Merge branch 'for-linus-4.4' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

Pull btrfs updates from Chris Mason:
 "We have a lot of subvolume quota improvements in here, along with big
  piles of cleanups from Dave Sterba and Anand Jain and others.

  Josef pitched in a batch of allocator fixes based on production use
  here at FB.  We found that mount -o ssd_spread greatly improved our
  performance on hardware raid5/6, but it exposed some CPU bottlenecks
  in the allocator.  These patches make a huge difference"

* 'for-linus-4.4' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (100 commits)
  Btrfs: fix hole punching when using the no-holes feature
  Btrfs: find_free_extent: Do not erroneously skip LOOP_CACHING_WAIT state
  btrfs: Fix a data space underflow warning
  btrfs: qgroup: Fix a rebase bug which will cause qgroup double free
  btrfs: qgroup: Fix a race in delayed_ref which leads to abort trans
  btrfs: clear PF_NOFREEZE in cleaner_kthread()
  btrfs: qgroup: Don't copy extent buffer to do qgroup rescan
  btrfs: add balance filters limits, stripes and usage to supported mask
  btrfs: extend balance filter usage to take minimum and maximum
  btrfs: add balance filter for stripes
  btrfs: extend balance filter limit to take minimum and maximum
  btrfs: fix use after free iterating extrefs
  btrfs: check unsupported filters in balance arguments
  Btrfs: fix regression running delayed references when using qgroups
  Btrfs: fix regression when running delayed references
  Btrfs: don't do extra bitmap search in one bit case
  Btrfs: keep track of largest extent in bitmaps
  Btrfs: don't keep trying to build clusters if we are fragmented
  Btrfs: cut down on loops through the allocator
  Btrfs: don't continue setting up space cache when enospc
  ...

fs/btrfs/backref.c

@@ -362,6 +362,12 @@ static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 		goto out;
 	}
 
+	if (btrfs_test_is_dummy_root(root)) {
+		srcu_read_unlock(&fs_info->subvol_srcu, index);
+		ret = -ENOENT;
+		goto out;
+	}
+
 	if (path->search_commit_root)
 		root_level = btrfs_header_level(root->commit_root);
 	else if (time_seq == (u64)-1)

fs/btrfs/check-integrity.c

@@ -667,7 +667,7 @@ static int btrfsic_process_superblock(struct btrfsic_state *state,
 	selected_super = kzalloc(sizeof(*selected_super), GFP_NOFS);
 	if (NULL == selected_super) {
 		printk(KERN_INFO "btrfsic: error, kmalloc failed!\n");
-		return -1;
+		return -ENOMEM;
 	}
 
 	list_for_each_entry(device, dev_head, dev_list) {
@@ -845,8 +845,8 @@ static int btrfsic_process_superblock_dev_mirror(
 		superblock_tmp->never_written = 0;
 		superblock_tmp->mirror_num = 1 + superblock_mirror_num;
 		if (state->print_mask & BTRFSIC_PRINT_MASK_SUPERBLOCK_WRITE)
-			printk_in_rcu(KERN_INFO "New initial S-block (bdev %p, %s)"
-				     " @%llu (%s/%llu/%d)\n",
+			btrfs_info_in_rcu(device->dev_root->fs_info,
+				"new initial S-block (bdev %p, %s) @%llu (%s/%llu/%d)",
 				     superblock_bdev,
 				     rcu_str_deref(device->name), dev_bytenr,
 				     dev_state->name, dev_bytenr,
@@ -1660,7 +1660,7 @@ static int btrfsic_read_block(struct btrfsic_state *state,
 					  sizeof(*block_ctx->pagev)) *
 					 num_pages, GFP_NOFS);
 	if (!block_ctx->mem_to_free)
-		return -1;
+		return -ENOMEM;
 	block_ctx->datav = block_ctx->mem_to_free;
 	block_ctx->pagev = (struct page **)(block_ctx->datav + num_pages);
 	for (i = 0; i < num_pages; i++) {

fs/btrfs/compression.c

@@ -745,11 +745,13 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 	return ret;
 }
 
-static struct list_head comp_idle_workspace[BTRFS_COMPRESS_TYPES];
-static spinlock_t comp_workspace_lock[BTRFS_COMPRESS_TYPES];
-static int comp_num_workspace[BTRFS_COMPRESS_TYPES];
-static atomic_t comp_alloc_workspace[BTRFS_COMPRESS_TYPES];
-static wait_queue_head_t comp_workspace_wait[BTRFS_COMPRESS_TYPES];
+static struct {
+	struct list_head idle_ws;
+	spinlock_t ws_lock;
+	int num_ws;
+	atomic_t alloc_ws;
+	wait_queue_head_t ws_wait;
+} btrfs_comp_ws[BTRFS_COMPRESS_TYPES];
 
 static const struct btrfs_compress_op * const btrfs_compress_op[] = {
 	&btrfs_zlib_compress,
@@ -761,10 +763,10 @@ void __init btrfs_init_compress(void)
 	int i;
 
 	for (i = 0; i < BTRFS_COMPRESS_TYPES; i++) {
-		INIT_LIST_HEAD(&comp_idle_workspace[i]);
-		spin_lock_init(&comp_workspace_lock[i]);
-		atomic_set(&comp_alloc_workspace[i], 0);
-		init_waitqueue_head(&comp_workspace_wait[i]);
+		INIT_LIST_HEAD(&btrfs_comp_ws[i].idle_ws);
+		spin_lock_init(&btrfs_comp_ws[i].ws_lock);
+		atomic_set(&btrfs_comp_ws[i].alloc_ws, 0);
+		init_waitqueue_head(&btrfs_comp_ws[i].ws_wait);
 	}
 }
 
@@ -778,38 +780,38 @@ static struct list_head *find_workspace(int type)
 	int cpus = num_online_cpus();
 	int idx = type - 1;
 
-	struct list_head *idle_workspace	= &comp_idle_workspace[idx];
-	spinlock_t *workspace_lock		= &comp_workspace_lock[idx];
-	atomic_t *alloc_workspace		= &comp_alloc_workspace[idx];
-	wait_queue_head_t *workspace_wait	= &comp_workspace_wait[idx];
-	int *num_workspace			= &comp_num_workspace[idx];
+	struct list_head *idle_ws	= &btrfs_comp_ws[idx].idle_ws;
+	spinlock_t *ws_lock		= &btrfs_comp_ws[idx].ws_lock;
+	atomic_t *alloc_ws		= &btrfs_comp_ws[idx].alloc_ws;
+	wait_queue_head_t *ws_wait	= &btrfs_comp_ws[idx].ws_wait;
+	int *num_ws			= &btrfs_comp_ws[idx].num_ws;
 again:
-	spin_lock(workspace_lock);
-	if (!list_empty(idle_workspace)) {
-		workspace = idle_workspace->next;
+	spin_lock(ws_lock);
+	if (!list_empty(idle_ws)) {
+		workspace = idle_ws->next;
 		list_del(workspace);
-		(*num_workspace)--;
-		spin_unlock(workspace_lock);
+		(*num_ws)--;
+		spin_unlock(ws_lock);
 		return workspace;
 
 	}
-	if (atomic_read(alloc_workspace) > cpus) {
+	if (atomic_read(alloc_ws) > cpus) {
 		DEFINE_WAIT(wait);
 
-		spin_unlock(workspace_lock);
-		prepare_to_wait(workspace_wait, &wait, TASK_UNINTERRUPTIBLE);
-		if (atomic_read(alloc_workspace) > cpus && !*num_workspace)
+		spin_unlock(ws_lock);
+		prepare_to_wait(ws_wait, &wait, TASK_UNINTERRUPTIBLE);
+		if (atomic_read(alloc_ws) > cpus && !*num_ws)
 			schedule();
-		finish_wait(workspace_wait, &wait);
+		finish_wait(ws_wait, &wait);
 		goto again;
 	}
-	atomic_inc(alloc_workspace);
-	spin_unlock(workspace_lock);
+	atomic_inc(alloc_ws);
+	spin_unlock(ws_lock);
 
