*/
#define BTRFS_EXTENT_CSUM_OBJECTID -10ULL
+/* For storing free space cache */
+#define BTRFS_FREE_SPACE_OBJECTID -11ULL
+
/* dummy objectid represents multiple objectids */
#define BTRFS_MULTIPLE_OBJECTIDS -255ULL
/* additional stripes go here */
} __attribute__ ((__packed__));
+#define BTRFS_FREE_SPACE_EXTENT 1
+#define BTRFS_FREE_SPACE_BITMAP 2
+
+struct btrfs_free_space_entry {
+ __le64 offset;
+ __le64 bytes;
+ u8 type;
+} __attribute__ ((__packed__));
+
+struct btrfs_free_space_header {
+ struct btrfs_disk_key location;
+ __le64 generation;
+ __le64 num_entries;
+ __le64 num_bitmaps;
+} __attribute__ ((__packed__));
+
static inline unsigned long btrfs_chunk_item_size(int num_stripes)
{
BUG_ON(num_stripes == 0);
char label[BTRFS_LABEL_SIZE];
+ __le64 cache_generation;
+
/* future expansion */
- __le64 reserved[32];
+ __le64 reserved[31];
u8 sys_chunk_array[BTRFS_SYSTEM_CHUNK_ARRAY_SIZE];
} __attribute__ ((__packed__));
* ones specified below then we will fail to mount
*/
#define BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF (1ULL << 0)
-#define BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL (2ULL << 0)
+#define BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL (1ULL << 1)
+#define BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS (1ULL << 2)
#define BTRFS_FEATURE_COMPAT_SUPP 0ULL
#define BTRFS_FEATURE_COMPAT_RO_SUPP 0ULL
-#define BTRFS_FEATURE_INCOMPAT_SUPP \
- (BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF | \
- BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL)
+#define BTRFS_FEATURE_INCOMPAT_SUPP \
+ (BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF | \
+ BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL | \
+ BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS)
/*
* A leaf is full of items. offset and size tell us where to find
struct btrfs_space_info {
u64 flags;
- u64 total_bytes; /* total bytes in the space */
+ u64 total_bytes; /* total bytes in the space,
+ this doesn't take mirrors into account */
u64 bytes_used; /* total bytes used,
this does't take mirrors into account */
u64 bytes_pinned; /* total bytes pinned, will be freed when the
u64 bytes_may_use; /* number of bytes that may be used for
delalloc/allocations */
u64 disk_used; /* total bytes used on disk */
+ u64 disk_total; /* total bytes on disk, takes mirrors into
+ account */
int full; /* indicates that we cannot allocate any more
chunks for this space */
BTRFS_CACHE_FINISHED = 2,
};
+enum btrfs_disk_cache_state {
+ BTRFS_DC_WRITTEN = 0,
+ BTRFS_DC_ERROR = 1,
+ BTRFS_DC_CLEAR = 2,
+ BTRFS_DC_SETUP = 3,
+ BTRFS_DC_NEED_WRITE = 4,
+};
+
struct btrfs_caching_control {
struct list_head list;
struct mutex mutex;
struct btrfs_key key;
struct btrfs_block_group_item item;
struct btrfs_fs_info *fs_info;
+ struct inode *inode;
spinlock_t lock;
u64 pinned;
u64 reserved;
int extents_thresh;
int free_extents;
int total_bitmaps;
- int ro;
- int dirty;
+ int ro:1;
+ int dirty:1;
+ int iref:1;
+
+ int disk_cache_state;
/* cache tracking stuff */
int cached;
struct btrfs_workers endio_meta_workers;
struct btrfs_workers endio_meta_write_workers;
struct btrfs_workers endio_write_workers;
+ struct btrfs_workers endio_freespace_worker;
struct btrfs_workers submit_workers;
/*
* fixup workers take dirty pages that didn't properly go through
#define BTRFS_MOUNT_NOSSD (1 << 9)
#define BTRFS_MOUNT_DISCARD (1 << 10)
#define BTRFS_MOUNT_FORCE_COMPRESS (1 << 11)
+#define BTRFS_MOUNT_SPACE_CACHE (1 << 12)
+#define BTRFS_MOUNT_CLEAR_CACHE (1 << 13)
#define btrfs_clear_opt(o, opt) ((o) &= ~BTRFS_MOUNT_##opt)
#define btrfs_set_opt(o, opt) ((o) |= BTRFS_MOUNT_##opt)
write_eb_member(eb, item, struct btrfs_dir_item, location, key);
}
+BTRFS_SETGET_FUNCS(free_space_entries, struct btrfs_free_space_header,
+ num_entries, 64);
+BTRFS_SETGET_FUNCS(free_space_bitmaps, struct btrfs_free_space_header,
+ num_bitmaps, 64);
+BTRFS_SETGET_FUNCS(free_space_generation, struct btrfs_free_space_header,
+ generation, 64);
+
+static inline void btrfs_free_space_key(struct extent_buffer *eb,
+ struct btrfs_free_space_header *h,
+ struct btrfs_disk_key *key)
+{
+ read_eb_member(eb, h, struct btrfs_free_space_header, location, key);
+}
+
+static inline void btrfs_set_free_space_key(struct extent_buffer *eb,
+ struct btrfs_free_space_header *h,
+ struct btrfs_disk_key *key)
+{
+ write_eb_member(eb, h, struct btrfs_free_space_header, location, key);
+}
+
/* struct btrfs_disk_key */
BTRFS_SETGET_STACK_FUNCS(disk_key_objectid, struct btrfs_disk_key,
objectid, 64);
incompat_flags, 64);
BTRFS_SETGET_STACK_FUNCS(super_csum_type, struct btrfs_super_block,
csum_type, 16);
+BTRFS_SETGET_STACK_FUNCS(super_cache_generation, struct btrfs_super_block,
+ cache_generation, 64);
static inline int btrfs_super_csum_size(struct btrfs_super_block *s)
{
return file->f_path.dentry;
}
+static inline bool btrfs_mixed_space_info(struct btrfs_space_info *space_info)
+{
+ return ((space_info->flags & BTRFS_BLOCK_GROUP_METADATA) &&
+ (space_info->flags & BTRFS_BLOCK_GROUP_DATA));
+}
+
/* extent-tree.c */
void btrfs_put_block_group(struct btrfs_block_group_cache *cache);
int btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
void btrfs_free_reserved_data_space(struct inode *inode, u64 bytes);
int btrfs_trans_reserve_metadata(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
- int num_items, int *retries);
+ int num_items);
void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans,
struct btrfs_root *root);
int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans,
int btrfs_block_rsv_add(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_block_rsv *block_rsv,
- u64 num_bytes, int *retries);
+ u64 num_bytes);
int btrfs_block_rsv_check(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_block_rsv *block_rsv,
struct btrfs_block_group_cache *cache);
int btrfs_set_block_group_rw(struct btrfs_root *root,
struct btrfs_block_group_cache *cache);
+void btrfs_put_block_group_cache(struct btrfs_fs_info *info);
/* ctree.c */
int btrfs_bin_search(struct extent_buffer *eb, struct btrfs_key *key,
int level, int *slot);
u32 min_type);
int btrfs_start_delalloc_inodes(struct btrfs_root *root, int delay_iput);
- int btrfs_start_one_delalloc_inode(struct btrfs_root *root, int delay_iput);
+ int btrfs_start_one_delalloc_inode(struct btrfs_root *root, int delay_iput,
+ int sync);
int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end,
struct extent_state **cached_state);
int btrfs_writepages(struct address_space *mapping,
pgoff_t offset, pgoff_t last_index);
int btrfs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
int btrfs_readpage(struct file *file, struct page *page);
-void btrfs_delete_inode(struct inode *inode);
+void btrfs_evict_inode(struct inode *inode);
void btrfs_put_inode(struct inode *inode);
int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc);
void btrfs_dirty_inode(struct inode *inode);
struct inode *btrfs_alloc_inode(struct super_block *sb);
void btrfs_destroy_inode(struct inode *inode);
-void btrfs_drop_inode(struct inode *inode);
+int btrfs_drop_inode(struct inode *inode);
int btrfs_init_cachep(void);
void btrfs_destroy_cachep(void);
long btrfs_ioctl_trans_end(struct file *file);
int btrfs_prealloc_file_range(struct inode *inode, int mode,
u64 start, u64 num_bytes, u64 min_size,
loff_t actual_len, u64 *alloc_hint);
+int btrfs_prealloc_file_range_trans(struct inode *inode,
+ struct btrfs_trans_handle *trans, int mode,
+ u64 start, u64 num_bytes, u64 min_size,
+ loff_t actual_len, u64 *alloc_hint);
extern const struct dentry_operations btrfs_dentry_operations;
/* ioctl.c */
return NULL;
}
+ /* We're loading it the fast way, so we don't have a caching_ctl. */
+ if (!cache->caching_ctl) {
+ spin_unlock(&cache->lock);
+ return NULL;
+ }
+
ctl = cache->caching_ctl;
atomic_inc(&ctl->count);
spin_unlock(&cache->lock);
return 0;
}
-static int cache_block_group(struct btrfs_block_group_cache *cache)
+static int cache_block_group(struct btrfs_block_group_cache *cache,
+ struct btrfs_trans_handle *trans,
+ int load_cache_only)
{
struct btrfs_fs_info *fs_info = cache->fs_info;
struct btrfs_caching_control *caching_ctl;
if (cache->cached != BTRFS_CACHE_NO)
return 0;
+ /*
+ * We can't do the read from on-disk cache during a commit since we need
+ * to have the normal tree locking.
