struct address_space_operations *aops;
-
-			ret = -EPERM;
-			aops = d_backing_inode(object->dentry)->i_mapping->a_ops;
-			if (!aops->bmap)
-				goto check_error;
-			if (object->dentry->d_sb->s_blocksize > PAGE_SIZE)
-				goto check_error;
-
-			object->backer = object->dentry;
-		} else {
-			BUG(); // TODO: open file in data-class subdir
-		}
+	/* We need to open a file interface onto a data file now as we can't do
+	 * it on demand because writeback called from do_exit() sees
+	 * current->fs == NULL - which breaks d_path() called from ext4 open.
+	 */
+	path.mnt = cache->mnt;
+	path.dentry = dentry;
+	file = open_with_fake_path(&path, O_RDWR | O_LARGEFILE | O_DIRECT,
+				   d_backing_inode(dentry), cache->cache_cred);
+	if (IS_ERR(file)) {
+		trace_cachefiles_vfs_error(object, d_backing_inode(dentry),
+					   PTR_ERR(file),
+					   cachefiles_trace_open_error);
+		goto error;
 	}
 
-	object->new = 0;
-	fscache_obtained_object(&object->fscache);
-
-	_leave(" = 0 [%lu]", d_backing_inode(object->dentry)->i_ino);
-	return 0;
-
-no_space_error:
-	fscache_object_mark_killed(&object->fscache, FSCACHE_OBJECT_NO_SPACE);
-create_error:
-	_debug("create error %d", ret);
-	if (ret == -EIO)
-		cachefiles_io_error(cache, "Create/mkdir failed");
-	goto error;
+	if (unlikely(!file->f_op->read_iter) ||
+	    unlikely(!file->f_op->write_iter)) {
+		pr_notice("Cache does not support read_iter and write_iter\n");
+		goto error_fput;
+	}
+	_debug("file -> %pd positive", dentry);
 
-mark_active_timed_out:
-	_debug("mark active timed out");
-	goto release_dentry;
+	ret = cachefiles_check_auxdata(object, file);
+	if (ret < 0)
+		goto check_failed;
 
-check_error:
-	_debug("check error %d", ret);
-	cachefiles_mark_object_inactive(
-		cache, object, d_backing_inode(object->dentry)->i_blocks);
-release_dentry:
-	dput(object->dentry);
-	object->dentry = NULL;
-	goto error_out;
-
-delete_error:
-	_debug("delete error %d", ret);
-	goto error_out2;
+	object->file = file;
 
-lookup_error:
-	_debug("lookup error %ld", PTR_ERR(next));
-	ret = PTR_ERR(next);
-	if (ret == -EIO)
-		cachefiles_io_error(cache, "Lookup failed");
-	next = NULL;
+	/* 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(&file->f_path);
+	dput(dentry);
+	return true;
+
+check_failed:
+	fscache_cookie_lookup_negative(object->cookie);
+	cachefiles_unmark_inode_in_use(object, file);
+	if (ret == -ESTALE) {
+		fput(file);
+		dput(dentry);
+		return cachefiles_create_file(object);
+	}
+error_fput:
+	fput(file);
 error:
-	inode_unlock(d_inode(dir));
-	dput(next);
-error_out2:
-	dput(dir);
-error_out:
-	_leave(" = error %d", -ret);
-	return ret;
+	dput(dentry);
+	return false;
 }
 
 /*
- * get a subdirectory
+ * walk from the parent object to the child object through the backing
+ * filesystem, creating directories as we go
  */
-struct dentry *cachefiles_get_directory(struct cachefiles_cache *cache,
-					struct dentry *dir,
-					const char *dirname)
+bool cachefiles_look_up_object(struct cachefiles_object *object)
 {
-	struct dentry *subdir;
-	struct path path;
+	struct cachefiles_volume *volume = object->volume;
+	struct dentry *dentry, *fan = volume->fanout[(u8)object->cookie->key_hash];
 	int ret;
 
-	_enter(",,%s", dirname);
-
-	/* search the current directory for the element name */
-	inode_lock(d_inode(dir));
-
-retry:
-	subdir = lookup_one_len(dirname, dir, strlen(dirname));
-	if (IS_ERR(subdir)) {
-		if (PTR_ERR(subdir) == -ENOMEM)
-			goto nomem_d_alloc;
-		goto lookup_error;
+	_enter("OBJ%x,%s,", object->debug_id, object->d_name);
+
+	/* Look up path "cache/vol/fanout/file". */
+	ret = cachefiles_inject_read_error();
+	if (ret == 0)
+		dentry = lookup_positive_unlocked(object->d_name, fan,
+						  object->d_name_len);
+	else
+		dentry = ERR_PTR(ret);
+	trace_cachefiles_lookup(object, dentry);
+	if (IS_ERR(dentry)) {
+		if (dentry == ERR_PTR(-ENOENT))
+			goto new_file;
+		if (dentry == ERR_PTR(-EIO))
+			cachefiles_io_error_obj(object, "Lookup failed");
+		return false;
+	}
+
+	if (!d_is_reg(dentry)) {
+		pr_err("%pd is not a file\n", dentry);
+		inode_lock_nested(d_inode(fan), I_MUTEX_PARENT);
+		ret = cachefiles_bury_object(volume->cache, object, fan, dentry,
+					     FSCACHE_OBJECT_IS_WEIRD);
+		dput(dentry);
+		if (ret < 0)
+			return false;
+		goto new_file;
 	}
 
-	_debug("subdir -> %pd %s",
-	       subdir, d_backing_inode(subdir) ? "positive" : "negative");
+	if (!cachefiles_open_file(object, dentry))
+		return false;
 
-	/* we need to create the subdir if it doesn't exist yet */
-	if (d_is_negative(subdir)) {
-		ret = cachefiles_has_space(cache, 1, 0);
-		if (ret < 0)
-			goto mkdir_error;
+	_leave(" = t [%lu]", file_inode(object->file)->i_ino);
+	return true;
 
-		_debug("attempt mkdir");
+new_file:
+	fscache_cookie_lookup_negative(object->cookie);
+	return cachefiles_create_file(object);
+}
 
-		path.mnt = cache->mnt;
-		path.dentry = dir;
-		ret = security_path_mkdir(&path, subdir, 0700);
-		if (ret < 0)
-			goto mkdir_error;
-		ret = vfs_mkdir(&init_user_ns, d_inode(dir), subdir, 0700);
-		if (ret < 0)
-			goto mkdir_error;
+/*
+ * Attempt to link a temporary file into its rightful place in the cache.
+ */
+bool cachefiles_commit_tmpfile(struct cachefiles_cache *cache,
+			       struct cachefiles_object *object)
+{
+	struct cachefiles_volume *volume = object->volume;
+	struct dentry *dentry, *fan = volume->fanout[(u8)object->cookie->key_hash];
+	bool success = false;
+	int ret;
 
-		if (unlikely(d_unhashed(subdir))) {
-			dput(subdir);
-			goto retry;
+	_enter(",%pD", object->file);
+
+	inode_lock_nested(d_inode(fan), I_MUTEX_PARENT);
+	ret = cachefiles_inject_read_error();
+	if (ret == 0)
+		dentry = lookup_one_len(object->d_name, fan, object->d_name_len);
+	else
+		dentry = ERR_PTR(ret);
+	if (IS_ERR(dentry)) {
+		trace_cachefiles_vfs_error(object, d_inode(fan), PTR_ERR(dentry),
+					   cachefiles_trace_lookup_error);
+		_debug("lookup fail %ld", PTR_ERR(dentry));
+		goto out_unlock;
+	}
+
+	if (!d_is_negative(dentry)) {
+		if (d_backing_inode(dentry) == file_inode(object->file)) {
+			success = true;
+			goto out_dput;
 		}
-		ASSERT(d_backing_inode(subdir));
 