 	workspace = btrfs_compress_op[idx]->alloc_workspace();
 	if (IS_ERR(workspace)) {
-		atomic_dec(alloc_workspace);
-		wake_up(workspace_wait);
+		atomic_dec(alloc_ws);
+		wake_up(ws_wait);
 	}
 	return workspace;
 }
@@ -821,27 +823,30 @@ static struct list_head *find_workspace(int type)
 static void free_workspace(int type, struct list_head *workspace)
 {
 	int idx = type - 1;
-	struct list_head *idle_workspace	= &comp_idle_workspace[idx];
-	spinlock_t *workspace_lock		= &comp_workspace_lock[idx];
-	atomic_t *alloc_workspace		= &comp_alloc_workspace[idx];
-	wait_queue_head_t *workspace_wait	= &comp_workspace_wait[idx];
-	int *num_workspace			= &comp_num_workspace[idx];
-
-	spin_lock(workspace_lock);
-	if (*num_workspace < num_online_cpus()) {
-		list_add(workspace, idle_workspace);
-		(*num_workspace)++;
-		spin_unlock(workspace_lock);
+	struct list_head *idle_ws	= &btrfs_comp_ws[idx].idle_ws;
+	spinlock_t *ws_lock		= &btrfs_comp_ws[idx].ws_lock;
+	atomic_t *alloc_ws		= &btrfs_comp_ws[idx].alloc_ws;
+	wait_queue_head_t *ws_wait	= &btrfs_comp_ws[idx].ws_wait;
+	int *num_ws			= &btrfs_comp_ws[idx].num_ws;
+
+	spin_lock(ws_lock);
+	if (*num_ws < num_online_cpus()) {
+		list_add(workspace, idle_ws);
+		(*num_ws)++;
+		spin_unlock(ws_lock);
 		goto wake;
 	}
-	spin_unlock(workspace_lock);
+	spin_unlock(ws_lock);
 
 	btrfs_compress_op[idx]->free_workspace(workspace);
-	atomic_dec(alloc_workspace);
+	atomic_dec(alloc_ws);
 wake:
+	/*
+	 * Make sure counter is updated before we wake up waiters.
+	 */
 	smp_mb();
-	if (waitqueue_active(workspace_wait))
-		wake_up(workspace_wait);
+	if (waitqueue_active(ws_wait))
+		wake_up(ws_wait);
 }
 
 /*
@@ -853,11 +858,11 @@ static void free_workspaces(void)
 	int i;
 
 	for (i = 0; i < BTRFS_COMPRESS_TYPES; i++) {
-		while (!list_empty(&comp_idle_workspace[i])) {
-			workspace = comp_idle_workspace[i].next;
+		while (!list_empty(&btrfs_comp_ws[i].idle_ws)) {
+			workspace = btrfs_comp_ws[i].idle_ws.next;
 			list_del(workspace);
 			btrfs_compress_op[i]->free_workspace(workspace);
-			atomic_dec(&comp_alloc_workspace[i]);
+			atomic_dec(&btrfs_comp_ws[i].alloc_ws);
 		}
 	}
 }

fs/btrfs/ctree.c

@@ -1011,7 +1011,7 @@ static noinline int update_ref_for_cow(struct btrfs_trans_handle *trans,
 			return ret;
 		if (refs == 0) {
 			ret = -EROFS;
-			btrfs_std_error(root->fs_info, ret);
+			btrfs_std_error(root->fs_info, ret, NULL);
 			return ret;
 		}
 	} else {
@@ -1927,7 +1927,7 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 		child = read_node_slot(root, mid, 0);
 		if (!child) {
 			ret = -EROFS;
-			btrfs_std_error(root->fs_info, ret);
+			btrfs_std_error(root->fs_info, ret, NULL);
 			goto enospc;
 		}
 
@@ -2030,7 +2030,7 @@ static noinline int balance_level(struct btrfs_trans_handle *trans,
 		 */
 		if (!left) {
 			ret = -EROFS;
-			btrfs_std_error(root->fs_info, ret);
+			btrfs_std_error(root->fs_info, ret, NULL);
 			goto enospc;
 		}
 		wret = balance_node_right(trans, root, mid, left);
@@ -4940,8 +4940,8 @@ int btrfs_del_items(struct btrfs_trans_handle *trans, struct btrfs_root *root,
 {
 	struct extent_buffer *leaf;
 	struct btrfs_item *item;
-	int last_off;
-	int dsize = 0;
+	u32 last_off;
+	u32 dsize = 0;
 	int ret = 0;
 	int wret;
 	int i;

fs/btrfs/delayed-inode.c

@@ -463,6 +463,10 @@ static int __btrfs_add_delayed_deletion_item(struct btrfs_delayed_node *node,
 static void finish_one_item(struct btrfs_delayed_root *delayed_root)
 {
 	int seq = atomic_inc_return(&delayed_root->items_seq);
+
+	/*
+	 * atomic_dec_return implies a barrier for waitqueue_active
+	 */
 	if ((atomic_dec_return(&delayed_root->items) <
 	    BTRFS_DELAYED_BACKGROUND || seq % BTRFS_DELAYED_BATCH == 0) &&
 	    waitqueue_active(&delayed_root->wait))

fs/btrfs/delayed-ref.h

@@ -68,7 +68,6 @@ struct btrfs_delayed_ref_node {
 
 	unsigned int action:8;
 	unsigned int type:8;
-	unsigned int no_quota:1;
 	/* is this node still in the rbtree? */
 	unsigned int is_head:1;
 	unsigned int in_tree:1;
@@ -112,6 +111,17 @@ struct btrfs_delayed_ref_head {
 	 */
 	int total_ref_mod;
 
+	/*
+	 * For qgroup reserved space freeing.
+	 *
+	 * ref_root and reserved will be recorded after
+	 * BTRFS_ADD_DELAYED_EXTENT is called.
+	 * And will be used to free reserved qgroup space at
+	 * run_delayed_refs() time.
+	 */
+	u64 qgroup_ref_root;
+	u64 qgroup_reserved;
+
 	/*
 	 * when a new extent is allocated, it is just reserved in memory
 	 * The actual extent isn't inserted into the extent allocation tree
@@ -233,15 +243,16 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 			       struct btrfs_trans_handle *trans,
 			       u64 bytenr, u64 num_bytes, u64 parent,
 			       u64 ref_root, int level, int action,
-			       struct btrfs_delayed_extent_op *extent_op,
-			       int no_quota);
+			       struct btrfs_delayed_extent_op *extent_op);
 int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 			       struct btrfs_trans_handle *trans,
 			       u64 bytenr, u64 num_bytes,
 			       u64 parent, u64 ref_root,
-			       u64 owner, u64 offset, int action,
-			       struct btrfs_delayed_extent_op *extent_op,
-			       int no_quota);
+			       u64 owner, u64 offset, u64 reserved, int action,
+			       struct btrfs_delayed_extent_op *extent_op);
+int btrfs_add_delayed_qgroup_reserve(struct btrfs_fs_info *fs_info,
+				     struct btrfs_trans_handle *trans,
+				     u64 ref_root, u64 bytenr, u64 num_bytes);
 int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info,
 				struct btrfs_trans_handle *trans,
 				u64 bytenr, u64 num_bytes,

fs/btrfs/dev-replace.c

@@ -327,19 +327,6 @@ int btrfs_dev_replace_start(struct btrfs_root *root,
 	    args->start.tgtdev_name[0] == '\0')
 		return -EINVAL;
 
-	/*
-	 * Here we commit the transaction to make sure commit_total_bytes
-	 * of all the devices are updated.
-	 */
-	trans = btrfs_attach_transaction(root);
-	if (!IS_ERR(trans)) {
-		ret = btrfs_commit_transaction(trans, root);
-		if (ret)
-			return ret;
-	} else if (PTR_ERR(trans) != -ENOENT) {
-		return PTR_ERR(trans);
-	}
-
 	/* the disk copy procedure reuses the scrub code */
 	mutex_lock(&fs_info->volume_mutex);
 	ret = btrfs_dev_replace_find_srcdev(root, args->start.srcdevid,
@@ -356,6 +343,19 @@ int btrfs_dev_replace_start(struct btrfs_root *root,
 	if (ret)
 		return ret;
 
+	/*
+	 * Here we commit the transaction to make sure commit_total_bytes
+	 * of all the devices are updated.
+	 */
+	trans = btrfs_attach_transaction(root);
+	if (!IS_ERR(trans)) {
+		ret = btrfs_commit_transaction(trans, root);
+		if (ret)
+			return ret;
+	} else if (PTR_ERR(trans) != -ENOENT) {
+		return PTR_ERR(trans);
+	}
+
 	btrfs_dev_replace_lock(dev_replace);
 	switch (dev_replace->replace_state) {
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_NEVER_STARTED:
@@ -375,12 +375,8 @@ int btrfs_dev_replace_start(struct btrfs_root *root,
 	WARN_ON(!tgt_device);
 	dev_replace->tgtdev = tgt_device;
 
-	ret = btrfs_kobj_add_device(tgt_device->fs_devices, tgt_device);
-	if (ret)
-		btrfs_err(root->fs_info, "kobj add dev failed %d\n", ret);
-
-	printk_in_rcu(KERN_INFO
-		      "BTRFS: dev_replace from %s (devid %llu) to %s started\n",
+	btrfs_info_in_rcu(root->fs_info,
+		      "dev_replace from %s (devid %llu) to %s started",
 		      src_device->missing ? "<missing disk>" :
 		        rcu_str_deref(src_device->name),
 		      src_device->devid,
@@ -401,6 +397,10 @@ int btrfs_dev_replace_start(struct btrfs_root *root,
 	args->result = BTRFS_IOCTL_DEV_REPLACE_RESULT_NO_ERROR;
 	btrfs_dev_replace_unlock(dev_replace);
 
