#include "math.h"
#include "dev-replace.h"
#include "sysfs.h"
+#include "tree-checker.h"
const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES] = {
[BTRFS_RAID_RAID10] = {
mutex_init(&fs_devs->device_list_mutex);
INIT_LIST_HEAD(&fs_devs->devices);
- INIT_LIST_HEAD(&fs_devs->resized_devices);
INIT_LIST_HEAD(&fs_devs->alloc_list);
INIT_LIST_HEAD(&fs_devs->fs_list);
if (fsid)
void btrfs_free_device(struct btrfs_device *device)
{
+ WARN_ON(!list_empty(&device->post_commit_list));
rcu_string_free(device->name);
+ extent_io_tree_release(&device->alloc_state);
bio_put(device->flush_bio);
kfree(device);
}
INIT_LIST_HEAD(&dev->dev_list);
INIT_LIST_HEAD(&dev->dev_alloc_list);
- INIT_LIST_HEAD(&dev->resized_list);
+ INIT_LIST_HEAD(&dev->post_commit_list);
spin_lock_init(&dev->io_lock);
btrfs_device_data_ordered_init(dev);
INIT_RADIX_TREE(&dev->reada_zones, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
INIT_RADIX_TREE(&dev->reada_extents, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
+ extent_io_tree_init(NULL, &dev->alloc_state, 0, NULL);
return dev;
}
mutex_unlock(&uuid_mutex);
}
-static void free_device_rcu(struct rcu_head *head)
-{
- struct btrfs_device *device;
-
- device = container_of(head, struct btrfs_device, rcu);
- btrfs_free_device(device);
-}
-
static void btrfs_close_bdev(struct btrfs_device *device)
{
if (!device->bdev)
list_replace_rcu(&device->dev_list, &new_device->dev_list);
new_device->fs_devices = device->fs_devices;
- call_rcu(&device->rcu, free_device_rcu);
+ synchronize_rcu();
+ btrfs_free_device(device);
}
static int close_fs_devices(struct btrfs_fs_devices *fs_devices)
return device;
}
-static int contains_pending_extent(struct btrfs_transaction *transaction,
- struct btrfs_device *device,
- u64 *start, u64 len)
+/*
+ * Try to find a chunk that intersects [start, start + len] range and when one
+ * such is found, record the end of it in *start
+ */
+static bool contains_pending_extent(struct btrfs_device *device, u64 *start,
+ u64 len)
{
- struct btrfs_fs_info *fs_info = device->fs_info;
- struct extent_map *em;
- struct list_head *search_list = &fs_info->pinned_chunks;
- int ret = 0;
- u64 physical_start = *start;
+ u64 physical_start, physical_end;
- if (transaction)
- search_list = &transaction->pending_chunks;
-again:
- list_for_each_entry(em, search_list, list) {
- struct map_lookup *map;
- int i;
+ lockdep_assert_held(&device->fs_info->chunk_mutex);
- map = em->map_lookup;
- for (i = 0; i < map->num_stripes; i++) {
- u64 end;
+ if (!find_first_extent_bit(&device->alloc_state, *start,
+ &physical_start, &physical_end,
+ CHUNK_ALLOCATED, NULL)) {
- if (map->stripes[i].dev != device)
- continue;
- if (map->stripes[i].physical >= physical_start + len ||
- map->stripes[i].physical + em->orig_block_len <=
- physical_start)
- continue;
- /*
- * Make sure that while processing the pinned list we do
- * not override our *start with a lower value, because
- * we can have pinned chunks that fall within this
- * device hole and that have lower physical addresses
- * than the pending chunks we processed before. If we
- * do not take this special care we can end up getting
- * 2 pending chunks that start at the same physical
- * device offsets because the end offset of a pinned
- * chunk can be equal to the start offset of some
- * pending chunk.
