From 80f94f29f677d32a1edabc055acfe91c00803592 Mon Sep 17 00:00:00 2001
From: David Howells <dhowells@redhat.com>
Date: Thu, 21 Oct 2021 08:59:46 +0100
Subject: cachefiles: Provide a function to check how much space there is

Provide a function to check how much space there is.  This also flips the
state on the cache and will signal the daemon to inform it of the change
and to ask it to do some culling if necessary.

We will also need to subtract the amount of data currently being written to
the cache (cache->b_writing) from the amount of available space to avoid
hitting ENOSPC accidentally.

Signed-off-by: David Howells <dhowells@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
cc: linux-cachefs@redhat.com
Link: https://lore.kernel.org/r/163819629322.215744.13457425294680841213.stgit@warthog.procyon.org.uk/ # v1
Link: https://lore.kernel.org/r/163906930100.143852.1681026700865762069.stgit@warthog.procyon.org.uk/ # v2
Link: https://lore.kernel.org/r/163967140058.1823006.7781243664702837128.stgit@warthog.procyon.org.uk/ # v3
Link: https://lore.kernel.org/r/164021539957.640689.12477177372616805706.stgit@warthog.procyon.org.uk/ # v4
---
 fs/cachefiles/cache.c | 103 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 103 insertions(+)
 create mode 100644 fs/cachefiles/cache.c

(limited to 'fs/cachefiles/cache.c')

diff --git a/fs/cachefiles/cache.c b/fs/cachefiles/cache.c
new file mode 100644
index 000000000000..73636f89eefa
--- /dev/null
+++ b/fs/cachefiles/cache.c
@@ -0,0 +1,103 @@
+// 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"
+
+/*
+ * 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)
+{
+	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 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;
+
+begin_cull:
+	if (!test_and_set_bit(CACHEFILES_CULLING, &cache->flags)) {
+		_debug("### CULL CACHE ###");
+		cachefiles_state_changed(cache);
+	}
+
+	_leave(" = %d", ret);
+	return ret;
+}
-- 
cgit v1.2.3


From d1065b0a6fd9397edd3094c56b777d0d8ec1290d Mon Sep 17 00:00:00 2001
From: David Howells <dhowells@redhat.com>
Date: Fri, 26 Nov 2021 14:29:06 +0000
Subject: cachefiles: Implement cache registration and withdrawal

Do the following:

 (1) Fill out cachefiles_daemon_add_cache() so that it sets up the cache
     directories and registers the cache with cachefiles.

 (2) Add a function to do the top-level part of cache withdrawal and
     unregistration.

 (3) Add a function to sync a cache.

Signed-off-by: David Howells <dhowells@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
cc: linux-cachefs@redhat.com
Link: https://lore.kernel.org/r/163819633175.215744.10857127598041268340.stgit@warthog.procyon.org.uk/ # v1
Link: https://lore.kernel.org/r/163906935445.143852.15545194974036410029.stgit@warthog.procyon.org.uk/ # v2
Link: https://lore.kernel.org/r/163967142904.1823006.244055483596047072.stgit@warthog.procyon.org.uk/ # v3
Link: https://lore.kernel.org/r/164021543872.640689.14370017789605073222.stgit@warthog.procyon.org.uk/ # v4
---
 fs/cachefiles/Makefile    |   1 +
 fs/cachefiles/cache.c     | 207 ++++++++++++++++++++++++++++++++++++++++++++++
 fs/cachefiles/daemon.c    |   8 +-
 fs/cachefiles/interface.c |  18 ++++
 fs/cachefiles/internal.h  |   9 ++
 5 files changed, 240 insertions(+), 3 deletions(-)
 create mode 100644 fs/cachefiles/interface.c

