Re: [Linux-NTFS-Dev] Oops with corrupted NTFS image

["Vegard Nossum" <vegard.nossum@gmail.com>]

On Mon, Jul 7, 2008 at 2:24 PM, Eric Sesterhenn <snakebyte@gmx.de> wrote:
> (gdb) l *(ntfs_read_locked_inode+0x16c)
> 0xc030adbc is in ntfs_read_locked_inode (fs/ntfs/time.h:90).
> 85      static inline struct timespec ntfs2utc(const sle64 time)
> 86      {
> 87              struct timespec ts;
> 88
> 89              /* Subtract the NTFS time offset. */
> 90              u64 t = (u64)(sle64_to_cpu(time) - NTFS_TIME_OFFSET);
> 91              /*
> 92               * Convert the time to 1-second intervals and the remainder to
> 93               * 1-nano-second intervals.
> 94               */
> (gdb) quit
>
> Not sure why this happens. I checked out a fresh git tree to
> make sure my tree isnt broken or something. Might gcc be bogus
> or the debug information and the bug happens in reality somewhere else?

Are you sure you didn't recompile/relink vmlinux after getting the
error? If not, maybe it's gdb which gets confused (somehow) by the
inlining.

Your 'Code:' line decodes to these instructions:

   0:   8b 58 08                mov    0x8(%eax),%ebx
   3:   8b 70 0c                mov    0xc(%eax),%esi

And I find this in my own compiled vmlinux at:

c025bcc1:       8b 58 08                mov    0x8(%eax),%ebx
c025bcc4:       8b 70 0c                mov    0xc(%eax),%esi

which is at...

    $ addr2line -e vmlinux -i c025bcc1
    fs/ntfs/inode.c:670

which is...

        vi->i_mtime = ntfs2utc(si->last_data_change_time);

which is probably what is causing the NULL pointer dereference.


Vegard

-- 
"The animistic metaphor of the bug that maliciously sneaked in while
the programmer was not looking is intellectually dishonest as it
disguises that the error is the programmer's own creation."
	-- E. W. Dijkstra, EWD1036