-		_debug("mkdir -> %pd{ino=%lu}",
-		       subdir, d_backing_inode(subdir)->i_ino);
-	}
-
-	inode_unlock(d_inode(dir));
-
-	/* we need to make sure the subdir is a directory */
-	ASSERT(d_backing_inode(subdir));
+		ret = cachefiles_unlink(volume->cache, object, fan, dentry,
+					FSCACHE_OBJECT_IS_STALE);
+		if (ret < 0)
+			goto out_dput;
 
-	if (!d_can_lookup(subdir)) {
-		pr_err("%s is not a directory\n", dirname);
-		ret = -EIO;
-		goto check_error;
+		dput(dentry);
+		ret = cachefiles_inject_read_error();
+		if (ret == 0)
+			dentry = lookup_one_len(object->d_name, fan, object->d_name_len);
+		else
+			dentry = ERR_PTR(ret);
+		if (IS_ERR(dentry)) {
+			trace_cachefiles_vfs_error(object, d_inode(fan), PTR_ERR(dentry),
+						   cachefiles_trace_lookup_error);
+			_debug("lookup fail %ld", PTR_ERR(dentry));
+			goto out_unlock;
+		}
 	}
 
-	ret = -EPERM;
-	if (!(d_backing_inode(subdir)->i_opflags & IOP_XATTR) ||
-	    !d_backing_inode(subdir)->i_op->lookup ||
-	    !d_backing_inode(subdir)->i_op->mkdir ||
-	    !d_backing_inode(subdir)->i_op->create ||
-	    !d_backing_inode(subdir)->i_op->rename ||
-	    !d_backing_inode(subdir)->i_op->rmdir ||
-	    !d_backing_inode(subdir)->i_op->unlink)
-		goto check_error;
-
-	_leave(" = [%lu]", d_backing_inode(subdir)->i_ino);
-	return subdir;
-
-check_error:
-	dput(subdir);
-	_leave(" = %d [check]", ret);
-	return ERR_PTR(ret);
-
-mkdir_error:
-	inode_unlock(d_inode(dir));
-	dput(subdir);
-	pr_err("mkdir %s failed with error %d\n", dirname, ret);
-	return ERR_PTR(ret);
-
-lookup_error:
-	inode_unlock(d_inode(dir));
-	ret = PTR_ERR(subdir);
-	pr_err("Lookup %s failed with error %d\n", dirname, ret);
-	return ERR_PTR(ret);
-
-nomem_d_alloc:
-	inode_unlock(d_inode(dir));
-	_leave(" = -ENOMEM");
-	return ERR_PTR(-ENOMEM);
+	ret = cachefiles_inject_read_error();
+	if (ret == 0)
+		ret = vfs_link(object->file->f_path.dentry, &init_user_ns,
+			       d_inode(fan), dentry, NULL);
+	if (ret < 0) {
+		trace_cachefiles_vfs_error(object, d_inode(fan), ret,
+					   cachefiles_trace_link_error);
+		_debug("link fail %d", ret);
+	} else {
+		trace_cachefiles_link(object, file_inode(object->file));
+		spin_lock(&object->lock);
+		/* TODO: Do we want to switch the file pointer to the new dentry? */
+		clear_bit(CACHEFILES_OBJECT_USING_TMPFILE, &object->flags);
+		spin_unlock(&object->lock);
+		success = true;
+	}
+
+out_dput:
+	dput(dentry);
+out_unlock:
+	inode_unlock(d_inode(fan));
+	_leave(" = %u", success);
+	return success;
 }
 
 /*
- * find out if an object is in use or not
- * - if finds object and it's not in use:
- *   - returns a pointer to the object and a reference on it
- *   - returns with the directory locked
+ * Look up an inode to be checked or culled.  Return -EBUSY if the inode is
+ * marked in use.
  */
-static struct dentry *cachefiles_check_active(struct cachefiles_cache *cache,
-					      struct dentry *dir,
-					      char *filename)
+static struct dentry *cachefiles_lookup_for_cull(struct cachefiles_cache *cache,
+						 struct dentry *dir,
+						 char *filename)
 {
-	struct cachefiles_object *object;
-	struct rb_node *_n;
 	struct dentry *victim;
-	int ret;
-
-	//_enter(",%pd/,%s",
-	//       dir, filename);
+	int ret = -ENOENT;
 
-	/* look up the victim */
 	inode_lock_nested(d_inode(dir), I_MUTEX_PARENT);
 
 	victim = lookup_one_len(filename, dir, strlen(filename));
 	if (IS_ERR(victim))
 		goto lookup_error;
-
-	//_debug("victim -> %pd %s",
-	//       victim, d_backing_inode(victim) ? "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 (d_is_negative(victim)) {
-		inode_unlock(d_inode(dir));
-		dput(victim);
-		_leave(" = -ENOENT [absent]");
-		return ERR_PTR(-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);
-
-	//_leave(" = %pd", victim);
+	if (d_is_negative(victim))
+		goto lookup_put;
+	if (d_inode(victim)->i_flags & S_KERNEL_FILE)
+		goto lookup_busy;
 	return victim;
 
-object_in_use:
-	read_unlock(&cache->active_lock);
+lookup_busy:
+	ret = -EBUSY;
+lookup_put:
 	inode_unlock(d_inode(dir));
 	dput(victim);
-	//_leave(" = -EBUSY [in use]");
-	return ERR_PTR(-EBUSY);
+	return ERR_PTR(ret);
 
 lookup_error:
 	inode_unlock(d_inode(dir));
 	ret = PTR_ERR(victim);
-	if (ret == -ENOENT) {
-		/* file or dir now absent - probably retired by netfs */
-		_leave(" = -ESTALE [absent]");
-		return ERR_PTR(-ESTALE);
-	}
+	if (ret == -ENOENT)
+		return ERR_PTR(-ESTALE); /* Probably got retired by the netfs */
 
 	if (ret == -EIO) {
 		cachefiles_io_error(cache, "Lookup failed");
@@ -931,46 +766,46 @@ lookup_error:
 		ret = -EIO;
 	}
 
-	_leave(" = %d", ret);
 	return ERR_PTR(ret);
 }
 
 /*
- * cull an object if it's not in use
+ * 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 dentry *victim;
+	struct inode *inode;
 	int ret;
 
 	_enter(",%pd/,%s", dir, filename);
 
-	victim = cachefiles_check_active(cache, dir, filename);
+	victim = cachefiles_lookup_for_cull(cache, dir, filename);
 	if (IS_ERR(victim))
 		return PTR_ERR(victim);
 
-	_debug("victim -> %pd %s",
-	       victim, d_backing_inode(victim) ? "positive" : "negative");
-
-	/* 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);
+	/* check to see if someone is using this object */
+	inode = d_inode(victim);
+	inode_lock(inode);
+	if (inode->i_flags & S_KERNEL_FILE) {
+		ret = -EBUSY;
+	} else {
+		/* Stop the cache from picking it back up */
+		inode->i_flags |= S_KERNEL_FILE;
+		ret = 0;
+	}
+	inode_unlock(inode);
 	if (ret < 0)
 		goto error_unlock;
 
-	/*  actually remove the victim (drops the dir mutex) */
-	_debug("bury");
-
-	ret = cachefiles_bury_object(cache, NULL, dir, victim, false,
+	ret = cachefiles_bury_object(cache, NULL, dir, victim,
 				     FSCACHE_OBJECT_WAS_CULLED);
 	if (ret < 0)
 		goto error;
 
+	fscache_count_culled();
 	dput(victim);
 	_leave(" = 0");
 	return 0;
@@ -979,11 +814,8 @@ error_unlock:
 	inode_unlock(d_inode(dir));
 error:
 	dput(victim);
-	if (ret == -ENOENT) {
-		/* file or dir now absent - probably retired by netfs */
-		_leave(" = -ESTALE [absent]");
-		return -ESTALE;
-	}
+	if (ret == -ENOENT)
+		return -ESTALE; /* Probably got retired by the netfs */
 
 	if (ret != -ENOMEM) {
 		pr_err("Internal error: %d\n", ret);
@@ -995,7 +827,7 @@ error:
 }
 
