2 * Copyright (C) 2007 Oracle. All rights reserved.
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public
6 * License v2 as published by the Free Software Foundation.
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
13 * You should have received a copy of the GNU General Public
14 * License along with this program; if not, write to the
15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
16 * Boston, MA 021110-1307, USA.
18 #include <linux/sched.h>
19 #include <linux/bio.h>
20 #include <linux/buffer_head.h>
21 #include <asm/div64.h>
23 #include "extent_map.h"
25 #include "transaction.h"
26 #include "print-tree.h"
37 struct btrfs_bio_stripe stripes[];
40 #define map_lookup_size(n) (sizeof(struct map_lookup) + \
41 (sizeof(struct btrfs_bio_stripe) * (n)))
43 static DEFINE_MUTEX(uuid_mutex);
44 static LIST_HEAD(fs_uuids);
46 int btrfs_cleanup_fs_uuids(void)
48 struct btrfs_fs_devices *fs_devices;
49 struct list_head *uuid_cur;
50 struct list_head *devices_cur;
51 struct btrfs_device *dev;
53 list_for_each(uuid_cur, &fs_uuids) {
54 fs_devices = list_entry(uuid_cur, struct btrfs_fs_devices,
56 while(!list_empty(&fs_devices->devices)) {
57 devices_cur = fs_devices->devices.next;
58 dev = list_entry(devices_cur, struct btrfs_device,
60 printk("uuid cleanup finds %s\n", dev->name);
63 close_bdev_excl(dev->bdev);
65 list_del(&dev->dev_list);
72 static struct btrfs_device *__find_device(struct list_head *head, u64 devid)
74 struct btrfs_device *dev;
75 struct list_head *cur;
77 list_for_each(cur, head) {
78 dev = list_entry(cur, struct btrfs_device, dev_list);
79 if (dev->devid == devid)
85 static struct btrfs_fs_devices *find_fsid(u8 *fsid)
87 struct list_head *cur;
88 struct btrfs_fs_devices *fs_devices;
90 list_for_each(cur, &fs_uuids) {
91 fs_devices = list_entry(cur, struct btrfs_fs_devices, list);
92 if (memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE) == 0)
98 static int device_list_add(const char *path,
99 struct btrfs_super_block *disk_super,
100 u64 devid, struct btrfs_fs_devices **fs_devices_ret)
102 struct btrfs_device *device;
103 struct btrfs_fs_devices *fs_devices;
104 u64 found_transid = btrfs_super_generation(disk_super);
106 fs_devices = find_fsid(disk_super->fsid);
108 fs_devices = kmalloc(sizeof(*fs_devices), GFP_NOFS);
111 INIT_LIST_HEAD(&fs_devices->devices);
112 list_add(&fs_devices->list, &fs_uuids);
113 memcpy(fs_devices->fsid, disk_super->fsid, BTRFS_FSID_SIZE);
114 fs_devices->latest_devid = devid;
115 fs_devices->latest_trans = found_transid;
116 fs_devices->lowest_devid = (u64)-1;
117 fs_devices->num_devices = 0;
120 device = __find_device(&fs_devices->devices, devid);
123 device = kzalloc(sizeof(*device), GFP_NOFS);
125 /* we can safely leave the fs_devices entry around */
128 device->devid = devid;
129 device->barriers = 1;
130 spin_lock_init(&device->io_lock);
131 device->name = kstrdup(path, GFP_NOFS);
136 list_add(&device->dev_list, &fs_devices->devices);
137 fs_devices->num_devices++;
140 if (found_transid > fs_devices->latest_trans) {
141 fs_devices->latest_devid = devid;
142 fs_devices->latest_trans = found_transid;
144 if (fs_devices->lowest_devid > devid) {
145 fs_devices->lowest_devid = devid;
146 printk("lowest devid now %Lu\n", devid);
148 *fs_devices_ret = fs_devices;
152 int btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
154 struct list_head *head = &fs_devices->devices;
155 struct list_head *cur;
156 struct btrfs_device *device;
158 mutex_lock(&uuid_mutex);
159 list_for_each(cur, head) {
160 device = list_entry(cur, struct btrfs_device, dev_list);
162 close_bdev_excl(device->bdev);
163 printk("close devices closes %s\n", device->name);
167 mutex_unlock(&uuid_mutex);
171 int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
172 int flags, void *holder)
174 struct block_device *bdev;
