On Wed, 20 Jul 2005, Andrew Clausen wrote:
> On Wed, Jul 20, 2005 at 10:17:29AM +0200, [EMAIL PROTECTED] wrote:
> > Resizing (esp. shrinking) a file system has the same issues.
>
> No, the resizers have been designed to have at worst a very small
> chance of data loss in the case of power failure.
>
> Why do you think it would be an issue? Most resizing algorithms involve
> writing to free space, or updating block references where the source and
> destination blocks have the same data.
>
> In the case of the FAT resizer, it is possible to have both the
> old and new filesystems being valid simultaneously! (Provided the
> FATs, etc. don't overlap)
If you update block references then they can't cover both filesystems at
the same time. Consequently they can't be valid simultaneously.
However if you rewrite everything on the free space then it should be ok,
unless the kernel, the disk or its controller still cached reordered data
when the power loss happened, after user exited Parted.
Szaka
_______________________________________________
Bug-parted mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/bug-parted
