#include "volumes.h"
#include "locking.h"
#include "free-space-cache.h"
+#include "math.h"
#undef SCRAMBLE_DELAYED_REFS
RESERVE_ALLOC_NO_ACCOUNT = 2,
};
-static int update_block_group(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
+static int update_block_group(struct btrfs_root *root,
u64 bytenr, u64 num_bytes, int alloc);
static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
}
static int cache_block_group(struct btrfs_block_group_cache *cache,
- struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
int load_cache_only)
{
DEFINE_WAIT(wait);
cache->cached = BTRFS_CACHE_FAST;
spin_unlock(&cache->lock);
- /*
- * We can't do the read from on-disk cache during a commit since we need
- * to have the normal tree locking. Also if we are currently trying to
- * allocate blocks for the tree root we can't do the fast caching since
- * we likely hold important locks.
- */
if (fs_info->mount_opt & BTRFS_MOUNT_SPACE_CACHE) {
ret = load_free_space_cache(fs_info, cache);
rcu_read_unlock();
}
-static u64 div_factor(u64 num, int factor)
-{
- if (factor == 10)
- return num;
- num *= factor;
- do_div(num, 10);
- return num;
-}
-
-static u64 div_factor_fine(u64 num, int factor)
-{
- if (factor == 100)
- return num;
- num *= factor;
- do_div(num, 100);
- return num;
-}
-
u64 btrfs_find_block_group(struct btrfs_root *root,
u64 search_start, u64 search_hint, int owner)
{
/* Tell the block device(s) that the sectors can be discarded */
- ret = btrfs_map_block(&root->fs_info->mapping_tree, REQ_DISCARD,
+ ret = btrfs_map_block(root->fs_info, REQ_DISCARD,
bytenr, &num_bytes, &bbio, 0);
/* Error condition is -ENOMEM */
if (!ret) {
node->num_bytes);
}
}
- mutex_unlock(&head->mutex);
return ret;
}
* process of being added. Don't run this ref yet.
*/
list_del_init(&locked_ref->cluster);
- mutex_unlock(&locked_ref->mutex);
+ btrfs_delayed_ref_unlock(locked_ref);
locked_ref = NULL;
delayed_refs->num_heads_ready++;
spin_unlock(&delayed_refs->lock);
ref = &locked_ref->node;
if (extent_op && must_insert_reserved) {
- kfree(extent_op);
+ btrfs_free_delayed_extent_op(extent_op);
extent_op = NULL;
}
ret = run_delayed_extent_op(trans, root,
ref, extent_op);
- kfree(extent_op);
+ btrfs_free_delayed_extent_op(extent_op);
if (ret) {
- printk(KERN_DEBUG "btrfs: run_delayed_extent_op returned %d\n", ret);
+ printk(KERN_DEBUG
+ "btrfs: run_delayed_extent_op "
+ "returned %d\n", ret);
spin_lock(&delayed_refs->lock);
+ btrfs_delayed_ref_unlock(locked_ref);
return ret;
}
goto next;
}
-
- list_del_init(&locked_ref->cluster);
- locked_ref = NULL;
}
ref->in_tree = 0;
rb_erase(&ref->rb_node, &delayed_refs->root);
delayed_refs->num_entries--;
- if (locked_ref) {
+ if (!btrfs_delayed_ref_is_head(ref)) {
/*
* when we play the delayed ref, also correct the
* ref_mod on head
ret = run_one_delayed_ref(trans, root, ref, extent_op,
must_insert_reserved);
- btrfs_put_delayed_ref(ref);
- kfree(extent_op);
- count++;
-
+ btrfs_free_delayed_extent_op(extent_op);
if (ret) {
- printk(KERN_DEBUG "btrfs: run_one_delayed_ref returned %d\n", ret);
+ btrfs_delayed_ref_unlock(locked_ref);
+ btrfs_put_delayed_ref(ref);
+ printk(KERN_DEBUG
+ "btrfs: run_one_delayed_ref returned %d\n", ret);
spin_lock(&delayed_refs->lock);
return ret;
}
+ /*
+ * If this node is a head, that means all the refs in this head
+ * have been dealt with, and we will pick the next head to deal
+ * with, so we must unlock the head and drop it from the cluster
+ * list before we release it.