+ */
+ if (!trans->transaction->in_commit) {
+ spin_lock(&cache->lock);
+ if (cache->cached != BTRFS_CACHE_NO) {
+ spin_unlock(&cache->lock);
+ return 0;
+ }
+ cache->cached = BTRFS_CACHE_STARTED;
+ spin_unlock(&cache->lock);
+
+ ret = load_free_space_cache(fs_info, cache);
+
+ spin_lock(&cache->lock);
+ if (ret == 1) {
+ cache->cached = BTRFS_CACHE_FINISHED;
+ cache->last_byte_to_unpin = (u64)-1;
+ } else {
+ cache->cached = BTRFS_CACHE_NO;
+ }
+ spin_unlock(&cache->lock);
+ if (ret == 1)
+ return 0;
+ }
+
+ if (load_cache_only)
+ return 0;
+
caching_ctl = kzalloc(sizeof(*caching_ctl), GFP_KERNEL);
BUG_ON(!caching_ctl);
rcu_read_lock();
list_for_each_entry_rcu(found, head, list) {
- if (found->flags == flags) {
+ if (found->flags & flags) {
rcu_read_unlock();
return found;
}
return cache;
}
+static int cache_save_setup(struct btrfs_block_group_cache *block_group,
+ struct btrfs_trans_handle *trans,
+ struct btrfs_path *path)
+{
+ struct btrfs_root *root = block_group->fs_info->tree_root;
+ struct inode *inode = NULL;
+ u64 alloc_hint = 0;
+ int num_pages = 0;
+ int retries = 0;
+ int ret = 0;
+
+ /*
+ * If this block group is smaller than 100 megs don't bother caching the
+ * block group.
+ */
+ if (block_group->key.offset < (100 * 1024 * 1024)) {
+ spin_lock(&block_group->lock);
+ block_group->disk_cache_state = BTRFS_DC_WRITTEN;
+ spin_unlock(&block_group->lock);
+ return 0;
+ }
+
+again:
+ inode = lookup_free_space_inode(root, block_group, path);
+ if (IS_ERR(inode) && PTR_ERR(inode) != -ENOENT) {
+ ret = PTR_ERR(inode);
+ btrfs_release_path(root, path);
+ goto out;
+ }
+
+ if (IS_ERR(inode)) {
+ BUG_ON(retries);
+ retries++;
+
+ if (block_group->ro)
+ goto out_free;
+
+ ret = create_free_space_inode(root, trans, block_group, path);
+ if (ret)
+ goto out_free;
+ goto again;
+ }
+
+ /*
+ * We want to set the generation to 0, that way if anything goes wrong
+ * from here on out we know not to trust this cache when we load up next
+ * time.
+ */
+ BTRFS_I(inode)->generation = 0;
+ ret = btrfs_update_inode(trans, root, inode);
+ WARN_ON(ret);
+
+ if (i_size_read(inode) > 0) {
+ ret = btrfs_truncate_free_space_cache(root, trans, path,
+ inode);
+ if (ret)
+ goto out_put;
+ }
+
+ spin_lock(&block_group->lock);
+ if (block_group->cached != BTRFS_CACHE_FINISHED) {
+ spin_unlock(&block_group->lock);
+ goto out_put;
+ }
+ spin_unlock(&block_group->lock);
+
+ num_pages = (int)div64_u64(block_group->key.offset, 1024 * 1024 * 1024);
+ if (!num_pages)
+ num_pages = 1;
+
+ /*
+ * Just to make absolutely sure we have enough space, we're going to
+ * preallocate 12 pages worth of space for each block group. In
+ * practice we ought to use at most 8, but we need extra space so we can
+ * add our header and have a terminator between the extents and the
+ * bitmaps.
+ */
+ num_pages *= 16;
+ num_pages *= PAGE_CACHE_SIZE;
+
+ ret = btrfs_check_data_free_space(inode, num_pages);
+ if (ret)
+ goto out_put;
+
+ ret = btrfs_prealloc_file_range_trans(inode, trans, 0, 0, num_pages,
+ num_pages, num_pages,
+ &alloc_hint);
+ btrfs_free_reserved_data_space(inode, num_pages);
+out_put:
+ iput(inode);
+out_free:
+ btrfs_release_path(root, path);
+out:
+ spin_lock(&block_group->lock);
+ if (ret)
+ block_group->disk_cache_state = BTRFS_DC_ERROR;
+ else
+ block_group->disk_cache_state = BTRFS_DC_SETUP;
+ spin_unlock(&block_group->lock);
+
+ return ret;
+}
+
int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans,
struct btrfs_root *root)
{
if (!path)
return -ENOMEM;
+again:
+ while (1) {
+ cache = btrfs_lookup_first_block_group(root->fs_info, last);
+ while (cache) {
+ if (cache->disk_cache_state == BTRFS_DC_CLEAR)
+ break;
+ cache = next_block_group(root, cache);
+ }
+ if (!cache) {
+ if (last == 0)
+ break;
+ last = 0;
+ continue;
+ }
+ err = cache_save_setup(cache, trans, path);
+ last = cache->key.objectid + cache->key.offset;
+ btrfs_put_block_group(cache);
+ }
+
while (1) {
if (last == 0) {
err = btrfs_run_delayed_refs(trans, root,
cache = btrfs_lookup_first_block_group(root->fs_info, last);
while (cache) {
+ if (cache->disk_cache_state == BTRFS_DC_CLEAR) {
+ btrfs_put_block_group(cache);
+ goto again;
+ }
+
if (cache->dirty)
break;
cache = next_block_group(root, cache);
continue;
}
+ if (cache->disk_cache_state == BTRFS_DC_SETUP)
+ cache->disk_cache_state = BTRFS_DC_NEED_WRITE;
cache->dirty = 0;
last = cache->key.objectid + cache->key.offset;
btrfs_put_block_group(cache);
}
+ while (1) {
+ /*
+ * I don't think this is needed since we're just marking our
+ * preallocated extent as written, but just in case it can't
+ * hurt.
+ */
+ if (last == 0) {
+ err = btrfs_run_delayed_refs(trans, root,
+ (unsigned long)-1);
+ BUG_ON(err);
+ }
+
+ cache = btrfs_lookup_first_block_group(root->fs_info, last);
+ while (cache) {
+ /*
+ * Really this shouldn't happen, but it could if we
+ * couldn't write the entire preallocated extent and
+ * splitting the extent resulted in a new block.
+ */
+ if (cache->dirty) {
+ btrfs_put_block_group(cache);
+ goto again;
+ }
+ if (cache->disk_cache_state == BTRFS_DC_NEED_WRITE)
+ break;
+ cache = next_block_group(root, cache);
+ }
+ if (!cache) {
+ if (last == 0)
+ break;
+ last = 0;
+ continue;
+ }
+
+ btrfs_write_out_cache(root, trans, cache, path);
+
+ /*
+ * If we didn't have an error then the cache state is still
+ * NEED_WRITE, so we can set it to WRITTEN.
+ */
+ if (cache->disk_cache_state == BTRFS_DC_NEED_WRITE)
+ cache->disk_cache_state = BTRFS_DC_WRITTEN;
+ last = cache->key.objectid + cache->key.offset;
+ btrfs_put_block_group(cache);
+ }
+
btrfs_free_path(path);
return 0;
}
if (found) {
spin_lock(&found->lock);
found->total_bytes += total_bytes;
+ found->disk_total += total_bytes * factor;
found->bytes_used += bytes_used;
found->disk_used += bytes_used * factor;
found->full = 0;
BTRFS_BLOCK_GROUP_SYSTEM |
BTRFS_BLOCK_GROUP_METADATA);
found->total_bytes = total_bytes;
+ found->disk_total = total_bytes * factor;
found->bytes_used = bytes_used;
found->disk_used = bytes_used * factor;
found->bytes_pinned = 0;
struct btrfs_space_info *data_sinfo;
struct btrfs_root *root = BTRFS_I(inode)->root;
u64 used;
- int ret = 0, committed = 0;
+ int ret = 0, committed = 0, alloc_chunk = 1;
/* make sure bytes are sectorsize aligned */
bytes = (bytes + root->sectorsize - 1) & ~((u64)root->sectorsize - 1);
+ if (root == root->fs_info->tree_root) {
+ alloc_chunk = 0;
+ committed = 1;
+ }
+
data_sinfo = BTRFS_I(inode)->space_info;
if (!data_sinfo)
goto alloc;
* if we don't have enough free bytes in this space then we need
* to alloc a new chunk.
