cachefiles: Implement culling daemon commands

Implement the ability for the userspace daemon to try and cull a file or
directory in the cache.  Two daemon commands are implemented:

 (1) The "inuse" command.  This queries if a file is in use or whether it
     can be deleted.  It checks the S_KERNEL_FILE flag on the inode
     referred to by the specified filename.

 (2) The "cull" command.  This asks for a file or directory to be removed,
     where removal means either unlinking it or moving it to the graveyard
     directory for userspace to dismantle.

Changes
=======
ver #2:
 - Fix logging of wrong error[1].
 - Need to unmark an inode we've moved to the graveyard before unlocking.

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/163819643179.215744.13641580295708315695.stgit@warthog.procyon.org.uk/ # v1
Link: https://lore.kernel.org/r/163906945705.143852.8177595531814485350.stgit@warthog.procyon.org.uk/ # v2
Link: https://lore.kernel.org/r/163967155792.1823006.1088936326902550910.stgit@warthog.procyon.org.uk/ # v3
Link: https://lore.kernel.org/r/164021555037.640689.9472627499842585255.stgit@warthog.procyon.org.uk/ # v4

1 files changed, 307 insertions, 0 deletions

diff --git a/fs/cachefiles/namei.c b/fs/cachefiles/namei.c
index db60a671c3fc..e87c401239f1 100644
--- a/fs/cachefiles/namei.c
+++ b/fs/cachefiles/namei.c
@@ -217,3 +217,310 @@ void cachefiles_put_directory(struct dentry *dir)
 		dput(dir);
 	}
 }
+
+/*
+ * 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;
+
+	trace_cachefiles_unlink(object, dentry, why);
+	ret = security_path_unlink(&path, dentry);
+	if (ret < 0) {
+		cachefiles_io_error(cache, "Unlink security error");
+		return ret;
+	}
+
+	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
+ */
+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;
+	char nbuffer[8 + 8 + 1];
+	int ret;
+
+	_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)) {
+		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));
+		_leave(" = %d", ret);
+		return ret;
+	}
+
+	/* directories have to be moved to the graveyard */
+	_debug("move stale object to graveyard");
+	inode_unlock(d_inode(dir));
+
+try_again:
+	/* first step is to make up a grave dentry in the graveyard */
+	sprintf(nbuffer, "%08x%08x",
+		(uint32_t) ktime_get_real_seconds(),
+		(uint32_t) atomic_inc_return(&cache->gravecounter));
+
+	/* do the multiway lock magic */
+	trap = lock_rename(cache->graveyard, dir);
+
+	/* do some checks before getting the grave dentry */
+	if (rep->d_parent != dir || IS_DEADDIR(d_inode(rep))) {
+		/* the entry was probably culled when we dropped the parent dir
+		 * lock */
+		unlock_rename(cache->graveyard, dir);
+		_leave(" = 0 [culled?]");
+		return 0;
+	}
+
+	if (!d_can_lookup(cache->graveyard)) {
+		unlock_rename(cache->graveyard, dir);
+		cachefiles_io_error(cache, "Graveyard no longer a directory");
+		return -EIO;
+	}
+
+	if (trap == rep) {
+		unlock_rename(cache->graveyard, dir);
+		cachefiles_io_error(cache, "May not make directory loop");
+		return -EIO;
+	}
+
+	if (d_mountpoint(rep)) {
+		unlock_rename(cache->graveyard, dir);
+		cachefiles_io_error(cache, "Mountpoint in cache");
+		return -EIO;
+	}
+
+	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));
+		return -EIO;
+	}
+
+	if (d_is_positive(grave)) {
+		unlock_rename(cache->graveyard, dir);
+		dput(grave);
+		grave = NULL;
+		cond_resched();
+		goto try_again;
+	}
+
+	if (d_mountpoint(grave)) {
+		unlock_rename(cache->graveyard, dir);