+ */
+ if (btrfs_delayed_ref_is_head(ref)) {
+ list_del_init(&locked_ref->cluster);
+ btrfs_delayed_ref_unlock(locked_ref);
+ locked_ref = NULL;
+ }
+ btrfs_put_delayed_ref(ref);
+ count++;
next:
cond_resched();
spin_lock(&delayed_refs->lock);
ret = run_clustered_refs(trans, root, &cluster);
if (ret < 0) {
+ btrfs_release_ref_cluster(&cluster);
spin_unlock(&delayed_refs->lock);
btrfs_abort_transaction(trans, root, ret);
return ret;
struct btrfs_delayed_extent_op *extent_op;
int ret;
- extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
+ extent_op = btrfs_alloc_delayed_extent_op();
if (!extent_op)
return -ENOMEM;
ret = btrfs_add_delayed_extent_op(root->fs_info, trans, bytenr,
num_bytes, extent_op);
if (ret)
- kfree(extent_op);
+ btrfs_free_delayed_extent_op(extent_op);
return ret;
}
int wait_for_alloc = 0;
int ret = 0;
+ /* Don't re-enter if we're already allocating a chunk */
+ if (trans->allocating_chunk)
+ return -ENOSPC;
+
space_info = __find_space_info(extent_root->fs_info, flags);
if (!space_info) {
ret = update_space_info(extent_root->fs_info, flags,
goto again;
}
+ trans->allocating_chunk = true;
+
/*
* If we have mixed data/metadata chunks we want to make sure we keep
* allocating mixed chunks instead of individual chunks.
check_system_chunk(trans, extent_root, flags);
ret = btrfs_alloc_chunk(trans, extent_root, flags);
+ trans->allocating_chunk = false;
if (ret < 0 && ret != -ENOSPC)
goto out;
static int can_overcommit(struct btrfs_root *root,
struct btrfs_space_info *space_info, u64 bytes,
- int flush)
+ enum btrfs_reserve_flush_enum flush)
{
u64 profile = btrfs_get_alloc_profile(root, 0);
u64 avail;
avail >>= 1;
/*
- * If we aren't flushing don't let us overcommit too much, say
- * 1/8th of the space. If we can flush, let it overcommit up to
- * 1/2 of the space.
+ * If we aren't flushing all things, let us overcommit up to
+ * 1/2th of the space. If we can flush, don't let us overcommit
+ * too much, let it overcommit up to 1/8 of the space.
*/
- if (flush)
+ if (flush == BTRFS_RESERVE_FLUSH_ALL)
avail >>= 3;
else
avail >>= 1;
return 0;
}
+static inline int writeback_inodes_sb_nr_if_idle_safe(struct super_block *sb,
+ unsigned long nr_pages,
+ enum wb_reason reason)
+{
+ /* the flusher is dealing with the dirty inodes now. */
+ if (writeback_in_progress(sb->s_bdi))
+ return 1;
+
+ if (down_read_trylock(&sb->s_umount)) {
+ writeback_inodes_sb_nr(sb, nr_pages, reason);
+ up_read(&sb->s_umount);
+ return 1;
+ }
+
+ return 0;
+}
+
+void btrfs_writeback_inodes_sb_nr(struct btrfs_root *root,
+ unsigned long nr_pages)
+{
+ struct super_block *sb = root->fs_info->sb;
+ int started;
+
+ /* If we can not start writeback, just sync all the delalloc file. */
+ started = writeback_inodes_sb_nr_if_idle_safe(sb, nr_pages,
+ WB_REASON_FS_FREE_SPACE);
+ if (!started) {
+ /*
+ * We needn't worry the filesystem going from r/w to r/o though
+ * we don't acquire ->s_umount mutex, because the filesystem
+ * should guarantee the delalloc inodes list be empty after
+ * the filesystem is readonly(all dirty pages are written to
+ * the disk).