(limited to 'fs/cachefiles/cache.c')

diff --git a/fs/cachefiles/Makefile b/fs/cachefiles/Makefile
index e0b092ca077f..92af5daee8ce 100644
--- a/fs/cachefiles/Makefile
+++ b/fs/cachefiles/Makefile
@@ -6,6 +6,7 @@
 cachefiles-y := \
 	cache.o \
 	daemon.o \
+	interface.o \
 	main.o \
 	namei.o \
 	security.o
diff --git a/fs/cachefiles/cache.c b/fs/cachefiles/cache.c
index 73636f89eefa..0462e7af87fb 100644
--- a/fs/cachefiles/cache.c
+++ b/fs/cachefiles/cache.c
@@ -10,6 +10,166 @@
 #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 (mnt_user_ns(path.mnt) != &init_user_ns) {
+		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_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
@@ -101,3 +261,50 @@ begin_cull:
 	_leave(" = %d", ret);
 	return ret;
 }
+
+/*
+ * 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 */
+	// PLACEHOLDER: Withdraw objects
+	fscache_wait_for_objects(fscache);
+
+	// PLACEHOLDER: Withdraw volume
+	cachefiles_sync_cache(cache);
+	cache->cache = NULL;
+	fscache_relinquish_cache(fscache);
+}
diff --git a/fs/cachefiles/daemon.c b/fs/cachefiles/daemon.c
index 7d4691614cec..a449ee661987 100644
--- a/fs/cachefiles/daemon.c
+++ b/fs/cachefiles/daemon.c
@@ -702,6 +702,7 @@ static int cachefiles_daemon_bind(struct cachefiles_cache *cache, char *args)
 
 	pr_warn("Cache is disabled for development\n");
 	return -ENOANO; // Don't allow the cache to operate yet
+	//return cachefiles_add_cache(cache);
 }
 
 /*
@@ -711,10 +712,11 @@ static void cachefiles_daemon_unbind(struct cachefiles_cache *cache)
 {
 	_enter("");
 
-	if (test_bit(CACHEFILES_READY, &cache->flags)) {
-		// PLACEHOLDER: Withdraw cache
-	}
+	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);
diff --git a/fs/cachefiles/interface.c b/fs/cachefiles/interface.c
new file mode 100644
index 000000000000..564ea8fa6641
--- /dev/null
+++ b/fs/cachefiles/interface.c
@@ -0,0 +1,18 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/* FS-Cache interface to CacheFiles
+ *
+ * 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"
+
+const struct fscache_cache_ops cachefiles_cache_ops = {
+	.name			= "cachefiles",
+};
diff --git a/fs/cachefiles/internal.h b/fs/cachefiles/internal.h
index 48768a3ab105..77e874c2bbe7 100644
--- a/fs/cachefiles/internal.h
+++ b/fs/cachefiles/internal.h
@@ -32,6 +32,8 @@ struct cachefiles_object {
 struct cachefiles_cache {
 	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 */
 	const struct cred		*cache_cred;	/* security override for accessing cache */
 	struct mutex			daemon_mutex;	/* command serialisation mutex */
@@ -78,8 +80,10 @@ static inline void cachefiles_state_changed(struct cachefiles_cache *cache)
 /*
  * cache.c
  */
+extern int cachefiles_add_cache(struct cachefiles_cache *cache);
 extern int cachefiles_has_space(struct cachefiles_cache *cache,
 				unsigned fnr, unsigned bnr);
+extern void cachefiles_withdraw_cache(struct cachefiles_cache *cache);
 
 /*
  * daemon.c
@@ -125,6 +129,11 @@ 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;
+
 /*
  * namei.c
  */
-- 
cgit v1.2.3


From fe2140e2f57fef8562e0f9b7cd447d2b08dc2f35 Mon Sep 17 00:00:00 2001
From: David Howells <dhowells@redhat.com>
Date: Thu, 21 Oct 2021 09:55:21 +0100
Subject: cachefiles: Implement volume support

Implement support for creating the directory layout for a volume on disk
and setting up and withdrawing volume caching.

Each volume has a directory named for the volume key under the root of the
cache (prefixed with an 'I' to indicate to cachefilesd that it's an index)
and then creates a bunch of hash bucket subdirectories under that (named as
'@' plus a hex number) in which cookie files will be created.