 /*
- * find out if an object is in use or not
+ * Find out if an object is in use or not
  * - called only by cache manager daemon
  * - returns -EBUSY or 0 to indicate whether an object is in use or not
  */
@@ -1003,16 +835,13 @@ int cachefiles_check_in_use(struct cachefiles_cache *cache, struct dentry *dir,
 			    char *filename)
 {
 	struct dentry *victim;
+	int ret = 0;
 
-	//_enter(",%pd/,%s",
-	//       dir, filename);
-
-	victim = cachefiles_check_active(cache, dir, filename);
+	victim = cachefiles_lookup_for_cull(cache, dir, filename);
 	if (IS_ERR(victim))
 		return PTR_ERR(victim);
 
 	inode_unlock(d_inode(dir));
 	dput(victim);
-	//_leave(" = 0");
-	return 0;
+	return ret;
 }
diff --git a/fs/cachefiles/rdwr.c b/fs/cachefiles/rdwr.c
deleted file mode 100644
index fcf4f3b72923..000000000000
--- a/fs/cachefiles/rdwr.c
+++ /dev/null
@@ -1,972 +0,0 @@
-// SPDX-License-Identifier: GPL-2.0-or-later
-/* Storage object read/write
- *
- * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
- * Written by David Howells (dhowells@redhat.com)
- */
-
-#include <linux/mount.h>
-#include <linux/slab.h>
-#include <linux/file.h>
-#include <linux/swap.h>
-#include "internal.h"
-
-/*
- * detect wake up events generated by the unlocking of pages in which we're
- * interested
- * - we use this to detect read completion of backing pages
- * - the caller holds the waitqueue lock
- */
-static int cachefiles_read_waiter(wait_queue_entry_t *wait, unsigned mode,
-				  int sync, void *_key)
-{
-	struct cachefiles_one_read *monitor =
-		container_of(wait, struct cachefiles_one_read, monitor);
-	struct cachefiles_object *object;
-	struct fscache_retrieval *op = monitor->op;
-	struct wait_page_key *key = _key;
-	struct folio *folio = wait->private;
-
-	ASSERT(key);
-
-	_enter("{%lu},%u,%d,{%p,%u}",
-	       monitor->netfs_page->index, mode, sync,
-	       key->folio, key->bit_nr);
-
-	if (key->folio != folio || key->bit_nr != PG_locked)
-		return 0;
-
-	_debug("--- monitor %p %lx ---", folio, folio->flags);
-
-	if (!folio_test_uptodate(folio) && !folio_test_error(folio)) {
-		/* unlocked, not uptodate and not erronous? */
-		_debug("page probably truncated");
-	}
-
-	/* remove from the waitqueue */
-	list_del(&wait->entry);
-
-	/* move onto the action list and queue for FS-Cache thread pool */
-	ASSERT(op);
-
-	/* We need to temporarily bump the usage count as we don't own a ref
-	 * here otherwise cachefiles_read_copier() may free the op between the
-	 * monitor being enqueued on the op->to_do list and the op getting
-	 * enqueued on the work queue.
-	 */
-	fscache_get_retrieval(op);
-
-	object = container_of(op->op.object, struct cachefiles_object, fscache);
-	spin_lock(&object->work_lock);
-	list_add_tail(&monitor->op_link, &op->to_do);
-	fscache_enqueue_retrieval(op);
-	spin_unlock(&object->work_lock);
-
-	fscache_put_retrieval(op);
-	return 0;
-}
-
-/*
- * handle a probably truncated page
- * - check to see if the page is still relevant and reissue the read if
- *   possible
- * - return -EIO on error, -ENODATA if the page is gone, -EINPROGRESS if we
- *   must wait again and 0 if successful
- */
-static int cachefiles_read_reissue(struct cachefiles_object *object,
-				   struct cachefiles_one_read *monitor)
-{
-	struct address_space *bmapping = d_backing_inode(object->backer)->i_mapping;
-	struct page *backpage = monitor->back_page, *backpage2;
-	int ret;
-
-	_enter("{ino=%lx},{%lx,%lx}",
-	       d_backing_inode(object->backer)->i_ino,
-	       backpage->index, backpage->flags);
-
-	/* skip if the page was truncated away completely */
-	if (backpage->mapping != bmapping) {
-		_leave(" = -ENODATA [mapping]");
-		return -ENODATA;
-	}
-
-	backpage2 = find_get_page(bmapping, backpage->index);
-	if (!backpage2) {
-		_leave(" = -ENODATA [gone]");
-		return -ENODATA;
-	}
-
-	if (backpage != backpage2) {
-		put_page(backpage2);
-		_leave(" = -ENODATA [different]");
-		return -ENODATA;
-	}
-
-	/* the page is still there and we already have a ref on it, so we don't
-	 * need a second */
-	put_page(backpage2);
-
-	INIT_LIST_HEAD(&monitor->op_link);
-	folio_add_wait_queue(page_folio(backpage), &monitor->monitor);
-
-	if (trylock_page(backpage)) {
-		ret = -EIO;
-		if (PageError(backpage))
-			goto unlock_discard;
-		ret = 0;
-		if (PageUptodate(backpage))
-			goto unlock_discard;
-
-		_debug("reissue read");
-		ret = bmapping->a_ops->readpage(NULL, backpage);
-		if (ret < 0)
-			goto discard;
-	}
-
-	/* 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 (trylock_page(backpage)) {
-		_debug("jumpstart %p {%lx}", backpage, backpage->flags);
-		unlock_page(backpage);
-	}
-
-	/* it'll reappear on the todo list */
-	_leave(" = -EINPROGRESS");
-	return -EINPROGRESS;
-
-unlock_discard:
-	unlock_page(backpage);
-discard:
-	spin_lock_irq(&object->work_lock);
-	list_del(&monitor->op_link);
-	spin_unlock_irq(&object->work_lock);
-	_leave(" = %d", ret);
-	return ret;
-}
-
-/*
- * copy data from backing pages to netfs pages to complete a read operation
- * - driven by FS-Cache's thread pool
- */
-static void cachefiles_read_copier(struct fscache_operation *_op)
-{
-	struct cachefiles_one_read *monitor;
-	struct cachefiles_object *object;
-	struct fscache_retrieval *op;
-	int error, max;
-
-	op = container_of(_op, struct fscache_retrieval, op);
-	object = container_of(op->op.object,
-			      struct cachefiles_object, fscache);
-
-	_enter("{ino=%lu}", d_backing_inode(object->backer)->i_ino);
-
-	max = 8;
-	spin_lock_irq(&object->work_lock);
-
-	while (!list_empty(&op->to_do)) {
-		monitor = list_entry(op->to_do.next,
-				     struct cachefiles_one_read, op_link);
-		list_del(&monitor->op_link);
-
-		spin_unlock_irq(&object->work_lock);
-
-		_debug("- copy {%lu}", monitor->back_page->index);
-
-	recheck:
-		if (test_bit(FSCACHE_COOKIE_INVALIDATING,
-			     &object->fscache.cookie->flags)) {
-			error = -ESTALE;
-		} else if (PageUptodate(monitor->back_page)) {
-			copy_highpage(monitor->netfs_page, monitor->back_page);
-			fscache_mark_page_cached(monitor->op,
-						 monitor->netfs_page);
-			error = 0;
-		} else if (!PageError(monitor->back_page)) {
-			/* the page has probably been truncated */
-			error = cachefiles_read_reissue(object, monitor);
-			if (error == -EINPROGRESS)
-				goto next;
-			goto recheck;
-		} else {
-			cachefiles_io_error_obj(
-				object,
-				"Readpage failed on backing file %lx",
-				(unsigned long) monitor->back_page->flags);
-			error = -EIO;
-		}
-
-		put_page(monitor->back_page);
-
-		fscache_end_io(op, monitor->netfs_page, error);
-		put_page(monitor->netfs_page);
-		fscache_retrieval_complete(op, 1);
-		fscache_put_retrieval(op);
-		kfree(monitor);
-
-	next:
-		/* let the thread pool have some air occasionally */
-		max--;
-		if (max < 0 || need_resched()) {
-			if (!list_empty(&op->to_do))
-				fscache_enqueue_retrieval(op);
-			_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,
-					    struct fscache_retrieval *op,
-					    struct page *netpage)
-{
-	struct cachefiles_one_read *monitor;
-	struct address_space *bmapping;
-	struct page *newpage, *backpage;
-	int ret;
-
-	_enter("");
-
-	_debug("read back %p{%lu,%d}",
-	       netpage, netpage->index, page_count(netpage));
-
-	monitor = kzalloc(sizeof(*monitor), cachefiles_gfp);
