21 files changed, 569 insertions, 466 deletions

diff --git a/fs/btrfs/Makefile b/fs/btrfs/Makefile
index 9adf5e4f7e9..94212844a9b 100644
--- a/fs/btrfs/Makefile
+++ b/fs/btrfs/Makefile
@@ -1,25 +1,10 @@
-ifneq ($(KERNELRELEASE),)
-# kbuild part of makefile
 
 obj-$(CONFIG_BTRFS_FS) := btrfs.o
-btrfs-y := super.o ctree.o extent-tree.o print-tree.o root-tree.o dir-item.o \
+
+btrfs-y += super.o ctree.o extent-tree.o print-tree.o root-tree.o dir-item.o \
 	   file-item.o inode-item.o inode-map.o disk-io.o \
 	   transaction.o inode.o file.o tree-defrag.o \
 	   extent_map.o sysfs.o struct-funcs.o xattr.o ordered-data.o \
 	   extent_io.o volumes.o async-thread.o ioctl.o locking.o orphan.o \
 	   ref-cache.o export.o tree-log.o acl.o free-space-cache.o zlib.o \
 	   compression.o delayed-ref.o
-else
-
-# Normal Makefile
-
-KERNELDIR := /lib/modules/`uname -r`/build
-all:
-	$(MAKE) -C $(KERNELDIR) M=`pwd` CONFIG_BTRFS_FS=m modules
-
-modules_install:
-	$(MAKE) -C $(KERNELDIR) M=`pwd` modules_install
-clean:
-	$(MAKE) -C $(KERNELDIR) M=`pwd` clean
-
-endif
diff --git a/fs/btrfs/acl.c b/fs/btrfs/acl.c
index 7fdd184a528..cbba000dccb 100644
--- a/fs/btrfs/acl.c
+++ b/fs/btrfs/acl.c
@@ -60,15 +60,20 @@ static struct posix_acl *btrfs_get_acl(struct inode *inode, int type)
 		return ERR_PTR(-EINVAL);
 	}
 
+	/* Handle the cached NULL acl case without locking */
+	acl = ACCESS_ONCE(*p_acl);
+	if (!acl)
+		return acl;
+
 	spin_lock(&inode->i_lock);
-	if (*p_acl != BTRFS_ACL_NOT_CACHED)
-		acl = posix_acl_dup(*p_acl);
+	acl = *p_acl;
+	if (acl != BTRFS_ACL_NOT_CACHED)
+		acl = posix_acl_dup(acl);
 	spin_unlock(&inode->i_lock);
 
-	if (acl)
+	if (acl != BTRFS_ACL_NOT_CACHED)
 		return acl;
 
-
 	size = __btrfs_getxattr(inode, name, "", 0);
 	if (size > 0) {
 		value = kzalloc(size, GFP_NOFS);
@@ -80,9 +85,12 @@ static struct posix_acl *btrfs_get_acl(struct inode *inode, int type)
 			btrfs_update_cached_acl(inode, p_acl, acl);
 		}
 		kfree(value);
-	} else if (size == -ENOENT) {
+	} else if (size == -ENOENT || size == -ENODATA || size == 0) {
+		/* FIXME, who returns -ENOENT?  I think nobody */
 		acl = NULL;
 		btrfs_update_cached_acl(inode, p_acl, acl);
+	} else {
+		acl = ERR_PTR(-EIO);
 	}
 
 	return acl;
diff --git a/fs/btrfs/async-thread.c b/fs/btrfs/async-thread.c
index 51bfdfc8fcd..502c3d61de6 100644
--- a/fs/btrfs/async-thread.c
+++ b/fs/btrfs/async-thread.c
@@ -25,6 +25,7 @@
 #define WORK_QUEUED_BIT 0
 #define WORK_DONE_BIT 1
 #define WORK_ORDER_DONE_BIT 2
+#define WORK_HIGH_PRIO_BIT 3
 
 /*
  * container for the kthread task pointer and the list of pending work
@@ -36,6 +37,7 @@ struct btrfs_worker_thread {
 
 	/* list of struct btrfs_work that are waiting for service */
 	struct list_head pending;
+	struct list_head prio_pending;
 
 	/* list of worker threads from struct btrfs_workers */
 	struct list_head worker_list;
@@ -103,10 +105,16 @@ static noinline int run_ordered_completions(struct btrfs_workers *workers,
 
 	spin_lock_irqsave(&workers->lock, flags);
 