+ */
+ btrfs_start_delalloc_inodes(root, 0);
+ btrfs_wait_ordered_extents(root, 0);
+ }
+}
+
/*
* shrink metadata reservation for delalloc
*/
long time_left;
unsigned long nr_pages = (2 * 1024 * 1024) >> PAGE_CACHE_SHIFT;
int loops = 0;
+ enum btrfs_reserve_flush_enum flush;
trans = (struct btrfs_trans_handle *)current->journal_info;
block_rsv = &root->fs_info->delalloc_block_rsv;
while (delalloc_bytes && loops < 3) {
max_reclaim = min(delalloc_bytes, to_reclaim);
nr_pages = max_reclaim >> PAGE_CACHE_SHIFT;
- writeback_inodes_sb_nr_if_idle(root->fs_info->sb, nr_pages,
- WB_REASON_FS_FREE_SPACE);
-
+ btrfs_writeback_inodes_sb_nr(root, nr_pages);
/*
* We need to wait for the async pages to actually start before
* we do anything.
wait_event(root->fs_info->async_submit_wait,
!atomic_read(&root->fs_info->async_delalloc_pages));
+ if (!trans)
+ flush = BTRFS_RESERVE_FLUSH_ALL;
+ else
+ flush = BTRFS_RESERVE_NO_FLUSH;
spin_lock(&space_info->lock);
- if (can_overcommit(root, space_info, orig, !trans)) {
+ if (can_overcommit(root, space_info, orig, flush)) {
spin_unlock(&space_info->lock);
break;
}
*/
static int reserve_metadata_bytes(struct btrfs_root *root,
struct btrfs_block_rsv *block_rsv,
- u64 orig_bytes, int flush)
+ u64 orig_bytes,
+ enum btrfs_reserve_flush_enum flush)
{
struct btrfs_space_info *space_info = block_rsv->space_info;
u64 used;
ret = 0;
spin_lock(&space_info->lock);
/*
- * We only want to wait if somebody other than us is flushing and we are
- * actually alloed to flush.
+ * We only want to wait if somebody other than us is flushing and we
+ * are actually allowed to flush all things.
*/
- while (flush && !flushing && space_info->flush) {
+ while (flush == BTRFS_RESERVE_FLUSH_ALL && !flushing &&
+ space_info->flush) {
spin_unlock(&space_info->lock);
/*
* If we have a trans handle we can't wait because the flusher
* Couldn't make our reservation, save our place so while we're trying
* to reclaim space we can actually use it instead of somebody else
* stealing it from us.
+ *
+ * We make the other tasks wait for the flush only when we can flush
+ * all things.
*/
- if (ret && flush) {
+ if (ret && flush != BTRFS_RESERVE_NO_FLUSH) {
flushing = true;
space_info->flush = 1;
}
spin_unlock(&space_info->lock);
- if (!ret || !flush)
+ if (!ret || flush == BTRFS_RESERVE_NO_FLUSH)
goto out;
ret = flush_space(root, space_info, num_bytes, orig_bytes,
flush_state);
flush_state++;
+
+ /*
+ * If we are FLUSH_LIMIT, we can not flush delalloc, or the deadlock
+ * would happen. So skip delalloc flush.
+ */
+ if (flush == BTRFS_RESERVE_FLUSH_LIMIT &&
+ (flush_state == FLUSH_DELALLOC ||
+ flush_state == FLUSH_DELALLOC_WAIT))
+ flush_state = ALLOC_CHUNK;
+
if (!ret)
goto again;
- else if (flush_state <= COMMIT_TRANS)
+ else if (flush == BTRFS_RESERVE_FLUSH_LIMIT &&
+ flush_state < COMMIT_TRANS)
+ goto again;
+ else if (flush == BTRFS_RESERVE_FLUSH_ALL &&
+ flush_state <= COMMIT_TRANS)
goto again;
out:
kfree(rsv);
}
-static inline int __block_rsv_add(struct btrfs_root *root,
- struct btrfs_block_rsv *block_rsv,