+	ret = btrfs_sysfs_add_device_link(tgt_device->fs_devices, tgt_device);
+	if (ret)
+		btrfs_err(root->fs_info, "kobj add dev failed %d\n", ret);
+
 	btrfs_wait_ordered_roots(root->fs_info, -1);
 
 	/* force writing the updated state information to disk */
@@ -454,8 +454,7 @@ static void btrfs_rm_dev_replace_blocked(struct btrfs_fs_info *fs_info)
 static void btrfs_rm_dev_replace_unblocked(struct btrfs_fs_info *fs_info)
 {
 	clear_bit(BTRFS_FS_STATE_DEV_REPLACING, &fs_info->fs_state);
-	if (waitqueue_active(&fs_info->replace_wait))
-		wake_up(&fs_info->replace_wait);
+	wake_up(&fs_info->replace_wait);
 }
 
 static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
@@ -523,8 +522,8 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 								src_device,
 								tgt_device);
 	} else {
-		printk_in_rcu(KERN_ERR
-			      "BTRFS: btrfs_scrub_dev(%s, %llu, %s) failed %d\n",
+		btrfs_err_in_rcu(root->fs_info,
+			      "btrfs_scrub_dev(%s, %llu, %s) failed %d",
 			      src_device->missing ? "<missing disk>" :
 			        rcu_str_deref(src_device->name),
 			      src_device->devid,
@@ -540,8 +539,8 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 		return scrub_ret;
 	}
 
-	printk_in_rcu(KERN_INFO
-		      "BTRFS: dev_replace from %s (devid %llu) to %s finished\n",
+	btrfs_info_in_rcu(root->fs_info,
+		      "dev_replace from %s (devid %llu) to %s finished",
 		      src_device->missing ? "<missing disk>" :
 		        rcu_str_deref(src_device->name),
 		      src_device->devid,
@@ -586,7 +585,7 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 	mutex_unlock(&uuid_mutex);
 
 	/* replace the sysfs entry */
-	btrfs_kobj_rm_device(fs_info->fs_devices, src_device);
+	btrfs_sysfs_rm_device_link(fs_info->fs_devices, src_device);
 	btrfs_rm_dev_replace_free_srcdev(fs_info, src_device);
 
 	/* write back the superblocks */
@@ -809,8 +808,8 @@ static int btrfs_dev_replace_kthread(void *data)
 		progress = status_args->status.progress_1000;
 		kfree(status_args);
 		progress = div_u64(progress, 10);
-		printk_in_rcu(KERN_INFO
-			"BTRFS: continuing dev_replace from %s (devid %llu) to %s @%u%%\n",
+		btrfs_info_in_rcu(fs_info,
+			"continuing dev_replace from %s (devid %llu) to %s @%u%%",
 			dev_replace->srcdev->missing ? "<missing disk>" :
 			rcu_str_deref(dev_replace->srcdev->name),
 			dev_replace->srcdev->devid,

fs/btrfs/disk-io.h

@@ -60,6 +60,8 @@ void close_ctree(struct btrfs_root *root);
 int write_ctree_super(struct btrfs_trans_handle *trans,
 		      struct btrfs_root *root, int max_mirrors);
 struct buffer_head *btrfs_read_dev_super(struct block_device *bdev);
+int btrfs_read_dev_one_super(struct block_device *bdev, int copy_num,
+			struct buffer_head **bh_ret);
 int btrfs_commit_super(struct btrfs_root *root);
 struct extent_buffer *btrfs_find_tree_block(struct btrfs_fs_info *fs_info,
 					    u64 bytenr);

fs/btrfs/extent_io.h

@@ -2,6 +2,7 @@
 #define __EXTENTIO__
 
 #include <linux/rbtree.h>
+#include "ulist.h"
 
 /* bits for the extent state */
 #define EXTENT_DIRTY		(1U << 0)
@@ -18,6 +19,7 @@
 #define EXTENT_NEED_WAIT	(1U << 13)
 #define EXTENT_DAMAGED		(1U << 14)
 #define EXTENT_NORESERVE	(1U << 15)
+#define EXTENT_QGROUP_RESERVED	(1U << 16)
 #define EXTENT_IOBITS		(EXTENT_LOCKED | EXTENT_WRITEBACK)
 #define EXTENT_CTLBITS		(EXTENT_DO_ACCOUNTING | EXTENT_FIRST_DELALLOC)
 
@@ -161,6 +163,17 @@ struct extent_buffer {
 #endif
 };
 
+/*
+ * Structure to record how many bytes and which ranges are set/cleared
+ */
+struct extent_changeset {
+	/* How many bytes are set/cleared in this operation */
+	u64 bytes_changed;
+
+	/* Changed ranges */
+	struct ulist *range_changed;
+};
+
 static inline void extent_set_compress_type(unsigned long *bio_flags,
 					    int compress_type)
 {
@@ -210,11 +223,17 @@ int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
 		   struct extent_state *cached_state);
 int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 		      unsigned bits, gfp_t mask);
+int clear_record_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
+			     unsigned bits, gfp_t mask,
+			     struct extent_changeset *changeset);
 int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
 		     unsigned bits, int wake, int delete,
 		     struct extent_state **cached, gfp_t mask);
 int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 		    unsigned bits, gfp_t mask);
+int set_record_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
+			   unsigned bits, gfp_t mask,
+			   struct extent_changeset *changeset);
 int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
 		   unsigned bits, u64 *failed_start,
 		   struct extent_state **cached_state, gfp_t mask);

fs/btrfs/free-space-cache.c

@@ -450,9 +450,9 @@ static int io_ctl_check_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
 
 	gen = io_ctl->cur;
 	if (le64_to_cpu(*gen) != generation) {
-		printk_ratelimited(KERN_ERR "BTRFS: space cache generation "
-				   "(%Lu) does not match inode (%Lu)\n", *gen,
-				   generation);
+		btrfs_err_rl(io_ctl->root->fs_info,
+			"space cache generation (%llu) does not match inode (%llu)",
+				*gen, generation);
 		io_ctl_unmap_page(io_ctl);
 		return -EIO;
 	}
@@ -506,8 +506,8 @@ static int io_ctl_check_crc(struct btrfs_io_ctl *io_ctl, int index)
 			      PAGE_CACHE_SIZE - offset);
 	btrfs_csum_final(crc, (char *)&crc);
 	if (val != crc) {
-		printk_ratelimited(KERN_ERR "BTRFS: csum mismatch on free "
-				   "space cache\n");
+		btrfs_err_rl(io_ctl->root->fs_info,
+			"csum mismatch on free space cache");
 		io_ctl_unmap_page(io_ctl);
 		return -EIO;
 	}
@@ -1215,7 +1215,7 @@ int btrfs_wait_cache_io(struct btrfs_root *root,
  * @offset - the offset for the key we'll insert
  *
  * This function writes out a free space cache struct to disk for quick recovery
- * on mount.  This will return 0 if it was successfull in writing the cache out,
+ * on mount.  This will return 0 if it was successful in writing the cache out,
  * or an errno if it was not.
  */
 static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
@@ -1730,7 +1730,7 @@ static void bitmap_set_bits(struct btrfs_free_space_ctl *ctl,
  */
 static int search_bitmap(struct btrfs_free_space_ctl *ctl,
 			 struct btrfs_free_space *bitmap_info, u64 *offset,
-			 u64 *bytes)
+			 u64 *bytes, bool for_alloc)
 {
 	unsigned long found_bits = 0;
 	unsigned long max_bits = 0;
@@ -1738,11 +1738,26 @@ static int search_bitmap(struct btrfs_free_space_ctl *ctl,
 	unsigned long next_zero;
 	unsigned long extent_bits;
 
+	/*
+	 * Skip searching the bitmap if we don't have a contiguous section that
+	 * is large enough for this allocation.
+	 */
+	if (for_alloc &&
+	    bitmap_info->max_extent_size &&
+	    bitmap_info->max_extent_size < *bytes) {
+		*bytes = bitmap_info->max_extent_size;
+		return -1;
+	}
+
 	i = offset_to_bit(bitmap_info->offset, ctl->unit,
 			  max_t(u64, *offset, bitmap_info->offset));
 	bits = bytes_to_bits(*bytes, ctl->unit);
 
 	for_each_set_bit_from(i, bitmap_info->bitmap, BITS_PER_BITMAP) {
+		if (for_alloc && bits == 1) {
+			found_bits = 1;
+			break;
+		}
 		next_zero = find_next_zero_bit(bitmap_info->bitmap,
 					       BITS_PER_BITMAP, i);
 		extent_bits = next_zero - i;
@@ -1762,6 +1777,7 @@ static int search_bitmap(struct btrfs_free_space_ctl *ctl,
 	}
 
