diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index e616f8872e69bb0cf3b9a3f369ba49eeca57f2de..12394a90d60fb7115d5b8feee0e7df701f9e743d 100644
--- a/fs/btrfs/btrfs_inode.h
+++ b/fs/btrfs/btrfs_inode.h
@@ -37,6 +37,7 @@
 #define BTRFS_INODE_IN_DEFRAG			3
 #define BTRFS_INODE_DELALLOC_META_RESERVED	4
 #define BTRFS_INODE_HAS_ORPHAN_ITEM		5
+#define BTRFS_INODE_HAS_ASYNC_EXTENT		6
 
 /* in memory btrfs inode */
 struct btrfs_inode {
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 06075043da5dce7968f35bba88842985868a418f..7a090fb4eb988a03d6c5b827ecffffca2a64ac50 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -1398,20 +1398,23 @@ static int run_delalloc_range(struct inode *inode, struct page *locked_page,
 	int ret;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 
-	if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW)
+	if (BTRFS_I(inode)->flags & BTRFS_INODE_NODATACOW) {
 		ret = run_delalloc_nocow(inode, locked_page, start, end,
 					 page_started, 1, nr_written);
-	else if (BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC)
+	} else if (BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC) {
 		ret = run_delalloc_nocow(inode, locked_page, start, end,
 					 page_started, 0, nr_written);
-	else if (!btrfs_test_opt(root, COMPRESS) &&
-		 !(BTRFS_I(inode)->force_compress) &&
-		 !(BTRFS_I(inode)->flags & BTRFS_INODE_COMPRESS))
+	} else if (!btrfs_test_opt(root, COMPRESS) &&
+		   !(BTRFS_I(inode)->force_compress) &&
+		   !(BTRFS_I(inode)->flags & BTRFS_INODE_COMPRESS)) {
 		ret = cow_file_range(inode, locked_page, start, end,
 				      page_started, nr_written, 1);
-	else
+	} else {
+		set_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
+			&BTRFS_I(inode)->runtime_flags);
 		ret = cow_file_range_async(inode, locked_page, start, end,
 					   page_started, nr_written);
+	}
 	return ret;
 }
 
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index 9e138cdc36c5eb7d66bf80dfc37829878eeaa6e2..643335a4fe3c6699a894c19d01e79bed9ef631c7 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -627,7 +627,27 @@ void btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len)
 	/* start IO across the range first to instantiate any delalloc
 	 * extents
 	 */
-	filemap_write_and_wait_range(inode->i_mapping, start, orig_end);
+	filemap_fdatawrite_range(inode->i_mapping, start, orig_end);
+
+	/*
+	 * So with compression we will find and lock a dirty page and clear the
+	 * first one as dirty, setup an async extent, and immediately return
+	 * with the entire range locked but with nobody actually marked with
+	 * writeback.  So we can't just filemap_write_and_wait_range() and
+	 * expect it to work since it will just kick off a thread to do the
+	 * actual work.  So we need to call filemap_fdatawrite_range _again_
+	 * since it will wait on the page lock, which won't be unlocked until
+	 * after the pages have been marked as writeback and so we're good to go
+	 * from there.  We have to do this otherwise we'll miss the ordered
+	 * extents and that results in badness.  Please Josef, do not think you
+	 * know better and pull this out at some point in the future, it is
+	 * right and you are wrong.
+	 */
+	if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
+		     &BTRFS_I(inode)->runtime_flags))
+		filemap_fdatawrite_range(inode->i_mapping, start, orig_end);
+
+	filemap_fdatawait_range(inode->i_mapping, start, orig_end);
 
 	end = orig_end;
 	found = 0;