*/
- if (!data_sinfo->full) {
+ if (!data_sinfo->full && alloc_chunk) {
u64 alloc_target;
data_sinfo->force_alloc = 1;
rcu_read_unlock();
}
- static int should_alloc_chunk(struct btrfs_space_info *sinfo,
- u64 alloc_bytes)
+ static int should_alloc_chunk(struct btrfs_space_info *sinfo, u64 alloc_bytes)
{
u64 num_bytes = sinfo->total_bytes - sinfo->bytes_readonly;
alloc_bytes < div_factor(num_bytes, 8))
return 0;
+ if (num_bytes > 256 * 1024 * 1024 &&
+ sinfo->bytes_used < div_factor(num_bytes, 3))
+ return 0;
+
return 1;
}
}
spin_unlock(&space_info->lock);
+ /*
+ * If we have mixed data/metadata chunks we want to make sure we keep
+ * allocating mixed chunks instead of individual chunks.
+ */
+ if (btrfs_mixed_space_info(space_info))
+ flags |= (BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA);
+
/*
* if we're doing a data chunk, go ahead and make sure that
* we keep a reasonable number of metadata chunks allocated in the
return ret;
}
- static int maybe_allocate_chunk(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct btrfs_space_info *sinfo, u64 num_bytes)
- {
- int ret;
- int end_trans = 0;
-
- if (sinfo->full)
- return 0;
-
- spin_lock(&sinfo->lock);
- ret = should_alloc_chunk(sinfo, num_bytes + 2 * 1024 * 1024);
- spin_unlock(&sinfo->lock);
- if (!ret)
- return 0;
-
- if (!trans) {
- trans = btrfs_join_transaction(root, 1);
- BUG_ON(IS_ERR(trans));
- end_trans = 1;
- }
-
- ret = do_chunk_alloc(trans, root->fs_info->extent_root,
- num_bytes + 2 * 1024 * 1024,
- get_alloc_profile(root, sinfo->flags), 0);
-
- if (end_trans)
- btrfs_end_transaction(trans, root);
-
- return ret == 1 ? 1 : 0;
- }
-
/*
* shrink metadata reservation for delalloc
*/
static int shrink_delalloc(struct btrfs_trans_handle *trans,
- struct btrfs_root *root, u64 to_reclaim)
+ struct btrfs_root *root, u64 to_reclaim, int sync)
{
struct btrfs_block_rsv *block_rsv;
+ struct btrfs_space_info *space_info;
u64 reserved;
u64 max_reclaim;
u64 reclaimed = 0;
+ int no_reclaim = 0;
int pause = 1;
int ret;
block_rsv = &root->fs_info->delalloc_block_rsv;
- spin_lock(&block_rsv->lock);
- reserved = block_rsv->reserved;
- spin_unlock(&block_rsv->lock);
+ space_info = block_rsv->space_info;
+ spin_lock(&space_info->lock);
+ reserved = space_info->bytes_reserved;
+ spin_unlock(&space_info->lock);
if (reserved == 0)
return 0;
max_reclaim = min(reserved, to_reclaim);
while (1) {
- ret = btrfs_start_one_delalloc_inode(root, trans ? 1 : 0);
+ ret = btrfs_start_one_delalloc_inode(root, trans ? 1 : 0, sync);
if (!ret) {
+ if (no_reclaim > 2)
+ break;
+ no_reclaim++;
__set_current_state(TASK_INTERRUPTIBLE);
schedule_timeout(pause);
pause <<= 1;
if (pause > HZ / 10)
pause = HZ / 10;
} else {
+ no_reclaim = 0;
pause = 1;
}
- spin_lock(&block_rsv->lock);
- if (reserved > block_rsv->reserved)
- reclaimed = reserved - block_rsv->reserved;
- reserved = block_rsv->reserved;
- spin_unlock(&block_rsv->lock);
+ spin_lock(&space_info->lock);
+ if (reserved > space_info->bytes_reserved)
+ reclaimed += reserved - space_info->bytes_reserved;
+ reserved = space_info->bytes_reserved;
+ spin_unlock(&space_info->lock);
if (reserved == 0 || reclaimed >= max_reclaim)
break;
return reclaimed >= to_reclaim;
}
- static int should_retry_reserve(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct btrfs_block_rsv *block_rsv,
- u64 num_bytes, int *retries)
+ /*
+ * Retries tells us how many times we've called reserve_metadata_bytes. The
+ * idea is if this is the first call (retries == 0) then we will add to our
+ * reserved count if we can't make the allocation in order to hold our place
+ * while we go and try and free up space. That way for retries > 1 we don't try
+ * and add space, we just check to see if the amount of unused space is >= the
+ * total space, meaning that our reservation is valid.
+ *
+ * However if we don't intend to retry this reservation, pass -1 as retries so
+ * that it short circuits this logic.
+ */
+ static int reserve_metadata_bytes(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_block_rsv *block_rsv,
+ u64 orig_bytes, int flush)
{
struct btrfs_space_info *space_info = block_rsv->space_info;
- int ret;
+ u64 unused;
+ u64 num_bytes = orig_bytes;
+ int retries = 0;
+ int ret = 0;
+ bool reserved = false;
+ bool committed = false;
- if ((*retries) > 2)
- return -ENOSPC;
+ again:
+ ret = -ENOSPC;
+ if (reserved)
+ num_bytes = 0;
- ret = maybe_allocate_chunk(trans, root, space_info, num_bytes);
- if (ret)
- return 1;
+ spin_lock(&space_info->lock);
+ unused = space_info->bytes_used + space_info->bytes_reserved +
+ space_info->bytes_pinned + space_info->bytes_readonly +
+ space_info->bytes_may_use;
- if (trans && trans->transaction->in_commit)
- return -ENOSPC;
+ /*
+ * The idea here is that we've not already over-reserved the block group
+ * then we can go ahead and save our reservation first and then start
+ * flushing if we need to. Otherwise if we've already overcommitted
+ * lets start flushing stuff first and then come back and try to make
+ * our reservation.
+ */
+ if (unused <= space_info->total_bytes) {
+ unused -= space_info->total_bytes;
+ if (unused >= num_bytes) {
+ if (!reserved)
+ space_info->bytes_reserved += orig_bytes;
+ ret = 0;
+ } else {
+ /*
+ * Ok set num_bytes to orig_bytes since we aren't
+ * overocmmitted, this way we only try and reclaim what
+ * we need.
+ */
+ num_bytes = orig_bytes;
+ }
+ } else {
+ /*
+ * Ok we're over committed, set num_bytes to the overcommitted
+ * amount plus the amount of bytes that we need for this
+ * reservation.
+ */
+ num_bytes = unused - space_info->total_bytes +
+ (orig_bytes * (retries + 1));
+ }
- ret = shrink_delalloc(trans, root, num_bytes);
- if (ret)
- return ret;
+ /*
+ * 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.
+ */
+ if (ret && !reserved) {
+ space_info->bytes_reserved += orig_bytes;
+ reserved = true;
+ }
- spin_lock(&space_info->lock);
- if (space_info->bytes_pinned < num_bytes)
- ret = 1;
spin_unlock(&space_info->lock);
- if (ret)
- return -ENOSPC;
- (*retries)++;
-
- if (trans)
- return -EAGAIN;
+ if (!ret)
+ return 0;
- trans = btrfs_join_transaction(root, 1);
- BUG_ON(IS_ERR(trans));
- ret = btrfs_commit_transaction(trans, root);
- BUG_ON(ret);
+ if (!flush)
+ goto out;
- return 1;
- }
+ /*
+ * We do synchronous shrinking since we don't actually unreserve
+ * metadata until after the IO is completed.
+ */
+ ret = shrink_delalloc(trans, root, num_bytes, 1);
+ if (ret > 0)
+ return 0;
+ else if (ret < 0)
+ goto out;
- static int reserve_metadata_bytes(struct btrfs_block_rsv *block_rsv,
- u64 num_bytes)
- {
- struct btrfs_space_info *space_info = block_rsv->space_info;
- u64 unused;
- int ret = -ENOSPC;
+ /*
+ * So if we were overcommitted it's possible that somebody else flushed
+ * out enough space and we simply didn't have enough space to reclaim,
+ * so go back around and try again.