 	*bytes = (u64)(max_bits) * ctl->unit;
+	bitmap_info->max_extent_size = *bytes;
 	return -1;
 }
 
@@ -1813,7 +1829,7 @@ find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
 		if (entry->bitmap) {
 			u64 size = *bytes;
 
-			ret = search_bitmap(ctl, entry, &tmp, &size);
+			ret = search_bitmap(ctl, entry, &tmp, &size, true);
 			if (!ret) {
 				*offset = tmp;
 				*bytes = size;
@@ -1874,7 +1890,8 @@ static noinline int remove_from_bitmap(struct btrfs_free_space_ctl *ctl,
 	search_start = *offset;
 	search_bytes = ctl->unit;
 	search_bytes = min(search_bytes, end - search_start + 1);
-	ret = search_bitmap(ctl, bitmap_info, &search_start, &search_bytes);
+	ret = search_bitmap(ctl, bitmap_info, &search_start, &search_bytes,
+			    false);
 	if (ret < 0 || search_start != *offset)
 		return -EINVAL;
 
@@ -1919,7 +1936,7 @@ static noinline int remove_from_bitmap(struct btrfs_free_space_ctl *ctl,
 		search_start = *offset;
 		search_bytes = ctl->unit;
 		ret = search_bitmap(ctl, bitmap_info, &search_start,
-				    &search_bytes);
+				    &search_bytes, false);
 		if (ret < 0 || search_start != *offset)
 			return -EAGAIN;
 
@@ -1943,6 +1960,12 @@ static u64 add_bytes_to_bitmap(struct btrfs_free_space_ctl *ctl,
 
 	bitmap_set_bits(ctl, info, offset, bytes_to_set);
 
+	/*
+	 * We set some bytes, we have no idea what the max extent size is
+	 * anymore.
+	 */
+	info->max_extent_size = 0;
+
 	return bytes_to_set;
 
 }
@@ -1951,12 +1974,19 @@ static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
 		      struct btrfs_free_space *info)
 {
 	struct btrfs_block_group_cache *block_group = ctl->private;
+	bool forced = false;
+
+#ifdef CONFIG_BTRFS_DEBUG
+	if (btrfs_should_fragment_free_space(block_group->fs_info->extent_root,
+					     block_group))
+		forced = true;
+#endif
 
 	/*
 	 * If we are below the extents threshold then we can add this as an
 	 * extent, and don't have to deal with the bitmap
 	 */
-	if (ctl->free_extents < ctl->extents_thresh) {
+	if (!forced && ctl->free_extents < ctl->extents_thresh) {
 		/*
 		 * If this block group has some small extents we don't want to
 		 * use up all of our free slots in the cache with them, we want
@@ -2661,7 +2691,7 @@ static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group_cache *block_group,
 	search_start = min_start;
 	search_bytes = bytes;
 
-	err = search_bitmap(ctl, entry, &search_start, &search_bytes);
+	err = search_bitmap(ctl, entry, &search_start, &search_bytes, true);
 	if (err) {
 		if (search_bytes > *max_extent_size)
 			*max_extent_size = search_bytes;
@@ -2775,6 +2805,7 @@ static int btrfs_bitmap_cluster(struct btrfs_block_group_cache *block_group,
 	unsigned long want_bits;
 	unsigned long min_bits;
 	unsigned long found_bits;
+	unsigned long max_bits = 0;
 	unsigned long start = 0;
 	unsigned long total_found = 0;
 	int ret;
@@ -2784,6 +2815,13 @@ static int btrfs_bitmap_cluster(struct btrfs_block_group_cache *block_group,
 	want_bits = bytes_to_bits(bytes, ctl->unit);
 	min_bits = bytes_to_bits(min_bytes, ctl->unit);
 
+	/*
+	 * Don't bother looking for a cluster in this bitmap if it's heavily
+	 * fragmented.
+	 */
+	if (entry->max_extent_size &&
+	    entry->max_extent_size < cont1_bytes)
+		return -ENOSPC;
 again:
 	found_bits = 0;
 	for_each_set_bit_from(i, entry->bitmap, BITS_PER_BITMAP) {
@@ -2791,13 +2829,19 @@ static int btrfs_bitmap_cluster(struct btrfs_block_group_cache *block_group,
 					       BITS_PER_BITMAP, i);
 		if (next_zero - i >= min_bits) {
 			found_bits = next_zero - i;
+			if (found_bits > max_bits)
+				max_bits = found_bits;
 			break;
 		}
+		if (next_zero - i > max_bits)
+			max_bits = next_zero - i;
 		i = next_zero;
 	}
 
-	if (!found_bits)
+	if (!found_bits) {
+		entry->max_extent_size = (u64)max_bits * ctl->unit;
 		return -ENOSPC;
+	}
 
 	if (!total_found) {
 		start = i;
@@ -3056,6 +3100,7 @@ void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster)
 	spin_lock_init(&cluster->refill_lock);
 	cluster->root = RB_ROOT;
 	cluster->max_size = 0;
+	cluster->fragmented = false;
 	INIT_LIST_HEAD(&cluster->block_group_list);
 	cluster->block_group = NULL;
 }
@@ -3223,7 +3268,7 @@ static int trim_bitmaps(struct btrfs_block_group_cache *block_group,
 		}
 
 		bytes = minlen;
-		ret2 = search_bitmap(ctl, entry, &start, &bytes);
+		ret2 = search_bitmap(ctl, entry, &start, &bytes, false);
 		if (ret2 || start >= end) {
 			spin_unlock(&ctl->tree_lock);
 			mutex_unlock(&ctl->cache_writeout_mutex);
@@ -3376,7 +3421,7 @@ u64 btrfs_find_ino_for_alloc(struct btrfs_root *fs_root)
 		u64 count = 1;
 		int ret;
 
-		ret = search_bitmap(ctl, entry, &offset, &count);
+		ret = search_bitmap(ctl, entry, &offset, &count, true);
 		/* Logic error; Should be empty if it can't find anything */
 		ASSERT(!ret);
 
@@ -3532,6 +3577,7 @@ int test_add_free_space_entry(struct btrfs_block_group_cache *cache,
 		spin_lock(&ctl->tree_lock);
 		info->offset = offset;
 		info->bytes = bytes;
+		info->max_extent_size = 0;
 		ret = link_free_space(ctl, info);
 		spin_unlock(&ctl->tree_lock);
 		if (ret)
@@ -3559,6 +3605,7 @@ int test_add_free_space_entry(struct btrfs_block_group_cache *cache,
 	}
 
 	bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes);
+
 	bytes -= bytes_added;
 	offset += bytes_added;
 	spin_unlock(&ctl->tree_lock);
@@ -3602,7 +3649,7 @@ int test_check_exists(struct btrfs_block_group_cache *cache,
 
 		bit_off = offset;
 		bit_bytes = ctl->unit;
-		ret = search_bitmap(ctl, info, &bit_off, &bit_bytes);
+		ret = search_bitmap(ctl, info, &bit_off, &bit_bytes, false);
 		if (!ret) {
 			if (bit_off == offset) {
 				ret = 1;

fs/btrfs/free-space-cache.h

@@ -23,6 +23,7 @@ struct btrfs_free_space {
 	struct rb_node offset_index;
 	u64 offset;
 	u64 bytes;
+	u64 max_extent_size;
 	unsigned long *bitmap;
 	struct list_head list;
 };

fs/btrfs/inode-item.c

@@ -157,7 +157,7 @@ static int btrfs_del_inode_extref(struct btrfs_trans_handle *trans,
 	 */
 	if (!btrfs_find_name_in_ext_backref(path, ref_objectid,
 					    name, name_len, &extref)) {
-		btrfs_std_error(root->fs_info, -ENOENT);
+		btrfs_std_error(root->fs_info, -ENOENT, NULL);
 		ret = -EROFS;
 		goto out;
 	}

fs/btrfs/inode-map.c

@@ -488,17 +488,17 @@ int btrfs_save_ino_cache(struct btrfs_root *root,
 	/* Just to make sure we have enough space */
 	prealloc += 8 * PAGE_CACHE_SIZE;
 