175 struct list_head *head = &fs_devices->devices;
176 struct list_head *cur;
177 struct btrfs_device *device;
180 mutex_lock(&uuid_mutex);
181 list_for_each(cur, head) {
182 device = list_entry(cur, struct btrfs_device, dev_list);
183 bdev = open_bdev_excl(device->name, flags, holder);
186 printk("open %s failed\n", device->name);
190 if (device->devid == fs_devices->latest_devid)
191 fs_devices->latest_bdev = bdev;
192 if (device->devid == fs_devices->lowest_devid) {
193 fs_devices->lowest_bdev = bdev;
197 mutex_unlock(&uuid_mutex);
200 mutex_unlock(&uuid_mutex);
201 btrfs_close_devices(fs_devices);
205 int btrfs_scan_one_device(const char *path, int flags, void *holder,
206 struct btrfs_fs_devices **fs_devices_ret)
208 struct btrfs_super_block *disk_super;
209 struct block_device *bdev;
210 struct buffer_head *bh;
215 mutex_lock(&uuid_mutex);
217 printk("scan one opens %s\n", path);
218 bdev = open_bdev_excl(path, flags, holder);
221 printk("open failed\n");
226 ret = set_blocksize(bdev, 4096);
229 bh = __bread(bdev, BTRFS_SUPER_INFO_OFFSET / 4096, 4096);
234 disk_super = (struct btrfs_super_block *)bh->b_data;
235 if (strncmp((char *)(&disk_super->magic), BTRFS_MAGIC,
236 sizeof(disk_super->magic))) {
237 printk("no btrfs found on %s\n", path);
241 devid = le64_to_cpu(disk_super->dev_item.devid);
242 transid = btrfs_super_generation(disk_super);
243 printk("found device %Lu transid %Lu on %s\n", devid, transid, path);
244 ret = device_list_add(path, disk_super, devid, fs_devices_ret);
249 close_bdev_excl(bdev);
251 mutex_unlock(&uuid_mutex);
256 * this uses a pretty simple search, the expectation is that it is
257 * called very infrequently and that a given device has a small number
260 static int find_free_dev_extent(struct btrfs_trans_handle *trans,
261 struct btrfs_device *device,
262 struct btrfs_path *path,
263 u64 num_bytes, u64 *start)
265 struct btrfs_key key;
266 struct btrfs_root *root = device->dev_root;
267 struct btrfs_dev_extent *dev_extent = NULL;
270 u64 search_start = 0;
271 u64 search_end = device->total_bytes;
275 struct extent_buffer *l;
280 /* FIXME use last free of some kind */
282 /* we don't want to overwrite the superblock on the drive,
283 * so we make sure to start at an offset of at least 1MB
285 search_start = max((u64)1024 * 1024, search_start);
286 key.objectid = device->devid;
287 key.offset = search_start;
288 key.type = BTRFS_DEV_EXTENT_KEY;
289 ret = btrfs_search_slot(trans, root, &key, path, 0, 0);
292 ret = btrfs_previous_item(root, path, 0, key.type);
296 btrfs_item_key_to_cpu(l, &key, path->slots[0]);
299 slot = path->slots[0];
300 if (slot >= btrfs_header_nritems(l)) {
301 ret = btrfs_next_leaf(root, path);
308 if (search_start >= search_end) {
312 *start = search_start;
316 *start = last_byte > search_start ?
317 last_byte : search_start;
318 if (search_end <= *start) {
324 btrfs_item_key_to_cpu(l, &key, slot);
326 if (key.objectid < device->devid)
329 if (key.objectid > device->devid)
332 if (key.offset >= search_start && key.offset > last_byte &&
334 if (last_byte < search_start)
335 last_byte = search_start;
336 hole_size = key.offset - last_byte;
337 if (key.offset > last_byte &&
338 hole_size >= num_bytes) {
343 if (btrfs_key_type(&key) != BTRFS_DEV_EXTENT_KEY) {
348 dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
349 last_byte = key.offset + btrfs_dev_extent_length(l, dev_extent);
355 /* we have to make sure we didn't find an extent that has already
356 * been allocated by the map tree or the original allocation
358 btrfs_release_path(root, path);
359 BUG_ON(*start < search_start);
361 if (*start + num_bytes > search_end) {
365 /* check for pending inserts here */
369 btrfs_release_path(root, path);
373 int btrfs_alloc_dev_extent(struct btrfs_trans_handle *trans,
374 struct btrfs_device *device,
