Re: mballoc errors

[Valerie Clement <valerie.clement@bull.net>]

Eric Sandeen wrote:
> This report just came in on the fedora list...
> 
> EXT4-fs error (device dm-0): ext4_mb_generate_buddy: EXT4-fs: group 717:
> 23410 blocks in bitmap, 23411 in gd
> 
> EXT4-fs error (device dm-0): ext4_mb_generate_buddy: EXT4-fs: group 721:
> 19309 blocks in bitmap, 19333 in gd
> 
> has anyone else seen this problem in testing?  I guess it means
> freespace accounting got out of sync...?
> 
I've got no problem running tests on filesystems with a 4K blocksize.
But since I work on FS with 1K blocksize, I've got the following errors
while running fsstress:

EXT4-fs error (device sdc1): ext4_ext_find_extent: bad header in inode #4407333: 
invalid magic - magic dada, entries 56026, max 56026(0), depth 56026(0)

or

EXT4-fs error (device sdc1): ext4_valid_block_bitmap: Invalid block bitmap - 
block_group = 6767, block = 55435267

I already found a bug in the patch ext4-fallocate-full-fs-ENOSPC-handling.
In the function, ext4_ext_zeroout()
       /* convert ee_pblock in 512 byte sector */
       ee_pblock = ee_pblock << (blkbits >> 9);
should be
       ee_pblock = ee_pblock << (blkbits - 9);

But after doing this change, the problems occur more quickly.
I'm not sure that the function ext4_ext_zeroout() is correct if
blocksize < pagesize.
Aneesh, could you check that please?
Thanks,
   Valerie


How I reproduce the problem:
# mkfs.ext3 -I256 -E test_fs -b 1024 /dev/sdc1
# mount -t ext4dev /dev/sdc1 /mnt/test
# fsstress -d /mnt/test -n1000 -p1000