-	if (!monitor)
-		goto nomem;
-
-	monitor->netfs_page = netpage;
-	monitor->op = fscache_get_retrieval(op);
-
-	init_waitqueue_func_entry(&monitor->monitor, cachefiles_read_waiter);
-
-	/* attempt to get hold of the backing page */
-	bmapping = d_backing_inode(object->backer)->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(cachefiles_gfp);
-			if (!newpage)
-				goto nomem_monitor;
-		}
-
-		ret = add_to_page_cache_lru(newpage, bmapping,
-					    netpage->index, cachefiles_gfp);
-		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 start
-	 * it reading */
-installed_new_backing_page:
-	_debug("- new %p", newpage);
-
-	backpage = newpage;
-	newpage = NULL;
-
-read_backing_page:
-	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 */
-	get_page(monitor->netfs_page);
-	get_page(backpage);
-	monitor->back_page = backpage;
-	monitor->monitor.private = backpage;
-	folio_add_wait_queue(page_folio(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 (trylock_page(backpage)) {
-		_debug("jumpstart %p {%lx}", backpage, backpage->flags);
-		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) {
-		put_page(newpage);
-		newpage = NULL;
-	}
-
-	if (PageError(backpage))
-		goto io_error;
-
-	if (PageUptodate(backpage))
-		goto backing_page_already_uptodate;
-
-	if (!trylock_page(backpage))
-		goto monitor_backing_page;
-	_debug("read %p {%lx}", backpage, backpage->flags);
-	goto read_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");
-
-	fscache_mark_page_cached(op, netpage);
-
-	copy_highpage(netpage, backpage);
-	fscache_end_io(op, netpage, 0);
-	fscache_retrieval_complete(op, 1);
-
-success:
-	_debug("success");
-	ret = 0;
-
-out:
-	if (backpage)
-		put_page(backpage);
-	if (monitor) {
-		fscache_put_retrieval(monitor->op);
-		kfree(monitor);
-	}
-	_leave(" = %d", ret);
-	return ret;
-
-read_error:
-	_debug("read error %d", ret);
-	if (ret == -ENOMEM) {
-		fscache_retrieval_complete(op, 1);
-		goto out;
-	}
-io_error:
-	cachefiles_io_error_obj(object, "Page read error on backing file");
-	fscache_retrieval_complete(op, 1);
-	ret = -ENOBUFS;
-	goto out;
-
-nomem_page:
-	put_page(newpage);
-nomem_monitor:
-	fscache_put_retrieval(monitor->op);
-	kfree(monitor);
-nomem:
-	fscache_retrieval_complete(op, 1);
-	_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
- */
-int cachefiles_read_or_alloc_page(struct fscache_retrieval *op,
-				  struct page *page,
-				  gfp_t gfp)
-{
-	struct cachefiles_object *object;
-	struct cachefiles_cache *cache;
-	struct inode *inode;
-	sector_t block;
-	unsigned shift;
-	int ret, ret2;
-
-	object = container_of(op->op.object,
-			      struct cachefiles_object, fscache);
-	cache = container_of(object->fscache.cache,
-			     struct cachefiles_cache, cache);
-
-	_enter("{%p},{%lx},,,", object, page->index);
-
-	if (!object->backer)
-		goto enobufs;
-
-	inode = d_backing_inode(object->backer);
-	ASSERT(S_ISREG(inode->i_mode));
-
-	/* calculate the shift required to use bmap */
-	shift = PAGE_SHIFT - inode->i_sb->s_blocksize_bits;
-
-	op->op.flags &= FSCACHE_OP_KEEP_FLAGS;
-	op->op.flags |= FSCACHE_OP_ASYNC;
-	op->op.processor = cachefiles_read_copier;
-
-	/* 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
-	 */
-	block = page->index;
-	block <<= shift;
-
-	ret2 = bmap(inode, &block);
-	ASSERT(ret2 == 0);
-
-	_debug("%llx -> %llx",
-	       (unsigned long long) (page->index << shift),
-	       (unsigned long long) block);
-
-	if (block) {
-		/* submit the apparently valid page to the backing fs to be
-		 * read from disk */
-		ret = cachefiles_read_backing_file_one(object, op, page);
-	} else if (cachefiles_has_space(cache, 0, 1) == 0) {
-		/* there's space in the cache we can use */
-		fscache_mark_page_cached(op, page);
-		fscache_retrieval_complete(op, 1);
-		ret = -ENODATA;
-	} else {
-		goto enobufs;
-	}
-
-	_leave(" = %d", ret);
-	return ret;
-
-enobufs:
-	fscache_retrieval_complete(op, 1);
-	_leave(" = -ENOBUFS");
-	return -ENOBUFS;
-}
-
-/*
- * 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,
-					struct fscache_retrieval *op,
-					struct list_head *list)
-{
-	struct cachefiles_one_read *monitor = NULL;
-	struct address_space *bmapping = d_backing_inode(object->backer)->i_mapping;
-	struct page *newpage = NULL, *netpage, *_n, *backpage = NULL;
-	int ret = 0;
-
-	_enter("");
-
-	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), cachefiles_gfp);
-			if (!monitor)
-				goto nomem;
-
-			monitor->op = fscache_get_retrieval(op);
-			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(cachefiles_gfp);
-				if (!newpage)
-					goto nomem;
-			}
-
-			ret = add_to_page_cache_lru(newpage, bmapping,
-						    netpage->index,
-						    cachefiles_gfp);
-			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 start it reading */
-	installed_new_backing_page:
-		_debug("- new %p", newpage);
-
-		backpage = newpage;
-		newpage = NULL;
-
-	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_lru(netpage, op->mapping,
-					    netpage->index, cachefiles_gfp);
-		if (ret < 0) {
-			if (ret == -EEXIST) {
-				put_page(backpage);
-				backpage = NULL;
-				put_page(netpage);
-				netpage = NULL;
-				fscache_retrieval_complete(op, 1);
-				continue;
-			}
-			goto nomem;
-		}
-
-		/* install a monitor */
-		get_page(netpage);
-		monitor->netfs_page = netpage;
-
-		get_page(backpage);
-		monitor->back_page = backpage;
-		monitor->monitor.private = backpage;
-		folio_add_wait_queue(page_folio(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 (trylock_page(backpage)) {
-			_debug("2unlock %p {%lx}", backpage, backpage->flags);
-			unlock_page(backpage);
-		}
-
-		put_page(backpage);
-		backpage = NULL;
-
-		put_page(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 (!trylock_page(backpage))
-			goto monitor_backing_page;
-
-		if (PageError(backpage)) {
-			_debug("error %lx", backpage->flags);
-			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:
-		_debug("uptodate %lx", backpage->flags);
-		unlock_page(backpage);
-	backing_page_already_uptodate:
-		_debug("- uptodate");
-
-		ret = add_to_page_cache_lru(netpage, op->mapping,
-					    netpage->index, cachefiles_gfp);
-		if (ret < 0) {
-			if (ret == -EEXIST) {
-				put_page(backpage);
-				backpage = NULL;
-				put_page(netpage);
-				netpage = NULL;
-				fscache_retrieval_complete(op, 1);
-				continue;
-			}
-			goto nomem;
-		}
-
-		copy_highpage(netpage, backpage);
-
-		put_page(backpage);
-		backpage = NULL;
-
-		fscache_mark_page_cached(op, netpage);
-
-		/* the netpage is unlocked and marked up to date here */
-		fscache_end_io(op, netpage, 0);
-		put_page(netpage);
-		netpage = NULL;
-		fscache_retrieval_complete(op, 1);
-		continue;
-	}
-
-	netpage = NULL;
-
-	_debug("out");
-
-out:
-	/* tidy up */
-	if (newpage)
-		put_page(newpage);
-	if (netpage)
-		put_page(netpage);
-	if (backpage)
-		put_page(backpage);
-	if (monitor) {
-		fscache_put_retrieval(op);
-		kfree(monitor);
-	}
-
-	list_for_each_entry_safe(netpage, _n, list, lru) {
-		list_del(&netpage->lru);
-		put_page(netpage);
-		fscache_retrieval_complete(op, 1);
-	}
-
-	_leave(" = %d", ret);
-	return ret;
-
-nomem:
-	_debug("nomem");
-	ret = -ENOMEM;
-	goto record_page_complete;
-
-read_error:
-	_debug("read error %d", ret);
-	if (ret == -ENOMEM)