+ */
+ if (retries < 2) {
+ retries++;
+ goto again;
+ }
spin_lock(&space_info->lock);
- unused = space_info->bytes_used + space_info->bytes_reserved +
- space_info->bytes_pinned + space_info->bytes_readonly;
+ /*
+ * Not enough space to be reclaimed, don't bother committing the
+ * transaction.
+ */
+ if (space_info->bytes_pinned < orig_bytes)
+ ret = -ENOSPC;
+ spin_unlock(&space_info->lock);
+ if (ret)
+ goto out;
- if (unused < space_info->total_bytes)
- unused = space_info->total_bytes - unused;
- else
- unused = 0;
+ ret = -EAGAIN;
+ if (trans || committed)
+ goto out;
- if (unused >= num_bytes) {
- if (block_rsv->priority >= 10) {
- space_info->bytes_reserved += num_bytes;
- ret = 0;
- } else {
- if ((unused + block_rsv->reserved) *
- block_rsv->priority >=
- (num_bytes + block_rsv->reserved) * 10) {
- space_info->bytes_reserved += num_bytes;
- ret = 0;
- }
- }
+ ret = -ENOSPC;
+ trans = btrfs_join_transaction(root, 1);
+ if (IS_ERR(trans))
+ goto out;
+ ret = btrfs_commit_transaction(trans, root);
+ if (!ret) {
+ trans = NULL;
+ committed = true;
+ goto again;
+ }
+
+ out:
+ if (reserved) {
+ spin_lock(&space_info->lock);
+ space_info->bytes_reserved -= orig_bytes;
+ spin_unlock(&space_info->lock);
}
- spin_unlock(&space_info->lock);
return ret;
}
int btrfs_block_rsv_add(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct btrfs_block_rsv *block_rsv,
- u64 num_bytes, int *retries)
+ u64 num_bytes)
{
int ret;
if (num_bytes == 0)
return 0;
- again:
- ret = reserve_metadata_bytes(block_rsv, num_bytes);
+
+ ret = reserve_metadata_bytes(trans, root, block_rsv, num_bytes, 1);
if (!ret) {
block_rsv_add_bytes(block_rsv, num_bytes, 1);
return 0;
}
- ret = should_retry_reserve(trans, root, block_rsv, num_bytes, retries);
- if (ret > 0)
- goto again;
-
return ret;
}
return 0;
if (block_rsv->refill_used) {
- ret = reserve_metadata_bytes(block_rsv, num_bytes);
+ ret = reserve_metadata_bytes(trans, root, block_rsv,
+ num_bytes, 0);
if (!ret) {
block_rsv_add_bytes(block_rsv, num_bytes, 0);
return 0;
sinfo = __find_space_info(fs_info, BTRFS_BLOCK_GROUP_METADATA);
spin_lock(&sinfo->lock);
+ if (sinfo->flags & BTRFS_BLOCK_GROUP_DATA)
+ data_used = 0;
meta_used = sinfo->bytes_used;
spin_unlock(&sinfo->lock);
block_rsv->size = num_bytes;
num_bytes = sinfo->bytes_used + sinfo->bytes_pinned +
- sinfo->bytes_reserved + sinfo->bytes_readonly;
+ sinfo->bytes_reserved + sinfo->bytes_readonly +
+ sinfo->bytes_may_use;
if (sinfo->total_bytes > num_bytes) {
num_bytes = sinfo->total_bytes - num_bytes;
int btrfs_trans_reserve_metadata(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
- int num_items, int *retries)
+ int num_items)
{
u64 num_bytes;
int ret;
num_bytes = calc_trans_metadata_size(root, num_items);
ret = btrfs_block_rsv_add(trans, root, &root->fs_info->trans_block_rsv,
- num_bytes, retries);
+ num_bytes);
if (!ret) {
trans->bytes_reserved += num_bytes;
trans->block_rsv = &root->fs_info->trans_block_rsv;
struct btrfs_block_rsv *block_rsv = &root->fs_info->delalloc_block_rsv;
u64 to_reserve;
int nr_extents;
- int retries = 0;
int ret;
if (btrfs_transaction_in_commit(root->fs_info))
schedule_timeout(1);
num_bytes = ALIGN(num_bytes, root->sectorsize);
- again:
+
spin_lock(&BTRFS_I(inode)->accounting_lock);
nr_extents = atomic_read(&BTRFS_I(inode)->outstanding_extents) + 1;
if (nr_extents > BTRFS_I(inode)->reserved_extents) {
nr_extents = 0;
to_reserve = 0;
}
+ spin_unlock(&BTRFS_I(inode)->accounting_lock);
to_reserve += calc_csum_metadata_size(inode, num_bytes);
- ret = reserve_metadata_bytes(block_rsv, to_reserve);
- if (ret) {
- spin_unlock(&BTRFS_I(inode)->accounting_lock);
- ret = should_retry_reserve(NULL, root, block_rsv, to_reserve,
- &retries);
- if (ret > 0)
- goto again;
+ ret = reserve_metadata_bytes(NULL, root, block_rsv, to_reserve, 1);
+ if (ret)
return ret;
- }
+ spin_lock(&BTRFS_I(inode)->accounting_lock);
BTRFS_I(inode)->reserved_extents += nr_extents;
atomic_inc(&BTRFS_I(inode)->outstanding_extents);
spin_unlock(&BTRFS_I(inode)->accounting_lock);
block_rsv_add_bytes(block_rsv, to_reserve, 1);
if (block_rsv->size > 512 * 1024 * 1024)
- shrink_delalloc(NULL, root, to_reserve);
+ shrink_delalloc(NULL, root, to_reserve, 0);
return 0;
}
struct btrfs_root *root,
u64 bytenr, u64 num_bytes, int alloc)
{
- struct btrfs_block_group_cache *cache;
+ struct btrfs_block_group_cache *cache = NULL;
struct btrfs_fs_info *info = root->fs_info;
- int factor;
u64 total = num_bytes;
u64 old_val;
u64 byte_in_group;
+ int factor;
/* block accounting for super block */
spin_lock(&info->delalloc_lock);
factor = 2;
else
factor = 1;
+ /*
+ * If this block group has free space cache written out, we
+ * need to make sure to load it if we are removing space. This
+ * is because we need the unpinning stage to actually add the
+ * space back to the block group, otherwise we will leak space.
+ */
+ if (!alloc && cache->cached == BTRFS_CACHE_NO)
+ cache_block_group(cache, trans, 1);
+
byte_in_group = bytenr - cache->key.objectid;
WARN_ON(byte_in_group > cache->key.offset);
spin_lock(&cache->space_info->lock);
spin_lock(&cache->lock);
+
+ if (btrfs_super_cache_generation(&info->super_copy) != 0 &&
+ cache->disk_cache_state < BTRFS_DC_CLEAR)
+ cache->disk_cache_state = BTRFS_DC_CLEAR;
+
cache->dirty = 1;
old_val = btrfs_block_group_used(&cache->item);
num_bytes = min(total, cache->key.offset - byte_in_group);
bool found_uncached_bg = false;
bool failed_cluster_refill = false;
bool failed_alloc = false;
+ bool use_cluster = true;
u64 ideal_cache_percent = 0;
u64 ideal_cache_offset = 0;
return -ENOSPC;
}
+ /*
+ * If the space info is for both data and metadata it means we have a
+ * small filesystem and we can't use the clustering stuff.