375 u64 chunk_tree, u64 chunk_objectid,
377 u64 num_bytes, u64 *start)
380 struct btrfs_path *path;
381 struct btrfs_root *root = device->dev_root;
382 struct btrfs_dev_extent *extent;
383 struct extent_buffer *leaf;
384 struct btrfs_key key;
386 path = btrfs_alloc_path();
390 ret = find_free_dev_extent(trans, device, path, num_bytes, start);
395 key.objectid = device->devid;
397 key.type = BTRFS_DEV_EXTENT_KEY;
398 ret = btrfs_insert_empty_item(trans, root, path, &key,
402 leaf = path->nodes[0];
403 extent = btrfs_item_ptr(leaf, path->slots[0],
404 struct btrfs_dev_extent);
405 btrfs_set_dev_extent_chunk_tree(leaf, extent, chunk_tree);
406 btrfs_set_dev_extent_chunk_objectid(leaf, extent, chunk_objectid);
407 btrfs_set_dev_extent_chunk_offset(leaf, extent, chunk_offset);
409 write_extent_buffer(leaf, root->fs_info->chunk_tree_uuid,
410 (unsigned long)btrfs_dev_extent_chunk_tree_uuid(extent),
413 btrfs_set_dev_extent_length(leaf, extent, num_bytes);
414 btrfs_mark_buffer_dirty(leaf);
416 btrfs_free_path(path);
420 static int find_next_chunk(struct btrfs_root *root, u64 objectid, u64 *offset)
422 struct btrfs_path *path;
424 struct btrfs_key key;
425 struct btrfs_chunk *chunk;
426 struct btrfs_key found_key;
428 path = btrfs_alloc_path();
431 key.objectid = objectid;
432 key.offset = (u64)-1;
433 key.type = BTRFS_CHUNK_ITEM_KEY;
435 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
441 ret = btrfs_previous_item(root, path, 0, BTRFS_CHUNK_ITEM_KEY);
445 btrfs_item_key_to_cpu(path->nodes[0], &found_key,
447 if (found_key.objectid != objectid)
450 chunk = btrfs_item_ptr(path->nodes[0], path->slots[0],
452 *offset = found_key.offset +
453 btrfs_chunk_length(path->nodes[0], chunk);
458 btrfs_free_path(path);
462 static int find_next_devid(struct btrfs_root *root, struct btrfs_path *path,
466 struct btrfs_key key;
467 struct btrfs_key found_key;
469 key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
470 key.type = BTRFS_DEV_ITEM_KEY;
471 key.offset = (u64)-1;
473 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
479 ret = btrfs_previous_item(root, path, BTRFS_DEV_ITEMS_OBJECTID,
484 btrfs_item_key_to_cpu(path->nodes[0], &found_key,
486 *objectid = found_key.offset + 1;
490 btrfs_release_path(root, path);
495 * the device information is stored in the chunk root
496 * the btrfs_device struct should be fully filled in
498 int btrfs_add_device(struct btrfs_trans_handle *trans,
499 struct btrfs_root *root,
500 struct btrfs_device *device)
503 struct btrfs_path *path;
504 struct btrfs_dev_item *dev_item;
505 struct extent_buffer *leaf;
506 struct btrfs_key key;
510 root = root->fs_info->chunk_root;
512 path = btrfs_alloc_path();
516 ret = find_next_devid(root, path, &free_devid);
520 key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
521 key.type = BTRFS_DEV_ITEM_KEY;
522 key.offset = free_devid;
524 ret = btrfs_insert_empty_item(trans, root, path, &key,
529 leaf = path->nodes[0];
530 dev_item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dev_item);
532 device->devid = free_devid;
533 btrfs_set_device_id(leaf, dev_item, device->devid);
534 btrfs_set_device_type(leaf, dev_item, device->type);
535 btrfs_set_device_io_align(leaf, dev_item, device->io_align);
536 btrfs_set_device_io_width(leaf, dev_item, device->io_width);
537 btrfs_set_device_sector_size(leaf, dev_item, device->sector_size);
538 btrfs_set_device_total_bytes(leaf, dev_item, device->total_bytes);
539 btrfs_set_device_bytes_used(leaf, dev_item, device->bytes_used);
540 btrfs_set_device_group(leaf, dev_item, 0);
541 btrfs_set_device_seek_speed(leaf, dev_item, 0);
542 btrfs_set_device_bandwidth(leaf, dev_item, 0);
544 ptr = (unsigned long)btrfs_device_uuid(dev_item);
545 write_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
546 btrfs_mark_buffer_dirty(leaf);
