Re: SUC[C]ESS!!! - was [Re: Progress report Re: Invoking ddrescue]

[David Wright <deblis@lionunicorn.co.uk>]

On Tue 15 Nov 2016 at 12:39:49 (+0100), Thomas Schmitt wrote:
> Richard Owlett wrote:
> > ddrescue has run to completion without _reported_ errors for all partitions
> > of the drive. I understand that does *NOT* guarantee that the files are not
> > corrupt.
> 
> A coarse test would be to mount the partitions and to let some archiver
> crawl the tree to read the content of each data file.
> 
> Create a mount point for the partition to be tested:
> 
>   mkdir /mnt/partition
> 
> Then with each of the copied partition files /mnt/my_sdb6/my_sdc* do:
> 
>   mount -o loop,ro,nodev,noexec,nosuid /mnt/my_sdb6/my_sdc1 /mnt/partition
> 
>   tar cf - /mnt/partition | wc -c
> 
>   umount /mnt/partition
> 
> If tar does not report i/o errors and wc tells a plausible byte count,
> then all might be well.
> 
> mounting has to be done as superuser, of course.
> man 8 mount paragraph "Mount options for ntfs" says that you will have
> to do the tar part as superuser, unless you assign it at mount time
> to a less powerful user id. I read from the manual something like:
>   -o loop,ro,nodev,noexec,nosuid,uid=NORMAL_USER_ID,umask=400
> but have no ntfs to test whether this makes all file readable for the
> desktop user of whom you obtained the number NORMAL_USER_ID by:
>   echo $UID

That umask looks odd. It should be set to the bits whose permission
you want to withhold. umask=022 would be a sensible choice as it
allows you to remount rw later, but still protect all files from
being written to except by the owner. (Yes, yes, I know, noone else
uses this machine...)

> Not so easy to test:
> - The file content may still be damaged on application level (e.g. zeros
>   where other data should be).
> - Important parts of the tree may be missing.

Aren't we getting carried away a bit? We have a "defective" drive
(also described as "flaky") that has been copied satisfactorily, is
reportedly "unreliable" and in "bad shape", and has "unidentified
problems" on the first partition which was running windows.
IIRC the only problem reported was that XP refused to boot.
The OP has signed his disclaimer (above) so there's no need to
be pessimistic about the data at this stage, is there.

Cheers,
David.