+ */
+ if (btrfs_mixed_space_info(space_info))
+ use_cluster = false;
+
if (orig_root->ref_cows || empty_size)
allowed_chunk_alloc = 1;
- if (data & BTRFS_BLOCK_GROUP_METADATA) {
+ if (data & BTRFS_BLOCK_GROUP_METADATA && use_cluster) {
last_ptr = &root->fs_info->meta_alloc_cluster;
if (!btrfs_test_opt(root, SSD))
empty_cluster = 64 * 1024;
}
- if ((data & BTRFS_BLOCK_GROUP_DATA) && btrfs_test_opt(root, SSD)) {
+ if ((data & BTRFS_BLOCK_GROUP_DATA) && use_cluster &&
+ btrfs_test_opt(root, SSD)) {
last_ptr = &root->fs_info->data_alloc_cluster;
}
if (unlikely(block_group->cached == BTRFS_CACHE_NO)) {
u64 free_percent;
+ ret = cache_block_group(block_group, trans, 1);
+ if (block_group->cached == BTRFS_CACHE_FINISHED)
+ goto have_block_group;
+
free_percent = btrfs_block_group_used(&block_group->item);
free_percent *= 100;
free_percent = div64_u64(free_percent,
if (loop > LOOP_CACHING_NOWAIT ||
(loop > LOOP_FIND_IDEAL &&
atomic_read(&space_info->caching_threads) < 2)) {
- ret = cache_block_group(block_group);
+ ret = cache_block_group(block_group, trans, 0);
BUG_ON(ret);
}
found_uncached_bg = true;
u64 num_bytes = ins->offset;
block_group = btrfs_lookup_block_group(root->fs_info, ins->objectid);
- cache_block_group(block_group);
+ cache_block_group(block_group, trans, 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(block_rsv, blocksize);
+ ret = reserve_metadata_bytes(trans, root, block_rsv,
+ blocksize, 0);
if (ret)
return ERR_PTR(ret);
return block_rsv;
if (!ret)
return block_rsv;
- WARN_ON(1);
- printk(KERN_INFO"block_rsv size %llu reserved %llu freed %llu %llu\n",
- block_rsv->size, block_rsv->reserved,
- block_rsv->freed[0], block_rsv->freed[1]);
-
return ERR_PTR(-ENOSPC);
}
return ret;
}
+void btrfs_put_block_group_cache(struct btrfs_fs_info *info)
+{
+ struct btrfs_block_group_cache *block_group;
+ u64 last = 0;
+
+ while (1) {
+ struct inode *inode;
+
+ block_group = btrfs_lookup_first_block_group(info, last);
+ while (block_group) {
+ spin_lock(&block_group->lock);
+ if (block_group->iref)
+ break;
+ spin_unlock(&block_group->lock);
+ block_group = next_block_group(info->tree_root,
+ block_group);
+ }
+ if (!block_group) {
+ if (last == 0)
+ break;
+ last = 0;
+ continue;
+ }
+
+ inode = block_group->inode;
+ block_group->iref = 0;
+ block_group->inode = NULL;
+ spin_unlock(&block_group->lock);
+ iput(inode);
+ last = block_group->key.objectid + block_group->key.offset;
+ btrfs_put_block_group(block_group);
+ }
+}
+
int btrfs_free_block_groups(struct btrfs_fs_info *info)
{
struct btrfs_block_group_cache *block_group;
struct btrfs_key key;
struct btrfs_key found_key;
struct extent_buffer *leaf;
+ int need_clear = 0;
+ u64 cache_gen;
root = info->extent_root;
key.objectid = 0;
if (!path)
return -ENOMEM;
+ cache_gen = btrfs_super_cache_generation(&root->fs_info->super_copy);
+ if (cache_gen != 0 &&
+ btrfs_super_generation(&root->fs_info->super_copy) != cache_gen)
+ need_clear = 1;
+ if (btrfs_test_opt(root, CLEAR_CACHE))
+ need_clear = 1;
+ if (!btrfs_test_opt(root, SPACE_CACHE) && cache_gen)
+ printk(KERN_INFO "btrfs: disk space caching is enabled\n");
+
while (1) {
ret = find_first_block_group(root, path, &key);
if (ret > 0)
INIT_LIST_HEAD(&cache->list);
INIT_LIST_HEAD(&cache->cluster_list);
+ if (need_clear)
+ cache->disk_cache_state = BTRFS_DC_CLEAR;
+
/*
* we only want to have 32k of ram per block group for keeping
* track of free space, and if we pass 1/2 of that we want to
cache->key.offset = size;
cache->key.type = BTRFS_BLOCK_GROUP_ITEM_KEY;
cache->sectorsize = root->sectorsize;
+ cache->fs_info = root->fs_info;
/*
* we only want to have 32k of ram per block group for keeping track
struct btrfs_path *path;
struct btrfs_block_group_cache *block_group;
struct btrfs_free_cluster *cluster;
+ struct btrfs_root *tree_root = root->fs_info->tree_root;
struct btrfs_key key;
+ struct inode *inode;
int ret;
+ int factor;
root = root->fs_info->extent_root;
BUG_ON(!block_group);
BUG_ON(!block_group->ro);
+ memcpy(&key, &block_group->key, sizeof(key));
+ if (block_group->flags & (BTRFS_BLOCK_GROUP_DUP |
+ BTRFS_BLOCK_GROUP_RAID1 |
+ BTRFS_BLOCK_GROUP_RAID10))
+ factor = 2;
+ else
+ factor = 1;
+
/* make sure this block group isn't part of an allocation cluster */
cluster = &root->fs_info->data_alloc_cluster;
spin_lock(&cluster->refill_lock);
path = btrfs_alloc_path();
BUG_ON(!path);
+ inode = lookup_free_space_inode(root, block_group, path);
+ if (!IS_ERR(inode)) {
+ btrfs_orphan_add(trans, inode);
+ clear_nlink(inode);
+ /* One for the block groups ref */
+ spin_lock(&block_group->lock);
+ if (block_group->iref) {
+ block_group->iref = 0;
+ block_group->inode = NULL;
+ spin_unlock(&block_group->lock);
+ iput(inode);
+ } else {
+ spin_unlock(&block_group->lock);
+ }
+ /* One for our lookup ref */
+ iput(inode);
+ }
+
+ key.objectid = BTRFS_FREE_SPACE_OBJECTID;
+ key.offset = block_group->key.objectid;
+ key.type = 0;
+
+ ret = btrfs_search_slot(trans, tree_root, &key, path, -1, 1);
+ if (ret < 0)
+ goto out;
+ if (ret > 0)
+ btrfs_release_path(tree_root, path);
+ if (ret == 0) {
+ ret = btrfs_del_item(trans, tree_root, path);
+ if (ret)
+ goto out;
+ btrfs_release_path(tree_root, path);
+ }
+
spin_lock(&root->fs_info->block_group_cache_lock);
rb_erase(&block_group->cache_node,
&root->fs_info->block_group_cache_tree);
spin_lock(&block_group->space_info->lock);
block_group->space_info->total_bytes -= block_group->key.offset;
block_group->space_info->bytes_readonly -= block_group->key.offset;
+ block_group->space_info->disk_total -= block_group->key.offset * factor;
spin_unlock(&block_group->space_info->lock);
+ memcpy(&key, &block_group->key, sizeof(key));
+
btrfs_clear_space_info_full(root->fs_info);
btrfs_put_block_group(block_group);
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
int ret = 0;
+ BUG_ON(root == root->fs_info->tree_root);
trans = btrfs_join_transaction(root, 1);
BUG_ON(!trans);
btrfs_set_trans_block_group(trans, inode);
int type;
int nocow;
int check_prev = 1;
+ bool nolock = false;
path = btrfs_alloc_path();
BUG_ON(!path);
- trans = btrfs_join_transaction(root, 1);
+ if (root == root->fs_info->tree_root) {
+ nolock = true;
+ trans = btrfs_join_transaction_nolock(root, 1);
+ } else {
+ trans = btrfs_join_transaction(root, 1);
+ }
BUG_ON(!trans);
cow_start = (u64)-1;
BUG_ON(ret);
}
- ret = btrfs_end_transaction(trans, root);
- BUG_ON(ret);
+ if (nolock) {
+ ret = btrfs_end_transaction_nolock(trans, root);
+ BUG_ON(ret);
+ } else {
+ ret = btrfs_end_transaction(trans, root);
+ BUG_ON(ret);
+ }
btrfs_free_path(path);
return 0;
}
if (!(state->state & EXTENT_DELALLOC) && (*bits & EXTENT_DELALLOC)) {
struct btrfs_root *root = BTRFS_I(inode)->root;
u64 len = state->end + 1 - state->start;
+ int do_list = (root->root_key.objectid !=
+ BTRFS_ROOT_TREE_OBJECTID);
if (*bits & EXTENT_FIRST_DELALLOC)
*bits &= ~EXTENT_FIRST_DELALLOC;
spin_lock(&root->fs_info->delalloc_lock);
BTRFS_I(inode)->delalloc_bytes += len;
root->fs_info->delalloc_bytes += len;
- if (list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
+ if (do_list && list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
list_add_tail(&BTRFS_I(inode)->delalloc_inodes,
&root->fs_info->delalloc_inodes);
}
if ((state->state & EXTENT_DELALLOC) && (*bits & EXTENT_DELALLOC)) {
struct btrfs_root *root = BTRFS_I(inode)->root;
u64 len = state->end + 1 - state->start;
+ int do_list = (root->root_key.objectid !=
+ BTRFS_ROOT_TREE_OBJECTID);
if (*bits & EXTENT_FIRST_DELALLOC)
*bits &= ~EXTENT_FIRST_DELALLOC;
if (*bits & EXTENT_DO_ACCOUNTING)
btrfs_delalloc_release_metadata(inode, len);
- if (root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID)
+ if (root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID
+ && do_list)
btrfs_free_reserved_data_space(inode, len);
spin_lock(&root->fs_info->delalloc_lock);
root->fs_info->delalloc_bytes -= len;
BTRFS_I(inode)->delalloc_bytes -= len;
- if (BTRFS_I(inode)->delalloc_bytes == 0 &&
+ if (do_list && BTRFS_I(inode)->delalloc_bytes == 0 &&
!list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
list_del_init(&BTRFS_I(inode)->delalloc_inodes);
}