Signed-off-by: David Howells <dhowells@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
cc: linux-cachefs@redhat.com
Link: https://lore.kernel.org/r/163819635314.215744.13081522301564537723.stgit@warthog.procyon.org.uk/ # v1
Link: https://lore.kernel.org/r/163906936397.143852.17788457778396467161.stgit@warthog.procyon.org.uk/ # v2
Link: https://lore.kernel.org/r/163967143860.1823006.7185205806080225038.stgit@warthog.procyon.org.uk/ # v3
Link: https://lore.kernel.org/r/164021545212.640689.5064821392307582927.stgit@warthog.procyon.org.uk/ # v4
---
 fs/cachefiles/Makefile    |   3 +-
 fs/cachefiles/cache.c     |  28 ++++++++++-
 fs/cachefiles/daemon.c    |   2 +
 fs/cachefiles/interface.c |   2 +
 fs/cachefiles/internal.h  |  20 ++++++++
 fs/cachefiles/volume.c    | 118 ++++++++++++++++++++++++++++++++++++++++++++++
 6 files changed, 171 insertions(+), 2 deletions(-)
 create mode 100644 fs/cachefiles/volume.c

(limited to 'fs/cachefiles/cache.c')

diff --git a/fs/cachefiles/Makefile b/fs/cachefiles/Makefile
index 92af5daee8ce..d67210ece9cd 100644
--- a/fs/cachefiles/Makefile
+++ b/fs/cachefiles/Makefile
@@ -9,7 +9,8 @@ cachefiles-y := \
 	interface.o \
 	main.o \
 	namei.o \
-	security.o
+	security.o \
+	volume.o
 
 cachefiles-$(CONFIG_CACHEFILES_ERROR_INJECTION) += error_inject.o
 
diff --git a/fs/cachefiles/cache.c b/fs/cachefiles/cache.c
index 0462e7af87fb..c4b9280ca0cd 100644
--- a/fs/cachefiles/cache.c
+++ b/fs/cachefiles/cache.c
@@ -262,6 +262,32 @@ begin_cull:
 	return ret;
 }
 
+/*
+ * 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.
  */
@@ -303,7 +329,7 @@ void cachefiles_withdraw_cache(struct cachefiles_cache *cache)
 	// PLACEHOLDER: Withdraw objects
 	fscache_wait_for_objects(fscache);
 
-	// PLACEHOLDER: Withdraw volume
+	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 a449ee661987..337597a4e30c 100644
--- a/fs/cachefiles/daemon.c
+++ b/fs/cachefiles/daemon.c
@@ -105,6 +105,8 @@ static int cachefiles_daemon_open(struct inode *inode, struct file *file)
 
 	mutex_init(&cache->daemon_mutex);
 	init_waitqueue_head(&cache->daemon_pollwq);
+	INIT_LIST_HEAD(&cache->volumes);
+	spin_lock_init(&cache->object_list_lock);
 
 	/* set default caching limits
 	 * - limit at 1% free space and/or free files
diff --git a/fs/cachefiles/interface.c b/fs/cachefiles/interface.c
index 564ea8fa6641..1793e46bd3e7 100644
--- a/fs/cachefiles/interface.c
+++ b/fs/cachefiles/interface.c
@@ -15,4 +15,6 @@
 
 const struct fscache_cache_ops cachefiles_cache_ops = {
 	.name			= "cachefiles",
+	.acquire_volume		= cachefiles_acquire_volume,
+	.free_volume		= cachefiles_free_volume,
 };
diff --git a/fs/cachefiles/internal.h b/fs/cachefiles/internal.h
index 77e874c2bbe7..ab0e9307be7b 100644
--- a/fs/cachefiles/internal.h
+++ b/fs/cachefiles/internal.h
@@ -19,6 +19,17 @@
 struct cachefiles_cache;
 struct cachefiles_object;
 
+/*
+ * 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 */
+};
+
 /*
  * Data file records.
  */
@@ -35,6 +46,8 @@ struct cachefiles_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 */
+	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 */
@@ -163,6 +176,13 @@ static inline void cachefiles_end_secure(struct cachefiles_cache *cache,
 	revert_creds(saved_cred);
 }
 
+/*
+ * 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);
+
 /*
  * Error handling
  */
diff --git a/fs/cachefiles/volume.c b/fs/cachefiles/volume.c
new file mode 100644
index 000000000000..4a14f5e72764
--- /dev/null
+++ b/fs/cachefiles/volume.c
@@ -0,0 +1,118 @@
+// 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;
+	int 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;
+
+	vdentry = cachefiles_get_directory(cache, cache->store, name, NULL);
+	if (IS_ERR(vdentry))
+		goto error_name;
+	volume->dentry = vdentry;
+
+	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]);
+	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_free_volume(volume);
+}
-- 
cgit v1.2.3


From 1f08c925e7a38002bde509e66f6f891468848511 Mon Sep 17 00:00:00 2001
From: David Howells <dhowells@redhat.com>
Date: Thu, 21 Oct 2021 08:50:10 +0100
Subject: cachefiles: Implement backing file wrangling

Implement the wrangling of backing files, including the following pieces:

 (1) Lookup and creation of a file on disk, using a tmpfile if the file
     isn't yet present.  The file is then opened, sized for DIO and the
     file handle is attached to the cachefiles_object struct.  The inode is
     marked to indicate that it's in use by a kernel service.

