aboutsummaryrefslogtreecommitdiff
path: root/fs/btrfs
diff options
context:
space:
mode:
Diffstat (limited to 'fs/btrfs')
-rw-r--r--fs/btrfs/ctree.h4
-rw-r--r--fs/btrfs/disk-io.c2
-rw-r--r--fs/btrfs/extent-tree.c30
-rw-r--r--fs/btrfs/free-space-cache.c73
-rw-r--r--fs/btrfs/inode.c3
-rw-r--r--fs/btrfs/relocation.c9
-rw-r--r--fs/btrfs/transaction.c22
-rw-r--r--fs/btrfs/transaction.h1
-rw-r--r--fs/btrfs/zlib.c6
9 files changed, 119 insertions, 31 deletions
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 215ef8cae82..837435ce84c 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -825,6 +825,7 @@ struct btrfs_fs_info {
struct mutex drop_mutex;
struct mutex volume_mutex;
struct mutex tree_reloc_mutex;
+ struct rw_semaphore extent_commit_sem;
/*
* this protects the ordered operations list only while we are
@@ -959,9 +960,6 @@ struct btrfs_root {
/* the node lock is held while changing the node pointer */
spinlock_t node_lock;
- /* taken when updating the commit root */
- struct rw_semaphore commit_root_sem;
-
struct extent_buffer *commit_root;
struct btrfs_root *log_root;
struct btrfs_root *reloc_root;
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 7dcaa813886..e83be2e4602 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -909,7 +909,6 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
spin_lock_init(&root->inode_lock);
mutex_init(&root->objectid_mutex);
mutex_init(&root->log_mutex);
- init_rwsem(&root->commit_root_sem);
init_waitqueue_head(&root->log_writer_wait);
init_waitqueue_head(&root->log_commit_wait[0]);
init_waitqueue_head(&root->log_commit_wait[1]);
@@ -1640,6 +1639,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
mutex_init(&fs_info->cleaner_mutex);
mutex_init(&fs_info->volume_mutex);
mutex_init(&fs_info->tree_reloc_mutex);
+ init_rwsem(&fs_info->extent_commit_sem);
btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
btrfs_init_free_cluster(&fs_info->data_alloc_cluster);
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index fadf69a2764..72a2b9c28e9 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -265,10 +265,6 @@ static int caching_kthread(void *data)
atomic_inc(&block_group->space_info->caching_threads);
last = max_t(u64, block_group->key.objectid, BTRFS_SUPER_INFO_OFFSET);
-again:
- /* need to make sure the commit_root doesn't disappear */
- down_read(&fs_info->extent_root->commit_root_sem);
-
/*
* We don't want to deadlock with somebody trying to allocate a new
* extent for the extent root while also trying to search the extent
@@ -282,6 +278,10 @@ again:
key.objectid = last;
key.offset = 0;
btrfs_set_key_type(&key, BTRFS_EXTENT_ITEM_KEY);
+again:
+ /* need to make sure the commit_root doesn't disappear */
+ down_read(&fs_info->extent_commit_sem);
+
ret = btrfs_search_slot(NULL, fs_info->extent_root, &key, path, 0, 0);
if (ret < 0)
goto err;
@@ -302,10 +302,24 @@ again:
else if (ret)
break;
- if (need_resched()) {
+ if (need_resched() ||
+ btrfs_transaction_in_commit(fs_info)) {
+ leaf = path->nodes[0];
+
+ /* this shouldn't happen, but if the
+ * leaf is empty just move on.
+ */
+ if (btrfs_header_nritems(leaf) == 0)
+ break;
+ /*
+ * we need to copy the key out so that
+ * we are sure the next search advances
+ * us forward in the btree.
+ */
+ btrfs_item_key_to_cpu(leaf, &key, 0);
btrfs_release_path(fs_info->extent_root, path);
- up_read(&fs_info->extent_root->commit_root_sem);
- cond_resched();
+ up_read(&fs_info->extent_commit_sem);
+ schedule_timeout(1);
goto again;
}
@@ -345,7 +359,7 @@ next:
err:
btrfs_free_path(path);
- up_read(&fs_info->extent_root->commit_root_sem);
+ up_read(&fs_info->extent_commit_sem);
atomic_dec(&block_group->space_info->caching_threads);
wake_up(&block_group->caching_q);
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index af99b78b288..5edcee3a617 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -414,11 +414,29 @@ static noinline int remove_from_bitmap(struct btrfs_block_group_cache *block_gro
u64 *offset, u64 *bytes)
{
u64 end;
+ u64 search_start, search_bytes;
+ int ret;
again:
end = bitmap_info->offset +
(u64)(BITS_PER_BITMAP * block_group->sectorsize) - 1;
+ /*
+ * XXX - this can go away after a few releases.
