Not really, log space is a circular buffer, and can be over written as soon
as the metadata itself has gone to disk. The file data itself is also not
protected against reuse. On a busy system the log can wrap several times a
second.
Steve
>
> Hmmmm...
>
> It strikes me that it would be easier to implement an undelete function
> using a journalling filesystem than not. I theorise that if you logged
> what you'd deleted, then a [relatively speaking] simple reverse run of
> the relevant log entries _might_ get you your file back.
>
> Pseudo Code would be:
>
> 1) extract from the log where foo file actually was
> 2) forcibly lock all those blocks/cylinders or whatever
> 3) put it back together
> 4) write foo_file somewhere else
>
> Naturally because of the nature of the cached writing system and such
> it's not a guaranteed thing, but it would be better than having to
> wander around the file system following inodes etc manually...
>
> DSL
> --
> "And the winner is
> InUnifiedCanadianAboriginalSyllabics"
> - Larry Wall et al in Programming Perl
|