 (2) Invalidation of an object, creating a tmpfile and switching the file
     pointer in the cachefiles object.

 (3) Committing a file to disk, including setting the coherency xattr on it
     and, if necessary, creating a hard link to it.

     Note that this would be a good place to use Omar Sandoval's vfs_link()
     with AT_LINK_REPLACE[1] as I may have to unlink an old file before I
     can link a tmpfile into place.

 (4) Withdrawal of open objects when a cache is being withdrawn or a cookie
     is relinquished.  This involves committing or discarding the file.

Changes
=======
ver #2:
 - Fix logging of wrong error[1].

Signed-off-by: David Howells <dhowells@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
cc: linux-cachefs@redhat.com
Link: https://lore.kernel.org/r/20211203094950.GA2480@kili/ [1]
Link: https://lore.kernel.org/r/163819644097.215744.4505389616742411239.stgit@warthog.procyon.org.uk/ # v1
Link: https://lore.kernel.org/r/163906949512.143852.14222856795032602080.stgit@warthog.procyon.org.uk/ # v2
Link: https://lore.kernel.org/r/163967158526.1823006.17482695321424642675.stgit@warthog.procyon.org.uk/ # v3
Link: https://lore.kernel.org/r/164021557060.640689.16373541458119269871.stgit@warthog.procyon.org.uk/ # v4
---
 fs/cachefiles/cache.c     |  32 ++++-
 fs/cachefiles/daemon.c    |   1 +
 fs/cachefiles/interface.c | 260 +++++++++++++++++++++++++++++++++++++
 fs/cachefiles/internal.h  |   9 ++
 fs/cachefiles/namei.c     | 318 ++++++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 619 insertions(+), 1 deletion(-)

(limited to 'fs/cachefiles/cache.c')

diff --git a/fs/cachefiles/cache.c b/fs/cachefiles/cache.c
index c4b9280ca0cd..e2cbbc08bad9 100644
--- a/fs/cachefiles/cache.c
+++ b/fs/cachefiles/cache.c
@@ -262,6 +262,36 @@ begin_cull:
 	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.
  */
@@ -326,7 +356,7 @@ void cachefiles_withdraw_cache(struct cachefiles_cache *cache)
 	/* 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 */
-	// PLACEHOLDER: Withdraw objects
+	cachefiles_withdraw_objects(cache);
 	fscache_wait_for_objects(fscache);
 
 	cachefiles_withdraw_volumes(cache);
diff --git a/fs/cachefiles/daemon.c b/fs/cachefiles/daemon.c
index 985c3f3e6767..61e8740d01be 100644
--- a/fs/cachefiles/daemon.c
+++ b/fs/cachefiles/daemon.c
@@ -106,6 +106,7 @@ static int cachefiles_daemon_open(struct inode *inode, struct file *file)
 	mutex_init(&cache->daemon_mutex);
 	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
diff --git a/fs/cachefiles/interface.c b/fs/cachefiles/interface.c
index 68bb7b6c4945..e47c52c34071 100644
--- a/fs/cachefiles/interface.c
+++ b/fs/cachefiles/interface.c
@@ -99,8 +99,268 @@ void cachefiles_put_object(struct cachefiles_object *object,
 	_leave("");
 }
 