+		dput(grave);
+		cachefiles_io_error(cache, "Mountpoint in graveyard");
+		return -EIO;
+	}
+
+	/* target should not be an ancestor of source */
+	if (trap == grave) {
+		unlock_rename(cache->graveyard, dir);
+		dput(grave);
+		cachefiles_io_error(cache, "May not make directory loop");
+		return -EIO;
+	}
+
+	/* attempt the rename */
+	path.mnt = cache->mnt;
+	path.dentry = dir;
+	path_to_graveyard.mnt = cache->mnt;
+	path_to_graveyard.dentry = cache->graveyard;
+	ret = security_path_rename(&path, rep, &path_to_graveyard, grave, 0);
+	if (ret < 0) {
+		cachefiles_io_error(cache, "Rename security error %d", ret);
+	} else {
+		struct renamedata rd = {
+			.old_mnt_userns	= &init_user_ns,
+			.old_dir	= d_inode(dir),
+			.old_dentry	= rep,
+			.new_mnt_userns	= &init_user_ns,
+			.new_dir	= d_inode(cache->graveyard),
+			.new_dentry	= grave,
+		};
+		trace_cachefiles_rename(object, rep, grave, why);
+		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);
+	}
+
+	__cachefiles_unmark_inode_in_use(object, rep);
+	unlock_rename(cache->graveyard, dir);
+	dput(grave);
+	_leave(" = 0");
+	return 0;
+}
+
+/*
+ * Look up an inode to be checked or culled.  Return -EBUSY if the inode is
+ * marked in use.
+ */
+static struct dentry *cachefiles_lookup_for_cull(struct cachefiles_cache *cache,
+						 struct dentry *dir,
+						 char *filename)
+{
+	struct dentry *victim;
+	int ret = -ENOENT;
+
+	inode_lock_nested(d_inode(dir), I_MUTEX_PARENT);
+
+	victim = lookup_one_len(filename, dir, strlen(filename));
+	if (IS_ERR(victim))
+		goto lookup_error;
+	if (d_is_negative(victim))
+		goto lookup_put;
+	if (d_inode(victim)->i_flags & S_KERNEL_FILE)
+		goto lookup_busy;
+	return victim;
+
+lookup_busy:
+	ret = -EBUSY;
+lookup_put:
+	inode_unlock(d_inode(dir));
+	dput(victim);
+	return ERR_PTR(ret);
+
+lookup_error:
+	inode_unlock(d_inode(dir));
+	ret = PTR_ERR(victim);
+	if (ret == -ENOENT)
+		return ERR_PTR(-ESTALE); /* Probably got retired by the netfs */
+
+	if (ret == -EIO) {
+		cachefiles_io_error(cache, "Lookup failed");
+	} else if (ret != -ENOMEM) {
+		pr_err("Internal error: %d\n", ret);
+		ret = -EIO;
+	}
+
+	return ERR_PTR(ret);
+}
+
+/*
+ * Cull an object if it's not in use
+ * - called only by cache manager daemon
+ */
+int cachefiles_cull(struct cachefiles_cache *cache, struct dentry *dir,
+		    char *filename)
+{
+	struct dentry *victim;
+	struct inode *inode;
+	int ret;
+
+	_enter(",%pd/,%s", dir, filename);
+
+	victim = cachefiles_lookup_for_cull(cache, dir, filename);
+	if (IS_ERR(victim))
+		return PTR_ERR(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;
+
+	ret = cachefiles_bury_object(cache, NULL, dir, victim,
+				     FSCACHE_OBJECT_WAS_CULLED);
+	if (ret < 0)
+		goto error;
+
+	dput(victim);
+	_leave(" = 0");
+	return 0;
+
+error_unlock:
+	inode_unlock(d_inode(dir));
+error:
+	dput(victim);
+	if (ret == -ENOENT)
+		return -ESTALE; /* Probably got retired by the netfs */
+
+	if (ret != -ENOMEM) {
+		pr_err("Internal error: %d\n", ret);
+		ret = -EIO;
+	}
+
+	_leave(" = %d", ret);
+	return ret;
+}
+
+/*
+ * 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
+ */
+int cachefiles_check_in_use(struct cachefiles_cache *cache, struct dentry *dir,
+			    char *filename)
+{
+	struct dentry *victim;
+	int ret = 0;
+
+	victim = cachefiles_lookup_for_cull(cache, dir, filename);
+	if (IS_ERR(victim))
+		return PTR_ERR(victim);
+
+	inode_unlock(d_inode(dir));
+	dput(victim);
+	return ret;
+}