Re: [PATCH] fix for htree corruption. Was: [2.6] Perl weirdness with ext3 and HTREE

[chrisl@vmware.com]

On Tue, Aug 19, 2003 at 03:55:06AM -0700, Andrew Morton wrote:
> chrisl@vmware.com wrote:
> 
> Could you please regenerate a full, single diff against a known kernel
> version?  That patch generated 100% rejects for me...
> 
> 
Sorry for that. Here is the patch again. It is against the 2.6-test3.
It also apply cleanly on current BK tree.

Chris


===== fs/ext3/namei.c 1.44 vs edited =====
--- 1.44/fs/ext3/namei.c	Sun Jun 29 23:49:04 2003
+++ edited/fs/ext3/namei.c	Tue Aug 19 03:28:52 2003
@@ -1304,7 +1304,8 @@
 	data1 = bh2->b_data;
 
 	/* The 0th block becomes the root, move the dirents out */
-	de = (struct ext3_dir_entry_2 *) &root->info;
+	de = &root->dotdot;
+	de = (struct ext3_dir_entry_2 *) ((char *)de + de->rec_len);
 	len = ((char *) root) + blocksize - (char *) de;
 	memcpy (data1, de, len);
 	de = (struct ext3_dir_entry_2 *) data1;
@@ -2006,9 +2007,9 @@
 	 * recovery. */
 	inode->i_size = 0;
 	ext3_orphan_add(handle, inode);
+	inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME;
 	ext3_mark_inode_dirty(handle, inode);
 	dir->i_nlink--;
-	inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME;
 	ext3_update_dx_flag(dir);
 	ext3_mark_inode_dirty(handle, dir);
 
@@ -2060,8 +2061,8 @@
 	inode->i_nlink--;
 	if (!inode->i_nlink)
 		ext3_orphan_add(handle, inode);
-	ext3_mark_inode_dirty(handle, inode);
 	inode->i_ctime = dir->i_ctime;
+	ext3_mark_inode_dirty(handle, inode);
 	retval = 0;
 
 end_unlink:
@@ -2220,7 +2221,6 @@
 			goto end_rename;
 	} else {
 		BUFFER_TRACE(new_bh, "get write access");
-		BUFFER_TRACE(new_bh, "get_write_access");
 		ext3_journal_get_write_access(handle, new_bh);
 		new_de->inode = le32_to_cpu(old_inode->i_ino);
 		if (EXT3_HAS_INCOMPAT_FEATURE(new_dir->i_sb,