+/*
+ * 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 int cachefiles_adjust_size(struct cachefiles_object *object)
+{
+	struct iattr newattrs;
+	struct file *file = object->file;
+	uint64_t ni_size;
+	loff_t oi_size;
+	int ret;
+
+	ni_size = object->cookie->object_size;
+	ni_size = round_up(ni_size, CACHEFILES_DIO_BLOCK_SIZE);
+
+	_enter("{OBJ%x},[%llu]",
+	       object->debug_id, (unsigned long long) ni_size);
+
+	if (!file)
+		return -ENOBUFS;
+
+	oi_size = i_size_read(file_inode(file));
+	if (oi_size == ni_size)
+		return 0;
+
+	inode_lock(file_inode(file));
+
+	/* 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;
+	}
+
+	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);
+
+truncate_failed:
+	inode_unlock(file_inode(file));
+
+	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;
+	}
+
+	_leave(" = %d", ret);
+	return ret;
+}
+
+/*
+ * Attempt to look up the nominated node in this cache
+ */
+static bool cachefiles_lookup_cookie(struct fscache_cookie *cookie)
+{
+	struct cachefiles_object *object;
+	struct cachefiles_cache *cache = cookie->volume->cache->cache_priv;
+	const struct cred *saved_cred;
+	bool success;
+
+	object = cachefiles_alloc_object(cookie);
+	if (!object)
+		goto fail;
+
+	_enter("{OBJ%x}", object->debug_id);
+
+	if (!cachefiles_cook_key(object))
+		goto fail_put;
+
+	cookie->cache_priv = object;
+
+	cachefiles_begin_secure(cache, &saved_cred);
+
+	success = cachefiles_look_up_object(object);
+	if (!success)
+		goto fail_withdraw;
+
+	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);
+
+	cachefiles_end_secure(cache, saved_cred);
+	_leave(" = t");
+	return true;
+
+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;
+
+fail_put:
+	cachefiles_put_object(object, cachefiles_obj_put_alloc_fail);
+fail:
+	return false;
+}
+
+/*
+ * Commit changes to the object as we drop it.
+ */
+static void cachefiles_commit_object(struct cachefiles_object *object,
+				     struct cachefiles_cache *cache)
+{
+	bool update = false;
+
+	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);
+
+	if (test_bit(CACHEFILES_OBJECT_USING_TMPFILE, &object->flags))
+		cachefiles_commit_tmpfile(cache, object);
+}
+
+/*
+ * 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);
+	}
+
+	cachefiles_unmark_inode_in_use(object, object->file);
+	if (object->file) {
+		fput(object->file);
+		object->file = NULL;
+	}
+}
+
+/*
+ * Withdraw caching for a cookie.
+ */
+static void cachefiles_withdraw_cookie(struct fscache_cookie *cookie)
+{
+	struct cachefiles_object *object = cookie->cache_priv;
+	struct cachefiles_cache *cache = object->volume->cache;
+	const struct cred *saved_cred;
+
+	_enter("o=%x", object->debug_id);
+	cachefiles_see_object(object, cachefiles_obj_see_withdraw_cookie);
+
+	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);
+	}
+
+	if (object->file) {
+		cachefiles_begin_secure(cache, &saved_cred);
+		cachefiles_clean_up_object(object, cache);
+		cachefiles_end_secure(cache, saved_cred);
+	}
+
+	cookie->cache_priv = NULL;
+	cachefiles_put_object(object, cachefiles_obj_put_detach);
+}
+
+/*
+ * Invalidate the storage associated with a cookie.
+ */
+static bool cachefiles_invalidate_cookie(struct fscache_cookie *cookie)
+{
+	struct cachefiles_object *object = cookie->cache_priv;
+	struct file *new_file, *old_file;
+	bool old_tmpfile;
+
+	_enter("o=%x,[%llu]", object->debug_id, object->cookie->object_size);
+
+	old_tmpfile = test_bit(CACHEFILES_OBJECT_USING_TMPFILE, &object->flags);
+
+	if (!object->file) {
+		fscache_resume_after_invalidation(cookie);
+		_leave(" = t [light]");
+		return true;
+	}
+
+	new_file = cachefiles_create_tmpfile(object);
+	if (IS_ERR(new_file))
+		goto failed;
+
+	/* Substitute the VFS target */
+	_debug("sub");
+	spin_lock(&object->lock);
+
+	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);
+
+	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];
+
+			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);
+	}
+
+	_leave(" = t");
+	return true;
+
+failed:
+	_leave(" = f");
+	return false;
+}
+
 const struct fscache_cache_ops cachefiles_cache_ops = {
 	.name			= "cachefiles",
 	.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,
+	.prepare_to_write	= cachefiles_prepare_to_write,
 };
diff --git a/fs/cachefiles/internal.h b/fs/cachefiles/internal.h
index 654dbd51b965..d7aae04edc61 100644
--- a/fs/cachefiles/internal.h
+++ b/fs/cachefiles/internal.h
@@ -16,6 +16,8 @@
 #include <linux/cred.h>
 #include <linux/security.h>
 