-		goto record_page_complete;
-io_error:
-	cachefiles_io_error_obj(object, "Page read error on backing file");
-	ret = -ENOBUFS;
-record_page_complete:
-	fscache_retrieval_complete(op, 1);
-	goto out;
-}
-
-/*
- * read a list of pages from the cache or allocate blocks in which to store
- * them
- */
-int cachefiles_read_or_alloc_pages(struct fscache_retrieval *op,
-				   struct list_head *pages,
-				   unsigned *nr_pages,
-				   gfp_t gfp)
-{
-	struct cachefiles_object *object;
-	struct cachefiles_cache *cache;
-	struct list_head backpages;
-	struct pagevec pagevec;
-	struct inode *inode;
-	struct page *page, *_n;
-	unsigned shift, nrbackpages;
-	int ret, ret2, space;
-
-	object = container_of(op->op.object,
-			      struct cachefiles_object, fscache);
-	cache = container_of(object->fscache.cache,
-			     struct cachefiles_cache, cache);
-
-	_enter("{OBJ%x,%d},,%d,,",
-	       object->fscache.debug_id, atomic_read(&op->op.usage),
-	       *nr_pages);
-
-	if (!object->backer)
-		goto all_enobufs;
-
-	space = 1;
-	if (cachefiles_has_space(cache, 0, *nr_pages) < 0)
-		space = 0;
-
-	inode = d_backing_inode(object->backer);
-	ASSERT(S_ISREG(inode->i_mode));
-
-	/* calculate the shift required to use bmap */
-	shift = PAGE_SHIFT - inode->i_sb->s_blocksize_bits;
-
-	pagevec_init(&pagevec);
-
-	op->op.flags &= FSCACHE_OP_KEEP_FLAGS;
-	op->op.flags |= FSCACHE_OP_ASYNC;
-	op->op.processor = cachefiles_read_copier;
-
-	INIT_LIST_HEAD(&backpages);
-	nrbackpages = 0;
-
-	ret = space ? -ENODATA : -ENOBUFS;
-	list_for_each_entry_safe(page, _n, pages, lru) {
-		sector_t 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
-		 */
-		block = page->index;
-		block <<= shift;
-
-		ret2 = bmap(inode, &block);
-		ASSERT(ret2 == 0);
-
-		_debug("%llx -> %llx",
-		       (unsigned long long) (page->index << shift),
-		       (unsigned long long) 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) {
-			fscache_mark_pages_cached(op, &pagevec);
-			fscache_retrieval_complete(op, 1);
-			ret = -ENODATA;
-		} else {
-			fscache_retrieval_complete(op, 1);
-		}
-	}
-
-	if (pagevec_count(&pagevec) > 0)
-		fscache_mark_pages_cached(op, &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, op, &backpages);
-		if (ret2 == -ENOMEM || ret2 == -EINTR)
-			ret = ret2;
-	}
-
-	_leave(" = %d [nr=%u%s]",
-	       ret, *nr_pages, list_empty(pages) ? " empty" : "");
-	return ret;
-
-all_enobufs:
-	fscache_retrieval_complete(op, *nr_pages);
-	return -ENOBUFS;
-}
-
-/*
- * allocate a block in the cache in which to store a page
- * - 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
- */
-int cachefiles_allocate_page(struct fscache_retrieval *op,
-			     struct page *page,
-			     gfp_t gfp)
-{
-	struct cachefiles_object *object;
-	struct cachefiles_cache *cache;
-	int ret;
-
-	object = container_of(op->op.object,
-			      struct cachefiles_object, fscache);
-	cache = container_of(object->fscache.cache,
-			     struct cachefiles_cache, cache);
-
-	_enter("%p,{%lx},", object, page->index);
-
-	ret = cachefiles_has_space(cache, 0, 1);
-	if (ret == 0)
-		fscache_mark_page_cached(op, page);
-	else
-		ret = -ENOBUFS;
-
-	fscache_retrieval_complete(op, 1);
-	_leave(" = %d", ret);
-	return ret;
-}
-
-/*
- * allocate blocks in the cache in which to store a set of pages
- * - cache withdrawal is prevented by the caller
- * - returns -EINTR if interrupted
- * - returns -ENOMEM if ran out of memory
- * - returns -ENOBUFS if some buffers couldn't be made available
- * - returns -ENOBUFS if some pages are beyond EOF
- * - otherwise:
- *   - -ENODATA will be returned
- * - metadata will be retained for any page marked
- */
-int cachefiles_allocate_pages(struct fscache_retrieval *op,
-			      struct list_head *pages,
-			      unsigned *nr_pages,
-			      gfp_t gfp)
-{
-	struct cachefiles_object *object;
-	struct cachefiles_cache *cache;
-	struct pagevec pagevec;
-	struct page *page;
-	int ret;
-
-	object = container_of(op->op.object,
-			      struct cachefiles_object, fscache);
-	cache = container_of(object->fscache.cache,
-			     struct cachefiles_cache, cache);
-
-	_enter("%p,,,%d,", object, *nr_pages);
-
-	ret = cachefiles_has_space(cache, 0, *nr_pages);
-	if (ret == 0) {
-		pagevec_init(&pagevec);
-
-		list_for_each_entry(page, pages, lru) {
-			if (pagevec_add(&pagevec, page) == 0)
-				fscache_mark_pages_cached(op, &pagevec);
-		}
-
-		if (pagevec_count(&pagevec) > 0)
-			fscache_mark_pages_cached(op, &pagevec);
-		ret = -ENODATA;
-	} else {
-		ret = -ENOBUFS;
-	}
-
-	fscache_retrieval_complete(op, *nr_pages);
-	_leave(" = %d", ret);
-	return ret;
-}
-
-/*
- * 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
- */
-int cachefiles_write_page(struct fscache_storage *op, struct page *page)
-{
-	struct cachefiles_object *object;
-	struct cachefiles_cache *cache;
-	struct file *file;
-	struct path path;
-	loff_t pos, eof;
-	size_t len;
-	void *data;
-	int ret = -ENOBUFS;
-
-	ASSERT(op != NULL);
-	ASSERT(page != NULL);
-
-	object = container_of(op->op.object,
-			      struct cachefiles_object, fscache);
-
-	_enter("%p,%p{%lx},,,", object, page, page->index);
-
-	if (!object->backer) {
-		_leave(" = -ENOBUFS");
-		return -ENOBUFS;
-	}
-
-	ASSERT(d_is_reg(object->backer));
-
-	cache = container_of(object->fscache.cache,
-			     struct cachefiles_cache, cache);
-
-	pos = (loff_t)page->index << PAGE_SHIFT;
-
-	/* We mustn't write more data than we have, so we have to beware of a
-	 * partial page at EOF.
-	 */
-	eof = object->fscache.store_limit_l;
-	if (pos >= eof)
-		goto error;
-
-	/* write the page to the backing filesystem and let it store it in its
-	 * own time */
-	path.mnt = cache->mnt;
-	path.dentry = object->backer;
-	file = dentry_open(&path, O_RDWR | O_LARGEFILE, cache->cache_cred);
-	if (IS_ERR(file)) {
-		ret = PTR_ERR(file);
-		goto error_2;
-	}
-
-	len = PAGE_SIZE;
-	if (eof & ~PAGE_MASK) {
-		if (eof - pos < PAGE_SIZE) {
-			_debug("cut short %llx to %llx",
-			       pos, eof);
-			len = eof - pos;
-			ASSERTCMP(pos + len, ==, eof);
-		}
-	}
-
-	data = kmap(page);
-	ret = kernel_write(file, data, len, &pos);
-	kunmap(page);
-	fput(file);
-	if (ret != len)
-		goto error_eio;
-
-	_leave(" = 0");
-	return 0;
-
-error_eio:
-	ret = -EIO;
-error_2:
-	if (ret == -EIO)
-		cachefiles_io_error_obj(object,
-					"Write page to backing file failed");
-error:
-	_leave(" = -ENOBUFS [%d]", ret);
-	return -ENOBUFS;
-}
-
-/*
- * detach a backing block from a page
- * - cache withdrawal is prevented by the caller
- */
-void cachefiles_uncache_page(struct fscache_object *_object, struct page *page)
-	__releases(&object->fscache.cookie->lock)
-{
-	struct cachefiles_object *object;
-
-	object = container_of(_object, struct cachefiles_object, fscache);
-
-	_enter("%p,{%lu}", object, page->index);
-
-	spin_unlock(&object->fscache.cookie->lock);
-}
diff --git a/fs/cachefiles/security.c b/fs/cachefiles/security.c
index aec13fd94692..fe777164f1d8 100644
--- a/fs/cachefiles/security.c
+++ b/fs/cachefiles/security.c
@@ -1,7 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 /* CacheFiles security management
  *
- * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
+ * Copyright (C) 2007, 2021 Red Hat, Inc. All Rights Reserved.
  * Written by David Howells (dhowells@redhat.com)
  */
 