-	while (!list_empty(&workers->order_list)) {
-		work = list_entry(workers->order_list.next,
-				  struct btrfs_work, order_list);
-
+	while (1) {
+		if (!list_empty(&workers->prio_order_list)) {
+			work = list_entry(workers->prio_order_list.next,
+					  struct btrfs_work, order_list);
+		} else if (!list_empty(&workers->order_list)) {
+			work = list_entry(workers->order_list.next,
+					  struct btrfs_work, order_list);
+		} else {
+			break;
+		}
 		if (!test_bit(WORK_DONE_BIT, &work->flags))
 			break;
 
@@ -143,8 +151,14 @@ static int worker_loop(void *arg)
 	do {
 		spin_lock_irq(&worker->lock);
 again_locked:
-		while (!list_empty(&worker->pending)) {
-			cur = worker->pending.next;
+		while (1) {
+			if (!list_empty(&worker->prio_pending))
+				cur = worker->prio_pending.next;
+			else if (!list_empty(&worker->pending))
+				cur = worker->pending.next;
+			else
+				break;
+
 			work = list_entry(cur, struct btrfs_work, list);
 			list_del(&work->list);
 			clear_bit(WORK_QUEUED_BIT, &work->flags);
@@ -163,7 +177,6 @@ again_locked:
 
 			spin_lock_irq(&worker->lock);
 			check_idle_worker(worker);
-
 		}
 		if (freezing(current)) {
 			worker->working = 0;
@@ -178,7 +191,8 @@ again_locked:
 				 * jump_in?
 				 */
 				smp_mb();
-				if (!list_empty(&worker->pending))
+				if (!list_empty(&worker->pending) ||
+				    !list_empty(&worker->prio_pending))
 					continue;
 
 				/*
@@ -191,7 +205,8 @@ again_locked:
 				 */
 				schedule_timeout(1);
 				smp_mb();
-				if (!list_empty(&worker->pending))
+				if (!list_empty(&worker->pending) ||
+				    !list_empty(&worker->prio_pending))
 					continue;
 
 				if (kthread_should_stop())
@@ -200,7 +215,8 @@ again_locked:
 				/* still no more work?, sleep for real */
 				spin_lock_irq(&worker->lock);
 				set_current_state(TASK_INTERRUPTIBLE);
-				if (!list_empty(&worker->pending))
+				if (!list_empty(&worker->pending) ||
+				    !list_empty(&worker->prio_pending))
 					goto again_locked;
 
 				/*
@@ -248,6 +264,7 @@ void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max)
 	INIT_LIST_HEAD(&workers->worker_list);
 	INIT_LIST_HEAD(&workers->idle_list);
 	INIT_LIST_HEAD(&workers->order_list);
+	INIT_LIST_HEAD(&workers->prio_order_list);
 	spin_lock_init(&workers->lock);
 	workers->max_workers = max;
 	workers->idle_thresh = 32;
@@ -273,6 +290,7 @@ int btrfs_start_workers(struct btrfs_workers *workers, int num_workers)
 		}
 
 		INIT_LIST_HEAD(&worker->pending);
+		INIT_LIST_HEAD(&worker->prio_pending);
 		INIT_LIST_HEAD(&worker->worker_list);
 		spin_lock_init(&worker->lock);
 		atomic_set(&worker->num_pending, 0);
@@ -396,7 +414,10 @@ int btrfs_requeue_work(struct btrfs_work *work)
 		goto out;
 
 	spin_lock_irqsave(&worker->lock, flags);
-	list_add_tail(&work->list, &worker->pending);
+	if (test_bit(WORK_HIGH_PRIO_BIT, &work->flags))
+		list_add_tail(&work->list, &worker->prio_pending);
+	else
+		list_add_tail(&work->list, &worker->pending);
 	atomic_inc(&worker->num_pending);
 
 	/* by definition we're busy, take ourselves off the idle
@@ -422,6 +443,11 @@ out:
 	return 0;
 }
 
+void btrfs_set_work_high_prio(struct btrfs_work *work)
+{
+	set_bit(WORK_HIGH_PRIO_BIT, &work->flags);
+}
+
 /*
  * places a struct btrfs_work into the pending queue of one of the kthreads
  */
@@ -438,7 +464,12 @@ int btrfs_queue_worker(struct btrfs_workers *workers, struct btrfs_work *work)
 	worker = find_worker(workers);
 	if (workers->ordered) {
 		spin_lock_irqsave(&workers->lock, flags);
-		list_add_tail(&work->order_list, &workers->order_list);
+		if (test_bit(WORK_HIGH_PRIO_BIT, &work->flags)) {
+			list_add_tail(&work->order_list,
+				      &workers->prio_order_list);
+		} else {
+			list_add_tail(&work->order_list, &workers->order_list);
+		}
 		spin_unlock_irqrestore(&workers->lock, flags);
 	} else {
 		INIT_LIST_HEAD(&work->order_list);
@@ -446,7 +477,10 @@ int btrfs_queue_worker(struct btrfs_workers *workers, struct btrfs_work *work)
 
 	spin_lock_irqsave(&worker->lock, flags);
 
