exofs: Define on-disk per-inode optional layout attribute

* Layouts describe the way a file is spread on multiple devices.
  The layout information is stored in the objects attribute introduced
  in this patch.

* There can be multiple generating function for the layout.
  Currently defined:
    - No attribute present - use below moving-window on global
      device table, all devices.
      (This is the only one currently used in exofs)
    - an obj_id generated moving window - the obj_id is a randomizing
      factor in the otherwise global map layout.
    - An explicit layout stored, including a data_map and a device
      index list.
    - More might be defined in future ...

* There are two attributes defined of the same structure:
  A-data-files-layout - This layout is used by data-files. If present
                        at a directory, all files of that directory will
                        be created with this layout.
  A-meta-data-layout - This layout is used by a directory and other
                       meta-data information. Also inherited at creation
                       of subdirectories.

* At creation time inodes are created with the layout specified above.
  A usermode utility may change the creation layout on a give directory
  or file. Which in the case of directories, will also apply to newly
  created files/subdirectories, children of that directory.
  In the simple unaltered case of a newly created exofs, no layout
  attributes are present, and all layouts adhere to the layout specified
  at the device-table.

* In case of a future file system loaded in an old exofs-driver.
  At iget(), the generating_function is inspected and if not supported
  will return an IO error to the application and the inode will not
  be loaded. So not to damage any data.
  Note: After this patch we do not yet support any type of layout
        only the RAID0 patch that enables striping at the super-block
        level will add support for RAID0 layouts above. This way we
        are past and future compatible and fully bisectable.

* Access to the device table is done by an accessor since
  it will change according to above information.

Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>

fs/exofs/common.h

@@ -55,6 +55,8 @@
 /* exofs Application specific page/attribute */
 # define EXOFS_APAGE_FS_DATA	(OSD_APAGE_APP_DEFINED_FIRST + 3)
 # define EXOFS_ATTR_INODE_DATA	1
+# define EXOFS_ATTR_INODE_FILE_LAYOUT	2
+# define EXOFS_ATTR_INODE_DIR_LAYOUT	3
 
 /*
  * The maximum number of files we can have is limited by the size of the
@@ -206,4 +208,41 @@ enum {
 	(((name_len) + offsetof(struct exofs_dir_entry, name)  + \
 	  EXOFS_DIR_ROUND) & ~EXOFS_DIR_ROUND)
 
+/*
+ * The on-disk (optional) layout structure.
+ * sits in an EXOFS_ATTR_INODE_FILE_LAYOUT or EXOFS_ATTR_INODE_DIR_LAYOUT
+ * attribute, attached to any inode, usually to a directory.
+ */
+
+enum exofs_inode_layout_gen_functions {
+	LAYOUT_MOVING_WINDOW = 0,
+	LAYOUT_IMPLICT = 1,
+};
+
+struct exofs_on_disk_inode_layout {
+	__le16 gen_func; /* One of enum exofs_inode_layout_gen_functions */
+	__le16 pad;
+	union {
+		/* gen_func == LAYOUT_MOVING_WINDOW (default) */
+		struct exofs_layout_sliding_window {
+			__le32 num_devices; /* first n devices in global-table*/
+		} sliding_window __packed;
+
+		/* gen_func == LAYOUT_IMPLICT */
+		struct exofs_layout_implict_list {
+			struct exofs_dt_data_map data_map;
+			/* Variable array of size data_map.cb_num_comps. These
+			 * are device indexes of the devices in the global table
+			 */
+			__le32 dev_indexes[];
+		} implict __packed;
+	};
+} __packed;
+
+static inline size_t exofs_on_disk_inode_layout_size(unsigned max_devs)
+{
+	return sizeof(struct exofs_on_disk_inode_layout) +
+		max_devs * sizeof(__le32);
+}
+
 #endif /*ifndef __EXOFS_COM_H__*/

fs/exofs/exofs.h

@@ -185,6 +185,12 @@ static inline struct exofs_i_info *exofs_i(struct inode *inode)
 	return container_of(inode, struct exofs_i_info, vfs_inode);
 }
 