skip_sum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
- ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
+ if (root == root->fs_info->tree_root)
+ ret = btrfs_bio_wq_end_io(root->fs_info, bio, 2);
+ else
+ ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
BUG_ON(ret);
- if (!(rw & (1 << BIO_RW))) {
+ if (!(rw & REQ_WRITE)) {
if (bio_flags & EXTENT_BIO_COMPRESSED) {
return btrfs_submit_compressed_read(inode, bio,
mirror_num, bio_flags);
struct extent_state *cached_state = NULL;
int compressed = 0;
int ret;
+ bool nolock = false;
ret = btrfs_dec_test_ordered_pending(inode, &ordered_extent, start,
end - start + 1);
return 0;
BUG_ON(!ordered_extent);
+ nolock = (root == root->fs_info->tree_root);
+
if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) {
BUG_ON(!list_empty(&ordered_extent->list));
ret = btrfs_ordered_update_i_size(inode, 0, ordered_extent);
if (!ret) {
- trans = btrfs_join_transaction(root, 1);
+ if (nolock)
+ trans = btrfs_join_transaction_nolock(root, 1);
+ else
+ trans = btrfs_join_transaction(root, 1);
+ BUG_ON(!trans);
btrfs_set_trans_block_group(trans, inode);
trans->block_rsv = &root->fs_info->delalloc_block_rsv;
ret = btrfs_update_inode(trans, root, inode);
ordered_extent->file_offset + ordered_extent->len - 1,
0, &cached_state, GFP_NOFS);
- trans = btrfs_join_transaction(root, 1);
+ if (nolock)
+ trans = btrfs_join_transaction_nolock(root, 1);
+ else
+ trans = btrfs_join_transaction(root, 1);
btrfs_set_trans_block_group(trans, inode);
trans->block_rsv = &root->fs_info->delalloc_block_rsv;
ordered_extent->len);
BUG_ON(ret);
} else {
+ BUG_ON(root == root->fs_info->tree_root);
ret = insert_reserved_file_extent(trans, inode,
ordered_extent->file_offset,
ordered_extent->start,
ret = btrfs_update_inode(trans, root, inode);
BUG_ON(ret);
out:
- btrfs_delalloc_release_metadata(inode, ordered_extent->len);
- if (trans)
- btrfs_end_transaction(trans, root);
+ if (nolock) {
+ if (trans)
+ btrfs_end_transaction_nolock(trans, root);
+ } else {
+ btrfs_delalloc_release_metadata(inode, ordered_extent->len);
+ if (trans)
+ btrfs_end_transaction(trans, root);
+ }
+
/* once for us */
btrfs_put_ordered_extent(ordered_extent);
/* once for the tree */
bio->bi_size = 0;
bio_add_page(bio, page, failrec->len, start - page_offset(page));
- if (failed_bio->bi_rw & (1 << BIO_RW))
+ if (failed_bio->bi_rw & REQ_WRITE)
rw = WRITE;
else
rw = READ;
dir->i_mtime = dir->i_ctime = CURRENT_TIME;
ret = btrfs_update_inode(trans, root, dir);
BUG_ON(ret);
- dir->i_sb->s_dirt = 1;
btrfs_free_path(path);
return 0;
BUG_ON(new_size > 0 && min_type != BTRFS_EXTENT_DATA_KEY);
- if (root->ref_cows)
+ if (root->ref_cows || root == root->fs_info->tree_root)
btrfs_drop_extent_cache(inode, new_size & (~mask), (u64)-1, 0);
path = btrfs_alloc_path();
} else {
break;
}
- if (found_extent && root->ref_cows) {
+ if (found_extent && (root->ref_cows ||
+ root == root->fs_info->tree_root)) {
btrfs_set_path_blocking(path);
ret = btrfs_free_extent(trans, root, extent_start,
extent_num_bytes, 0,
if (err)
return err;
}
- attr->ia_valid &= ~ATTR_SIZE;
- if (attr->ia_valid)
- err = inode_setattr(inode, attr);
+ if (attr->ia_valid) {
+ setattr_copy(inode, attr);
+ mark_inode_dirty(inode);
+
+ if (attr->ia_valid & ATTR_MODE)
+ err = btrfs_acl_chmod(inode);
+ }
- if (!err && ((attr->ia_valid & ATTR_MODE)))
- err = btrfs_acl_chmod(inode);
return err;
}
-void btrfs_delete_inode(struct inode *inode)
+void btrfs_evict_inode(struct inode *inode)
{
struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(inode)->root;
int ret;
truncate_inode_pages(&inode->i_data, 0);
+ if (inode->i_nlink && (btrfs_root_refs(&root->root_item) != 0 ||
+ root == root->fs_info->tree_root))
+ goto no_delete;
+
if (is_bad_inode(inode)) {
btrfs_orphan_del(NULL, inode);
goto no_delete;
}
+ /* do we really want it for ->i_nlink > 0 and zero btrfs_root_refs? */
btrfs_wait_ordered_range(inode, 0, (u64)-1);
if (root->fs_info->log_root_recovering) {
btrfs_end_transaction(trans, root);
btrfs_btree_balance_dirty(root, nr);
no_delete:
- clear_inode(inode);
+ end_writeback(inode);
return;
}
p = &parent->rb_right;
else {
WARN_ON(!(entry->vfs_inode.i_state &
- (I_WILL_FREE | I_FREEING | I_CLEAR)));
+ (I_WILL_FREE | I_FREEING)));
rb_erase(parent, &root->inode_tree);
RB_CLEAR_NODE(parent);
spin_unlock(&root->inode_lock);
}
spin_unlock(&root->inode_lock);
- if (empty && btrfs_root_refs(&root->root_item) == 0) {
+ /*
+ * Free space cache has inodes in the tree root, but the tree root has a
+ * root_refs of 0, so this could end up dropping the tree root as a
+ * snapshot, so we need the extra !root->fs_info->tree_root check to
+ * make sure we don't drop it.
+ */
+ if (empty && btrfs_root_refs(&root->root_item) == 0 &&
+ root != root->fs_info->tree_root) {
synchronize_srcu(&root->fs_info->subvol_srcu);
spin_lock(&root->inode_lock);
empty = RB_EMPTY_ROOT(&root->inode_tree);
if (atomic_read(&inode->i_count) > 1)
d_prune_aliases(inode);
/*
- * btrfs_drop_inode will remove it from
+ * btrfs_drop_inode will have it removed from
* the inode cache when its usage count
* hits zero.
*/
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_trans_handle *trans;
int ret = 0;
+ bool nolock = false;
if (BTRFS_I(inode)->dummy_inode)
return 0;
+ smp_mb();
+ nolock = (root->fs_info->closing && root == root->fs_info->tree_root);
+
if (wbc->sync_mode == WB_SYNC_ALL) {
- trans = btrfs_join_transaction(root, 1);
+ if (nolock)
+ trans = btrfs_join_transaction_nolock(root, 1);
+ else
+ trans = btrfs_join_transaction(root, 1);
btrfs_set_trans_block_group(trans, inode);
- ret = btrfs_commit_transaction(trans, root);
+ if (nolock)
+ ret = btrfs_end_transaction_nolock(trans, root);
+ else
+ ret = btrfs_commit_transaction(trans, root);
}
return ret;
}
struct bio_vec *bvec = bio->bi_io_vec;
u64 start;
int skip_sum;
- int write = rw & (1 << BIO_RW);
+ int write = rw & REQ_WRITE;
int ret = 0;
skip_sum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM;
spin_unlock(&root->fs_info->ordered_extent_lock);
}
+ if (root == root->fs_info->tree_root) {
+ struct btrfs_block_group_cache *block_group;
+
+ block_group = btrfs_lookup_block_group(root->fs_info,
+ BTRFS_I(inode)->block_group);
+ if (block_group && block_group->inode == inode) {
+ spin_lock(&block_group->lock);
+ block_group->inode = NULL;
+ spin_unlock(&block_group->lock);
+ btrfs_put_block_group(block_group);
+ } else if (block_group) {
+ btrfs_put_block_group(block_group);
+ }
+ }
+
spin_lock(&root->orphan_lock);
if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
printk(KERN_INFO "BTRFS: inode %lu still on the orphan list\n",
kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode));
}
-void btrfs_drop_inode(struct inode *inode)
+int btrfs_drop_inode(struct inode *inode)
{
struct btrfs_root *root = BTRFS_I(inode)->root;
- if (inode->i_nlink > 0 && btrfs_root_refs(&root->root_item) == 0)
- generic_delete_inode(inode);
+
+ if (btrfs_root_refs(&root->root_item) == 0 &&
+ root != root->fs_info->tree_root)
+ return 1;
else
- generic_drop_inode(inode);
+ return generic_drop_inode(inode);
}
static void init_once(void *foo)
return 0;
}
- int btrfs_start_one_delalloc_inode(struct btrfs_root *root, int delay_iput)
+ int btrfs_start_one_delalloc_inode(struct btrfs_root *root, int delay_iput,
+ int sync)
{
struct btrfs_inode *binode;
struct inode *inode = NULL;
spin_unlock(&root->fs_info->delalloc_lock);
if (inode) {
- write_inode_now(inode, 0);
+ if (sync) {
+ filemap_write_and_wait(inode->i_mapping);
+ /*
+ * We have to do this because compression doesn't
+ * actually set PG_writeback until it submits the pages
+ * for IO, which happens in an async thread, so we could
+ * race and not actually wait for any writeback pages
+ * because they've not been submitted yet. Technically
+ * this could still be the case for the ordered stuff
+ * since the async thread may not have started to do its
+ * work yet. If this becomes the case then we need to
+ * figure out a way to make sure that in writepage we
+ * wait for any async pages to be submitted before
+ * returning so that fdatawait does what its supposed to
+ * do.