-	list_add_tail(&work->list, &worker->pending);
+	if (test_bit(WORK_HIGH_PRIO_BIT, &work->flags))
+		list_add_tail(&work->list, &worker->prio_pending);
+	else
+		list_add_tail(&work->list, &worker->pending);
 	atomic_inc(&worker->num_pending);
 	check_busy_worker(worker);
 
diff --git a/fs/btrfs/async-thread.h b/fs/btrfs/async-thread.h
index 31be4ed8b63..1b511c109db 100644
--- a/fs/btrfs/async-thread.h
+++ b/fs/btrfs/async-thread.h
@@ -85,6 +85,7 @@ struct btrfs_workers {
 	 * of work items waiting for completion
 	 */
 	struct list_head order_list;
+	struct list_head prio_order_list;
 
 	/* lock for finding the next worker thread to queue on */
 	spinlock_t lock;
@@ -98,4 +99,5 @@ int btrfs_start_workers(struct btrfs_workers *workers, int num_workers);
 int btrfs_stop_workers(struct btrfs_workers *workers);
 void btrfs_init_workers(struct btrfs_workers *workers, char *name, int max);
 int btrfs_requeue_work(struct btrfs_work *work);
+void btrfs_set_work_high_prio(struct btrfs_work *work);
 #endif
diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index e5b2533b691..a99f1c2a710 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -1325,12 +1325,12 @@ static noinline int reada_for_balance(struct btrfs_root *root,
 	int ret = 0;
 	int blocksize;
 
-	parent = path->nodes[level - 1];
+	parent = path->nodes[level + 1];
 	if (!parent)
 		return 0;
 
 	nritems = btrfs_header_nritems(parent);
-	slot = path->slots[level];
+	slot = path->slots[level + 1];
 	blocksize = btrfs_level_size(root, level);
 
 	if (slot > 0) {
@@ -1341,7 +1341,7 @@ static noinline int reada_for_balance(struct btrfs_root *root,
 			block1 = 0;
 		free_extent_buffer(eb);
 	}
-	if (slot < nritems) {
+	if (slot + 1 < nritems) {
 		block2 = btrfs_node_blockptr(parent, slot + 1);
 		gen = btrfs_node_ptr_generation(parent, slot + 1);
 		eb = btrfs_find_tree_block(root, block2, blocksize);
@@ -1351,7 +1351,11 @@ static noinline int reada_for_balance(struct btrfs_root *root,
 	}
 	if (block1 || block2) {
 		ret = -EAGAIN;
+
+		/* release the whole path */
 		btrfs_release_path(root, path);
+
+		/* read the blocks */
 		if (block1)
 			readahead_tree_block(root, block1, blocksize, 0);
 		if (block2)
@@ -1361,7 +1365,7 @@ static noinline int reada_for_balance(struct btrfs_root *root,
 			eb = read_tree_block(root, block1, blocksize, 0);
 			free_extent_buffer(eb);
 		}
-		if (block1) {
+		if (block2) {
 			eb = read_tree_block(root, block2, blocksize, 0);
 			free_extent_buffer(eb);
 		}
@@ -1481,12 +1485,15 @@ read_block_for_search(struct btrfs_trans_handle *trans,
 	 * of the btree by dropping locks before
 	 * we read.
 	 */
-	btrfs_release_path(NULL, p);
+	btrfs_unlock_up_safe(p, level + 1);
+	btrfs_set_path_blocking(p);
+
 	if (tmp)
 		free_extent_buffer(tmp);
 	if (p->reada)
 		reada_for_search(root, p, level, slot, key->objectid);
 
+	btrfs_release_path(NULL, p);
 	tmp = read_tree_block(root, blocknr, blocksize, gen);
 	if (tmp)
 		free_extent_buffer(tmp);
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index ad96495dedc..4414a5d9983 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -881,6 +881,9 @@ struct btrfs_fs_info {
 	u64 metadata_alloc_profile;
 	u64 system_alloc_profile;
 