- u64 num_bytes, int flush)
+int btrfs_block_rsv_add(struct btrfs_root *root,
+ struct btrfs_block_rsv *block_rsv, u64 num_bytes,
+ enum btrfs_reserve_flush_enum flush)
{
int ret;
return ret;
}
-int btrfs_block_rsv_add(struct btrfs_root *root,
- struct btrfs_block_rsv *block_rsv,
- u64 num_bytes)
-{
- return __block_rsv_add(root, block_rsv, num_bytes, 1);
-}
-
-int btrfs_block_rsv_add_noflush(struct btrfs_root *root,
- struct btrfs_block_rsv *block_rsv,
- u64 num_bytes)
-{
- return __block_rsv_add(root, block_rsv, num_bytes, 0);
-}
-
int btrfs_block_rsv_check(struct btrfs_root *root,
struct btrfs_block_rsv *block_rsv, int min_factor)
{
return ret;
}
-static inline int __btrfs_block_rsv_refill(struct btrfs_root *root,
- struct btrfs_block_rsv *block_rsv,
- u64 min_reserved, int flush)
+int btrfs_block_rsv_refill(struct btrfs_root *root,
+ struct btrfs_block_rsv *block_rsv, u64 min_reserved,
+ enum btrfs_reserve_flush_enum flush)
{
u64 num_bytes = 0;
int ret = -ENOSPC;
return ret;
}
-int btrfs_block_rsv_refill(struct btrfs_root *root,
- struct btrfs_block_rsv *block_rsv,
- u64 min_reserved)
-{
- return __btrfs_block_rsv_refill(root, block_rsv, min_reserved, 1);
-}
-
-int btrfs_block_rsv_refill_noflush(struct btrfs_root *root,
- struct btrfs_block_rsv *block_rsv,
- u64 min_reserved)
-{
- return __btrfs_block_rsv_refill(root, block_rsv, min_reserved, 0);
-}
-
int btrfs_block_rsv_migrate(struct btrfs_block_rsv *src_rsv,
struct btrfs_block_rsv *dst_rsv,
u64 num_bytes)
u64 csum_bytes;
unsigned nr_extents = 0;
int extra_reserve = 0;
- int flush = 1;
- int ret;
+ enum btrfs_reserve_flush_enum flush = BTRFS_RESERVE_FLUSH_ALL;
+ int ret = 0;
+ bool delalloc_lock = true;
- /* Need to be holding the i_mutex here if we aren't free space cache */
- if (btrfs_is_free_space_inode(inode))
- flush = 0;
+ /* If we are a free space inode we need to not flush since we will be in
+ * the middle of a transaction commit. We also don't need the delalloc
+ * mutex since we won't race with anybody. We need this mostly to make
+ * lockdep shut its filthy mouth.
+ */
+ if (btrfs_is_free_space_inode(inode)) {
+ flush = BTRFS_RESERVE_NO_FLUSH;
+ delalloc_lock = false;
+ }
- if (flush && btrfs_transaction_in_commit(root->fs_info))
+ if (flush != BTRFS_RESERVE_NO_FLUSH &&
+ btrfs_transaction_in_commit(root->fs_info))
schedule_timeout(1);
- mutex_lock(&BTRFS_I(inode)->delalloc_mutex);
+ if (delalloc_lock)
+ mutex_lock(&BTRFS_I(inode)->delalloc_mutex);
+
num_bytes = ALIGN(num_bytes, root->sectorsize);
spin_lock(&BTRFS_I(inode)->lock);
csum_bytes = BTRFS_I(inode)->csum_bytes;
spin_unlock(&BTRFS_I(inode)->lock);
- if (root->fs_info->quota_enabled) {
+ if (root->fs_info->quota_enabled)
ret = btrfs_qgroup_reserve(root, num_bytes +
nr_extents * root->leafsize);
- if (ret) {
- mutex_unlock(&BTRFS_I(inode)->delalloc_mutex);
- return ret;
- }
- }
- ret = reserve_metadata_bytes(root, block_rsv, to_reserve, flush);
+ /*
+ * ret != 0 here means the qgroup reservation failed, we go straight to
+ * the shared error handling then.