-	ret = btrfs_delalloc_reserve_space(inode, prealloc);
+	ret = btrfs_delalloc_reserve_space(inode, 0, prealloc);
 	if (ret)
 		goto out_put;
 
 	ret = btrfs_prealloc_file_range_trans(inode, trans, 0, 0, prealloc,
 					      prealloc, prealloc, &alloc_hint);
 	if (ret) {
-		btrfs_delalloc_release_space(inode, prealloc);
+		btrfs_delalloc_release_space(inode, 0, prealloc);
 		goto out_put;
 	}
-	btrfs_free_reserved_data_space(inode, prealloc);
+	btrfs_free_reserved_data_space(inode, 0, prealloc);
 
 	ret = btrfs_write_out_ino_cache(root, trans, path, inode);
 out_put:

fs/btrfs/locking.c

@@ -79,6 +79,9 @@ void btrfs_clear_lock_blocking_rw(struct extent_buffer *eb, int rw)
 		write_lock(&eb->lock);
 		WARN_ON(atomic_read(&eb->spinning_writers));
 		atomic_inc(&eb->spinning_writers);
+		/*
+		 * atomic_dec_and_test implies a barrier for waitqueue_active
+		 */
 		if (atomic_dec_and_test(&eb->blocking_writers) &&
 		    waitqueue_active(&eb->write_lock_wq))
 			wake_up(&eb->write_lock_wq);
@@ -86,6 +89,9 @@ void btrfs_clear_lock_blocking_rw(struct extent_buffer *eb, int rw)
 		BUG_ON(atomic_read(&eb->blocking_readers) == 0);
 		read_lock(&eb->lock);
 		atomic_inc(&eb->spinning_readers);
+		/*
+		 * atomic_dec_and_test implies a barrier for waitqueue_active
+		 */
 		if (atomic_dec_and_test(&eb->blocking_readers) &&
 		    waitqueue_active(&eb->read_lock_wq))
 			wake_up(&eb->read_lock_wq);
@@ -229,6 +235,9 @@ void btrfs_tree_read_unlock_blocking(struct extent_buffer *eb)
 	}
 	btrfs_assert_tree_read_locked(eb);
 	WARN_ON(atomic_read(&eb->blocking_readers) == 0);
+	/*
+	 * atomic_dec_and_test implies a barrier for waitqueue_active
+	 */
 	if (atomic_dec_and_test(&eb->blocking_readers) &&
 	    waitqueue_active(&eb->read_lock_wq))
 		wake_up(&eb->read_lock_wq);
@@ -280,6 +289,9 @@ void btrfs_tree_unlock(struct extent_buffer *eb)
 	if (blockers) {
 		WARN_ON(atomic_read(&eb->spinning_writers));
 		atomic_dec(&eb->blocking_writers);
+		/*
+		 * Make sure counter is updated before we wake up waiters.
+		 */
 		smp_mb();
 		if (waitqueue_active(&eb->write_lock_wq))
 			wake_up(&eb->write_lock_wq);

fs/btrfs/ordered-data.c

@@ -345,6 +345,9 @@ int btrfs_dec_test_first_ordered_pending(struct inode *inode,
 
 	if (entry->bytes_left == 0) {
 		ret = test_and_set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags);
+		/*
+		 * Implicit memory barrier after test_and_set_bit
+		 */
 		if (waitqueue_active(&entry->wait))
 			wake_up(&entry->wait);
 	} else {
@@ -409,6 +412,9 @@ int btrfs_dec_test_ordered_pending(struct inode *inode,
 
 	if (entry->bytes_left == 0) {
 		ret = test_and_set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags);
+		/*
+		 * Implicit memory barrier after test_and_set_bit
+		 */
 		if (waitqueue_active(&entry->wait))
 			wake_up(&entry->wait);
 	} else {
@@ -484,15 +490,16 @@ void btrfs_wait_logged_extents(struct btrfs_trans_handle *trans,
 
 	spin_lock_irq(&log->log_extents_lock[index]);
 	while (!list_empty(&log->logged_list[index])) {
+		struct inode *inode;
 		ordered = list_first_entry(&log->logged_list[index],
 					   struct btrfs_ordered_extent,
 					   log_list);
 		list_del_init(&ordered->log_list);
+		inode = ordered->inode;
 		spin_unlock_irq(&log->log_extents_lock[index]);
 
 		if (!test_bit(BTRFS_ORDERED_IO_DONE, &ordered->flags) &&
 		    !test_bit(BTRFS_ORDERED_DIRECT, &ordered->flags)) {
-			struct inode *inode = ordered->inode;
 			u64 start = ordered->file_offset;
 			u64 end = ordered->file_offset + ordered->len - 1;
 
@@ -503,20 +510,25 @@ void btrfs_wait_logged_extents(struct btrfs_trans_handle *trans,
 						   &ordered->flags));
 
 		/*
-		 * If our ordered extent completed it means it updated the
-		 * fs/subvol and csum trees already, so no need to make the
-		 * current transaction's commit wait for it, as we end up
-		 * holding memory unnecessarily and delaying the inode's iput
-		 * until the transaction commit (we schedule an iput for the
-		 * inode when the ordered extent's refcount drops to 0), which
-		 * prevents it from being evictable until the transaction
-		 * commits.
+		 * In order to keep us from losing our ordered extent
+		 * information when committing the transaction we have to make
+		 * sure that any logged extents are completed when we go to
+		 * commit the transaction.  To do this we simply increase the
+		 * current transactions pending_ordered counter and decrement it
+		 * when the ordered extent completes.
 		 */
-		if (test_bit(BTRFS_ORDERED_COMPLETE, &ordered->flags))
-			btrfs_put_ordered_extent(ordered);
-		else
-			list_add_tail(&ordered->trans_list, &trans->ordered);
-
+		if (!test_bit(BTRFS_ORDERED_COMPLETE, &ordered->flags)) {
+			struct btrfs_ordered_inode_tree *tree;
+
+			tree = &BTRFS_I(inode)->ordered_tree;
+			spin_lock_irq(&tree->lock);
+			if (!test_bit(BTRFS_ORDERED_COMPLETE, &ordered->flags)) {
+				set_bit(BTRFS_ORDERED_PENDING, &ordered->flags);
+				atomic_inc(&trans->transaction->pending_ordered);
+			}
+			spin_unlock_irq(&tree->lock);
+		}
+		btrfs_put_ordered_extent(ordered);
 		spin_lock_irq(&log->log_extents_lock[index]);
 	}
 	spin_unlock_irq(&log->log_extents_lock[index]);
@@ -578,6 +590,7 @@ void btrfs_remove_ordered_extent(struct inode *inode,
 	struct btrfs_ordered_inode_tree *tree;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct rb_node *node;
+	bool dec_pending_ordered = false;
 
 	tree = &BTRFS_I(inode)->ordered_tree;
 	spin_lock_irq(&tree->lock);
@@ -587,8 +600,37 @@ void btrfs_remove_ordered_extent(struct inode *inode,
 	if (tree->last == node)
 		tree->last = NULL;
 	set_bit(BTRFS_ORDERED_COMPLETE, &entry->flags);
+	if (test_and_clear_bit(BTRFS_ORDERED_PENDING, &entry->flags))
+		dec_pending_ordered = true;
 	spin_unlock_irq(&tree->lock);
 
+	/*
+	 * The current running transaction is waiting on us, we need to let it
+	 * know that we're complete and wake it up.
+	 */
+	if (dec_pending_ordered) {
+		struct btrfs_transaction *trans;
+
+		/*
+		 * The checks for trans are just a formality, it should be set,
+		 * but if it isn't we don't want to deref/assert under the spin
+		 * lock, so be nice and check if trans is set, but ASSERT() so
+		 * if it isn't set a developer will notice.
+		 */
+		spin_lock(&root->fs_info->trans_lock);
+		trans = root->fs_info->running_transaction;
+		if (trans)
+			atomic_inc(&trans->use_count);
+		spin_unlock(&root->fs_info->trans_lock);
+
+		ASSERT(trans);
+		if (trans) {
+			if (atomic_dec_and_test(&trans->pending_ordered))
+				wake_up(&trans->pending_wait);
+			btrfs_put_transaction(trans);
+		}
+	}
+
 	spin_lock(&root->ordered_extent_lock);
 	list_del_init(&entry->root_extent_list);
 	root->nr_ordered_extents--;

fs/btrfs/ordered-data.h

@@ -73,6 +73,8 @@ struct btrfs_ordered_sum {
 
 #define BTRFS_ORDERED_LOGGED 10 /* Set when we've waited on this ordered extent
 				 * in the logging code. */
+#define BTRFS_ORDERED_PENDING 11 /* We are waiting for this ordered extent to
+				  * complete in the current transaction. */
 struct btrfs_ordered_extent {
 	/* logical offset in the file */
 	u64 file_offset;

fs/btrfs/props.c

@@ -49,18 +49,16 @@ static struct prop_handler prop_handlers[] = {
 		.extract = prop_compression_extract,
 		.inheritable = 1
 	},
-	{
-		.xattr_name = NULL
-	}
 };
 
 void __init btrfs_props_init(void)
 {
-	struct prop_handler *p;
+	int i;
 
 	hash_init(prop_handlers_ht);
 