550 btrfs_free_path(path);
553 int btrfs_update_device(struct btrfs_trans_handle *trans,
554 struct btrfs_device *device)
557 struct btrfs_path *path;
558 struct btrfs_root *root;
559 struct btrfs_dev_item *dev_item;
560 struct extent_buffer *leaf;
561 struct btrfs_key key;
563 root = device->dev_root->fs_info->chunk_root;
565 path = btrfs_alloc_path();
569 key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
570 key.type = BTRFS_DEV_ITEM_KEY;
571 key.offset = device->devid;
573 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
582 leaf = path->nodes[0];
583 dev_item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_dev_item);
585 btrfs_set_device_id(leaf, dev_item, device->devid);
586 btrfs_set_device_type(leaf, dev_item, device->type);
587 btrfs_set_device_io_align(leaf, dev_item, device->io_align);
588 btrfs_set_device_io_width(leaf, dev_item, device->io_width);
589 btrfs_set_device_sector_size(leaf, dev_item, device->sector_size);
590 btrfs_set_device_total_bytes(leaf, dev_item, device->total_bytes);
591 btrfs_set_device_bytes_used(leaf, dev_item, device->bytes_used);
592 btrfs_mark_buffer_dirty(leaf);
595 btrfs_free_path(path);
599 int btrfs_add_system_chunk(struct btrfs_trans_handle *trans,
600 struct btrfs_root *root,
601 struct btrfs_key *key,
602 struct btrfs_chunk *chunk, int item_size)
604 struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
605 struct btrfs_disk_key disk_key;
609 array_size = btrfs_super_sys_array_size(super_copy);
610 if (array_size + item_size > BTRFS_SYSTEM_CHUNK_ARRAY_SIZE)
613 ptr = super_copy->sys_chunk_array + array_size;
614 btrfs_cpu_key_to_disk(&disk_key, key);
615 memcpy(ptr, &disk_key, sizeof(disk_key));
616 ptr += sizeof(disk_key);
617 memcpy(ptr, chunk, item_size);
618 item_size += sizeof(disk_key);
619 btrfs_set_super_sys_array_size(super_copy, array_size + item_size);
623 int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
624 struct btrfs_root *extent_root, u64 *start,
625 u64 *num_bytes, u64 type)
628 struct btrfs_fs_info *info = extent_root->fs_info;
629 struct btrfs_root *chunk_root = extent_root->fs_info->chunk_root;
630 struct btrfs_stripe *stripes;
631 struct btrfs_device *device = NULL;
632 struct btrfs_chunk *chunk;
633 struct list_head private_devs;
634 struct list_head *dev_list = &extent_root->fs_info->fs_devices->devices;
635 struct list_head *cur;
636 struct extent_map_tree *em_tree;
637 struct map_lookup *map;
638 struct extent_map *em;
640 u64 calc_size = 1024 * 1024 * 1024;
641 u64 min_free = calc_size;
649 int stripe_len = 64 * 1024;
650 struct btrfs_key key;
652 if (list_empty(dev_list))
655 if (type & (BTRFS_BLOCK_GROUP_RAID0))
656 num_stripes = btrfs_super_num_devices(&info->super_copy);
657 if (type & (BTRFS_BLOCK_GROUP_DUP))
659 if (type & (BTRFS_BLOCK_GROUP_RAID1)) {
660 num_stripes = min_t(u64, 2,
661 btrfs_super_num_devices(&info->super_copy));
663 if (type & (BTRFS_BLOCK_GROUP_RAID10)) {
664 num_stripes = btrfs_super_num_devices(&info->super_copy);
667 num_stripes &= ~(u32)1;
671 INIT_LIST_HEAD(&private_devs);
672 cur = dev_list->next;
675 if (type & BTRFS_BLOCK_GROUP_DUP)
676 min_free = calc_size * 2;
678 /* build a private list of devices we will allocate from */
679 while(index < num_stripes) {
680 device = list_entry(cur, struct btrfs_device, dev_list);
682 avail = device->total_bytes - device->bytes_used;
684 if (avail > max_avail)
686 if (avail >= min_free) {
687 list_move_tail(&device->dev_list, &private_devs);
689 if (type & BTRFS_BLOCK_GROUP_DUP)
695 if (index < num_stripes) {
696 list_splice(&private_devs, dev_list);
697 if (!looped && max_avail > 0) {
699 calc_size = max_avail;
705 key.objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
706 key.type = BTRFS_CHUNK_ITEM_KEY;
707 ret = find_next_chunk(chunk_root, BTRFS_FIRST_CHUNK_TREE_OBJECTID,