+/*
+ * Given a layout, object_number and stripe_index return the associated global
+ * dev_index
+ */
+unsigned exofs_layout_od_id(struct exofs_layout *layout,
+			    osd_id obj_no, unsigned layout_index);
 /*
  * Maximum count of links to a file
  */

fs/exofs/inode.c

@@ -859,6 +859,15 @@ int exofs_setattr(struct dentry *dentry, struct iattr *iattr)
 	return error;
 }
 
+static const struct osd_attr g_attr_inode_file_layout = ATTR_DEF(
+	EXOFS_APAGE_FS_DATA,
+	EXOFS_ATTR_INODE_FILE_LAYOUT,
+	0);
+static const struct osd_attr g_attr_inode_dir_layout = ATTR_DEF(
+	EXOFS_APAGE_FS_DATA,
+	EXOFS_ATTR_INODE_DIR_LAYOUT,
+	0);
+
 /*
  * Read an inode from the OSD, and return it as is.  We also return the size
  * attribute in the 'obj_size' argument.
@@ -867,11 +876,16 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi,
 		    struct exofs_fcb *inode, uint64_t *obj_size)
 {
 	struct exofs_sb_info *sbi = sb->s_fs_info;
-	struct osd_attr attrs[2];
+	struct osd_attr attrs[] = {
+		[0] = g_attr_inode_data,
+		[1] = g_attr_inode_file_layout,
+		[2] = g_attr_inode_dir_layout,
+		[3] = g_attr_logical_length,
+	};
 	struct exofs_io_state *ios;
+	struct exofs_on_disk_inode_layout *layout;
 	int ret;
 
-	*obj_size = ~0;
 	ret = exofs_get_io_state(&sbi->layout, &ios);
 	if (unlikely(ret)) {
 		EXOFS_ERR("%s: exofs_get_io_state failed.\n", __func__);
@@ -882,8 +896,9 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi,
 	exofs_make_credential(oi->i_cred, &ios->obj);
 	ios->cred = oi->i_cred;
 
-	attrs[0] = g_attr_inode_data;
-	attrs[1] = g_attr_logical_length;
+	attrs[1].len = exofs_on_disk_inode_layout_size(sbi->layout.s_numdevs);
+	attrs[2].len = exofs_on_disk_inode_layout_size(sbi->layout.s_numdevs);
+
 	ios->in_attr = attrs;
 	ios->in_attr_len = ARRAY_SIZE(attrs);
 
@@ -900,12 +915,43 @@ static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi,
 	memcpy(inode, attrs[0].val_ptr, EXOFS_INO_ATTR_SIZE);
 
 	ret = extract_attr_from_ios(ios, &attrs[1]);
+	if (ret) {
+		EXOFS_ERR("%s: extract_attr of inode_data failed\n", __func__);
+		goto out;
+	}
+	if (attrs[1].len) {
+		layout = attrs[1].val_ptr;
+		if (layout->gen_func != cpu_to_le16(LAYOUT_MOVING_WINDOW)) {
+			EXOFS_ERR("%s: unsupported files layout %d\n",
+				__func__, layout->gen_func);
+			ret = -ENOTSUPP;
+			goto out;
+		}
+	}
+
+	ret = extract_attr_from_ios(ios, &attrs[2]);
+	if (ret) {
+		EXOFS_ERR("%s: extract_attr of inode_data failed\n", __func__);
+		goto out;
+	}
+	if (attrs[2].len) {
+		layout = attrs[2].val_ptr;
+		if (layout->gen_func != cpu_to_le16(LAYOUT_MOVING_WINDOW)) {
+			EXOFS_ERR("%s: unsupported meta-data layout %d\n",
+				__func__, layout->gen_func);
+			ret = -ENOTSUPP;
+			goto out;
+		}
+	}
+
+	*obj_size = ~0;
+	ret = extract_attr_from_ios(ios, &attrs[3]);
 	if (ret) {
 		EXOFS_ERR("%s: extract_attr of logical_length failed\n",
 			  __func__);
 		goto out;
 	}
-	*obj_size = get_unaligned_be64(attrs[1].val_ptr);
+	*obj_size = get_unaligned_be64(attrs[3].val_ptr);
 
