[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: UTF-8 problem with NTFS support?



Dave Burns <tburns <at> hawaii.edu> writes:
> 
> One directory seems to cause a problem:
> 
> ls /MyBook/paleo_enso/solar_forcing
> ls: reading directory /MyBook/paleo_enso/solar_forcing: Invalid or
> incomplete multibyte or wide character

Typically this would mean a non-UTF8 filename but please see below.
 
> Find apparently gives up entirely, returns nothing where I know there
> is something:
>  find /MyBook/paleo_enso/solar_forcing
> /MyBook/paleo_enso/solar_forcing
> 
> Also, this appears in the log:
> grep -i ntfs /var/log/messages
> Mar 23 08:03:41 c ntfs-3g[3165]: Incomplete multi-sector transfer:
> Input/output error
> Mar 23 08:03:41 c ntfs-3g[3165]: Skipping unrepresentable filename
> (inode 34473): Invalid or incomplete multibyte or wide character
> Mar 23 08:03:43 c ntfs-3g[3165]: Incomplete multi-sector transfer:
> Input/output error
> Mar 23 08:03:43 c ntfs-3g[3165]: Skipping unrepresentable filename
> (inode 34473): Invalid or incomplete multibyte or wide character

The "Incomplete multi-sector transfer" disk read error message means 
corrupted disk sectors (NTFS specific media CRC). The most common 
reason is hardware fault. Either permanent or temporary. 

CHKDSK /F/R DRIVE: should find, mark, replace bad disk sectors if 
it's permanent.

Due to some technical reason (online recovery, repair), NTFS-3G 
ignores this disk read error and trust uppers layers to online
fix or report the error. Unfortunately this happens in a very 
misleading way, as a filename corruption which can be misunderstood
as a filename characterset encoding problem.

I scheduled the misleading error message to be fixed in the next 
NTFS-3G driver. The correct error message should 

"ls: reading directory /MyBook/paleo_enso/solar_forcing: Input/Output
error". Details are at http://ntfs-3g.org/support.html#ioerror

> Googling all that leads me to a bug that was supposed to be fixed(? I
> think? bug is 'version 9', mine is ntfs-3g-2009.2.1-2):
> https://bugzilla.redhat.com/show_bug.cgi?id=467629

No. Your problem is unique and related to a hardware problem. 

Hardware problems are extremely common (and it will be even more 
so with bigger rotating and SSD drives) but they manifest in an
extremely diverse way. Thankfully NTFS has many consistency checks
and data redundancy hereby online recovery is often possible.

> The only workaround I can think of is somehow using the inode and
> mv/renaming the bogus files. But I don't know how to use mv with only
> an inode number (man page a bit terse).

Not the file but the filename in the directory index is corrupt.
CHKDSK /F /R DEVICE: should help.

Regards, Szaka

--
http://ntfs-3g.org



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]