712 chunk = kmalloc(btrfs_chunk_item_size(num_stripes), GFP_NOFS);
716 map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
722 stripes = &chunk->stripe;
724 if (type & (BTRFS_BLOCK_GROUP_RAID1 | BTRFS_BLOCK_GROUP_DUP))
725 *num_bytes = calc_size;
726 else if (type & BTRFS_BLOCK_GROUP_RAID10)
727 *num_bytes = calc_size * num_stripes / sub_stripes;
729 *num_bytes = calc_size * num_stripes;
732 printk("new chunk type %Lu start %Lu size %Lu\n", type, key.offset, *num_bytes);
733 while(index < num_stripes) {
734 struct btrfs_stripe *stripe;
735 BUG_ON(list_empty(&private_devs));
736 cur = private_devs.next;
737 device = list_entry(cur, struct btrfs_device, dev_list);
739 /* loop over this device again if we're doing a dup group */
740 if (!(type & BTRFS_BLOCK_GROUP_DUP) ||
741 (index == num_stripes - 1))
742 list_move_tail(&device->dev_list, dev_list);
744 ret = btrfs_alloc_dev_extent(trans, device,
745 info->chunk_root->root_key.objectid,
746 BTRFS_FIRST_CHUNK_TREE_OBJECTID, key.offset,
747 calc_size, &dev_offset);
749 printk("alloc chunk start %Lu size %Lu from dev %Lu type %Lu\n", key.offset, calc_size, device->devid, type);
750 device->bytes_used += calc_size;
751 ret = btrfs_update_device(trans, device);
754 map->stripes[index].dev = device;
755 map->stripes[index].physical = dev_offset;
756 stripe = stripes + index;
757 btrfs_set_stack_stripe_devid(stripe, device->devid);
758 btrfs_set_stack_stripe_offset(stripe, dev_offset);
759 memcpy(stripe->dev_uuid, device->uuid, BTRFS_UUID_SIZE);
760 physical = dev_offset;
763 BUG_ON(!list_empty(&private_devs));
765 /* key was set above */
766 btrfs_set_stack_chunk_length(chunk, *num_bytes);
767 btrfs_set_stack_chunk_owner(chunk, extent_root->root_key.objectid);
768 btrfs_set_stack_chunk_stripe_len(chunk, stripe_len);
769 btrfs_set_stack_chunk_type(chunk, type);
770 btrfs_set_stack_chunk_num_stripes(chunk, num_stripes);
771 btrfs_set_stack_chunk_io_align(chunk, stripe_len);
772 btrfs_set_stack_chunk_io_width(chunk, stripe_len);
773 btrfs_set_stack_chunk_sector_size(chunk, extent_root->sectorsize);
774 btrfs_set_stack_chunk_sub_stripes(chunk, sub_stripes);
775 map->sector_size = extent_root->sectorsize;
776 map->stripe_len = stripe_len;
777 map->io_align = stripe_len;
778 map->io_width = stripe_len;
780 map->num_stripes = num_stripes;
781 map->sub_stripes = sub_stripes;
783 ret = btrfs_insert_item(trans, chunk_root, &key, chunk,
784 btrfs_chunk_item_size(num_stripes));
786 *start = key.offset;;
788 em = alloc_extent_map(GFP_NOFS);
791 em->bdev = (struct block_device *)map;
792 em->start = key.offset;
793 em->len = *num_bytes;
798 em_tree = &extent_root->fs_info->mapping_tree.map_tree;
799 spin_lock(&em_tree->lock);
800 ret = add_extent_mapping(em_tree, em);
801 spin_unlock(&em_tree->lock);
807 void btrfs_mapping_init(struct btrfs_mapping_tree *tree)
809 extent_map_tree_init(&tree->map_tree, GFP_NOFS);
812 void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree)
814 struct extent_map *em;
817 spin_lock(&tree->map_tree.lock);
818 em = lookup_extent_mapping(&tree->map_tree, 0, (u64)-1);
820 remove_extent_mapping(&tree->map_tree, em);
821 spin_unlock(&tree->map_tree.lock);
827 /* once for the tree */
832 int btrfs_num_copies(struct btrfs_mapping_tree *map_tree, u64 logical, u64 len)
834 struct extent_map *em;
835 struct map_lookup *map;
836 struct extent_map_tree *em_tree = &map_tree->map_tree;
839 spin_lock(&em_tree->lock);
840 em = lookup_extent_mapping(em_tree, logical, len);
841 spin_unlock(&em_tree->lock);
844 BUG_ON(em->start > logical || em->start + em->len < logical);
845 map = (struct map_lookup *)em->bdev;
846 if (map->type & (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1))
847 ret = map->num_stripes;
848 else if (map->type & BTRFS_BLOCK_GROUP_RAID10)
849 ret = map->sub_stripes;