 out:
 	exofs_put_io_state(ios);

fs/exofs/ios.c

@@ -107,6 +107,19 @@ void exofs_put_io_state(struct exofs_io_state *ios)
 	}
 }
 
+unsigned exofs_layout_od_id(struct exofs_layout *layout,
+			    osd_id obj_no, unsigned layout_index)
+{
+	return layout_index;
+}
+
+static inline struct osd_dev *exofs_ios_od(struct exofs_io_state *ios,
+					   unsigned layout_index)
+{
+	return ios->layout->s_ods[
+		exofs_layout_od_id(ios->layout, ios->obj.id, layout_index)];
+}
+
 static void _sync_done(struct exofs_io_state *ios, void *p)
 {
 	struct completion *waiting = p;
@@ -242,7 +255,7 @@ int exofs_sbi_create(struct exofs_io_state *ios)
 	for (i = 0; i < ios->layout->s_numdevs; i++) {
 		struct osd_request *or;
 
-		or = osd_start_request(ios->layout->s_ods[i], GFP_KERNEL);
+		or = osd_start_request(exofs_ios_od(ios, i), GFP_KERNEL);
 		if (unlikely(!or)) {
 			EXOFS_ERR("%s: osd_start_request failed\n", __func__);
 			ret = -ENOMEM;
@@ -266,7 +279,7 @@ int exofs_sbi_remove(struct exofs_io_state *ios)
 	for (i = 0; i < ios->layout->s_numdevs; i++) {
 		struct osd_request *or;
 
-		or = osd_start_request(ios->layout->s_ods[i], GFP_KERNEL);
+		or = osd_start_request(exofs_ios_od(ios, i), GFP_KERNEL);
 		if (unlikely(!or)) {
 			EXOFS_ERR("%s: osd_start_request failed\n", __func__);
 			ret = -ENOMEM;
@@ -290,7 +303,7 @@ int exofs_sbi_write(struct exofs_io_state *ios)
 	for (i = 0; i < ios->layout->s_numdevs; i++) {
 		struct osd_request *or;
 
-		or = osd_start_request(ios->layout->s_ods[i], GFP_KERNEL);
+		or = osd_start_request(exofs_ios_od(ios, i), GFP_KERNEL);
 		if (unlikely(!or)) {
 			EXOFS_ERR("%s: osd_start_request failed\n", __func__);
 			ret = -ENOMEM;
@@ -361,7 +374,7 @@ int exofs_sbi_read(struct exofs_io_state *ios)
 	unsigned first_dev = (unsigned)ios->obj.id;
 
 	first_dev %= ios->layout->s_numdevs;
-	or = osd_start_request(ios->layout->s_ods[first_dev], GFP_KERNEL);
+	or = osd_start_request(exofs_ios_od(ios, first_dev), GFP_KERNEL);
 	if (unlikely(!or)) {
 		EXOFS_ERR("%s: osd_start_request failed\n", __func__);
 		return -ENOMEM;
@@ -442,7 +455,7 @@ int exofs_oi_truncate(struct exofs_i_info *oi, u64 size)
 	for (i = 0; i < sbi->layout.s_numdevs; i++) {
 		struct osd_request *or;
 
-		or = osd_start_request(sbi->layout.s_ods[i], GFP_KERNEL);
+		or = osd_start_request(exofs_ios_od(ios, i), GFP_KERNEL);
 		if (unlikely(!or)) {
 			EXOFS_ERR("%s: osd_start_request failed\n", __func__);
 			ret = -ENOMEM;