+#define CACHEFILES_DIO_BLOCK_SIZE 4096
+
 struct cachefiles_cache;
 struct cachefiles_object;
 
@@ -68,6 +70,7 @@ struct cachefiles_cache {
 	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 */
@@ -194,6 +197,9 @@ extern int cachefiles_bury_object(struct cachefiles_cache *cache,
 				  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,
@@ -205,6 +211,9 @@ extern int cachefiles_cull(struct cachefiles_cache *cache, struct dentry *dir,
 
 extern int cachefiles_check_in_use(struct cachefiles_cache *cache,
 				   struct dentry *dir, char *filename);
+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
diff --git a/fs/cachefiles/namei.c b/fs/cachefiles/namei.c
index e87c401239f1..b549e9f79c01 100644
--- a/fs/cachefiles/namei.c
+++ b/fs/cachefiles/namei.c
@@ -404,6 +404,324 @@ try_again:
 	return 0;
 }
 
+/*
+ * Delete a cache file.
+ */
+int cachefiles_delete_object(struct cachefiles_object *object,
+			     enum fscache_why_object_killed why)
+{
+	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->debug_id, object->file);
+
+	/* Stop the dentry being negated if it's only pinned by a file struct. */
+	dget(dentry);
+
+	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;
+}
+
+/*
+ * Create a temporary file and leave it unattached and un-xattr'd until the
+ * time comes to discard the object from memory.
+ */
+struct file *cachefiles_create_tmpfile(struct cachefiles_object *object)
+{
+	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;
+	uint64_t ni_size = object->cookie->object_size;
+	long ret;
+
+	ni_size = round_up(ni_size, CACHEFILES_DIO_BLOCK_SIZE);
+
+	cachefiles_begin_secure(cache, &saved_cred);
+
+	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;
+		}
+	}
+
+	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);
+	}
+
+out_dput:
+	dput(path.dentry);
+out:
+	cachefiles_end_secure(cache, saved_cred);
+	return file;
+}
+
+/*
+ * Create a new file.
+ */
+static bool cachefiles_create_file(struct cachefiles_object *object)
+{
+	struct file *file;
+	int ret;
+
+	ret = cachefiles_has_space(object->volume->cache, 1, 0);
+	if (ret < 0)
+		return false;
+
+	file = cachefiles_create_tmpfile(object);
+	if (IS_ERR(file))
+		return false;
+
+	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;
+}
+
+/*
+ * 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;
+
+	_enter("%pd", dentry);
+
+	if (!cachefiles_mark_inode_in_use(object, dentry))
+		return false;
+
+	/* 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;
+	}
+
+	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);
+
+	ret = cachefiles_check_auxdata(object, file);
+	if (ret < 0)
+		goto check_failed;
+
+	object->file = file;
+
+	/* 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:
+	dput(dentry);
+	return false;
+}
+
+/*
+ * walk from the parent object to the child object through the backing
+ * filesystem, creating directories as we go
+ */
+bool cachefiles_look_up_object(struct cachefiles_object *object)
+{
+	struct cachefiles_volume *volume = object->volume;
+	struct dentry *dentry, *fan = volume->fanout[(u8)object->cookie->key_hash];
+	int ret;
+
+	_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;
+	}
+
+	if (!cachefiles_open_file(object, dentry))
+		return false;
+
+	_leave(" = t [%lu]", file_inode(object->file)->i_ino);
+	return true;
+
+new_file:
+	fscache_cookie_lookup_negative(object->cookie);
+	return cachefiles_create_file(object);
+}
+
+/*
+ * 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;
+
+	_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;
+		}
+
+		ret = cachefiles_unlink(volume->cache, object, fan, dentry,
+					FSCACHE_OBJECT_IS_STALE);
+		if (ret < 0)
+			goto out_dput;
+
+		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 = 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;
+}
+
 /*
  * Look up an inode to be checked or culled.  Return -EBUSY if the inode is
  * marked in use.
-- 
cgit v1.2.3


From 3929eca769b5a231010b4978acc61c0735da198f Mon Sep 17 00:00:00 2001
From: David Howells <dhowells@redhat.com>
Date: Thu, 21 Oct 2021 21:58:29 +0100
Subject: fscache, cachefiles: Display stats of no-space events

Add stat counters of no-space events that caused caching not to happen and
display in /proc/fs/fscache/stats.