+	unsigned data_chunk_allocations;
+	unsigned metadata_ratio;
+
 	void *bdev_holder;
 };
 
@@ -2174,7 +2177,8 @@ int btrfs_check_file(struct btrfs_root *root, struct inode *inode);
 extern struct file_operations btrfs_file_operations;
 int btrfs_drop_extents(struct btrfs_trans_handle *trans,
 		       struct btrfs_root *root, struct inode *inode,
-		       u64 start, u64 end, u64 inline_limit, u64 *hint_block);
+		       u64 start, u64 end, u64 locked_end,
+		       u64 inline_limit, u64 *hint_block);
 int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
 			      struct btrfs_root *root,
 			      struct inode *inode, u64 start, u64 end);
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 92caa8035f3..0ff16d3331d 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -232,10 +232,14 @@ static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf,
 			memcpy(&found, result, csum_size);
 
 			read_extent_buffer(buf, &val, 0, csum_size);
-			printk(KERN_INFO "btrfs: %s checksum verify failed "
-			       "on %llu wanted %X found %X level %d\n",
-			       root->fs_info->sb->s_id,
-			       buf->start, val, found, btrfs_header_level(buf));
+			if (printk_ratelimit()) {
+				printk(KERN_INFO "btrfs: %s checksum verify "
+				       "failed on %llu wanted %X found %X "
+				       "level %d\n",
+				       root->fs_info->sb->s_id,
+				       (unsigned long long)buf->start, val, found,
+				       btrfs_header_level(buf));
+			}
 			if (result != (char *)&inline_result)
 				kfree(result);
 			return 1;
@@ -268,10 +272,13 @@ static int verify_parent_transid(struct extent_io_tree *io_tree,
 		ret = 0;
 		goto out;
 	}
-	printk("parent transid verify failed on %llu wanted %llu found %llu\n",
-	       (unsigned long long)eb->start,
-	       (unsigned long long)parent_transid,
-	       (unsigned long long)btrfs_header_generation(eb));
+	if (printk_ratelimit()) {
+		printk("parent transid verify failed on %llu wanted %llu "
+		       "found %llu\n",
+		       (unsigned long long)eb->start,
+		       (unsigned long long)parent_transid,
+		       (unsigned long long)btrfs_header_generation(eb));
+	}
 	ret = 1;
 	clear_extent_buffer_uptodate(io_tree, eb);
 out:
@@ -415,9 +422,12 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 
 	found_start = btrfs_header_bytenr(eb);
 	if (found_start != start) {
-		printk(KERN_INFO "btrfs bad tree block start %llu %llu\n",
-		       (unsigned long long)found_start,
-		       (unsigned long long)eb->start);
+		if (printk_ratelimit()) {
+			printk(KERN_INFO "btrfs bad tree block start "
+			       "%llu %llu\n",
+			       (unsigned long long)found_start,
+			       (unsigned long long)eb->start);
+		}
 		ret = -EIO;
 		goto err;
 	}
@@ -429,8 +439,10 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
 		goto err;
 	}
 	if (check_tree_block_fsid(root, eb)) {
-		printk(KERN_INFO "btrfs bad fsid on block %llu\n",
-		       (unsigned long long)eb->start);
+		if (printk_ratelimit()) {
+			printk(KERN_INFO "btrfs bad fsid on block %llu\n",
+			       (unsigned long long)eb->start);
+		}
 		ret = -EIO;
 		goto err;
 	}
@@ -579,19 +591,12 @@ int btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct inode *inode,
 	async->bio_flags = bio_flags;
 
 	atomic_inc(&fs_info->nr_async_submits);
+
+	if (rw & (1 << BIO_RW_SYNCIO))
+		btrfs_set_work_high_prio(&async->work);
+
 	btrfs_queue_worker(&fs_info->workers, &async->work);
-#if 0
-	int limit = btrfs_async_submit_limit(fs_info);
-	if (atomic_read(&fs_info->nr_async_submits) > limit) {
-		wait_event_timeout(fs_info->async_submit_wait,
-			   (atomic_read(&fs_info->nr_async_submits) < limit),
-			   HZ/10);
 
