afs: Fix directory handling

AFS directories are structured blobs that are downloaded just like files
and then parsed by the lookup and readdir code and, as such, are currently
handled in the pagecache like any other file, with the entire directory
content being thrown away each time the directory changes.

However, since the blob is a known structure and since the data version
counter on a directory increases by exactly one for each change committed
to that directory, we can actually edit the directory locally rather than
fetching it from the server after each locally-induced change.

What we can't do, though, is mix data from the server and data from the
client since the server is technically at liberty to rearrange or compress
a directory if it sees fit, provided it updates the data version number
when it does so and breaks the callback (ie. sends a notification).

Further, lookup with lookup-ahead, readdir and, when it arrives, local
editing are likely want to scan the whole of a directory.

So directory handling needs to be improved to maintain the coherency of the
directory blob prior to permitting local directory editing.

To this end:

 (1) If any directory page gets discarded, invalidate and reread the entire
     directory.

 (2) If readpage notes that if when it fetches a single page that the
     version number has changed, the entire directory is flagged for
     invalidation.

 (3) Read as much of the directory in one go as we can.

Note that this removes local caching of directories in fscache for the
moment as we can't pass the pages to fscache_read_or_alloc_pages() since
page->lru is in use by the LRU.
Signed-off-by: David Howells <dhowells@redhat.com>

fs/afs/file.c

@@ -187,10 +187,12 @@ void afs_put_read(struct afs_read *req)
 {
 	int i;
 
-	if (atomic_dec_and_test(&req->usage)) {
+	if (refcount_dec_and_test(&req->usage)) {
 		for (i = 0; i < req->nr_pages; i++)
 			if (req->pages[i])
 				put_page(req->pages[i]);
+		if (req->pages != req->array)
+			kfree(req->pages);
 		kfree(req);
 	}
 }
@@ -297,10 +299,11 @@ int afs_page_filler(void *data, struct page *page)
 		 * end of the file, the server will return a short read and the
 		 * unmarshalling code will clear the unfilled space.
 		 */
-		atomic_set(&req->usage, 1);
+		refcount_set(&req->usage, 1);
 		req->pos = (loff_t)page->index << PAGE_SHIFT;
 		req->len = PAGE_SIZE;
 		req->nr_pages = 1;
+		req->pages = req->array;
 		req->pages[0] = page;
 		get_page(page);
 
@@ -309,10 +312,6 @@ int afs_page_filler(void *data, struct page *page)
 		ret = afs_fetch_data(vnode, key, req);
 		afs_put_read(req);
 
-		if (ret >= 0 && S_ISDIR(inode->i_mode) &&
-		    !afs_dir_check_page(inode, page))
-			ret = -EIO;
-
 		if (ret < 0) {
 			if (ret == -ENOENT) {
 				_debug("got NOENT from server"
@@ -447,10 +446,11 @@ static int afs_readpages_one(struct file *file, struct address_space *mapping,
 	if (!req)
 		return -ENOMEM;
 
-	atomic_set(&req->usage, 1);
+	refcount_set(&req->usage, 1);
 	req->page_done = afs_readpages_page_done;
 	req->pos = first->index;
 	req->pos <<= PAGE_SHIFT;
+	req->pages = req->array;
 
 	/* Transfer the pages to the request.  We add them in until one fails
 	 * to add to the LRU and then we stop (as that'll make a hole in the

fs/afs/fsclient.c

@@ -101,10 +101,14 @@ void afs_update_inode_from_status(struct afs_vnode *vnode,
 			       vnode->fid.vid, vnode->fid.vnode,
 			       (unsigned long long) *expected_version);
 			vnode->invalid_before = status->data_version;
-			set_bit(AFS_VNODE_DIR_MODIFIED, &vnode->flags);
+			if (vnode->status.type == AFS_FTYPE_DIR) {
+				if (test_and_clear_bit(AFS_VNODE_DIR_VALID, &vnode->flags))
+					afs_stat_v(vnode, n_inval);
+			} else {
 				set_bit(AFS_VNODE_ZAP_DATA, &vnode->flags);
 			}
 		}
+	}
 
 	if (flags & (AFS_VNODE_DATA_CHANGED | AFS_VNODE_NOT_YET_SET)) {
 		inode_set_iversion_raw(&vnode->vfs_inode, status->data_version);
@@ -119,7 +123,7 @@ static int xdr_decode_AFSFetchStatus(const __be32 **_bp,
 				     struct afs_file_status *status,
 				     struct afs_vnode *vnode,
 				     const afs_dataversion_t *expected_version,
-				     afs_dataversion_t *_version)
+				     struct afs_read *read_req)
 {
 	const struct afs_xdr_AFSFetchStatus *xdr = (const void *)*_bp;
 	u64 data_version, size;
@@ -197,8 +201,11 @@ static int xdr_decode_AFSFetchStatus(const __be32 **_bp,
 		status->data_version = data_version;
 		flags |= AFS_VNODE_DATA_CHANGED;
 	}
-	if (_version)
-		*_version = data_version;
+
+	if (read_req) {
+		read_req->data_version = data_version;
+		read_req->file_size = size;
+	}
 