Signed-off-by: David Howells <dhowells@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
cc: linux-cachefs@redhat.com
Link: https://lore.kernel.org/r/163819653216.215744.17210522251617386509.stgit@warthog.procyon.org.uk/ # v1
Link: https://lore.kernel.org/r/163906958369.143852.7257100711818401748.stgit@warthog.procyon.org.uk/ # v2
Link: https://lore.kernel.org/r/163967166917.1823006.14842444049198947892.stgit@warthog.procyon.org.uk/ # v3
Link: https://lore.kernel.org/r/164021566184.640689.4417328329632709265.stgit@warthog.procyon.org.uk/ # v4
---
 fs/cachefiles/cache.c         | 18 +++++++++++++++---
 fs/cachefiles/daemon.c        |  2 +-
 fs/cachefiles/internal.h      | 11 +++++++++--
 fs/cachefiles/io.c            |  7 +++++--
 fs/cachefiles/namei.c         |  6 ++++--
 fs/fscache/stats.c            |  8 ++++++++
 include/linux/fscache-cache.h |  6 ++++++
 7 files changed, 48 insertions(+), 10 deletions(-)

(limited to 'fs/cachefiles/cache.c')

diff --git a/fs/cachefiles/cache.c b/fs/cachefiles/cache.c
index e2cbbc08bad9..809519286335 100644
--- a/fs/cachefiles/cache.c
+++ b/fs/cachefiles/cache.c
@@ -147,7 +147,7 @@ int cachefiles_add_cache(struct cachefiles_cache *cache)
 	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(cache, 0, 0, cachefiles_has_space_check);
 	cachefiles_end_secure(cache, saved_cred);
 	_leave(" = 0 [%px]", cache->cache);
 	return 0;
@@ -175,7 +175,8 @@ error_getsec:
  * cache
  */
 int cachefiles_has_space(struct cachefiles_cache *cache,
-			 unsigned fnr, unsigned bnr)
+			 unsigned fnr, unsigned bnr,
+			 enum cachefiles_has_space_for reason)
 {
 	struct kstatfs stats;
 	u64 b_avail, b_writing;
@@ -233,7 +234,7 @@ int cachefiles_has_space(struct cachefiles_cache *cache,
 	ret = -ENOBUFS;
 	if (stats.f_ffree < cache->fstop ||
 	    b_avail < cache->bstop)
-		goto begin_cull;
+		goto stop_and_begin_cull;
 
 	ret = 0;
 	if (stats.f_ffree < cache->fcull ||
@@ -252,6 +253,17 @@ int cachefiles_has_space(struct cachefiles_cache *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 ###");
diff --git a/fs/cachefiles/daemon.c b/fs/cachefiles/daemon.c
index 45af558a696e..40a792421fc1 100644
--- a/fs/cachefiles/daemon.c
+++ b/fs/cachefiles/daemon.c
@@ -170,7 +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);
diff --git a/fs/cachefiles/internal.h b/fs/cachefiles/internal.h
index abdd1b66f6b9..8dd54d9375b6 100644
--- a/fs/cachefiles/internal.h
+++ b/fs/cachefiles/internal.h
@@ -130,10 +130,17 @@ static inline void cachefiles_state_changed(struct cachefiles_cache *cache)
  * cache.c
  */
 extern int cachefiles_add_cache(struct cachefiles_cache *cache);
-extern int cachefiles_has_space(struct cachefiles_cache *cache,
-				unsigned fnr, unsigned bnr);
 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
  */
diff --git a/fs/cachefiles/io.c b/fs/cachefiles/io.c
index 6f4dce0cfc36..60b1eac2ce78 100644
--- a/fs/cachefiles/io.c
+++ b/fs/cachefiles/io.c
@@ -468,7 +468,8 @@ static int __cachefiles_prepare_write(struct netfs_cache_resources *cres,
 	 * 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) == 0)
+	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();
@@ -483,6 +484,7 @@ static int __cachefiles_prepare_write(struct netfs_cache_resources *cres,
 		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,
@@ -498,7 +500,8 @@ static int __cachefiles_prepare_write(struct netfs_cache_resources *cres,
 	return ret;
 