- */
- end = map->stripes[i].physical + em->orig_block_len;
- if (end > *start) {
- *start = end;
- ret = 1;
- }
+ if (in_range(physical_start, *start, len) ||
+ in_range(*start, physical_start,
+ physical_end - physical_start)) {
+ *start = physical_end + 1;
+ return true;
}
}
- if (search_list != &fs_info->pinned_chunks) {
- search_list = &fs_info->pinned_chunks;
- goto again;
- }
-
- return ret;
+ return false;
}
* But if we don't find suitable free space, it is used to store the size of
* the max free space.
*/
-int find_free_dev_extent_start(struct btrfs_transaction *transaction,
- struct btrfs_device *device, u64 num_bytes,
+int find_free_dev_extent_start(struct btrfs_device *device, u64 num_bytes,
u64 search_start, u64 *start, u64 *len)
{
struct btrfs_fs_info *fs_info = device->fs_info;
* Have to check before we set max_hole_start, otherwise
* we could end up sending back this offset anyway.
*/
- if (contains_pending_extent(transaction, device,
- &search_start,
+ if (contains_pending_extent(device, &search_start,
hole_size)) {
- if (key.offset >= search_start) {
+ if (key.offset >= search_start)
hole_size = key.offset - search_start;
- } else {
- WARN_ON_ONCE(1);
+ else
hole_size = 0;
- }
}
if (hole_size > max_hole_size) {
if (search_end > search_start) {
hole_size = search_end - search_start;
- if (contains_pending_extent(transaction, device, &search_start,
- hole_size)) {
+ if (contains_pending_extent(device, &search_start, hole_size)) {
btrfs_release_path(path);
goto again;
}
return ret;
}
-int find_free_dev_extent(struct btrfs_trans_handle *trans,
- struct btrfs_device *device, u64 num_bytes,
+int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes,
u64 *start, u64 *len)
{
/* FIXME use last free of some kind */
- return find_free_dev_extent_start(trans->transaction, device,
- num_bytes, 0, start, len);
+ return find_free_dev_extent_start(device, num_bytes, 0, start, len);
}
static int btrfs_free_dev_extent(struct btrfs_trans_handle *trans,
btrfs_scratch_superblocks(device->bdev, device->name->str);
btrfs_close_bdev(device);
- call_rcu(&device->rcu, free_device_rcu);
+ synchronize_rcu();
+ btrfs_free_device(device);
if (cur_devices->open_devices == 0) {
while (fs_devices) {
}
btrfs_close_bdev(srcdev);
- call_rcu(&srcdev->rcu, free_device_rcu);
+ synchronize_rcu();
+ btrfs_free_device(srcdev);
/* if this is no devs we rather delete the fs_devices */
if (!fs_devices->num_devices) {
btrfs_scratch_superblocks(tgtdev->bdev, tgtdev->name->str);
btrfs_close_bdev(tgtdev);
- call_rcu(&tgtdev->rcu, free_device_rcu);
+ synchronize_rcu();
+ btrfs_free_device(tgtdev);
}
static struct btrfs_device *btrfs_find_device_by_path(
{
struct btrfs_fs_info *fs_info = device->fs_info;
struct btrfs_super_block *super_copy = fs_info->super_copy;
- struct btrfs_fs_devices *fs_devices;
u64 old_total;
u64 diff;
return -EINVAL;
}
- fs_devices = fs_info->fs_devices;
-
btrfs_set_super_total_bytes(super_copy,
round_down(old_total + diff, fs_info->sectorsize));
device->fs_devices->total_rw_bytes += diff;
btrfs_device_set_total_bytes(device, new_size);
btrfs_device_set_disk_total_bytes(device, new_size);