-		wait_event_timeout(fs_info->async_submit_wait,
-			   (atomic_read(&fs_info->nr_async_bios) < limit),
-			   HZ/10);
-	}
-#endif
 	while (atomic_read(&fs_info->async_submit_draining) &&
 	      atomic_read(&fs_info->nr_async_submits)) {
 		wait_event(fs_info->async_submit_wait,
@@ -656,6 +661,7 @@ static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
 		return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
 				     mirror_num, 0);
 	}
+
 	/*
 	 * kthread helpers are used to submit writes so that checksumming
 	 * can happen in parallel across all CPUs
@@ -765,27 +771,6 @@ static void btree_invalidatepage(struct page *page, unsigned long offset)
 	}
 }
 
-#if 0
-static int btree_writepage(struct page *page, struct writeback_control *wbc)
-{
-	struct buffer_head *bh;
-	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
-	struct buffer_head *head;
-	if (!page_has_buffers(page)) {
-		create_empty_buffers(page, root->fs_info->sb->s_blocksize,
-					(1 << BH_Dirty)|(1 << BH_Uptodate));
-	}
-	head = page_buffers(page);
-	bh = head;
-	do {
-		if (buffer_dirty(bh))
-			csum_tree_block(root, bh, 0);
-		bh = bh->b_this_page;
-	} while (bh != head);
-	return block_write_full_page(page, btree_get_block, wbc);
-}
-#endif
-
 static struct address_space_operations btree_aops = {
 	.readpage	= btree_readpage,
 	.writepage	= btree_writepage,
@@ -1273,11 +1258,7 @@ static int btrfs_congested_fn(void *congested_data, int bdi_bits)
 	int ret = 0;
 	struct btrfs_device *device;
 	struct backing_dev_info *bdi;
-#if 0
-	if ((bdi_bits & (1 << BDI_write_congested)) &&
-	    btrfs_congested_async(info, 0))
-		return 1;
-#endif
+
 	list_for_each_entry(device, &info->fs_devices->devices, dev_list) {
 		if (!device->bdev)
 			continue;
@@ -1599,6 +1580,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	fs_info->btree_inode = new_inode(sb);
 	fs_info->btree_inode->i_ino = 1;
 	fs_info->btree_inode->i_nlink = 1;
+	fs_info->metadata_ratio = 8;
 
 	fs_info->thread_pool_size = min_t(unsigned long,
 					  num_online_cpus() + 2, 8);
@@ -1689,7 +1671,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	if (features) {
 		printk(KERN_ERR "BTRFS: couldn't mount because of "
 		       "unsupported optional features (%Lx).\n",
-		       features);
+		       (unsigned long long)features);
 		err = -EINVAL;
 		goto fail_iput;
 	}
@@ -1699,7 +1681,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	if (!(sb->s_flags & MS_RDONLY) && features) {
 		printk(KERN_ERR "BTRFS: couldn't mount RDWR because of "
 		       "unsupported option features (%Lx).\n",
-		       features);
+		       (unsigned long long)features);
 		err = -EINVAL;
 		goto fail_iput;
 	}
@@ -2095,10 +2077,10 @@ static int write_dev_supers(struct btrfs_device *device,
 				device->barriers = 0;
 				get_bh(bh);
 				lock_buffer(bh);
-				ret = submit_bh(WRITE, bh);
+				ret = submit_bh(WRITE_SYNC, bh);
 			}
 		} else {
-			ret = submit_bh(WRITE, bh);
+			ret = submit_bh(WRITE_SYNC, bh);
 		}
 
 		if (!ret && wait) {
@@ -2291,7 +2273,7 @@ int close_ctree(struct btrfs_root *root)
 
 	if (fs_info->delalloc_bytes) {
 		printk(KERN_INFO "btrfs: at unmount delalloc count %llu\n",
-		       fs_info->delalloc_bytes);
+		       (unsigned long long)fs_info->delalloc_bytes);
 	}
 	if (fs_info->total_ref_cache_size) {
 		printk(KERN_INFO "btrfs: at umount reference cache size %llu\n",
@@ -2328,16 +2310,6 @@ int close_ctree(struct btrfs_root *root)
 	btrfs_stop_workers(&fs_info->endio_write_workers);
 	btrfs_stop_workers(&fs_info->submit_workers);
 
-#if 0
-	while (!list_empty(&fs_info->hashers)) {
-		struct btrfs_hasher *hasher;
-		hasher = list_entry(fs_info->hashers.next, struct btrfs_hasher,
-				    hashers);
-		list_del(&hasher->hashers);
-		crypto_free_hash(&fs_info->hash_tfm);
-		kfree(hasher);
-	}
-#endif
 	btrfs_close_devices(fs_info->fs_devices);
 	btrfs_mapping_tree_free(&fs_info->mapping_tree);
 