-	for (p = &prop_handlers[0]; p->xattr_name; p++) {
+	for (i = 0; i < ARRAY_SIZE(prop_handlers); i++) {
+		struct prop_handler *p = &prop_handlers[i];
 		u64 h = btrfs_name_hash(p->xattr_name, strlen(p->xattr_name));
 
 		hash_add(prop_handlers_ht, &p->node, h);
@@ -301,15 +299,16 @@ static int inherit_props(struct btrfs_trans_handle *trans,
 			 struct inode *inode,
 			 struct inode *parent)
 {
-	const struct prop_handler *h;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	int ret;
+	int i;
 
 	if (!test_bit(BTRFS_INODE_HAS_PROPS,
 		      &BTRFS_I(parent)->runtime_flags))
 		return 0;
 
-	for (h = &prop_handlers[0]; h->xattr_name; h++) {
+	for (i = 0; i < ARRAY_SIZE(prop_handlers); i++) {
+		const struct prop_handler *h = &prop_handlers[i];
 		const char *value;
 		u64 num_bytes;
 

fs/btrfs/qgroup.c

@@ -1652,10 +1652,6 @@ static int qgroup_update_counters(struct btrfs_fs_info *fs_info,
 			}
 		}
 
-		/* For exclusive extent, free its reserved bytes too */
-		if (nr_old_roots == 0 && nr_new_roots == 1 &&
-		    cur_new_count == nr_new_roots)
-			qg->reserved -= num_bytes;
 		if (dirty)
 			qgroup_dirty(fs_info, qg);
 	}
@@ -2035,7 +2031,7 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
-int btrfs_qgroup_reserve(struct btrfs_root *root, u64 num_bytes)
+static int qgroup_reserve(struct btrfs_root *root, u64 num_bytes)
 {
 	struct btrfs_root *quota_root;
 	struct btrfs_qgroup *qgroup;
@@ -2116,14 +2112,13 @@ int btrfs_qgroup_reserve(struct btrfs_root *root, u64 num_bytes)
 	return ret;
 }
 
-void btrfs_qgroup_free(struct btrfs_root *root, u64 num_bytes)
+void btrfs_qgroup_free_refroot(struct btrfs_fs_info *fs_info,
+			       u64 ref_root, u64 num_bytes)
 {
 	struct btrfs_root *quota_root;
 	struct btrfs_qgroup *qgroup;
-	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct ulist_node *unode;
 	struct ulist_iterator uiter;
-	u64 ref_root = root->root_key.objectid;
 	int ret = 0;
 
 	if (!is_fstree(ref_root))
@@ -2169,6 +2164,11 @@ void btrfs_qgroup_free(struct btrfs_root *root, u64 num_bytes)
 	spin_unlock(&fs_info->qgroup_lock);
 }
 
+static inline void qgroup_free(struct btrfs_root *root, u64 num_bytes)
+{
+	return btrfs_qgroup_free_refroot(root->fs_info, root->objectid,
+					 num_bytes);
+}
 void assert_qgroups_uptodate(struct btrfs_trans_handle *trans)
 {
 	if (list_empty(&trans->qgroup_ref_list) && !trans->delayed_ref_elem.seq)
@@ -2188,10 +2188,10 @@ void assert_qgroups_uptodate(struct btrfs_trans_handle *trans)
  */
 static int
 qgroup_rescan_leaf(struct btrfs_fs_info *fs_info, struct btrfs_path *path,
-		   struct btrfs_trans_handle *trans,
-		   struct extent_buffer *scratch_leaf)
+		   struct btrfs_trans_handle *trans)
 {
 	struct btrfs_key found;
+	struct extent_buffer *scratch_leaf = NULL;
 	struct ulist *roots = NULL;
 	struct seq_list tree_mod_seq_elem = SEQ_LIST_INIT(tree_mod_seq_elem);
 	u64 num_bytes;
@@ -2229,7 +2229,15 @@ qgroup_rescan_leaf(struct btrfs_fs_info *fs_info, struct btrfs_path *path,
 	fs_info->qgroup_rescan_progress.objectid = found.objectid + 1;
 
 	btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
-	memcpy(scratch_leaf, path->nodes[0], sizeof(*scratch_leaf));
+	scratch_leaf = btrfs_clone_extent_buffer(path->nodes[0]);
+	if (!scratch_leaf) {
+		ret = -ENOMEM;
+		mutex_unlock(&fs_info->qgroup_rescan_lock);
+		goto out;
+	}
+	extent_buffer_get(scratch_leaf);
+	btrfs_tree_read_lock(scratch_leaf);
+	btrfs_set_lock_blocking_rw(scratch_leaf, BTRFS_READ_LOCK);
 	slot = path->slots[0];
 	btrfs_release_path(path);
 	mutex_unlock(&fs_info->qgroup_rescan_lock);
@@ -2255,6 +2263,10 @@ qgroup_rescan_leaf(struct btrfs_fs_info *fs_info, struct btrfs_path *path,
 			goto out;
 	}
 out:
+	if (scratch_leaf) {
+		btrfs_tree_read_unlock_blocking(scratch_leaf);
+		free_extent_buffer(scratch_leaf);
+	}
 	btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
 
 	return ret;
@@ -2266,16 +2278,12 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
 						     qgroup_rescan_work);
 	struct btrfs_path *path;
 	struct btrfs_trans_handle *trans = NULL;
-	struct extent_buffer *scratch_leaf = NULL;
 	int err = -ENOMEM;
 	int ret = 0;
 
 	path = btrfs_alloc_path();
 	if (!path)
 		goto out;
-	scratch_leaf = kmalloc(sizeof(*scratch_leaf), GFP_NOFS);
-	if (!scratch_leaf)
-		goto out;
 
 	err = 0;
 	while (!err) {
@@ -2287,8 +2295,7 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
 		if (!fs_info->quota_enabled) {
 			err = -EINTR;
 		} else {
-			err = qgroup_rescan_leaf(fs_info, path, trans,
-						 scratch_leaf);
+			err = qgroup_rescan_leaf(fs_info, path, trans);
 		}
 		if (err > 0)
 			btrfs_commit_transaction(trans, fs_info->fs_root);
@@ -2297,7 +2304,6 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
 	}
 
 out:
-	kfree(scratch_leaf);
 	btrfs_free_path(path);
 