+ */
+ btrfs_wait_ordered_range(inode, 0, (u64)-1);
+ } else {
+ filemap_flush(inode->i_mapping);
+ }
if (delay_iput)
btrfs_add_delayed_iput(inode);
else
return err;
}
-int btrfs_prealloc_file_range(struct inode *inode, int mode,
- u64 start, u64 num_bytes, u64 min_size,
- loff_t actual_len, u64 *alloc_hint)
+static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
+ u64 start, u64 num_bytes, u64 min_size,
+ loff_t actual_len, u64 *alloc_hint,
+ struct btrfs_trans_handle *trans)
{
- struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_key ins;
u64 cur_offset = start;
int ret = 0;
+ bool own_trans = true;
+ if (trans)
+ own_trans = false;
while (num_bytes > 0) {
- trans = btrfs_start_transaction(root, 3);
- if (IS_ERR(trans)) {
- ret = PTR_ERR(trans);
- break;
+ if (own_trans) {
+ trans = btrfs_start_transaction(root, 3);
+ if (IS_ERR(trans)) {
+ ret = PTR_ERR(trans);
+ break;
+ }
}
ret = btrfs_reserve_extent(trans, root, num_bytes, min_size,
0, *alloc_hint, (u64)-1, &ins, 1);
if (ret) {
- btrfs_end_transaction(trans, root);
+ if (own_trans)
+ btrfs_end_transaction(trans, root);
break;
}
ret = btrfs_update_inode(trans, root, inode);
BUG_ON(ret);
- btrfs_end_transaction(trans, root);
+ if (own_trans)
+ btrfs_end_transaction(trans, root);
}
return ret;
}
+int btrfs_prealloc_file_range(struct inode *inode, int mode,
+ u64 start, u64 num_bytes, u64 min_size,
+ loff_t actual_len, u64 *alloc_hint)
+{
+ return __btrfs_prealloc_file_range(inode, mode, start, num_bytes,
+ min_size, actual_len, alloc_hint,
+ NULL);
+}
+
+int btrfs_prealloc_file_range_trans(struct inode *inode,
+ struct btrfs_trans_handle *trans, int mode,
+ u64 start, u64 num_bytes, u64 min_size,
+ loff_t actual_len, u64 *alloc_hint)
+{
+ return __btrfs_prealloc_file_range(inode, mode, start, num_bytes,
+ min_size, actual_len, alloc_hint, trans);
+}
+
static long btrfs_fallocate(struct inode *inode, int mode,
loff_t offset, loff_t len)
{
#include "locking.h"
#include "btrfs_inode.h"
#include "async-thread.h"
+#include "free-space-cache.h"
/*
* backref_node, mapping_node and tree_block start with this
u64 search_start;
u64 extents_found;
- int block_rsv_retries;
-
unsigned int stage:8;
unsigned int create_reloc_tree:1;
unsigned int merge_reloc_tree:1;
LIST_HEAD(reloc_roots);
u64 num_bytes = 0;
int ret;
- int retries = 0;
mutex_lock(&root->fs_info->trans_mutex);
rc->merging_rsv_size += root->nodesize * (BTRFS_MAX_LEVEL - 1) * 2;
if (!err) {
num_bytes = rc->merging_rsv_size;
ret = btrfs_block_rsv_add(NULL, root, rc->block_rsv,
- num_bytes, &retries);
+ num_bytes);
if (ret)
err = ret;
}
btrfs_end_transaction(trans, rc->extent_root);
btrfs_block_rsv_release(rc->extent_root,
rc->block_rsv, num_bytes);
- retries = 0;
goto again;
}
}
num_bytes = calcu_metadata_size(rc, node, 1) * 2;
trans->block_rsv = rc->block_rsv;
- ret = btrfs_block_rsv_add(trans, root, rc->block_rsv, num_bytes,
- &rc->block_rsv_retries);
+ ret = btrfs_block_rsv_add(trans, root, rc->block_rsv, num_bytes);
if (ret) {
if (ret == -EAGAIN)
rc->commit_transaction = 1;
return ret;
}
- rc->block_rsv_retries = 0;
return 0;
}
return ret;
}
+static int delete_block_group_cache(struct btrfs_fs_info *fs_info,
+ struct inode *inode, u64 ino)
+{
+ struct btrfs_key key;
+ struct btrfs_path *path;
+ struct btrfs_root *root = fs_info->tree_root;
+ struct btrfs_trans_handle *trans;
+ unsigned long nr;
+ int ret = 0;
+
+ if (inode)
+ goto truncate;
+
+ key.objectid = ino;
+ key.type = BTRFS_INODE_ITEM_KEY;
+ key.offset = 0;
+
+ inode = btrfs_iget(fs_info->sb, &key, root, NULL);
+ if (!inode || IS_ERR(inode) || is_bad_inode(inode)) {
+ if (inode && !IS_ERR(inode))
+ iput(inode);
+ return -ENOENT;
+ }
+
+truncate:
+ path = btrfs_alloc_path();
+ if (!path) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ trans = btrfs_join_transaction(root, 0);
+ if (IS_ERR(trans)) {
+ btrfs_free_path(path);
+ goto out;
+ }
+
+ ret = btrfs_truncate_free_space_cache(root, trans, path, inode);
+
+ btrfs_free_path(path);
+ nr = trans->blocks_used;
+ btrfs_end_transaction(trans, root);
+ btrfs_btree_balance_dirty(root, nr);
+out:
+ iput(inode);
+ return ret;
+}
+
/*
* helper to add tree blocks for backref of type BTRFS_EXTENT_DATA_REF_KEY
* this function scans fs tree to find blocks reference the data extent
int counted;
int ret;
- path = btrfs_alloc_path();
- if (!path)
- return -ENOMEM;
-
ref_root = btrfs_extent_data_ref_root(leaf, ref);
ref_objectid = btrfs_extent_data_ref_objectid(leaf, ref);
ref_offset = btrfs_extent_data_ref_offset(leaf, ref);
ref_count = btrfs_extent_data_ref_count(leaf, ref);
+ /*
+ * This is an extent belonging to the free space cache, lets just delete
+ * it and redo the search.
+ */
+ if (ref_root == BTRFS_ROOT_TREE_OBJECTID) {
+ ret = delete_block_group_cache(rc->extent_root->fs_info,
+ NULL, ref_objectid);
+ if (ret != -ENOENT)
+ return ret;
+ ret = 0;
+ }
+
+ path = btrfs_alloc_path();
+ if (!path)
+ return -ENOMEM;
+
root = read_fs_root(rc->extent_root->fs_info, ref_root);
if (IS_ERR(root)) {
err = PTR_ERR(root);
* is no reservation in transaction handle.