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 178df4c67de..e4966444811 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -1844,10 +1844,14 @@ again:
 		printk(KERN_ERR "no space left, need %llu, %llu delalloc bytes"
 		       ", %llu bytes_used, %llu bytes_reserved, "
 		       "%llu bytes_pinned, %llu bytes_readonly, %llu may use"
-		       "%llu total\n", bytes, data_sinfo->bytes_delalloc,
-		       data_sinfo->bytes_used, data_sinfo->bytes_reserved,
-		       data_sinfo->bytes_pinned, data_sinfo->bytes_readonly,
-		       data_sinfo->bytes_may_use, data_sinfo->total_bytes);
+		       "%llu total\n", (unsigned long long)bytes,
+		       (unsigned long long)data_sinfo->bytes_delalloc,
+		       (unsigned long long)data_sinfo->bytes_used,
+		       (unsigned long long)data_sinfo->bytes_reserved,
+		       (unsigned long long)data_sinfo->bytes_pinned,
+		       (unsigned long long)data_sinfo->bytes_readonly,
+		       (unsigned long long)data_sinfo->bytes_may_use,
+		       (unsigned long long)data_sinfo->total_bytes);
 		return -ENOSPC;
 	}
 	data_sinfo->bytes_may_use += bytes;
@@ -1918,15 +1922,29 @@ void btrfs_delalloc_free_space(struct btrfs_root *root, struct inode *inode,
 	spin_unlock(&info->lock);
 }
 
+static void force_metadata_allocation(struct btrfs_fs_info *info)
+{
+	struct list_head *head = &info->space_info;
+	struct btrfs_space_info *found;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(found, head, list) {
+		if (found->flags & BTRFS_BLOCK_GROUP_METADATA)
+			found->force_alloc = 1;
+	}
+	rcu_read_unlock();
+}
+
 static int do_chunk_alloc(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *extent_root, u64 alloc_bytes,
 			  u64 flags, int force)
 {
 	struct btrfs_space_info *space_info;
+	struct btrfs_fs_info *fs_info = extent_root->fs_info;
 	u64 thresh;
 	int ret = 0;
 
-	mutex_lock(&extent_root->fs_info->chunk_mutex);
+	mutex_lock(&fs_info->chunk_mutex);
 
 	flags = btrfs_reduce_alloc_profile(extent_root, flags);
 
@@ -1958,6 +1976,18 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans,
 	}
 	spin_unlock(&space_info->lock);
 
+	/*
+	 * if we're doing a data chunk, go ahead and make sure that
+	 * we keep a reasonable number of metadata chunks allocated in the
+	 * FS as well.
+	 */
+	if (flags & BTRFS_BLOCK_GROUP_DATA) {
+		fs_info->data_chunk_allocations++;
+		if (!(fs_info->data_chunk_allocations %
+		      fs_info->metadata_ratio))
+			force_metadata_allocation(fs_info);
+	}
+
 	ret = btrfs_alloc_chunk(trans, extent_root, flags);
 	if (ret)
 		space_info->full = 1;
@@ -2798,9 +2828,12 @@ static void dump_space_info(struct btrfs_space_info *info, u64 bytes)
 				    info->bytes_pinned - info->bytes_reserved),
 	       (info->full) ? "" : "not ");
 	printk(KERN_INFO "space_info total=%llu, pinned=%llu, delalloc=%llu,"
-	       " may_use=%llu, used=%llu\n", info->total_bytes,
-	       info->bytes_pinned, info->bytes_delalloc, info->bytes_may_use,
-	       info->bytes_used);
+	       " may_use=%llu, used=%llu\n",
+	       (unsigned long long)info->total_bytes,
+	       (unsigned long long)info->bytes_pinned,
+	       (unsigned long long)info->bytes_delalloc,
+	       (unsigned long long)info->bytes_may_use,
+	       (unsigned long long)info->bytes_used);
 
 	down_read(&info->groups_sem);
 	list_for_each_entry(cache, &info->block_groups, list) {
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index eb2bee8b7fb..fe9eb990e44 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -17,12 +17,6 @@
 #include "ctree.h"
 #include "btrfs_inode.h"
 
-/* temporary define until extent_map moves out of btrfs */
-struct kmem_cache *btrfs_cache_create(const char *name, size_t size,
-				       unsigned long extra_flags,
-				       void (*ctor)(void *, struct kmem_cache *,
-						    unsigned long));
-
 static struct kmem_cache *extent_state_cache;
 static struct kmem_cache *extent_buffer_cache;
 