+ *
+ * since the only user of btrfs_remove_free_space is the tree logging
+ * stuff, and the only way to test that is under crash conditions, we
+ * want to have this debug stuff here just in case somethings not
+ * working. Search the bitmap for the space we are trying to use to
+ * make sure its actually there. If its not there then we need to stop
+ * because something has gone wrong.
+ */
+ search_start = *offset;
+ search_bytes = *bytes;
+ ret = search_bitmap(block_group, bitmap_info, &search_start,
+ &search_bytes);
+ BUG_ON(ret < 0 || search_start != *offset);
+
if (*offset > bitmap_info->offset && *offset + *bytes > end) {
bitmap_clear_bits(block_group, bitmap_info, *offset,
end - *offset + 1);
@@ -430,6 +448,7 @@ again:
}
if (*bytes) {
+ struct rb_node *next = rb_next(&bitmap_info->offset_index);
if (!bitmap_info->bytes) {
unlink_free_space(block_group, bitmap_info);
kfree(bitmap_info->bitmap);
@@ -438,16 +457,36 @@ again:
recalculate_thresholds(block_group);
}
- bitmap_info = tree_search_offset(block_group,
- offset_to_bitmap(block_group,
- *offset),
- 1, 0);
- if (!bitmap_info)
+ /*
+ * no entry after this bitmap, but we still have bytes to
+ * remove, so something has gone wrong.
+ */
+ if (!next)
return -EINVAL;
+ bitmap_info = rb_entry(next, struct btrfs_free_space,
+ offset_index);
+
+ /*
+ * if the next entry isn't a bitmap we need to return to let the
+ * extent stuff do its work.
+ */
if (!bitmap_info->bitmap)
return -EAGAIN;
+ /*
+ * Ok the next item is a bitmap, but it may not actually hold
+ * the information for the rest of this free space stuff, so
+ * look for it, and if we don't find it return so we can try
+ * everything over again.
+ */
+ search_start = *offset;
+ search_bytes = *bytes;
+ ret = search_bitmap(block_group, bitmap_info, &search_start,
+ &search_bytes);
+ if (ret < 0 || search_start != *offset)
+ return -EAGAIN;
+
goto again;
} else if (!bitmap_info->bytes) {
unlink_free_space(block_group, bitmap_info);
@@ -644,8 +683,17 @@ int btrfs_remove_free_space(struct btrfs_block_group_cache *block_group,
again:
info = tree_search_offset(block_group, offset, 0, 0);
if (!info) {
- WARN_ON(1);
- goto out_lock;
+ /*
+ * oops didn't find an extent that matched the space we wanted
+ * to remove, look for a bitmap instead
+ */
+ info = tree_search_offset(block_group,
+ offset_to_bitmap(block_group, offset),
+ 1, 0);
+ if (!info) {
+ WARN_ON(1);
+ goto out_lock;
+ }
}
if (info->bytes < bytes && rb_next(&info->offset_index)) {
@@ -957,8 +1005,15 @@ static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group_cache *block_group,
if (cluster->block_group != block_group)
goto out;
- entry = tree_search_offset(block_group, search_start, 0, 0);
-
+ /*
+ * search_start is the beginning of the bitmap, but at some point it may
+ * be a good idea to point to the actual start of the free area in the
+ * bitmap, so do the offset_to_bitmap trick anyway, and set bitmap_only
+ * to 1 to make sure we get the bitmap entry
+ */
+ entry = tree_search_offset(block_group,
+ offset_to_bitmap(block_group, search_start),
+ 1, 0);
if (!entry || !entry->bitmap)
goto out;
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 56fe83fa60c..272b9b2bea8 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -4785,8 +4785,7 @@ static int btrfs_rename(struct inode *old_dir, struct dentry *old_dentry,
* and the replacement file is large. Start IO on it now so
* we don't add too much work to the end of the transaction
*/