856 int btrfs_map_block(struct btrfs_mapping_tree *map_tree, int rw,
857 u64 logical, u64 *length,
858 struct btrfs_multi_bio **multi_ret, int mirror_num)
860 struct extent_map *em;
861 struct map_lookup *map;
862 struct extent_map_tree *em_tree = &map_tree->map_tree;
866 int stripes_allocated = 8;
867 int stripes_required = 1;
870 struct btrfs_multi_bio *multi = NULL;
872 if (multi_ret && !(rw & (1 << BIO_RW))) {
873 stripes_allocated = 1;
877 multi = kzalloc(btrfs_multi_bio_size(stripes_allocated),
883 spin_lock(&em_tree->lock);
884 em = lookup_extent_mapping(em_tree, logical, *length);
885 spin_unlock(&em_tree->lock);
887 printk("unable to find logical %Lu\n", logical);
891 BUG_ON(em->start > logical || em->start + em->len < logical);
892 map = (struct map_lookup *)em->bdev;
893 offset = logical - em->start;
895 if (mirror_num > map->num_stripes)
898 /* if our multi bio struct is too small, back off and try again */
899 if (rw & (1 << BIO_RW)) {
900 if (map->type & (BTRFS_BLOCK_GROUP_RAID1 |
901 BTRFS_BLOCK_GROUP_DUP)) {
902 stripes_required = map->num_stripes;
903 } else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
904 stripes_required = map->sub_stripes;
907 if (multi_ret && rw == WRITE &&
908 stripes_allocated < stripes_required) {
909 stripes_allocated = map->num_stripes;
916 * stripe_nr counts the total number of stripes we have to stride
917 * to get to this block
919 do_div(stripe_nr, map->stripe_len);
921 stripe_offset = stripe_nr * map->stripe_len;
922 BUG_ON(offset < stripe_offset);
924 /* stripe_offset is the offset of this block in its stripe*/
925 stripe_offset = offset - stripe_offset;
927 if (map->type & (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID1 |
928 BTRFS_BLOCK_GROUP_RAID10 |
929 BTRFS_BLOCK_GROUP_DUP)) {
930 /* we limit the length of each bio to what fits in a stripe */
931 *length = min_t(u64, em->len - offset,
932 map->stripe_len - stripe_offset);
934 *length = em->len - offset;
939 multi->num_stripes = 1;
941 if (map->type & BTRFS_BLOCK_GROUP_RAID1) {
942 if (rw & (1 << BIO_RW))
943 multi->num_stripes = map->num_stripes;
944 else if (mirror_num) {
945 stripe_index = mirror_num - 1;
949 struct btrfs_device *cur;
951 for (i = 0; i < map->num_stripes; i++) {
952 cur = map->stripes[i].dev;
953 spin_lock(&cur->io_lock);
954 if (cur->total_ios < least) {
955 least = cur->total_ios;
958 spin_unlock(&cur->io_lock);
961 } else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
962 if (rw & (1 << BIO_RW))
963 multi->num_stripes = map->num_stripes;
965 stripe_index = mirror_num - 1;
966 } else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
967 int factor = map->num_stripes / map->sub_stripes;
968 int orig_stripe_nr = stripe_nr;
970 stripe_index = do_div(stripe_nr, factor);
971 stripe_index *= map->sub_stripes;
973 if (rw & (1 << BIO_RW))
974 multi->num_stripes = map->sub_stripes;
976 stripe_index += mirror_num - 1;
978 stripe_index += orig_stripe_nr % map->sub_stripes;
981 * after this do_div call, stripe_nr is the number of stripes
982 * on this device we have to walk to find the data, and
983 * stripe_index is the number of our device in the stripe array
985 stripe_index = do_div(stripe_nr, map->num_stripes);
987 BUG_ON(stripe_index >= map->num_stripes);
989 for (i = 0; i < multi->num_stripes; i++) {
990 multi->stripes[i].physical =
991 map->stripes[stripe_index].physical + stripe_offset +
992 stripe_nr * map->stripe_len;
993 multi->stripes[i].dev = map->stripes[stripe_index].dev;
1002 #if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,23)
1003 static void end_bio_multi_stripe(struct bio *bio, int err)
1005 static int end_bio_multi_stripe(struct bio *bio,
1006 unsigned int bytes_done, int err)
1009 struct btrfs_multi_bio *multi = bio->bi_private;
1011 #if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23)