diff --git a/fs/cachefiles/volume.c b/fs/cachefiles/volume.c
new file mode 100644
index 000000000000..89df0ba8ba5e
--- /dev/null
+++ b/fs/cachefiles/volume.c
@@ -0,0 +1,139 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/* Volume handling.
+ *
+ * Copyright (C) 2021 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ */
+
+#include <linux/fs.h>
+#include <linux/slab.h>
+#include "internal.h"
+#include <trace/events/fscache.h>
+
+/*
+ * Allocate and set up a volume representation.  We make sure all the fanout
+ * directories are created and pinned.
+ */
+void cachefiles_acquire_volume(struct fscache_volume *vcookie)
+{
+	struct cachefiles_volume *volume;
+	struct cachefiles_cache *cache = vcookie->cache->cache_priv;
+	const struct cred *saved_cred;
+	struct dentry *vdentry, *fan;
+	size_t len;
+	char *name;
+	bool is_new = false;
+	int ret, n_accesses, i;
+
+	_enter("");
+
+	volume = kzalloc(sizeof(struct cachefiles_volume), GFP_KERNEL);
+	if (!volume)
+		return;
+	volume->vcookie = vcookie;
+	volume->cache = cache;
+	INIT_LIST_HEAD(&volume->cache_link);
+
+	cachefiles_begin_secure(cache, &saved_cred);
+
+	len = vcookie->key[0];
+	name = kmalloc(len + 3, GFP_NOFS);
+	if (!name)
+		goto error_vol;
+	name[0] = 'I';
+	memcpy(name + 1, vcookie->key + 1, len);
+	name[len + 1] = 0;
+
+retry:
+	vdentry = cachefiles_get_directory(cache, cache->store, name, &is_new);
+	if (IS_ERR(vdentry))
+		goto error_name;
+	volume->dentry = vdentry;
+
+	if (is_new) {
+		if (!cachefiles_set_volume_xattr(volume))
+			goto error_dir;
+	} else {
+		ret = cachefiles_check_volume_xattr(volume);
+		if (ret < 0) {
+			if (ret != -ESTALE)
+				goto error_dir;
+			inode_lock_nested(d_inode(cache->store), I_MUTEX_PARENT);
+			cachefiles_bury_object(cache, NULL, cache->store, vdentry,
+					       FSCACHE_VOLUME_IS_WEIRD);
+			cachefiles_put_directory(volume->dentry);
+			cond_resched();
+			goto retry;
+		}
+	}
+	
+	for (i = 0; i < 256; i++) {
+		sprintf(name, "@%02x", i);
+		fan = cachefiles_get_directory(cache, vdentry, name, NULL);
+		if (IS_ERR(fan))
+			goto error_fan;
+		volume->fanout[i] = fan;
+	}
+
+	cachefiles_end_secure(cache, saved_cred);
+
+	vcookie->cache_priv = volume;
+	n_accesses = atomic_inc_return(&vcookie->n_accesses); /* Stop wakeups on dec-to-0 */
+	trace_fscache_access_volume(vcookie->debug_id, 0,
+				    refcount_read(&vcookie->ref),
+				    n_accesses, fscache_access_cache_pin);
+
+	spin_lock(&cache->object_list_lock);
+	list_add(&volume->cache_link, &volume->cache->volumes);
+	spin_unlock(&cache->object_list_lock);
+
+	kfree(name);
+	return;
+
+error_fan:
+	for (i = 0; i < 256; i++)
+		cachefiles_put_directory(volume->fanout[i]);
+error_dir:
+	cachefiles_put_directory(volume->dentry);
+error_name:
+	kfree(name);
+error_vol:
+	kfree(volume);
+	cachefiles_end_secure(cache, saved_cred);
+}
+
+/*
+ * Release a volume representation.
+ */
+static void __cachefiles_free_volume(struct cachefiles_volume *volume)
+{
+	int i;
+
+	_enter("");
+
+	volume->vcookie->cache_priv = NULL;
+
+	for (i = 0; i < 256; i++)
+		cachefiles_put_directory(volume->fanout[i]);
+	cachefiles_put_directory(volume->dentry);
+	kfree(volume);
+}
+
+void cachefiles_free_volume(struct fscache_volume *vcookie)
+{
+	struct cachefiles_volume *volume = vcookie->cache_priv;
+
+	if (volume) {
+		spin_lock(&volume->cache->object_list_lock);
+		list_del_init(&volume->cache_link);
+		spin_unlock(&volume->cache->object_list_lock);
+		__cachefiles_free_volume(volume);
+	}
+}
+
+void cachefiles_withdraw_volume(struct cachefiles_volume *volume)
+{
+	fscache_withdraw_volume(volume->vcookie);
+	cachefiles_set_volume_xattr(volume);
+	__cachefiles_free_volume(volume);
+}
diff --git a/fs/cachefiles/xattr.c b/fs/cachefiles/xattr.c
index 9e82de668595..83f41bd0c3a9 100644
--- a/fs/cachefiles/xattr.c
+++ b/fs/cachefiles/xattr.c
@@ -1,7 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0-or-later
 /* CacheFiles extended attribute management
  *
- * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
+ * Copyright (C) 2021 Red Hat, Inc. All Rights Reserved.
  * Written by David Howells (dhowells@redhat.com)
  */
 