- if (new_inode && old_inode && S_ISREG(old_inode->i_mode) &&
- new_inode->i_size &&
+ if (new_inode && S_ISREG(old_inode->i_mode) && new_inode->i_size &&
old_inode->i_size > BTRFS_ORDERED_OPERATIONS_FLUSH_LIMIT)
filemap_flush(old_inode->i_mapping);
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index e71264d1c2c..c04f7f21260 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -2553,8 +2553,13 @@ int relocate_inode_pages(struct inode *inode, u64 start, u64 len)
last_index = (start + len - 1) >> PAGE_CACHE_SHIFT;
/* make sure the dirty trick played by the caller work */
- ret = invalidate_inode_pages2_range(inode->i_mapping,
- first_index, last_index);
+ while (1) {
+ ret = invalidate_inode_pages2_range(inode->i_mapping,
+ first_index, last_index);
+ if (ret != -EBUSY)
+ break;
+ schedule_timeout(HZ/10);
+ }
if (ret)
goto out_unlock;
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index e51d2bc532f..cdbb5022da5 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -42,10 +42,8 @@ static noinline void put_transaction(struct btrfs_transaction *transaction)
static noinline void switch_commit_root(struct btrfs_root *root)
{
- down_write(&root->commit_root_sem);
free_extent_buffer(root->commit_root);
root->commit_root = btrfs_root_node(root);
- up_write(&root->commit_root_sem);
}
/*
@@ -466,7 +464,10 @@ static int update_cowonly_root(struct btrfs_trans_handle *trans,
ret = btrfs_write_dirty_block_groups(trans, root);
BUG_ON(ret);
}
- switch_commit_root(root);
+
+ if (root != root->fs_info->extent_root)
+ switch_commit_root(root);
+
return 0;
}
@@ -499,6 +500,11 @@ static noinline int commit_cowonly_roots(struct btrfs_trans_handle *trans,
update_cowonly_root(trans, root);
}
+
+ down_write(&fs_info->extent_commit_sem);
+ switch_commit_root(fs_info->extent_root);
+ up_write(&fs_info->extent_commit_sem);
+
return 0;
}
@@ -851,6 +857,16 @@ static void update_super_roots(struct btrfs_root *root)
super->root_level = root_item->level;
}
+int btrfs_transaction_in_commit(struct btrfs_fs_info *info)
+{
+ int ret = 0;
+ spin_lock(&info->new_trans_lock);
+ if (info->running_transaction)
+ ret = info->running_transaction->in_commit;
+ spin_unlock(&info->new_trans_lock);
+ return ret;
+}
+
int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
struct btrfs_root *root)
{
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index 961c3ee5a2e..663c6740491 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -107,4 +107,5 @@ int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
struct btrfs_root *root);
int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
struct extent_io_tree *dirty_pages);
+int btrfs_transaction_in_commit(struct btrfs_fs_info *info);
#endif
diff --git a/fs/btrfs/zlib.c b/fs/btrfs/zlib.c
index ecfbce836d3..3e2b90eaa23 100644
--- a/fs/btrfs/zlib.c
+++ b/fs/btrfs/zlib.c
@@ -208,7 +208,7 @@ int btrfs_zlib_compress_pages(struct address_space *mapping,
*total_in = 0;
workspace = find_zlib_workspace();
- if (!workspace)
+ if (IS_ERR(workspace))
return -1;
if (Z_OK != zlib_deflateInit(&workspace->def_strm, 3)) {
@@ -366,7 +366,7 @@ int btrfs_zlib_decompress_biovec(struct page **pages_in,
char *kaddr;
workspace = find_zlib_workspace();
- if (!workspace)
+ if (IS_ERR(workspace))
return -ENOMEM;
data_in = kmap(pages_in[page_in_index]);
@@ -547,7 +547,7 @@ int btrfs_zlib_decompress(unsigned char *data_in,
return -ENOMEM;
workspace = find_zlib_workspace();
- if (!workspace)
+ if (IS_ERR(workspace))
return -ENOMEM;
workspace->inf_strm.next_in = data_in;