Btrfs: Data ordered fixes

* In btrfs_delete_inode, wait for ordered extents after calling
truncate_inode_pages.  This is much faster, and more correct

* Properly clear our the PageChecked bit everywhere we redirty the page.

* Change the writepage fixup handler to lock the page range and check to
see if an ordered extent had been inserted since the improperly dirtied
page was discovered

* Wait for ordered extents outside the transaction.  This isn't required
for locking rules but does improve transaction latencies

* Reduce contention on the alloc_mutex by dropping it while incrementing
refs on a node/leaf and while dropping refs on a leaf.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index c2b4a9c..0d87795 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -336,7 +336,7 @@
 		orig_end = start + len - 1;
 		wait_end = orig_end;
 	}
-
+again:
 	/* start IO across the range first to instantiate any delalloc
 	 * extents
 	 */
@@ -369,6 +369,14 @@
 			break;
 		end--;
 	}
+	if (test_range_bit(&BTRFS_I(inode)->io_tree, start, orig_end,
+			   EXTENT_ORDERED | EXTENT_DELALLOC, 0)) {
+		printk("inode %lu still ordered or delalloc after wait "
+		       "%llu %llu\n", inode->i_ino,
+		       (unsigned long long)start,
+		       (unsigned long long)orig_end);
+		goto again;
+	}
 }
 
 /*
@@ -545,7 +553,6 @@
 			sector_sums = &ordered_sum->sums;
 			for (i = 0; i < num_sectors; i++) {
 				if (sector_sums[i].offset == offset) {
-printk("find ordered sum inode %lu offset %Lu\n", inode->i_ino, offset);
 					*sum = sector_sums[i].sum;
 					ret = 0;
 					goto out;