btrfs_clear_space_info_full(device->fs_info);
- if (list_empty(&device->resized_list))
- list_add_tail(&device->resized_list,
- &fs_devices->resized_devices);
+ if (list_empty(&device->post_commit_list))
+ list_add_tail(&device->post_commit_list,
+ &trans->transaction->dev_update_list);
mutex_unlock(&fs_info->chunk_mutex);
return btrfs_update_device(trans, device);
return 0;
}
-static int should_balance_chunk(struct btrfs_fs_info *fs_info,
- struct extent_buffer *leaf,
+static int should_balance_chunk(struct extent_buffer *leaf,
struct btrfs_chunk *chunk, u64 chunk_offset)
{
+ struct btrfs_fs_info *fs_info = leaf->fs_info;
struct btrfs_balance_control *bctl = fs_info->balance_ctl;
struct btrfs_balance_args *bargs = NULL;
u64 chunk_type = btrfs_chunk_type(leaf, chunk);
spin_unlock(&fs_info->balance_lock);
}
- ret = should_balance_chunk(fs_info, leaf, chunk,
- found_key.offset);
+ ret = should_balance_chunk(leaf, chunk, found_key.offset);
btrfs_release_path(path);
if (!ret) {
int slot;
int failed = 0;
bool retried = false;
- bool checked_pending_chunks = false;
struct extent_buffer *l;
struct btrfs_key key;
struct btrfs_super_block *super_copy = fs_info->super_copy;
u64 old_total = btrfs_super_total_bytes(super_copy);
u64 old_size = btrfs_device_get_total_bytes(device);
u64 diff;
+ u64 start;
new_size = round_down(new_size, fs_info->sectorsize);
+ start = new_size;
diff = round_down(old_size - new_size, fs_info->sectorsize);
if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state))
path->reada = READA_BACK;
+ trans = btrfs_start_transaction(root, 0);
+ if (IS_ERR(trans)) {
+ btrfs_free_path(path);
+ return PTR_ERR(trans);
+ }
+
mutex_lock(&fs_info->chunk_mutex);
btrfs_device_set_total_bytes(device, new_size);
device->fs_devices->total_rw_bytes -= diff;
atomic64_sub(diff, &fs_info->free_chunk_space);
}
- mutex_unlock(&fs_info->chunk_mutex);
+
+ /*
+ * Once the device's size has been set to the new size, ensure all
+ * in-memory chunks are synced to disk so that the loop below sees them
+ * and relocates them accordingly.
+ */
+ if (contains_pending_extent(device, &start, diff)) {
+ mutex_unlock(&fs_info->chunk_mutex);
+ ret = btrfs_commit_transaction(trans);
+ if (ret)
+ goto done;
+ } else {
+ mutex_unlock(&fs_info->chunk_mutex);
+ btrfs_end_transaction(trans);
+ }
again:
key.objectid = device->devid;
}
mutex_lock(&fs_info->chunk_mutex);
-
- /*
- * We checked in the above loop all device extents that were already in
- * the device tree. However before we have updated the device's
- * total_bytes to the new size, we might have had chunk allocations that
- * have not complete yet (new block groups attached to transaction
- * handles), and therefore their device extents were not yet in the
- * device tree and we missed them in the loop above. So if we have any
- * pending chunk using a device extent that overlaps the device range
- * that we can not use anymore, commit the current transaction and
- * repeat the search on the device tree - this way we guarantee we will
- * not have chunks using device extents that end beyond 'new_size'.