@@ -15,310 +15,245 @@
 #include <linux/slab.h>
 #include "internal.h"
 
+#define CACHEFILES_COOKIE_TYPE_DATA 1
+
+struct cachefiles_xattr {
+	__be64	object_size;	/* Actual size of the object */
+	__be64	zero_point;	/* Size after which server has no data not written by us */
+	__u8	type;		/* Type of object */
+	__u8	content;	/* Content presence (enum cachefiles_content) */
+	__u8	data[];		/* netfs coherency data */
+} __packed;
+
 static const char cachefiles_xattr_cache[] =
 	XATTR_USER_PREFIX "CacheFiles.cache";
 
 /*
- * check the type label on an object
- * - done using xattrs
+ * set the state xattr on a cache file
  */
-int cachefiles_check_object_type(struct cachefiles_object *object)
+int cachefiles_set_object_xattr(struct cachefiles_object *object)
 {
-	struct dentry *dentry = object->dentry;
-	char type[3], xtype[3];
+	struct cachefiles_xattr *buf;
+	struct dentry *dentry;
+	struct file *file = object->file;
+	unsigned int len = object->cookie->aux_len;
 	int ret;
 
-	ASSERT(dentry);
-	ASSERT(d_backing_inode(dentry));
-
-	if (!object->fscache.cookie)
-		strcpy(type, "C3");
-	else
-		snprintf(type, 3, "%02x", object->fscache.cookie->def->type);
-
-	_enter("%x{%s}", object->fscache.debug_id, type);
+	if (!file)
+		return -ESTALE;
+	dentry = file->f_path.dentry;
 
-	/* attempt to install a type label directly */
-	ret = vfs_setxattr(&init_user_ns, dentry, cachefiles_xattr_cache, type,
-			   2, XATTR_CREATE);
-	if (ret == 0) {
-		_debug("SET"); /* we succeeded */
-		goto error;
-	}
+	_enter("%x,#%d", object->debug_id, len);
 
-	if (ret != -EEXIST) {
-		pr_err("Can't set xattr on %pd [%lu] (err %d)\n",
-		       dentry, d_backing_inode(dentry)->i_ino,
-		       -ret);
-		goto error;
-	}
+	buf = kmalloc(sizeof(struct cachefiles_xattr) + len, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
 
-	/* read the current type label */
-	ret = vfs_getxattr(&init_user_ns, dentry, cachefiles_xattr_cache, xtype,
-			   3);
+	buf->object_size	= cpu_to_be64(object->cookie->object_size);
+	buf->zero_point		= 0;
+	buf->type		= CACHEFILES_COOKIE_TYPE_DATA;
+	buf->content		= object->content_info;
+	if (test_bit(FSCACHE_COOKIE_LOCAL_WRITE, &object->cookie->flags))
+		buf->content	= CACHEFILES_CONTENT_DIRTY;
+	if (len > 0)
+		memcpy(buf->data, fscache_get_aux(object->cookie), len);
+
+	ret = cachefiles_inject_write_error();
+	if (ret == 0)
+		ret = vfs_setxattr(&init_user_ns, dentry, cachefiles_xattr_cache,
+				   buf, sizeof(struct cachefiles_xattr) + len, 0);
 	if (ret < 0) {
-		if (ret == -ERANGE)
-			goto bad_type_length;
-
-		pr_err("Can't read xattr on %pd [%lu] (err %d)\n",
-		       dentry, d_backing_inode(dentry)->i_ino,
-		       -ret);
-		goto error;
+		trace_cachefiles_vfs_error(object, file_inode(file), ret,
+					   cachefiles_trace_setxattr_error);
+		trace_cachefiles_coherency(object, file_inode(file)->i_ino,
+					   buf->content,
+					   cachefiles_coherency_set_fail);
+		if (ret != -ENOMEM)
+			cachefiles_io_error_obj(
+				object,
+				"Failed to set xattr with error %d", ret);
+	} else {
+		trace_cachefiles_coherency(object, file_inode(file)->i_ino,
+					   buf->content,
+					   cachefiles_coherency_set_ok);
 	}
 
-	/* 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:
+	kfree(buf);
 	_leave(" = %d", ret);
 	return ret;
-
-bad_type_length:
-	pr_err("Cache object %lu type xattr length incorrect\n",
-	       d_backing_inode(dentry)->i_ino);
-	ret = -EIO;
-	goto error;
-
-bad_type:
-	xtype[2] = 0;
-	pr_err("Cache object %pd [%lu] type %s not %s\n",
-	       dentry, d_backing_inode(dentry)->i_ino,
-	       xtype, type);
-	ret = -EIO;
-	goto error;
 }
 
 /*
- * set the state xattr on a cache file
+ * check the consistency between the backing cache and the FS-Cache cookie
  */
-int cachefiles_set_object_xattr(struct cachefiles_object *object,
-				struct cachefiles_xattr *auxdata)
+int cachefiles_check_auxdata(struct cachefiles_object *object, struct file *file)
 {
-	struct dentry *dentry = object->dentry;
-	int ret;
-
-	ASSERT(dentry);
-
-	_enter("%p,#%d", object, auxdata->len);
+	struct cachefiles_xattr *buf;
+	struct dentry *dentry = file->f_path.dentry;
+	unsigned int len = object->cookie->aux_len, tlen;
+	const void *p = fscache_get_aux(object->cookie);
+	enum cachefiles_coherency_trace why;
+	ssize_t xlen;
+	int ret = -ESTALE;
 
-	/* attempt to install the cache metadata directly */
-	_debug("SET #%u", auxdata->len);
+	tlen = sizeof(struct cachefiles_xattr) + len;
+	buf = kmalloc(tlen, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
 
-	clear_bit(FSCACHE_COOKIE_AUX_UPDATED, &object->fscache.cookie->flags);
-	ret = vfs_setxattr(&init_user_ns, dentry, cachefiles_xattr_cache,
-			   &auxdata->type, auxdata->len, XATTR_CREATE);
-	if (ret < 0 && ret != -ENOMEM)
-		cachefiles_io_error_obj(
-			object,
-			"Failed to set xattr with error %d", ret);
+	xlen = cachefiles_inject_read_error();
+	if (xlen == 0)
+		xlen = vfs_getxattr(&init_user_ns, dentry, cachefiles_xattr_cache, buf, tlen);
+	if (xlen != tlen) {
+		if (xlen < 0)
+			trace_cachefiles_vfs_error(object, file_inode(file), xlen,
+						   cachefiles_trace_getxattr_error);
+		if (xlen == -EIO)
+			cachefiles_io_error_obj(
+				object,
+				"Failed to read aux with error %zd", xlen);
+		why = cachefiles_coherency_check_xattr;
+	} else if (buf->type != CACHEFILES_COOKIE_TYPE_DATA) {
+		why = cachefiles_coherency_check_type;
+	} else if (memcmp(buf->data, p, len) != 0) {
+		why = cachefiles_coherency_check_aux;
+	} else if (be64_to_cpu(buf->object_size) != object->cookie->object_size) {
+		why = cachefiles_coherency_check_objsize;
+	} else if (buf->content == CACHEFILES_CONTENT_DIRTY) {
+		// TODO: Begin conflict resolution
+		pr_warn("Dirty object in cache\n");
+		why = cachefiles_coherency_check_dirty;
+	} else {
+		why = cachefiles_coherency_check_ok;
+		ret = 0;
+	}
 
-	_leave(" = %d", ret);
+	trace_cachefiles_coherency(object, file_inode(file)->i_ino,
+				   buf->content, why);
+	kfree(buf);
 	return ret;
 }
 
 /*
- * update the state xattr on a cache file
+ * remove the object's xattr to mark it stale
  */
-int cachefiles_update_object_xattr(struct cachefiles_object *object,
-				   struct cachefiles_xattr *auxdata)
+int cachefiles_remove_object_xattr(struct cachefiles_cache *cache,
+				   struct cachefiles_object *object,
+				   struct dentry *dentry)
 {
-	struct dentry *dentry = object->dentry;
 	int ret;
 
