svn commit: r243913 - stable/9/sys/fs/ext2fs

[Pedro F. Giffuni]

Author: pfg
Date: Wed Dec  5 22:47:45 2012
New Revision: 243913
URL: http://svnweb.freebsd.org/changeset/base/243913

Log:
  MFC	r243641:
  Partially bring r242520 to ext2fs.
  
  When a file is first being written, the dynamic block reallocation
  (implemented by ext2_reallocblks) relocates the file's blocks
  so as to cluster them together into a contiguous set of blocks on
  the disk.
  
  When the cluster crosses the boundary into the first indirect block,
  the first indirect block is initially allocated in a position
  immediately following the last direct block.  Block reallocation
  would usually destroy locality by moving the indirect block out of
  the way to keep the data blocks contiguous.
  
  The issue was diagnosed long ago by Bruce Evans on ffs and surfaced
  on ext2fs when block reallocaton was ported. This is only a partial
  solution based on the similarities with FFS. We still require more
  review of the allocation details that vary in ext2fs.
  
  Reported by:    bde

Modified:
  stable/9/sys/fs/ext2fs/ext2_alloc.c

Modified: stable/9/sys/fs/ext2fs/ext2_alloc.c
==============================================================================
--- stable/9/sys/fs/ext2fs/ext2_alloc.c	Wed Dec  5 22:32:12 2012	(r243912)
+++ stable/9/sys/fs/ext2fs/ext2_alloc.c	Wed Dec  5 22:47:45 2012	(r243913)
@@ -197,6 +197,18 @@ ext2_reallocblks(ap)
 			panic("ext2_reallocblks: non-cluster");
 #endif
 	/*
+	 * If the cluster crosses the boundary for the first indirect
+	 * block, leave space for the indirect block. Indirect blocks
+	 * are initially laid out in a position after the last direct
+	 * block. Block reallocation would usually destroy locality by
+	 * moving the indirect block out of the way to make room for
+	 * data blocks if we didn't compensate here. We should also do
+	 * this for other indirect block boundaries, but it is only
+	 * important for the first one.
+	 */
+	if (start_lbn < NDADDR && end_lbn >= NDADDR)
+		return (ENOSPC);
+	/*
 	 * If the latest allocation is in a new cylinder group, assume that
 	 * the filesystem has decided to move and do not force it back to
 	 * the previous cylinder group.