- */
- if (!checked_pending_chunks) {
- u64 start = new_size;
- u64 len = old_size - new_size;
-
- if (contains_pending_extent(trans->transaction, device,
- &start, len)) {
- mutex_unlock(&fs_info->chunk_mutex);
- checked_pending_chunks = true;
- failed = 0;
- retried = false;
- ret = btrfs_commit_transaction(trans);
- if (ret)
- goto done;
- goto again;
- }
- }
-
btrfs_device_set_disk_total_bytes(device, new_size);
- if (list_empty(&device->resized_list))
- list_add_tail(&device->resized_list,
- &fs_info->fs_devices->resized_devices);
+ if (list_empty(&device->post_commit_list))
+ list_add_tail(&device->post_commit_list,
+ &trans->transaction->dev_update_list);
WARN_ON(diff > old_total);
btrfs_set_super_total_bytes(super_copy,
btrfs_set_fs_incompat(info, RAID56);
}
-#define BTRFS_MAX_DEVS(info) ((BTRFS_MAX_ITEM_SIZE(info) \
- - sizeof(struct btrfs_chunk)) \
- / sizeof(struct btrfs_stripe) + 1)
-
-#define BTRFS_MAX_DEVS_SYS_CHUNK ((BTRFS_SYSTEM_CHUNK_ARRAY_SIZE \
- - 2 * sizeof(struct btrfs_disk_key) \
- - 2 * sizeof(struct btrfs_chunk)) \
- / sizeof(struct btrfs_stripe) + 1)
-
static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
u64 start, u64 type)
{
} else {
btrfs_err(info, "invalid chunk type 0x%llx requested",
type);
- BUG_ON(1);
+ BUG();
}
/* We don't want a chunk larger than 10% of writable space */
if (total_avail == 0)
continue;
- ret = find_free_dev_extent(trans, device,
+ ret = find_free_dev_extent(device,
max_stripe_size * dev_stripes,
&dev_offset, &max_avail);
if (ret && ret != -ENOSPC)
free_extent_map(em);
goto error;
}
-
- list_add_tail(&em->list, &trans->transaction->pending_chunks);
- refcount_inc(&em->refs);
write_unlock(&em_tree->lock);
ret = btrfs_make_block_group(trans, 0, type, start, chunk_size);
if (ret)
goto error_del_extent;
- for (i = 0; i < map->num_stripes; i++)
- btrfs_device_set_bytes_used(map->stripes[i].dev,
- map->stripes[i].dev->bytes_used + stripe_size);
+ for (i = 0; i < map->num_stripes; i++) {
+ struct btrfs_device *dev = map->stripes[i].dev;
+
+ btrfs_device_set_bytes_used(dev, dev->bytes_used + stripe_size);
+ if (list_empty(&dev->post_commit_list))
+ list_add_tail(&dev->post_commit_list,
+ &trans->transaction->dev_update_list);
+ }
atomic64_sub(stripe_size * map->num_stripes, &info->free_chunk_space);
free_extent_map(em);
/* One for the tree reference */
free_extent_map(em);
- /* One for the pending_chunks list reference */
- free_extent_map(em);
error:
kfree(devices_info);
return ret;
return dev;
}
-/* Return -EIO if any error, otherwise return 0. */
-static int btrfs_check_chunk_valid(struct btrfs_fs_info *fs_info,
- struct extent_buffer *leaf,
- struct btrfs_chunk *chunk, u64 logical)
-{
- u64 length;
- u64 stripe_len;
- u16 num_stripes;
- u16 sub_stripes;
- u64 type;
- u64 features;
- bool mixed = false;
-
- length = btrfs_chunk_length(leaf, chunk);
- stripe_len = btrfs_chunk_stripe_len(leaf, chunk);
- num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
- sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
- type = btrfs_chunk_type(leaf, chunk);
-
- if (!num_stripes) {
- btrfs_err(fs_info, "invalid chunk num_stripes: %u",
- num_stripes);
- return -EIO;
- }
