e2fsck is having a hard time with my ext2 filesystem after QNX has mounted it.

It appears that the group descriptor "bg_flags" field is being paved.

e2fsprogs defines the block desc as:
struct ext2_group_desc
{
   __u32 bg_block_bitmap;  /* Blocks bitmap block */
   __u32 bg_inode_bitmap;  /* Inodes bitmap block */
   __u32 bg_inode_table;      /* Inodes table block */
   __u16 bg_free_blocks_count;   /* Free blocks count */
   __u16 bg_free_inodes_count;   /* Free inodes count */
   __u16 bg_used_dirs_count;  /* Directories count */
   __u16 bg_flags;
   __u32 bg_reserved[2];
   __u16 bg_itable_unused; /* Unused inodes count */
   __u16 bg_checksum;      /* crc16(s_uuid+grouo_num+group_desc)*/
};

You folks (QNX) define it as:

typedef struct ext2_egroup
{
   uint32_t    g_block_bmp;   /* Location of block bmp */
   uint32_t    g_inode_bmp;   /* Location of inode bmp */
   uint32_t    g_inode_table; /* Location of inode table */

   uint16_t    g_free_blocks; /* Free blocks */
   uint16_t    g_free_inodes; /* Free inodes */
   uint16_t    g_used_dirs;   /* Number of inodes that are directories */
} ext2_egroup_t;

but do the I/O on a structure that is sizeof(struct ext2_group_desc):
public/sys/fs_ext2.h:#define E2FS_GROUP_DES_SIZE        32

It appears that ext2_GroupsSave+ext2_GroupMarshall effectively write stack trash into the bg_flags field.

I think all would be well if QNX preserved (or zeroed) the flags for write-back, but perhaps I am missing a trick...

dave

PS the flags seem to have to do with lazy initialization:
#define EXT2_BG_INODE_UNINIT  0x0001 /* Inode table/bitmap not initialized */
#define EXT2_BG_BLOCK_UNINIT  0x0002 /* Block bitmap not initialized */
#define EXT2_BG_INODE_ZEROED  0x0004 /* On-disk itable initialized to zero */

My assertion is that the stack trash is turning one or more of these on and confusing e2fsck.