+ */
+ if (ret == 0)
+ ret = reserve_metadata_bytes(root, block_rsv,
+ to_reserve, flush);
+
if (ret) {
u64 to_free = 0;
unsigned dropped;
btrfs_ino(inode),
to_free, 0);
}
- mutex_unlock(&BTRFS_I(inode)->delalloc_mutex);
+ if (root->fs_info->quota_enabled) {
+ btrfs_qgroup_free(root, num_bytes +
+ nr_extents * root->leafsize);
+ }
+ if (delalloc_lock)
+ mutex_unlock(&BTRFS_I(inode)->delalloc_mutex);
return ret;
}
}
BTRFS_I(inode)->reserved_extents += nr_extents;
spin_unlock(&BTRFS_I(inode)->lock);
- mutex_unlock(&BTRFS_I(inode)->delalloc_mutex);
+
+ if (delalloc_lock)
+ mutex_unlock(&BTRFS_I(inode)->delalloc_mutex);
if (to_reserve)
trace_btrfs_space_reservation(root->fs_info,"delalloc",
btrfs_free_reserved_data_space(inode, num_bytes);
}
-static int update_block_group(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
+static int update_block_group(struct btrfs_root *root,
u64 bytenr, u64 num_bytes, int alloc)
{
struct btrfs_block_group_cache *cache = NULL;
* space back to the block group, otherwise we will leak space.
*/
if (!alloc && cache->cached == BTRFS_CACHE_NO)
- cache_block_group(cache, trans, NULL, 1);
+ cache_block_group(cache, 1);
byte_in_group = bytenr - cache->key.objectid;
WARN_ON(byte_in_group > cache->key.offset);
* to one because the slow code to read in the free extents does check
* the pinned extents.
*/
- cache_block_group(cache, trans, root, 1);
+ cache_block_group(cache, 1);
pin_down_extent(root, cache, bytenr, num_bytes, 0);
{
struct btrfs_fs_info *fs_info = root->fs_info;
struct btrfs_block_group_cache *cache = NULL;
+ struct btrfs_space_info *space_info;
+ struct btrfs_block_rsv *global_rsv = &fs_info->global_block_rsv;
u64 len;
+ bool readonly;
while (start <= end) {
+ readonly = false;
if (!cache ||
start >= cache->key.objectid + cache->key.offset) {
if (cache)
}
start += len;
+ space_info = cache->space_info;
- spin_lock(&cache->space_info->lock);
+ spin_lock(&space_info->lock);
spin_lock(&cache->lock);
cache->pinned -= len;
- cache->space_info->bytes_pinned -= len;
- if (cache->ro)
- cache->space_info->bytes_readonly += len;
+ space_info->bytes_pinned -= len;
+ if (cache->ro) {
+ space_info->bytes_readonly += len;
+ readonly = true;
+ }
spin_unlock(&cache->lock);
- spin_unlock(&cache->space_info->lock);
+ if (!readonly && global_rsv->space_info == space_info) {
+ spin_lock(&global_rsv->lock);
+ if (!global_rsv->full) {
+ len = min(len, global_rsv->size -
+ global_rsv->reserved);
+ global_rsv->reserved += len;
+ space_info->bytes_may_use += len;
+ if (global_rsv->reserved >= global_rsv->size)
+ global_rsv->full = 1;
+ }
+ spin_unlock(&global_rsv->lock);
+ }
+ spin_unlock(&space_info->lock);
}
if (cache)
}
}
- ret = update_block_group(trans, root, bytenr, num_bytes, 0);
+ ret = update_block_group(root, bytenr, num_bytes, 0);
if (ret) {
btrfs_abort_transaction(trans, extent_root, ret);
goto out;
if (head->extent_op) {
if (!head->must_insert_reserved)
goto out;
- kfree(head->extent_op);
+ btrfs_free_delayed_extent_op(head->extent_op);
head->extent_op = NULL;
}
return 0;
}
-static int __get_block_group_index(u64 flags)
+int __get_raid_index(u64 flags)
{
int index;
static int get_block_group_index(struct btrfs_block_group_cache *cache)
{
- return __get_block_group_index(cache->flags);
+ return __get_raid_index(cache->flags);
}
enum btrfs_loop_type {
int empty_cluster = 2 * 1024 * 1024;
struct btrfs_space_info *space_info;
int loop = 0;
- int index = 0;
+ int index = __get_raid_index(data);
int alloc_type = (data & BTRFS_BLOCK_GROUP_DATA) ?
