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

[Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>]

On 07.07.2008 10:07, Anton Altaparmakov wrote:
> Hi,
>
> Thanks for the report.  Would you be able to try the below change to  
> your kernel and repeat?
>
> On 6 Jul 2008, at 18:21, Eric Sesterhenn wrote:
>   
>> 0xc030ad5c is in ntfs_read_locked_inode (fs/ntfs/time.h:95).
>> 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		 */
>> 95		ts.tv_nsec = do_div(t, 10000000) * 100;
>> 96		ts.tv_sec = t;
>> 97		return ts;
>> 98	}
>>     
>
> Please replace the ts.tv_nsec=... and ts.tv_sec=... with:
>
> 		if (t) {
> 			ts.tv_nsec = do_div(t, 10000000) * 100;
> 			ts.tv_sec = t;
> 		} else {
> 			ts.tv_nsec = 0;
> 			ts.tv_sec = 0;
> 		}
>   

AFAICS it is not the division that fails (unless a division itself can
cause an access to invalid memory), but either the write to ts or the
read from t. Dependign on compiler optimizations, the real root cause
may be related to an access to time.

Regards,
Carl-Daniel

-- 
http://www.hailfinger.org/