 	mutex_lock(&fs_info->qgroup_rescan_lock);
@@ -2486,3 +2492,190 @@ btrfs_qgroup_rescan_resume(struct btrfs_fs_info *fs_info)
 		btrfs_queue_work(fs_info->qgroup_rescan_workers,
 				 &fs_info->qgroup_rescan_work);
 }
+
+/*
+ * Reserve qgroup space for range [start, start + len).
+ *
+ * This function will either reserve space from related qgroups or doing
+ * nothing if the range is already reserved.
+ *
+ * Return 0 for successful reserve
+ * Return <0 for error (including -EQUOT)
+ *
+ * NOTE: this function may sleep for memory allocation.
+ */
+int btrfs_qgroup_reserve_data(struct inode *inode, u64 start, u64 len)
+{
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct extent_changeset changeset;
+	struct ulist_node *unode;
+	struct ulist_iterator uiter;
+	int ret;
+
+	if (!root->fs_info->quota_enabled || !is_fstree(root->objectid) ||
+	    len == 0)
+		return 0;
+
+	changeset.bytes_changed = 0;
+	changeset.range_changed = ulist_alloc(GFP_NOFS);
+	ret = set_record_extent_bits(&BTRFS_I(inode)->io_tree, start,
+			start + len -1, EXTENT_QGROUP_RESERVED, GFP_NOFS,
+			&changeset);
+	trace_btrfs_qgroup_reserve_data(inode, start, len,
+					changeset.bytes_changed,
+					QGROUP_RESERVE);
+	if (ret < 0)
+		goto cleanup;
+	ret = qgroup_reserve(root, changeset.bytes_changed);
+	if (ret < 0)
+		goto cleanup;
+
+	ulist_free(changeset.range_changed);
+	return ret;
+
+cleanup:
+	/* cleanup already reserved ranges */
+	ULIST_ITER_INIT(&uiter);
+	while ((unode = ulist_next(changeset.range_changed, &uiter)))
+		clear_extent_bit(&BTRFS_I(inode)->io_tree, unode->val,
+				 unode->aux, EXTENT_QGROUP_RESERVED, 0, 0, NULL,
+				 GFP_NOFS);
+	ulist_free(changeset.range_changed);
+	return ret;
+}
+
+static int __btrfs_qgroup_release_data(struct inode *inode, u64 start, u64 len,
+				       int free)
+{
+	struct extent_changeset changeset;
+	int trace_op = QGROUP_RELEASE;
+	int ret;
+
+	changeset.bytes_changed = 0;
+	changeset.range_changed = ulist_alloc(GFP_NOFS);
+	if (!changeset.range_changed)
+		return -ENOMEM;
+
+	ret = clear_record_extent_bits(&BTRFS_I(inode)->io_tree, start, 
+			start + len -1, EXTENT_QGROUP_RESERVED, GFP_NOFS,
+			&changeset);
+	if (ret < 0)
+		goto out;
+
+	if (free) {
+		qgroup_free(BTRFS_I(inode)->root, changeset.bytes_changed);
+		trace_op = QGROUP_FREE;
+	}
+	trace_btrfs_qgroup_release_data(inode, start, len,
+					changeset.bytes_changed, trace_op);
+out:
+	ulist_free(changeset.range_changed);
+	return ret;
+}
+
+/*
+ * Free a reserved space range from io_tree and related qgroups
+ *
+ * Should be called when a range of pages get invalidated before reaching disk.
+ * Or for error cleanup case.
+ *
+ * For data written to disk, use btrfs_qgroup_release_data().
+ *
+ * NOTE: This function may sleep for memory allocation.
+ */
+int btrfs_qgroup_free_data(struct inode *inode, u64 start, u64 len)
+{
+	return __btrfs_qgroup_release_data(inode, start, len, 1);
+}
+
+/*
+ * Release a reserved space range from io_tree only.
+ *
+ * Should be called when a range of pages get written to disk and corresponding
+ * FILE_EXTENT is inserted into corresponding root.
+ *
+ * Since new qgroup accounting framework will only update qgroup numbers at
+ * commit_transaction() time, its reserved space shouldn't be freed from
+ * related qgroups.
+ *
+ * But we should release the range from io_tree, to allow further write to be
+ * COWed.
+ *
+ * NOTE: This function may sleep for memory allocation.
+ */
+int btrfs_qgroup_release_data(struct inode *inode, u64 start, u64 len)
+{
+	return __btrfs_qgroup_release_data(inode, start, len, 0);
+}
+
+int btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes)
+{
+	int ret;
+
+	if (!root->fs_info->quota_enabled || !is_fstree(root->objectid) ||
+	    num_bytes == 0)
+		return 0;
+
+	BUG_ON(num_bytes != round_down(num_bytes, root->nodesize));
+	ret = qgroup_reserve(root, num_bytes);
+	if (ret < 0)
+		return ret;
+	atomic_add(num_bytes, &root->qgroup_meta_rsv);
+	return ret;
+}
+
+void btrfs_qgroup_free_meta_all(struct btrfs_root *root)
+{
+	int reserved;
+
+	if (!root->fs_info->quota_enabled || !is_fstree(root->objectid))
+		return;
+
+	reserved = atomic_xchg(&root->qgroup_meta_rsv, 0);
+	if (reserved == 0)
+		return;
+	qgroup_free(root, reserved);
+}
+
+void btrfs_qgroup_free_meta(struct btrfs_root *root, int num_bytes)
+{
+	if (!root->fs_info->quota_enabled || !is_fstree(root->objectid))
+		return;
+
+	BUG_ON(num_bytes != round_down(num_bytes, root->nodesize));
+	WARN_ON(atomic_read(&root->qgroup_meta_rsv) < num_bytes);
+	atomic_sub(num_bytes, &root->qgroup_meta_rsv);
+	qgroup_free(root, num_bytes);
+}
+
+/*
+ * Check qgroup reserved space leaking, normally at destory inode
+ * time
+ */
+void btrfs_qgroup_check_reserved_leak(struct inode *inode)
+{
+	struct extent_changeset changeset;
+	struct ulist_node *unode;
+	struct ulist_iterator iter;
+	int ret;
+
+	changeset.bytes_changed = 0;
+	changeset.range_changed = ulist_alloc(GFP_NOFS);
+	if (WARN_ON(!changeset.range_changed))
+		return;
+
+	ret = clear_record_extent_bits(&BTRFS_I(inode)->io_tree, 0, (u64)-1,
+			EXTENT_QGROUP_RESERVED, GFP_NOFS, &changeset);
+
+	WARN_ON(ret < 0);
+	if (WARN_ON(changeset.bytes_changed)) {
+		ULIST_ITER_INIT(&iter);
+		while ((unode = ulist_next(changeset.range_changed, &iter))) {
+			btrfs_warn(BTRFS_I(inode)->root->fs_info,
+				"leaking qgroup reserved space, ino: %lu, start: %llu, end: %llu",
+				inode->i_ino, unode->val, unode->aux);
+		}
+		qgroup_free(BTRFS_I(inode)->root, changeset.bytes_changed);
+	}
+	ulist_free(changeset.range_changed);
+}

fs/btrfs/qgroup.h

@@ -33,6 +33,13 @@ struct btrfs_qgroup_extent_record {
 	struct ulist *old_roots;
 };
 
+/*
+ * For qgroup event trace points only
+ */
+#define QGROUP_RESERVE		(1<<0)
+#define QGROUP_RELEASE		(1<<1)
+#define QGROUP_FREE		(1<<2)
+
 int btrfs_quota_enable(struct btrfs_trans_handle *trans,
 		       struct btrfs_fs_info *fs_info);
 int btrfs_quota_disable(struct btrfs_trans_handle *trans,
@@ -71,9 +78,18 @@ int btrfs_run_qgroups(struct btrfs_trans_handle *trans,
 int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans,
 			 struct btrfs_fs_info *fs_info, u64 srcid, u64 objectid,
 			 struct btrfs_qgroup_inherit *inherit);
-int btrfs_qgroup_reserve(struct btrfs_root *root, u64 num_bytes);
-void btrfs_qgroup_free(struct btrfs_root *root, u64 num_bytes);
-
+void btrfs_qgroup_free_refroot(struct btrfs_fs_info *fs_info,
+			       u64 ref_root, u64 num_bytes);
+/*
+ * TODO: Add proper trace point for it, as btrfs_qgroup_free() is
+ * called by everywhere, can't provide good trace for delayed ref case.
+ */
+static inline void btrfs_qgroup_free_delayed_ref(struct btrfs_fs_info *fs_info,
+						 u64 ref_root, u64 num_bytes)
+{
+	btrfs_qgroup_free_refroot(fs_info, ref_root, num_bytes);
+	trace_btrfs_qgroup_free_delayed_ref(ref_root, num_bytes);
+}
 void assert_qgroups_uptodate(struct btrfs_trans_handle *trans);
 
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
@@ -81,4 +97,13 @@ int btrfs_verify_qgroup_counts(struct btrfs_fs_info *fs_info, u64 qgroupid,
 			       u64 rfer, u64 excl);
 #endif
 
+/* New io_tree based accurate qgroup reserve API */
+int btrfs_qgroup_reserve_data(struct inode *inode, u64 start, u64 len);
+int btrfs_qgroup_release_data(struct inode *inode, u64 start, u64 len);
+int btrfs_qgroup_free_data(struct inode *inode, u64 start, u64 len);
+
+int btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes);
+void btrfs_qgroup_free_meta_all(struct btrfs_root *root);
+void btrfs_qgroup_free_meta(struct btrfs_root *root, int num_bytes);
+void btrfs_qgroup_check_reserved_leak(struct inode *inode);
 #endif /* __BTRFS_QGROUP__ */

fs/btrfs/raid56.c

@@ -810,7 +810,11 @@ static noinline void unlock_stripe(struct btrfs_raid_bio *rbio)
 			}
 
 			goto done_nolock;
-		} else  if (waitqueue_active(&h->wait)) {
+			/*
+			 * The barrier for this waitqueue_active is not needed,
+			 * we're protected by h->lock and can't miss a wakeup.
+			 */
+		} else if (waitqueue_active(&h->wait)) {
 			spin_unlock(&rbio->bio_list_lock);
 			spin_unlock_irqrestore(&h->lock, flags);
 			wake_up(&h->wait);

fs/btrfs/reada.c

@@ -569,7 +569,7 @@ static int reada_add_block(struct reada_control *rc, u64 logical,
 	rec = kzalloc(sizeof(*rec), GFP_NOFS);
 	if (!rec) {
 		reada_extent_put(root->fs_info, re);
-		return -1;
+		return -ENOMEM;
 	}
 
 	rec->rc = rc;
@@ -918,6 +918,7 @@ struct reada_control *btrfs_reada_add(struct btrfs_root *root,
 	u64 start;
 	u64 generation;
 	int level;
+	int ret;
 	struct extent_buffer *node;
 	static struct btrfs_key max_key = {
 		.objectid = (u64)-1,
@@ -943,9 +944,10 @@ struct reada_control *btrfs_reada_add(struct btrfs_root *root,
 	generation = btrfs_header_generation(node);
 	free_extent_buffer(node);
 