*/
ret = btrfs_block_rsv_add(NULL, rc->extent_root, rc->block_rsv,
- rc->extent_root->nodesize * 256,
- &rc->block_rsv_retries);
+ rc->extent_root->nodesize * 256);
if (ret)
return ret;
rc->extents_found = 0;
rc->nodes_relocated = 0;
rc->merging_rsv_size = 0;
- rc->block_rsv_retries = 0;
rc->create_reloc_tree = 1;
set_reloc_control(rc);
{
struct btrfs_fs_info *fs_info = extent_root->fs_info;
struct reloc_control *rc;
+ struct inode *inode;
+ struct btrfs_path *path;
int ret;
int rw = 0;
int err = 0;
rw = 1;
}
+ path = btrfs_alloc_path();
+ if (!path) {
+ err = -ENOMEM;
+ goto out;
+ }
+
+ inode = lookup_free_space_inode(fs_info->tree_root, rc->block_group,
+ path);
+ btrfs_free_path(path);
+
+ if (!IS_ERR(inode))
+ ret = delete_block_group_cache(fs_info, inode, 0);
+ else
+ ret = PTR_ERR(inode);
+
+ if (ret && ret != -ENOENT) {
+ err = ret;
+ goto out;
+ }
+
rc->data_inode = create_reloc_inode(fs_info, rc->block_group);
if (IS_ERR(rc->data_inode)) {
err = PTR_ERR(rc->data_inode);
Opt_nodatacow, Opt_max_inline, Opt_alloc_start, Opt_nobarrier, Opt_ssd,
Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl, Opt_compress,
Opt_compress_force, Opt_notreelog, Opt_ratio, Opt_flushoncommit,
- Opt_discard, Opt_err,
+ Opt_discard, Opt_space_cache, Opt_clear_cache, Opt_err,
};
static match_table_t tokens = {
{Opt_flushoncommit, "flushoncommit"},
{Opt_ratio, "metadata_ratio=%d"},
{Opt_discard, "discard"},
+ {Opt_space_cache, "space_cache"},
+ {Opt_clear_cache, "clear_cache"},
{Opt_err, NULL},
};
case Opt_discard:
btrfs_set_opt(info->mount_opt, DISCARD);
break;
+ case Opt_space_cache:
+ printk(KERN_INFO "btrfs: enabling disk space caching\n");
+ btrfs_set_opt(info->mount_opt, SPACE_CACHE);
+ case Opt_clear_cache:
+ printk(KERN_INFO "btrfs: force clearing of disk cache\n");
+ btrfs_set_opt(info->mount_opt, CLEAR_CACHE);
+ break;
case Opt_err:
printk(KERN_INFO "btrfs: unrecognized mount option "
"'%s'\n", p);
if (IS_ERR(root)) {
error = PTR_ERR(root);
deactivate_locked_super(s);
- goto error;
+ goto error_free_subvol_name;
}
/* if they gave us a subvolume name bind mount into that */
if (strcmp(subvol_name, ".")) {
deactivate_locked_super(s);
error = PTR_ERR(new_root);
dput(root);
- goto error_close_devices;
+ goto error_free_subvol_name;
}
if (!new_root->d_inode) {
dput(root);
dput(new_root);
deactivate_locked_super(s);
error = -ENXIO;
- goto error_close_devices;
+ goto error_free_subvol_name;
}
dput(root);
root = new_root;
btrfs_close_devices(fs_devices);
error_free_subvol_name:
kfree(subvol_name);
- error:
return error;
}
struct list_head *head = &root->fs_info->space_info;
struct btrfs_space_info *found;
u64 total_used = 0;
+ u64 total_used_data = 0;
int bits = dentry->d_sb->s_blocksize_bits;
__be32 *fsid = (__be32 *)root->fs_info->fsid;
rcu_read_lock();
- list_for_each_entry_rcu(found, head, list)
+ list_for_each_entry_rcu(found, head, list) {
+ if (found->flags & (BTRFS_BLOCK_GROUP_METADATA |
+ BTRFS_BLOCK_GROUP_SYSTEM))
+ total_used_data += found->disk_total;
+ else
+ total_used_data += found->disk_used;
total_used += found->disk_used;
+ }
rcu_read_unlock();
buf->f_namelen = BTRFS_NAME_LEN;
buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits;
buf->f_bfree = buf->f_blocks - (total_used >> bits);
- buf->f_bavail = buf->f_bfree;
+ buf->f_bavail = buf->f_blocks - (total_used_data >> bits);
buf->f_bsize = dentry->d_sb->s_blocksize;
buf->f_type = BTRFS_SUPER_MAGIC;
static const struct super_operations btrfs_super_ops = {
.drop_inode = btrfs_drop_inode,
- .delete_inode = btrfs_delete_inode,
+ .evict_inode = btrfs_evict_inode,
.put_super = btrfs_put_super,
.sync_fs = btrfs_sync_fs,
.show_options = btrfs_show_options,
TRANS_START,
TRANS_JOIN,
TRANS_USERSPACE,
+ TRANS_JOIN_NOLOCK,
};
static int may_wait_transaction(struct btrfs_root *root, int type)
{
struct btrfs_trans_handle *h;
struct btrfs_transaction *cur_trans;
- int retries = 0;
int ret;
again:
h = kmem_cache_alloc(btrfs_trans_handle_cachep, GFP_NOFS);
if (!h)
return ERR_PTR(-ENOMEM);
- mutex_lock(&root->fs_info->trans_mutex);
+ if (type != TRANS_JOIN_NOLOCK)
+ mutex_lock(&root->fs_info->trans_mutex);
if (may_wait_transaction(root, type))
wait_current_trans(root);
cur_trans = root->fs_info->running_transaction;
cur_trans->use_count++;
- mutex_unlock(&root->fs_info->trans_mutex);
+ if (type != TRANS_JOIN_NOLOCK)
+ mutex_unlock(&root->fs_info->trans_mutex);
h->transid = cur_trans->transid;
h->transaction = cur_trans;
}
if (num_items > 0) {
- ret = btrfs_trans_reserve_metadata(h, root, num_items,
- &retries);
+ ret = btrfs_trans_reserve_metadata(h, root, num_items);
if (ret == -EAGAIN) {
btrfs_commit_transaction(h, root);
goto again;
}
}
- mutex_lock(&root->fs_info->trans_mutex);
+ if (type != TRANS_JOIN_NOLOCK)
+ mutex_lock(&root->fs_info->trans_mutex);
record_root_in_trans(h, root);
- mutex_unlock(&root->fs_info->trans_mutex);
+ if (type != TRANS_JOIN_NOLOCK)
+ mutex_unlock(&root->fs_info->trans_mutex);
if (!current->journal_info && type != TRANS_USERSPACE)
current->journal_info = h;
return start_transaction(root, 0, TRANS_JOIN);
}
+struct btrfs_trans_handle *btrfs_join_transaction_nolock(struct btrfs_root *root,
+ int num_blocks)
+{
+ return start_transaction(root, 0, TRANS_JOIN_NOLOCK);
+}
+
struct btrfs_trans_handle *btrfs_start_ioctl_transaction(struct btrfs_root *r,
int num_blocks)
{
}
static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
- struct btrfs_root *root, int throttle)
+ struct btrfs_root *root, int throttle, int lock)
{
struct btrfs_transaction *cur_trans = trans->transaction;
struct btrfs_fs_info *info = root->fs_info;
btrfs_trans_release_metadata(trans, root);
- if (!root->fs_info->open_ioctl_trans &&
+ if (lock && !root->fs_info->open_ioctl_trans &&
should_end_transaction(trans, root))
trans->transaction->blocked = 1;
- if (cur_trans->blocked && !cur_trans->in_commit) {
+ if (lock && cur_trans->blocked && !cur_trans->in_commit) {
if (throttle)
return btrfs_commit_transaction(trans, root);
else
wake_up_process(info->transaction_kthread);
}
- mutex_lock(&info->trans_mutex);
+ if (lock)
+ mutex_lock(&info->trans_mutex);
WARN_ON(cur_trans != info->running_transaction);
WARN_ON(cur_trans->num_writers < 1);
cur_trans->num_writers--;
if (waitqueue_active(&cur_trans->writer_wait))
wake_up(&cur_trans->writer_wait);
put_transaction(cur_trans);
- mutex_unlock(&info->trans_mutex);
+ if (lock)
+ mutex_unlock(&info->trans_mutex);
if (current->journal_info == trans)
current->journal_info = NULL;
int btrfs_end_transaction(struct btrfs_trans_handle *trans,
struct btrfs_root *root)
{
- return __btrfs_end_transaction(trans, root, 0);
+ return __btrfs_end_transaction(trans, root, 0, 1);
}
int btrfs_end_transaction_throttle(struct btrfs_trans_handle *trans,
struct btrfs_root *root)
{
- return __btrfs_end_transaction(trans, root, 1);
+ return __btrfs_end_transaction(trans, root, 1, 1);
+}
+
+int btrfs_end_transaction_nolock(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root)
+{
+ return __btrfs_end_transaction(trans, root, 0, 0);
}
/*
struct extent_buffer *tmp;
struct extent_buffer *old;
int ret;
- int retries = 0;
u64 to_reserve = 0;
u64 index = 0;
u64 objectid;
if (to_reserve > 0) {
ret = btrfs_block_rsv_add(trans, root, &pending->block_rsv,
- to_reserve, &retries);
+ to_reserve);
if (ret) {
pending->error = ret;
goto fail;
super->root = root_item->bytenr;
super->generation = root_item->generation;
super->root_level = root_item->level;
+ if (super->cache_generation != 0 || btrfs_test_opt(root, SPACE_CACHE))
+ super->cache_generation = root_item->generation;
}
int btrfs_transaction_in_commit(struct btrfs_fs_info *info)