 	*_bp = (const void *)*_bp + sizeof(*xdr);
 
@@ -543,8 +550,7 @@ static int afs_deliver_fs_fetch_data(struct afs_call *call)
 
 		bp = call->buffer;
 		if (xdr_decode_AFSFetchStatus(&bp, &vnode->status, vnode,
-					      &vnode->status.data_version,
-					      &req->new_version) < 0)
+					      &vnode->status.data_version, req) < 0)
 			return -EBADMSG;
 		xdr_decode_AFSCallBack(call, vnode, &bp);
 		if (call->reply[1])
@@ -628,7 +634,7 @@ static int afs_fs_fetch_data64(struct afs_fs_cursor *fc, struct afs_read *req)
 	bp[6] = 0;
 	bp[7] = htonl(lower_32_bits(req->len));
 
-	atomic_inc(&req->usage);
+	refcount_inc(&req->usage);
 	call->cb_break = fc->cb_break;
 	afs_use_fs_server(call, fc->cbi);
 	trace_afs_make_fs_call(call, &vnode->fid);
@@ -671,7 +677,7 @@ int afs_fs_fetch_data(struct afs_fs_cursor *fc, struct afs_read *req)
 	bp[4] = htonl(lower_32_bits(req->pos));
 	bp[5] = htonl(lower_32_bits(req->len));
 
-	atomic_inc(&req->usage);
+	refcount_inc(&req->usage);
 	call->cb_break = fc->cb_break;
 	afs_use_fs_server(call, fc->cbi);
 	trace_afs_make_fs_call(call, &vnode->fid);

fs/afs/inode.c

@@ -53,11 +53,13 @@ static int afs_inode_init_from_status(struct afs_vnode *vnode, struct key *key)
 		inode->i_mode	= S_IFREG | vnode->status.mode;
 		inode->i_op	= &afs_file_inode_operations;
 		inode->i_fop	= &afs_file_operations;
+		inode->i_mapping->a_ops	= &afs_fs_aops;
 		break;
 	case AFS_FTYPE_DIR:
 		inode->i_mode	= S_IFDIR | vnode->status.mode;
 		inode->i_op	= &afs_dir_inode_operations;
 		inode->i_fop	= &afs_dir_file_operations;
+		inode->i_mapping->a_ops	= &afs_dir_aops;
 		break;
 	case AFS_FTYPE_SYMLINK:
 		/* Symlinks with a mode of 0644 are actually mountpoints. */
@@ -69,9 +71,11 @@ static int afs_inode_init_from_status(struct afs_vnode *vnode, struct key *key)
 			inode->i_mode	= S_IFDIR | 0555;
 			inode->i_op	= &afs_mntpt_inode_operations;
 			inode->i_fop	= &afs_mntpt_file_operations;
+			inode->i_mapping->a_ops	= &afs_fs_aops;
 		} else {
 			inode->i_mode	= S_IFLNK | vnode->status.mode;
 			inode->i_op	= &afs_symlink_inode_operations;
+			inode->i_mapping->a_ops	= &afs_fs_aops;
 		}
 		inode_nohighmem(inode);
 		break;
@@ -82,15 +86,9 @@ static int afs_inode_init_from_status(struct afs_vnode *vnode, struct key *key)
 	}
 
 	inode->i_blocks		= 0;
-	inode->i_mapping->a_ops	= &afs_fs_aops;
 	vnode->invalid_before	= vnode->status.data_version;
 
 	read_sequnlock_excl(&vnode->cb_lock);
-
-#ifdef CONFIG_AFS_FSCACHE
-	if (vnode->status.size > 0)
-		fscache_attr_changed(vnode->cache);
-#endif
 	return 0;
 }
 
@@ -247,6 +245,11 @@ static void afs_get_inode_cache(struct afs_vnode *vnode)
 	} __packed key;
 	struct afs_vnode_cache_aux aux;
 
+	if (vnode->status.type == AFS_FTYPE_DIR) {
+		vnode->cache = NULL;
+		return;
+	}
+
 	key.vnode_id		= vnode->fid.vnode;
 	key.unique		= vnode->fid.unique;
 	key.vnode_id_ext[0]	= 0;
@@ -338,10 +341,6 @@ struct inode *afs_iget(struct super_block *sb, struct key *key,
 
 	/* failure */
 bad_inode:
-#ifdef CONFIG_AFS_FSCACHE
-	fscache_relinquish_cookie(vnode->cache, NULL, ret == -ENOENT);
-	vnode->cache = NULL;
-#endif
 	iget_failed(inode);
 	_leave(" = %d [bad]", ret);
 	return ERR_PTR(ret);
@@ -355,9 +354,6 @@ void afs_zap_data(struct afs_vnode *vnode)
 {
 	_enter("{%x:%u}", vnode->fid.vid, vnode->fid.vnode);
 