-	if (reada_add_block(rc, start, &max_key, level, generation)) {
+	ret = reada_add_block(rc, start, &max_key, level, generation);
+	if (ret) {
 		kfree(rc);
-		return ERR_PTR(-ENOMEM);
+		return ERR_PTR(ret);
 	}
 
 	reada_start_machine(root->fs_info);

fs/btrfs/relocation.c

@@ -1716,7 +1716,7 @@ int replace_file_extents(struct btrfs_trans_handle *trans,
 		ret = btrfs_inc_extent_ref(trans, root, new_bytenr,
 					   num_bytes, parent,
 					   btrfs_header_owner(leaf),
-					   key.objectid, key.offset, 1);
+					   key.objectid, key.offset);
 		if (ret) {
 			btrfs_abort_transaction(trans, root, ret);
 			break;
@@ -1724,7 +1724,7 @@ int replace_file_extents(struct btrfs_trans_handle *trans,
 
 		ret = btrfs_free_extent(trans, root, bytenr, num_bytes,
 					parent, btrfs_header_owner(leaf),
-					key.objectid, key.offset, 1);
+					key.objectid, key.offset);
 		if (ret) {
 			btrfs_abort_transaction(trans, root, ret);
 			break;
@@ -1900,23 +1900,21 @@ int replace_path(struct btrfs_trans_handle *trans,
 
 		ret = btrfs_inc_extent_ref(trans, src, old_bytenr, blocksize,
 					path->nodes[level]->start,
-					src->root_key.objectid, level - 1, 0,
-					1);
+					src->root_key.objectid, level - 1, 0);
 		BUG_ON(ret);
 		ret = btrfs_inc_extent_ref(trans, dest, new_bytenr, blocksize,
 					0, dest->root_key.objectid, level - 1,
-					0, 1);
+					0);
 		BUG_ON(ret);
 
 		ret = btrfs_free_extent(trans, src, new_bytenr, blocksize,
 					path->nodes[level]->start,
-					src->root_key.objectid, level - 1, 0,
-					1);
+					src->root_key.objectid, level - 1, 0);
 		BUG_ON(ret);
 
 		ret = btrfs_free_extent(trans, dest, old_bytenr, blocksize,
 					0, dest->root_key.objectid, level - 1,
-					0, 1);
+					0);
 		BUG_ON(ret);
 
 		btrfs_unlock_up_safe(path, 0);
@@ -2418,7 +2416,7 @@ void merge_reloc_roots(struct reloc_control *rc)
 	}
 out:
 	if (ret) {
-		btrfs_std_error(root->fs_info, ret);
+		btrfs_std_error(root->fs_info, ret, NULL);
 		if (!list_empty(&reloc_roots))
 			free_reloc_roots(&reloc_roots);
 
@@ -2745,7 +2743,7 @@ static int do_relocation(struct btrfs_trans_handle *trans,
 						node->eb->start, blocksize,
 						upper->eb->start,
 						btrfs_header_owner(upper->eb),
-						node->level, 0, 1);
+						node->level, 0);
 			BUG_ON(ret);
 
 			ret = btrfs_drop_subtree(trans, root, eb, upper->eb);
@@ -3034,8 +3032,8 @@ int prealloc_file_extent_cluster(struct inode *inode,
 	BUG_ON(cluster->start != cluster->boundary[0]);
 	mutex_lock(&inode->i_mutex);
 
-	ret = btrfs_check_data_free_space(inode, cluster->end +
-					  1 - cluster->start, 0);
+	ret = btrfs_check_data_free_space(inode, cluster->start,
+					  cluster->end + 1 - cluster->start);
 	if (ret)
 		goto out;
 
@@ -3056,8 +3054,8 @@ int prealloc_file_extent_cluster(struct inode *inode,
 			break;
 		nr++;
 	}
-	btrfs_free_reserved_data_space(inode, cluster->end +
-				       1 - cluster->start);
+	btrfs_free_reserved_data_space(inode, cluster->start,
+				       cluster->end + 1 - cluster->start);
 out:
 	mutex_unlock(&inode->i_mutex);
 	return ret;

fs/btrfs/root-tree.c

@@ -45,12 +45,13 @@ static void btrfs_read_root_item(struct extent_buffer *eb, int slot,
 	if (!need_reset && btrfs_root_generation(item)
 		!= btrfs_root_generation_v2(item)) {
 		if (btrfs_root_generation_v2(item) != 0) {
-			printk(KERN_WARNING "BTRFS: mismatching "
+			btrfs_warn(eb->fs_info,
+					"mismatching "
 					"generation and generation_v2 "
 					"found in root item. This root "
 					"was probably mounted with an "
 					"older kernel. Resetting all "
-					"new fields.\n");
+					"new fields.");
 		}
 		need_reset = 1;
 	}
@@ -141,7 +142,7 @@ int btrfs_update_root(struct btrfs_trans_handle *trans, struct btrfs_root
 	int ret;
 	int slot;
 	unsigned long ptr;
-	int old_len;
+	u32 old_len;
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -283,7 +284,7 @@ int btrfs_find_orphan_roots(struct btrfs_root *tree_root)
 			trans = btrfs_join_transaction(tree_root);
 			if (IS_ERR(trans)) {
 				err = PTR_ERR(trans);
-				btrfs_error(tree_root->fs_info, err,
+				btrfs_std_error(tree_root->fs_info, err,
 					    "Failed to start trans to delete "
 					    "orphan item");
 				break;
@@ -292,7 +293,7 @@ int btrfs_find_orphan_roots(struct btrfs_root *tree_root)
 						    root_key.objectid);
 			btrfs_end_transaction(trans, tree_root);
 			if (err) {
-				btrfs_error(tree_root->fs_info, err,
+				btrfs_std_error(tree_root->fs_info, err,
 					    "Failed to delete root orphan "
 					    "item");
 				break;

fs/btrfs/sysfs.h

@@ -82,9 +82,9 @@ char *btrfs_printable_features(enum btrfs_feature_set set, u64 flags);
 extern const char * const btrfs_feature_set_names[3];
 extern struct kobj_type space_info_ktype;
 extern struct kobj_type btrfs_raid_ktype;
-int btrfs_kobj_add_device(struct btrfs_fs_devices *fs_devices,
+int btrfs_sysfs_add_device_link(struct btrfs_fs_devices *fs_devices,
 		struct btrfs_device *one_device);
-int btrfs_kobj_rm_device(struct btrfs_fs_devices *fs_devices,
+int btrfs_sysfs_rm_device_link(struct btrfs_fs_devices *fs_devices,
                 struct btrfs_device *one_device);
 int btrfs_sysfs_add_fsid(struct btrfs_fs_devices *fs_devs,
 				struct kobject *parent);

fs/btrfs/tests/free-space-tests.c

@@ -19,6 +19,7 @@
 #include <linux/slab.h>
 #include "btrfs-tests.h"
 #include "../ctree.h"
+#include "../disk-io.h"
 #include "../free-space-cache.h"
 
 #define BITS_PER_BITMAP		(PAGE_CACHE_SIZE * 8)
@@ -35,6 +36,12 @@ static struct btrfs_block_group_cache *init_test_block_group(void)
 		kfree(cache);
 		return NULL;
 	}
+	cache->fs_info = btrfs_alloc_dummy_fs_info();
+	if (!cache->fs_info) {
+		kfree(cache->free_space_ctl);
+		kfree(cache);
+		return NULL;
+	}
 
 	cache->key.objectid = 0;
 	cache->key.offset = 1024 * 1024 * 1024;
@@ -879,7 +886,8 @@ test_steal_space_from_bitmap_to_extent(struct btrfs_block_group_cache *cache)
 int btrfs_test_free_space_cache(void)
 {
 	struct btrfs_block_group_cache *cache;
-	int ret;
+	struct btrfs_root *root = NULL;
+	int ret = -ENOMEM;
 
 	test_msg("Running btrfs free space cache tests\n");
 
@@ -889,6 +897,17 @@ int btrfs_test_free_space_cache(void)
 		return 0;
 	}
 
+	root = btrfs_alloc_dummy_root();
+	if (!root)
+		goto out;
+
+	root->fs_info = btrfs_alloc_dummy_fs_info();
+	if (!root->fs_info)
+		goto out;
+
+	root->fs_info->extent_root = root;
+	cache->fs_info = root->fs_info;
+
 	ret = test_extents(cache);
 	if (ret)
 		goto out;
@@ -904,6 +923,7 @@ int btrfs_test_free_space_cache(void)
 	__btrfs_remove_free_space_cache(cache->free_space_ctl);
 	kfree(cache->free_space_ctl);
 	kfree(cache);
+	btrfs_free_dummy_root(root);
 	test_msg("Free space cache tests finished\n");
 	return ret;
 }