1018 if (atomic_dec_and_test(&multi->stripes_pending)) {
1019 bio->bi_private = multi->private;
1020 bio->bi_end_io = multi->end_io;
1022 if (!err && multi->error)
1026 #if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23)
1027 bio_endio(bio, bio->bi_size, err);
1029 bio_endio(bio, err);
1034 #if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23)
1039 int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
1042 struct btrfs_mapping_tree *map_tree;
1043 struct btrfs_device *dev;
1044 struct bio *first_bio = bio;
1045 u64 logical = bio->bi_sector << 9;
1048 struct bio_vec *bvec;
1049 struct btrfs_multi_bio *multi = NULL;
1055 bio_for_each_segment(bvec, bio, i) {
1056 length += bvec->bv_len;
1059 map_tree = &root->fs_info->mapping_tree;
1060 map_length = length;
1062 ret = btrfs_map_block(map_tree, rw, logical, &map_length, &multi,
1066 total_devs = multi->num_stripes;
1067 if (map_length < length) {
1068 printk("mapping failed logical %Lu bio len %Lu "
1069 "len %Lu\n", logical, length, map_length);
1072 multi->end_io = first_bio->bi_end_io;
1073 multi->private = first_bio->bi_private;
1074 atomic_set(&multi->stripes_pending, multi->num_stripes);
1076 while(dev_nr < total_devs) {
1077 if (total_devs > 1) {
1078 if (dev_nr < total_devs - 1) {
1079 bio = bio_clone(first_bio, GFP_NOFS);
1084 bio->bi_private = multi;
1085 bio->bi_end_io = end_bio_multi_stripe;
1087 bio->bi_sector = multi->stripes[dev_nr].physical >> 9;
1088 dev = multi->stripes[dev_nr].dev;
1089 bio->bi_bdev = dev->bdev;
1090 spin_lock(&dev->io_lock);
1092 spin_unlock(&dev->io_lock);
1093 submit_bio(rw, bio);
1096 if (total_devs == 1)
1101 struct btrfs_device *btrfs_find_device(struct btrfs_root *root, u64 devid)
1103 struct list_head *head = &root->fs_info->fs_devices->devices;
1105 return __find_device(head, devid);
1108 static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key,
1109 struct extent_buffer *leaf,
1110 struct btrfs_chunk *chunk)
1112 struct btrfs_mapping_tree *map_tree = &root->fs_info->mapping_tree;
1113 struct map_lookup *map;
1114 struct extent_map *em;
1122 logical = key->offset;
1123 length = btrfs_chunk_length(leaf, chunk);
1124 spin_lock(&map_tree->map_tree.lock);
1125 em = lookup_extent_mapping(&map_tree->map_tree, logical, 1);
1126 spin_unlock(&map_tree->map_tree.lock);
1128 /* already mapped? */
1129 if (em && em->start <= logical && em->start + em->len > logical) {
1130 free_extent_map(em);
1133 free_extent_map(em);
1136 map = kzalloc(sizeof(*map), GFP_NOFS);
1140 em = alloc_extent_map(GFP_NOFS);
1143 num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
1144 map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
1146 free_extent_map(em);
1150 em->bdev = (struct block_device *)map;
1151 em->start = logical;
1153 em->block_start = 0;
1155 map->num_stripes = num_stripes;
1156 map->io_width = btrfs_chunk_io_width(leaf, chunk);
1157 map->io_align = btrfs_chunk_io_align(leaf, chunk);
1158 map->sector_size = btrfs_chunk_sector_size(leaf, chunk);
1159 map->stripe_len = btrfs_chunk_stripe_len(leaf, chunk);
1160 map->type = btrfs_chunk_type(leaf, chunk);
1161 map->sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
1162 for (i = 0; i < num_stripes; i++) {
1163 map->stripes[i].physical =
1164 btrfs_stripe_offset_nr(leaf, chunk, i);
1165 devid = btrfs_stripe_devid_nr(leaf, chunk, i);
1166 map->stripes[i].dev = btrfs_find_device(root, devid);
1167 if (!map->stripes[i].dev) {
1169 free_extent_map(em);
1174 spin_lock(&map_tree->map_tree.lock);
1175 ret = add_extent_mapping(&map_tree->map_tree, em);
1176 spin_unlock(&map_tree->map_tree.lock);
1178 free_extent_map(em);
1183 static int fill_device_from_item(struct extent_buffer *leaf,
1184 struct btrfs_dev_item *dev_item,
1185 struct btrfs_device *device)
1189 device->devid = btrfs_device_id(leaf, dev_item);