- if (!IS_ALIGNED(logical, fs_info->sectorsize)) {
- btrfs_err(fs_info, "invalid chunk logical %llu", logical);
- return -EIO;
- }
- if (btrfs_chunk_sector_size(leaf, chunk) != fs_info->sectorsize) {
- btrfs_err(fs_info, "invalid chunk sectorsize %u",
- btrfs_chunk_sector_size(leaf, chunk));
- return -EIO;
- }
- if (!length || !IS_ALIGNED(length, fs_info->sectorsize)) {
- btrfs_err(fs_info, "invalid chunk length %llu", length);
- return -EIO;
- }
- if (!is_power_of_2(stripe_len) || stripe_len != BTRFS_STRIPE_LEN) {
- btrfs_err(fs_info, "invalid chunk stripe length: %llu",
- stripe_len);
- return -EIO;
- }
- if (~(BTRFS_BLOCK_GROUP_TYPE_MASK | BTRFS_BLOCK_GROUP_PROFILE_MASK) &
- type) {
- btrfs_err(fs_info, "unrecognized chunk type: %llu",
- ~(BTRFS_BLOCK_GROUP_TYPE_MASK |
- BTRFS_BLOCK_GROUP_PROFILE_MASK) &
- btrfs_chunk_type(leaf, chunk));
- return -EIO;
- }
-
- if ((type & BTRFS_BLOCK_GROUP_TYPE_MASK) == 0) {
- btrfs_err(fs_info, "missing chunk type flag: 0x%llx", type);
- return -EIO;
- }
-
- if ((type & BTRFS_BLOCK_GROUP_SYSTEM) &&
- (type & (BTRFS_BLOCK_GROUP_METADATA | BTRFS_BLOCK_GROUP_DATA))) {
- btrfs_err(fs_info,
- "system chunk with data or metadata type: 0x%llx", type);
- return -EIO;
- }
-
- features = btrfs_super_incompat_flags(fs_info->super_copy);
- if (features & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS)
- mixed = true;
-
- if (!mixed) {
- if ((type & BTRFS_BLOCK_GROUP_METADATA) &&
- (type & BTRFS_BLOCK_GROUP_DATA)) {
- btrfs_err(fs_info,
- "mixed chunk type in non-mixed mode: 0x%llx", type);
- return -EIO;
- }
- }
-
- if ((type & BTRFS_BLOCK_GROUP_RAID10 && sub_stripes != 2) ||
- (type & BTRFS_BLOCK_GROUP_RAID1 && num_stripes != 2) ||
- (type & BTRFS_BLOCK_GROUP_RAID5 && num_stripes < 2) ||
- (type & BTRFS_BLOCK_GROUP_RAID6 && num_stripes < 3) ||
- (type & BTRFS_BLOCK_GROUP_DUP && num_stripes != 2) ||
- ((type & BTRFS_BLOCK_GROUP_PROFILE_MASK) == 0 &&
- num_stripes != 1)) {
- btrfs_err(fs_info,
- "invalid num_stripes:sub_stripes %u:%u for profile %llu",
- num_stripes, sub_stripes,
- type & BTRFS_BLOCK_GROUP_PROFILE_MASK);
- return -EIO;
- }
-
- return 0;
-}
-
static void btrfs_report_missing_device(struct btrfs_fs_info *fs_info,
u64 devid, u8 *uuid, bool error)
{
devid, uuid);
}
+static u64 calc_stripe_length(u64 type, u64 chunk_len, int num_stripes)
+{
+ int index = btrfs_bg_flags_to_raid_index(type);
+ int ncopies = btrfs_raid_array[index].ncopies;
+ int data_stripes;
+
+ switch (type & BTRFS_BLOCK_GROUP_PROFILE_MASK) {
+ case BTRFS_BLOCK_GROUP_RAID5:
+ data_stripes = num_stripes - 1;
+ break;
+ case BTRFS_BLOCK_GROUP_RAID6:
+ data_stripes = num_stripes - 2;
+ break;
+ default:
+ data_stripes = num_stripes / ncopies;
+ break;
+ }
+ return div_u64(chunk_len, data_stripes);
+}
+
static int read_one_chunk(struct btrfs_fs_info *fs_info, struct btrfs_key *key,
struct extent_buffer *leaf,
struct btrfs_chunk *chunk)
length = btrfs_chunk_length(leaf, chunk);
num_stripes = btrfs_chunk_num_stripes(leaf, chunk);
- ret = btrfs_check_chunk_valid(fs_info, leaf, chunk, logical);
- if (ret)
- return ret;
+ /*
+ * Only need to verify chunk item if we're reading from sys chunk array,
+ * as chunk item in tree block is already verified by tree-checker.