@@ -50,20 +44,23 @@ struct extent_page_data {
 	/* tells writepage not to lock the state bits for this range
 	 * it still does the unlocking
 	 */
-	int extent_locked;
+	unsigned int extent_locked:1;
+
+	/* tells the submit_bio code to use a WRITE_SYNC */
+	unsigned int sync_io:1;
 };
 
 int __init extent_io_init(void)
 {
-	extent_state_cache = btrfs_cache_create("extent_state",
-					    sizeof(struct extent_state), 0,
-					    NULL);
+	extent_state_cache = kmem_cache_create("extent_state",
+			sizeof(struct extent_state), 0,
+			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
 	if (!extent_state_cache)
 		return -ENOMEM;
 
-	extent_buffer_cache = btrfs_cache_create("extent_buffers",
-					    sizeof(struct extent_buffer), 0,
-					    NULL);
+	extent_buffer_cache = kmem_cache_create("extent_buffers",
+			sizeof(struct extent_buffer), 0,
+			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
 	if (!extent_buffer_cache)
 		goto free_state_cache;
 	return 0;
@@ -1404,69 +1401,6 @@ out:
 	return total_bytes;
 }
 
-#if 0
-/*
- * helper function to lock both pages and extents in the tree.
- * pages must be locked first.
- */
-static int lock_range(struct extent_io_tree *tree, u64 start, u64 end)
-{
-	unsigned long index = start >> PAGE_CACHE_SHIFT;
-	unsigned long end_index = end >> PAGE_CACHE_SHIFT;
-	struct page *page;
-	int err;
-
-	while (index <= end_index) {
-		page = grab_cache_page(tree->mapping, index);
-		if (!page) {
-			err = -ENOMEM;
-			goto failed;
-		}
-		if (IS_ERR(page)) {
-			err = PTR_ERR(page);
-			goto failed;
-		}
-		index++;
-	}
-	lock_extent(tree, start, end, GFP_NOFS);
-	return 0;
-
-failed:
-	/*
-	 * we failed above in getting the page at 'index', so we undo here
-	 * up to but not including the page at 'index'
-	 */
-	end_index = index;
-	index = start >> PAGE_CACHE_SHIFT;
-	while (index < end_index) {
-		page = find_get_page(tree->mapping, index);
-		unlock_page(page);
-		page_cache_release(page);
-		index++;
-	}
-	return err;
-}
-
-/*
- * helper function to unlock both pages and extents in the tree.
- */
-static int unlock_range(struct extent_io_tree *tree, u64 start, u64 end)
-{
-	unsigned long index = start >> PAGE_CACHE_SHIFT;
-	unsigned long end_index = end >> PAGE_CACHE_SHIFT;
-	struct page *page;
-
-	while (index <= end_index) {
-		page = find_get_page(tree->mapping, index);
-		unlock_page(page);
-		page_cache_release(page);
-		index++;
-	}
-	unlock_extent(tree, start, end, GFP_NOFS);
-	return 0;
-}
-#endif
-
 /*
  * set the private field for a given byte offset in the tree.  If there isn't
  * an extent_state there already, this does nothing.
@@ -2101,6 +2035,16 @@ int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
 	return ret;
 }
 
+static noinline void update_nr_written(struct page *page,
+				      struct writeback_control *wbc,
+				      unsigned long nr_written)
+{
+	wbc->nr_to_write -= nr_written;
+	if (wbc->range_cyclic || (wbc->nr_to_write > 0 &&
+	    wbc->range_start == 0 && wbc->range_end == LLONG_MAX))
+		page->mapping->writeback_index = page->index + nr_written;
+}
+
 /*
  * the writepage semantics are similar to regular writepage.  extent
  * records are inserted to lock ranges in the tree, and as dirty areas
@@ -2136,8 +2080,14 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 	u64 delalloc_end;
 	int page_started;
 	int compressed;
+	int write_flags;
 	unsigned long nr_written = 0;
 