RESERVE_ALLOC_NO_ACCOUNT : RESERVE_ALLOC;
bool found_uncached_bg = false;
cached = block_group_cache_done(block_group);
if (unlikely(!cached)) {
found_uncached_bg = true;
- ret = cache_block_group(block_group, trans,
- orig_root, 0);
+ ret = cache_block_group(block_group, 0);
BUG_ON(ret < 0);
ret = 0;
}
btrfs_mark_buffer_dirty(path->nodes[0]);
btrfs_free_path(path);
- ret = update_block_group(trans, root, ins->objectid, ins->offset, 1);
+ ret = update_block_group(root, ins->objectid, ins->offset, 1);
if (ret) { /* -ENOENT, logic error */
printk(KERN_ERR "btrfs update block group failed for %llu "
"%llu\n", (unsigned long long)ins->objectid,
btrfs_mark_buffer_dirty(leaf);
btrfs_free_path(path);
- ret = update_block_group(trans, root, ins->objectid, ins->offset, 1);
+ ret = update_block_group(root, ins->objectid, ins->offset, 1);
if (ret) { /* -ENOENT, logic error */
printk(KERN_ERR "btrfs update block group failed for %llu "
"%llu\n", (unsigned long long)ins->objectid,
u64 num_bytes = ins->offset;
block_group = btrfs_lookup_block_group(root->fs_info, ins->objectid);
- cache_block_group(block_group, trans, NULL, 0);
+ cache_block_group(block_group, 0);
caching_ctl = get_caching_control(block_group);
if (!caching_ctl) {
block_rsv = get_block_rsv(trans, root);
if (block_rsv->size == 0) {
- ret = reserve_metadata_bytes(root, block_rsv, blocksize, 0);
+ ret = reserve_metadata_bytes(root, block_rsv, blocksize,
+ BTRFS_RESERVE_NO_FLUSH);
/*
* If we couldn't reserve metadata bytes try and use some from
* the global reserve.
static DEFINE_RATELIMIT_STATE(_rs,
DEFAULT_RATELIMIT_INTERVAL,
/*DEFAULT_RATELIMIT_BURST*/ 2);
- if (__ratelimit(&_rs)) {
- printk(KERN_DEBUG "btrfs: block rsv returned %d\n", ret);
- WARN_ON(1);
- }
- ret = reserve_metadata_bytes(root, block_rsv, blocksize, 0);
+ if (__ratelimit(&_rs))
+ WARN(1, KERN_DEBUG "btrfs: block rsv returned %d\n",
+ ret);
+ ret = reserve_metadata_bytes(root, block_rsv, blocksize,
+ BTRFS_RESERVE_NO_FLUSH);
if (!ret) {
return block_rsv;
} else if (ret && block_rsv != global_rsv) {
if (root_objectid != BTRFS_TREE_LOG_OBJECTID) {
struct btrfs_delayed_extent_op *extent_op;
- extent_op = kmalloc(sizeof(*extent_op), GFP_NOFS);
+ extent_op = btrfs_alloc_delayed_extent_op();
BUG_ON(!extent_op); /* -ENOMEM */
if (key)
memcpy(&extent_op->key, key, sizeof(extent_op->key));
&wc->flags[level]);
if (ret < 0) {
btrfs_tree_unlock_rw(eb, path->locks[level]);
+ path->locks[level] = 0;
return ret;
}
BUG_ON(wc->refs[level] == 0);
if (wc->refs[level] == 1) {
btrfs_tree_unlock_rw(eb, path->locks[level]);
+ path->locks[level] = 0;
return 1;
}
}
*/
target = get_restripe_target(root->fs_info, block_group->flags);
if (target) {
- index = __get_block_group_index(extended_to_chunk(target));
+ index = __get_raid_index(extended_to_chunk(target));
} else {
/*
* this is just a balance, so if we were marked as full
* check to make sure we can actually find a chunk with enough
* space to fit our block group in.
*/
- if (device->total_bytes > device->bytes_used + min_free) {
+ if (device->total_bytes > device->bytes_used + min_free &&
+ !device->is_tgtdev_for_dev_replace) {
ret = find_free_dev_extent(device, min_free,
&dev_offset, NULL);
if (!ret)
if (end - start >= range->minlen) {
if (!block_group_cache_done(cache)) {
- ret = cache_block_group(cache, NULL, root, 0);
+ ret = cache_block_group(cache, 0);
if (!ret)
wait_block_group_cache_done(cache);
}
projects / deliverable / linux.git / blobdiff