-	if (S_ISDIR(vnode->vfs_inode.i_mode))
-		afs_stat_v(vnode, n_inval);
-
 #ifdef CONFIG_AFS_FSCACHE
 	fscache_invalidate(vnode->cache);
 #endif
@@ -399,7 +395,7 @@ int afs_validate(struct afs_vnode *vnode, struct key *key)
 	if (test_bit(AFS_VNODE_CB_PROMISED, &vnode->flags)) {
 		if (vnode->cb_s_break != vnode->cb_interest->server->cb_s_break) {
 			vnode->cb_s_break = vnode->cb_interest->server->cb_s_break;
-		} else if (!test_bit(AFS_VNODE_DIR_MODIFIED, &vnode->flags) &&
+		} else if (test_bit(AFS_VNODE_DIR_VALID, &vnode->flags) &&
 			   !test_bit(AFS_VNODE_ZAP_DATA, &vnode->flags) &&
 			   vnode->cb_expires_at - 10 > now) {
 				valid = true;
@@ -445,8 +441,6 @@ int afs_validate(struct afs_vnode *vnode, struct key *key)
 	 * different */
 	if (test_and_clear_bit(AFS_VNODE_ZAP_DATA, &vnode->flags))
 		afs_zap_data(vnode);
-
-	clear_bit(AFS_VNODE_DIR_MODIFIED, &vnode->flags);
 	mutex_unlock(&vnode->validate_lock);
 valid:
 	_leave(" = 0");

fs/afs/internal.h

@@ -174,12 +174,14 @@ struct afs_read {
 	loff_t			len;		/* How much we're asking for */
 	loff_t			actual_len;	/* How much we're actually getting */
 	loff_t			remain;		/* Amount remaining */
-	afs_dataversion_t	new_version;	/* Version number returned by server */
-	atomic_t		usage;
+	loff_t			file_size;	/* File size returned by server */
+	afs_dataversion_t	data_version;	/* Version number returned by server */
+	refcount_t		usage;
 	unsigned int		index;		/* Which page we're reading into */
 	unsigned int		nr_pages;
 	void (*page_done)(struct afs_call *, struct afs_read *);
-	struct page		*pages[];
+	struct page		**pages;
+	struct page		*array[];
 };
 
 /*
@@ -267,6 +269,7 @@ struct afs_net {
 	atomic_t		n_lookup;	/* Number of lookups done */
 	atomic_t		n_reval;	/* Number of dentries needing revalidation */
 	atomic_t		n_inval;	/* Number of invalidations by the server */
+	atomic_t		n_relpg;	/* Number of invalidations by releasepage */
 	atomic_t		n_read_dir;	/* Number of directory pages read */
 };
 
@@ -491,7 +494,7 @@ struct afs_vnode {
 	unsigned long		flags;
 #define AFS_VNODE_CB_PROMISED	0		/* Set if vnode has a callback promise */
 #define AFS_VNODE_UNSET		1		/* set if vnode attributes not yet set */
-#define AFS_VNODE_DIR_MODIFIED	2		/* set if dir vnode's data modified */
+#define AFS_VNODE_DIR_VALID	2		/* Set if dir contents are valid */
 #define AFS_VNODE_ZAP_DATA	3		/* set if vnode's data should be invalidated */
 #define AFS_VNODE_DELETED	4		/* set if vnode deleted on server */
 #define AFS_VNODE_MOUNTPOINT	5		/* set if vnode is a mountpoint symlink */
@@ -671,9 +674,9 @@ extern bool afs_cm_incoming_call(struct afs_call *);
  */
 extern const struct file_operations afs_dir_file_operations;
 extern const struct inode_operations afs_dir_inode_operations;
+extern const struct address_space_operations afs_dir_aops;
 extern const struct dentry_operations afs_fs_dentry_operations;
 
-extern bool afs_dir_check_page(struct inode *, struct page *);
 extern void afs_d_release(struct dentry *);
 
 /*

fs/afs/proc.c

@@ -910,10 +910,11 @@ static int afs_proc_stats_show(struct seq_file *m, void *v)
 
 	seq_puts(m, "kAFS statistics\n");
 
-	seq_printf(m, "dir-mgmt: look=%u reval=%u inval=%u\n",
+	seq_printf(m, "dir-mgmt: look=%u reval=%u inval=%u relpg=%u\n",
 		   atomic_read(&net->n_lookup),
 		   atomic_read(&net->n_reval),
-		   atomic_read(&net->n_inval));
+		   atomic_read(&net->n_inval),
+		   atomic_read(&net->n_relpg));
 
 	seq_printf(m, "dir-data: rdpg=%u\n",
 		   atomic_read(&net->n_read_dir));

fs/afs/write.c

@@ -42,10 +42,11 @@ static int afs_fill_page(struct afs_vnode *vnode, struct key *key,
 	if (!req)
 		return -ENOMEM;
 
-	atomic_set(&req->usage, 1);
+	refcount_set(&req->usage, 1);
 	req->pos = pos;
 	req->len = len;
 	req->nr_pages = 1;
+	req->pages = req->array;
 	req->pages[0] = page;
 	get_page(page);