 check_space:
-	return cachefiles_has_space(cache, 0, *_len / PAGE_SIZE);
+	return cachefiles_has_space(cache, 0, *_len / PAGE_SIZE,
+				    cachefiles_has_space_for_write);
 }
 
 static int cachefiles_prepare_write(struct netfs_cache_resources *cres,
diff --git a/fs/cachefiles/namei.c b/fs/cachefiles/namei.c
index b549e9f79c01..ab3ca598acac 100644
--- a/fs/cachefiles/namei.c
+++ b/fs/cachefiles/namei.c
@@ -115,7 +115,8 @@ retry:
 
 	/* we need to create the subdir if it doesn't exist yet */
 	if (d_is_negative(subdir)) {
-		ret = cachefiles_has_space(cache, 1, 0);
+		ret = cachefiles_has_space(cache, 1, 0,
+					   cachefiles_has_space_for_create);
 		if (ret < 0)
 			goto mkdir_error;
 
@@ -513,7 +514,8 @@ static bool cachefiles_create_file(struct cachefiles_object *object)
 	struct file *file;
 	int ret;
 
-	ret = cachefiles_has_space(object->volume->cache, 1, 0);
+	ret = cachefiles_has_space(object->volume->cache, 1, 0,
+				   cachefiles_has_space_for_create);
 	if (ret < 0)
 		return false;
 
diff --git a/fs/fscache/stats.c b/fs/fscache/stats.c
index 798ee68b3e9d..db2f4e225dd9 100644
--- a/fs/fscache/stats.c
+++ b/fs/fscache/stats.c
@@ -42,6 +42,10 @@ atomic_t fscache_n_read;
 EXPORT_SYMBOL(fscache_n_read);
 atomic_t fscache_n_write;
 EXPORT_SYMBOL(fscache_n_write);
+atomic_t fscache_n_no_write_space;
+EXPORT_SYMBOL(fscache_n_no_write_space);
+atomic_t fscache_n_no_create_space;
+EXPORT_SYMBOL(fscache_n_no_create_space);
 
 /*
  * display the general statistics
@@ -82,6 +86,10 @@ int fscache_stats_show(struct seq_file *m, void *v)
 		   atomic_read(&fscache_n_relinquishes_retire),
 		   atomic_read(&fscache_n_relinquishes_dropped));
 
+	seq_printf(m, "NoSpace: nwr=%u ncr=%u\n",
+		   atomic_read(&fscache_n_no_write_space),
+		   atomic_read(&fscache_n_no_create_space));
+
 	seq_printf(m, "IO     : rd=%u wr=%u\n",
 		   atomic_read(&fscache_n_read),
 		   atomic_read(&fscache_n_write));
diff --git a/include/linux/fscache-cache.h b/include/linux/fscache-cache.h
index 3fa4902dc87c..007e47f38610 100644
--- a/include/linux/fscache-cache.h
+++ b/include/linux/fscache-cache.h
@@ -186,11 +186,17 @@ static inline void fscache_wait_for_objects(struct fscache_cache *cache)
 #ifdef CONFIG_FSCACHE_STATS
 extern atomic_t fscache_n_read;
 extern atomic_t fscache_n_write;
+extern atomic_t fscache_n_no_write_space;
+extern atomic_t fscache_n_no_create_space;
 #define fscache_count_read() atomic_inc(&fscache_n_read)
 #define fscache_count_write() atomic_inc(&fscache_n_write)
+#define fscache_count_no_write_space() atomic_inc(&fscache_n_no_write_space)
+#define fscache_count_no_create_space() atomic_inc(&fscache_n_no_create_space)
 #else
 #define fscache_count_read() do {} while(0)
 #define fscache_count_write() do {} while(0)
+#define fscache_count_no_write_space() do {} while(0)
+#define fscache_count_no_create_space() do {} while(0)
 #endif
 
 #endif /* _LINUX_FSCACHE_CACHE_H */
-- 
cgit v1.2.3