+	if (wbc->sync_mode == WB_SYNC_ALL)
+		write_flags = WRITE_SYNC_PLUG;
+	else
+		write_flags = WRITE;
+
 	WARN_ON(!PageLocked(page));
 	pg_offset = i_size & (PAGE_CACHE_SIZE - 1);
 	if (page->index > end_index ||
@@ -2164,6 +2114,12 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 	delalloc_end = 0;
 	page_started = 0;
 	if (!epd->extent_locked) {
+		/*
+		 * make sure the wbc mapping index is at least updated
+		 * to this page.
+		 */
+		update_nr_written(page, wbc, 0);
+
 		while (delalloc_end < page_end) {
 			nr_delalloc = find_lock_delalloc_range(inode, tree,
 						       page,
@@ -2185,7 +2141,13 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 		 */
 		if (page_started) {
 			ret = 0;
-			goto update_nr_written;
+			/*
+			 * we've unlocked the page, so we can't update
+			 * the mapping's writeback index, just update
+			 * nr_to_write.
+			 */
+			wbc->nr_to_write -= nr_written;
+			goto done_unlocked;
 		}
 	}
 	lock_extent(tree, start, page_end, GFP_NOFS);
@@ -2198,13 +2160,18 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 		if (ret == -EAGAIN) {
 			unlock_extent(tree, start, page_end, GFP_NOFS);
 			redirty_page_for_writepage(wbc, page);
+			update_nr_written(page, wbc, nr_written);
 			unlock_page(page);
 			ret = 0;
-			goto update_nr_written;
+			goto done_unlocked;
 		}
 	}
 
-	nr_written++;
+	/*
+	 * we don't want to touch the inode after unlocking the page,
+	 * so we update the mapping writeback index now
+	 */
+	update_nr_written(page, wbc, nr_written + 1);
 
 	end = page_end;
 	if (test_range_bit(tree, start, page_end, EXTENT_DELALLOC, 0))
@@ -2314,9 +2281,9 @@ static int __extent_writepage(struct page *page, struct writeback_control *wbc,
 				       (unsigned long long)end);
 			}
 
-			ret = submit_extent_page(WRITE, tree, page, sector,
-						 iosize, pg_offset, bdev,
-						 &epd->bio, max_nr,
+			ret = submit_extent_page(write_flags, tree, page,
+						 sector, iosize, pg_offset,
+						 bdev, &epd->bio, max_nr,
 						 end_bio_extent_writepage,
 						 0, 0, 0);
 			if (ret)
@@ -2336,11 +2303,8 @@ done:
 		unlock_extent(tree, unlock_start, page_end, GFP_NOFS);
 	unlock_page(page);
 
-update_nr_written:
-	wbc->nr_to_write -= nr_written;
-	if (wbc->range_cyclic || (wbc->nr_to_write > 0 &&
-	    wbc->range_start == 0 && wbc->range_end == LLONG_MAX))
-		page->mapping->writeback_index = page->index + nr_written;
+done_unlocked:
+
 	return 0;
 }
 
@@ -2460,15 +2424,23 @@ retry:
 	return ret;
 }
 
-static noinline void flush_write_bio(void *data)
+static void flush_epd_write_bio(struct extent_page_data *epd)
 {
-	struct extent_page_data *epd = data;
 	if (epd->bio) {
-		submit_one_bio(WRITE, epd->bio, 0, 0);
+		if (epd->sync_io)
+			submit_one_bio(WRITE_SYNC, epd->bio, 0, 0);
+		else
+			submit_one_bio(WRITE, epd->bio, 0, 0);
 		epd->bio = NULL;
 	}
 }
 
+static noinline void flush_write_bio(void *data)
+{
+	struct extent_page_data *epd = data;
+	flush_epd_write_bio(epd);
+}
+
 int extent_write_full_page(struct extent_io_tree *tree, struct page *page,
 			  get_extent_t *get_extent,
 			  struct writeback_control *wbc)
@@ -2480,23 +2452,22 @@ int extent_write_full_page(struct extent_io_tree *tree, struct page *page,
 		.tree = tree,
 		.get_extent = get_extent,
 		.extent_locked = 0,
+		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
 	};
 	struct writeback_control wbc_writepages = {
 		.bdi		= wbc->bdi,
-		.sync_mode	= WB_SYNC_NONE,
+		.sync_mode	= wbc->sync_mode,
 		.older_than_this = NULL,
 		.nr_to_write	= 64,
 		.range_start	= page_offset(page) + PAGE_CACHE_SIZE,
 		.range_end	= (loff_t)-1,
 	};
 
-
 	ret = __extent_writepage(page, wbc, &epd);
 
 	extent_write_cache_pages(tree, mapping, &wbc_writepages,
 				 __extent_writepage, &epd, flush_write_bio);
-	if (epd.bio)
-		submit_one_bio(WRITE, epd.bio, 0, 0);
+	flush_epd_write_bio(&epd);
 	return ret;
 }
 
@@ -2515,6 +2486,7 @@ int extent_write_locked_range(struct extent_io_tree *tree, struct inode *inode,