On Thu, Dec 22, 2011 at 05:30:22PM +0900, Akira Fujita wrote:

If we run the mke2fs with the -S option and the uninit_bg feature
simultaneously, the mke2fs marks blockgroups as uninitialized.
The e2fsck which run immediately after the mke2fs
removes all of the files.

To avoid this, the patch prohibits user from
setting the -S option and the uninit_bg feature simultaneously.

This is not the best fix.  The best fix is to clear the itable_unused
fields in the block group descriptors if mke2fs -S is set.  See below
for what I have in my tree.

					- Ted

commit 9b6a158524fe82202bef6d0d8a101b47e6c02b64
Author: Theodore Ts'o [off-list ref]
Date:   Thu Feb 16 23:16:34 2012 -0500

    mke2fs: allow file systems w/ uninit_bg to be recovered with mke2fs -S
    
    The command mke2fs -S is used as a last ditch recovery command to
    write new superblock and block group descriptors, but _not_ to destroy
    the inode table in hopes of recovering from a badly corrupted file
    system.  If the uninit_bg feature is enabled, we need to make sure to
    clear the unused inodes count field in the block group descriptors or
    else e2fsck -fy will leave the file system completely empty.
    
    Thanks to Akira Fujita for reporting this problem.
    
    Signed-off-by: "Theodore Ts'o" [off-list ref]

diff --git a/misc/mke2fs.c b/misc/mke2fs.c
index 08789c4..c70c1b4 100644
--- a/misc/mke2fs.c
+++ b/misc/mke2fs.c

@@ -2434,6 +2434,17 @@ int main (int argc, char *argv[])
 	if (super_only) {
 		fs->super->s_state |= EXT2_ERROR_FS;
 		fs->flags &= ~(EXT2_FLAG_IB_DIRTY|EXT2_FLAG_BB_DIRTY);
+		/* 
+		 * The command "mke2fs -S" is used to recover
+		 * corrupted file systems, so do not mark any of the
+		 * inodes as unused; we want e2fsck to consider all
+		 * inodes as potentially containing recoverable data.
+		 */
+		if (fs->super->s_feature_ro_compat &
+		    EXT4_FEATURE_RO_COMPAT_GDT_CSUM) {
+			for (i = 1; i < fs->group_desc_count; i++)
+				ext2fs_bg_itable_unused_set(fs, i, 0);
+		}
 	} else {
 		/* rsv must be a power of two (64kB is MD RAID sb alignment) */
 		blk64_t rsv = 65536 / fs->blocksize;