1190 device->total_bytes = btrfs_device_total_bytes(leaf, dev_item);
1191 device->bytes_used = btrfs_device_bytes_used(leaf, dev_item);
1192 device->type = btrfs_device_type(leaf, dev_item);
1193 device->io_align = btrfs_device_io_align(leaf, dev_item);
1194 device->io_width = btrfs_device_io_width(leaf, dev_item);
1195 device->sector_size = btrfs_device_sector_size(leaf, dev_item);
1197 ptr = (unsigned long)btrfs_device_uuid(dev_item);
1198 read_extent_buffer(leaf, device->uuid, ptr, BTRFS_UUID_SIZE);
1203 static int read_one_dev(struct btrfs_root *root,
1204 struct extent_buffer *leaf,
1205 struct btrfs_dev_item *dev_item)
1207 struct btrfs_device *device;
1210 devid = btrfs_device_id(leaf, dev_item);
1211 device = btrfs_find_device(root, devid);
1213 printk("warning devid %Lu not found already\n", devid);
1214 device = kzalloc(sizeof(*device), GFP_NOFS);
1217 list_add(&device->dev_list,
1218 &root->fs_info->fs_devices->devices);
1219 device->barriers = 1;
1220 spin_lock_init(&device->io_lock);
1223 fill_device_from_item(leaf, dev_item, device);
1224 device->dev_root = root->fs_info->dev_root;
1227 ret = btrfs_open_device(device);
1235 int btrfs_read_super_device(struct btrfs_root *root, struct extent_buffer *buf)
1237 struct btrfs_dev_item *dev_item;
1239 dev_item = (struct btrfs_dev_item *)offsetof(struct btrfs_super_block,
1241 return read_one_dev(root, buf, dev_item);
1244 int btrfs_read_sys_array(struct btrfs_root *root)
1246 struct btrfs_super_block *super_copy = &root->fs_info->super_copy;
1247 struct extent_buffer *sb = root->fs_info->sb_buffer;
1248 struct btrfs_disk_key *disk_key;
1249 struct btrfs_chunk *chunk;
1250 struct btrfs_key key;
1255 unsigned long sb_ptr;
1259 array_size = btrfs_super_sys_array_size(super_copy);
1262 * we do this loop twice, once for the device items and
1263 * once for all of the chunks. This way there are device
1264 * structs filled in for every chunk
1266 ptr = super_copy->sys_chunk_array;
1267 sb_ptr = offsetof(struct btrfs_super_block, sys_chunk_array);
1270 while (cur < array_size) {
1271 disk_key = (struct btrfs_disk_key *)ptr;
1272 btrfs_disk_key_to_cpu(&key, disk_key);
1274 len = sizeof(*disk_key);
1279 if (key.type == BTRFS_CHUNK_ITEM_KEY) {
1280 chunk = (struct btrfs_chunk *)sb_ptr;
1281 ret = read_one_chunk(root, &key, sb, chunk);
1283 num_stripes = btrfs_chunk_num_stripes(sb, chunk);
1284 len = btrfs_chunk_item_size(num_stripes);
1295 int btrfs_read_chunk_tree(struct btrfs_root *root)
1297 struct btrfs_path *path;
1298 struct extent_buffer *leaf;
1299 struct btrfs_key key;
1300 struct btrfs_key found_key;
1304 root = root->fs_info->chunk_root;
1306 path = btrfs_alloc_path();
1310 /* first we search for all of the device items, and then we
1311 * read in all of the chunk items. This way we can create chunk
1312 * mappings that reference all of the devices that are afound
1314 key.objectid = BTRFS_DEV_ITEMS_OBJECTID;
1318 ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1320 leaf = path->nodes[0];
1321 slot = path->slots[0];
1322 if (slot >= btrfs_header_nritems(leaf)) {
1323 ret = btrfs_next_leaf(root, path);
1330 btrfs_item_key_to_cpu(leaf, &found_key, slot);
1331 if (key.objectid == BTRFS_DEV_ITEMS_OBJECTID) {
1332 if (found_key.objectid != BTRFS_DEV_ITEMS_OBJECTID)
1334 if (found_key.type == BTRFS_DEV_ITEM_KEY) {
1335 struct btrfs_dev_item *dev_item;
1336 dev_item = btrfs_item_ptr(leaf, slot,
1337 struct btrfs_dev_item);
1338 ret = read_one_dev(root, leaf, dev_item);
1341 } else if (found_key.type == BTRFS_CHUNK_ITEM_KEY) {
1342 struct btrfs_chunk *chunk;
1343 chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);
1344 ret = read_one_chunk(root, &found_key, leaf, chunk);
1348 if (key.objectid == BTRFS_DEV_ITEMS_OBJECTID) {
1350 btrfs_release_path(root, path);
1354 btrfs_free_path(path);