+ */
+ if (leaf->start == BTRFS_SUPER_INFO_OFFSET) {
+ ret = btrfs_check_chunk_valid(fs_info, leaf, chunk, logical);
+ if (ret)
+ return ret;
+ }
read_lock(&map_tree->map_tree.lock);
em = lookup_extent_mapping(&map_tree->map_tree, logical, 1);
map->type = btrfs_chunk_type(leaf, chunk);
map->sub_stripes = btrfs_chunk_sub_stripes(leaf, chunk);
map->verified_stripes = 0;
+ em->orig_block_len = calc_stripe_length(map->type, em->len,
+ map->num_stripes);
for (i = 0; i < num_stripes; i++) {
map->stripes[i].physical =
btrfs_stripe_offset_nr(leaf, chunk, i);
}
/*
- * Update the size of all devices, which is used for writing out the
- * super blocks.
+ * Update the size and bytes used for each device where it changed. This is
+ * delayed since we would otherwise get errors while writing out the
+ * superblocks.
+ *
+ * Must be invoked during transaction commit.
*/
-void btrfs_update_commit_device_size(struct btrfs_fs_info *fs_info)
+void btrfs_commit_device_sizes(struct btrfs_transaction *trans)
{
- struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
struct btrfs_device *curr, *next;
- if (list_empty(&fs_devices->resized_devices))
- return;
-
- mutex_lock(&fs_devices->device_list_mutex);
- mutex_lock(&fs_info->chunk_mutex);
- list_for_each_entry_safe(curr, next, &fs_devices->resized_devices,
- resized_list) {
- list_del_init(&curr->resized_list);
- curr->commit_total_bytes = curr->disk_total_bytes;
- }
- mutex_unlock(&fs_info->chunk_mutex);
- mutex_unlock(&fs_devices->device_list_mutex);
-}
-
-/* Must be invoked during the transaction commit */
-void btrfs_update_commit_device_bytes_used(struct btrfs_transaction *trans)
-{
- struct btrfs_fs_info *fs_info = trans->fs_info;
- struct extent_map *em;
- struct map_lookup *map;
- struct btrfs_device *dev;
- int i;
+ ASSERT(trans->state == TRANS_STATE_COMMIT_DOING);
- if (list_empty(&trans->pending_chunks))
+ if (list_empty(&trans->dev_update_list))
return;
- /* In order to kick the device replace finish process */
- mutex_lock(&fs_info->chunk_mutex);
- list_for_each_entry(em, &trans->pending_chunks, list) {
- map = em->map_lookup;
-
- for (i = 0; i < map->num_stripes; i++) {
- dev = map->stripes[i].dev;
- dev->commit_bytes_used = dev->bytes_used;
- }
+ /*
+ * We don't need the device_list_mutex here. This list is owned by the
+ * transaction and the transaction must complete before the device is
+ * released.
+ */
+ mutex_lock(&trans->fs_info->chunk_mutex);
+ list_for_each_entry_safe(curr, next, &trans->dev_update_list,
+ post_commit_list) {
+ list_del_init(&curr->post_commit_list);
+ curr->commit_total_bytes = curr->disk_total_bytes;
+ curr->commit_bytes_used = curr->bytes_used;
}
- mutex_unlock(&fs_info->chunk_mutex);
+ mutex_unlock(&trans->fs_info->chunk_mutex);
}
void btrfs_set_fs_info_ptr(struct btrfs_fs_info *fs_info)
}
-static u64 calc_stripe_length(u64 type, u64 chunk_len, int num_stripes)
-{
- int index = btrfs_bg_flags_to_raid_index(type);
- int ncopies = btrfs_raid_array[index].ncopies;
- int data_stripes;
-
- switch (type & BTRFS_BLOCK_GROUP_PROFILE_MASK) {
- case BTRFS_BLOCK_GROUP_RAID5:
- data_stripes = num_stripes - 1;
- break;
- case BTRFS_BLOCK_GROUP_RAID6:
- data_stripes = num_stripes - 2;
- break;
- default:
- data_stripes = num_stripes / ncopies;
- break;
- }
- return div_u64(chunk_len, data_stripes);
-}
static int verify_one_dev_extent(struct btrfs_fs_info *fs_info,
u64 chunk_offset, u64 devid,