-	if (!dentry)
-		return -ESTALE;
-
-	_enter("%x,#%d", object->fscache.debug_id, auxdata->len);
-
-	/* attempt to install the cache metadata directly */
-	_debug("SET #%u", auxdata->len);
-
-	clear_bit(FSCACHE_COOKIE_AUX_UPDATED, &object->fscache.cookie->flags);
-	ret = vfs_setxattr(&init_user_ns, dentry, cachefiles_xattr_cache,
-			   &auxdata->type, auxdata->len, XATTR_REPLACE);
-	if (ret < 0 && ret != -ENOMEM)
-		cachefiles_io_error_obj(
-			object,
-			"Failed to update xattr with error %d", ret);
+	ret = cachefiles_inject_remove_error();
+	if (ret == 0)
+		ret = vfs_removexattr(&init_user_ns, dentry, cachefiles_xattr_cache);
+	if (ret < 0) {
+		trace_cachefiles_vfs_error(object, d_inode(dentry), ret,
+					   cachefiles_trace_remxattr_error);
+		if (ret == -ENOENT || ret == -ENODATA)
+			ret = 0;
+		else if (ret != -ENOMEM)
+			cachefiles_io_error(cache,
+					    "Can't remove xattr from %lu"
+					    " (error %d)",
+					    d_backing_inode(dentry)->i_ino, -ret);
+	}
 
 	_leave(" = %d", ret);
 	return ret;
 }
 
 /*
- * check the consistency between the backing cache and the FS-Cache cookie
+ * Stick a marker on the cache object to indicate that it's dirty.
  */
-int cachefiles_check_auxdata(struct cachefiles_object *object)
+void cachefiles_prepare_to_write(struct fscache_cookie *cookie)
 {
-	struct cachefiles_xattr *auxbuf;
-	enum fscache_checkaux validity;
-	struct dentry *dentry = object->dentry;
-	ssize_t xlen;
-	int ret;
-
-	ASSERT(dentry);
-	ASSERT(d_backing_inode(dentry));
-	ASSERT(object->fscache.cookie->def->check_aux);
-
-	auxbuf = kmalloc(sizeof(struct cachefiles_xattr) + 512, GFP_KERNEL);
-	if (!auxbuf)
-		return -ENOMEM;
+	const struct cred *saved_cred;
+	struct cachefiles_object *object = cookie->cache_priv;
+	struct cachefiles_cache *cache = object->volume->cache;
 
-	xlen = vfs_getxattr(&init_user_ns, dentry, cachefiles_xattr_cache,
-			    &auxbuf->type, 512 + 1);
-	ret = -ESTALE;
-	if (xlen < 1 ||
-	    auxbuf->type != object->fscache.cookie->def->type)
-		goto error;
+	_enter("c=%08x", object->cookie->debug_id);
 
-	xlen--;
-	validity = fscache_check_aux(&object->fscache, &auxbuf->data, xlen,
-				     i_size_read(d_backing_inode(dentry)));
-	if (validity != FSCACHE_CHECKAUX_OKAY)
-		goto error;
-
-	ret = 0;
-error:
-	kfree(auxbuf);
-	return ret;
+	if (!test_bit(CACHEFILES_OBJECT_USING_TMPFILE, &object->flags)) {
+		cachefiles_begin_secure(cache, &saved_cred);
+		cachefiles_set_object_xattr(object);
+		cachefiles_end_secure(cache, saved_cred);
+	}
 }
 
 /*
- * check the state xattr on a cache file
- * - return -ESTALE if the object should be deleted
+ * Set the state xattr on a volume directory.
  */
-int cachefiles_check_object_xattr(struct cachefiles_object *object,
-				  struct cachefiles_xattr *auxdata)
+bool cachefiles_set_volume_xattr(struct cachefiles_volume *volume)
 {
-	struct cachefiles_xattr *auxbuf;
-	struct dentry *dentry = object->dentry;
+	unsigned int len = volume->vcookie->coherency_len;
+	const void *p = volume->vcookie->coherency;
+	struct dentry *dentry = volume->dentry;
 	int ret;
 
-	_enter("%p,#%d", object, auxdata->len);
-
-	ASSERT(dentry);
-	ASSERT(d_backing_inode(dentry));
-
-	auxbuf = kmalloc(sizeof(struct cachefiles_xattr) + 512, cachefiles_gfp);
-	if (!auxbuf) {
-		_leave(" = -ENOMEM");
-		return -ENOMEM;
-	}
+	_enter("%x,#%d", volume->vcookie->debug_id, len);
 
-	/* read the current type label */
-	ret = vfs_getxattr(&init_user_ns, dentry, cachefiles_xattr_cache,
-			   &auxbuf->type, 512 + 1);
+	ret = cachefiles_inject_write_error();
+	if (ret == 0)
+		ret = vfs_setxattr(&init_user_ns, dentry, cachefiles_xattr_cache,
+				   p, len, 0);
 	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)",
-					d_backing_inode(dentry)->i_ino, -ret);
-		goto error;
+		trace_cachefiles_vfs_error(NULL, d_inode(dentry), ret,
+					   cachefiles_trace_setxattr_error);
+		trace_cachefiles_vol_coherency(volume, d_inode(dentry)->i_ino,
+					       cachefiles_coherency_vol_set_fail);
+		if (ret != -ENOMEM)
+			cachefiles_io_error(
+				volume->cache, "Failed to set xattr with error %d", ret);
+	} else {
+		trace_cachefiles_vol_coherency(volume, d_inode(dentry)->i_ino,
+					       cachefiles_coherency_vol_set_ok);
 	}
 
-	/* 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) {
-		enum fscache_checkaux result;
-		unsigned int dlen;
-
-		dlen = auxbuf->len - 1;
-
-		_debug("checkaux %s #%u",
-		       object->fscache.cookie->def->name, dlen);
-
-		result = fscache_check_aux(&object->fscache,
-					   &auxbuf->data, dlen,
-					   i_size_read(d_backing_inode(dentry)));
-
-		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 = vfs_setxattr(&init_user_ns, 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)",
-						d_backing_inode(dentry)->i_ino, -ret);
-			goto error;
-		}
-	}
-
-okay:
-	ret = 0;
-
-error:
-	kfree(auxbuf);
 	_leave(" = %d", ret);
-	return ret;
-
-bad_type_length:
-	pr_err("Cache object %lu xattr length incorrect\n",
-	       d_backing_inode(dentry)->i_ino);
-	ret = -EIO;
-	goto error;
-
-stale:
-	ret = -ESTALE;
-	goto error;
+	return ret == 0;
 }
 
 /*
- * remove the object's xattr to mark it stale
+ * Check the consistency between the backing cache and the volume cookie.
  */
-int cachefiles_remove_object_xattr(struct cachefiles_cache *cache,
-				   struct dentry *dentry)
+int cachefiles_check_volume_xattr(struct cachefiles_volume *volume)
 {
-	int ret;
+	struct cachefiles_xattr *buf;
+	struct dentry *dentry = volume->dentry;
+	unsigned int len = volume->vcookie->coherency_len;
+	const void *p = volume->vcookie->coherency;
+	enum cachefiles_coherency_trace why;
+	ssize_t xlen;
+	int ret = -ESTALE;
 
-	ret = vfs_removexattr(&init_user_ns, dentry, cachefiles_xattr_cache);
-	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)",
-					    d_backing_inode(dentry)->i_ino, -ret);
+	_enter("");
+
+	buf = kmalloc(len, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	xlen = cachefiles_inject_read_error();
+	if (xlen == 0)
+		xlen = vfs_getxattr(&init_user_ns, dentry, cachefiles_xattr_cache, buf, len);
+	if (xlen != len) {
+		if (xlen < 0) {
+			trace_cachefiles_vfs_error(NULL, d_inode(dentry), xlen,
+						   cachefiles_trace_getxattr_error);
+			if (xlen == -EIO)
+				cachefiles_io_error(
+					volume->cache,
+					"Failed to read xattr with error %zd", xlen);
+		}
+		why = cachefiles_coherency_vol_check_xattr;
+	} else if (memcmp(buf->data, p, len) != 0) {
+		why = cachefiles_coherency_vol_check_cmp;
+	} else {
+		why = cachefiles_coherency_vol_check_ok;
+		ret = 0;
 	}
 
+	trace_cachefiles_vol_coherency(volume, d_inode(dentry)->i_ino, why);
+	kfree(buf);
 	_leave(" = %d", ret);
 	return ret;
 }