>
> > Chris, this really means that Ext3 nesting is not just a "ref count"
> > issue, but that Ext3 journaling pretty much has the api for a filter to
> > say "include this in the transaction".  By positioning the filter
> > journaling statements correctly, it can be included in the first
> > transaction fragmet (for unlink, truncate) or in the last (most other
> > ops).
> >
>
> Yes, reiserfs could do it the same way, if we start putting a transaction
> handle pointer in the task struct.
>
> But, I thought you required some way of knowing when the transaction was
> broken up during an operation?  You don't really need it for
> truncate/unlink, since those are supposed to be finished, even after a
> crash in the middle.  file writes are different though, since you
> could end
> up with a write partially done that isn't included in your log.
>
> -chris
>

Yes, that's what I thought, but at present I can participate at two very
logical points: the end of a transaction and the beginning.

InterMezzo only has these two requirements:

        - check the outcome of the FS's operation and if successful include its
data in the transaction (e.g. journaling a "mkdir")
      - prefix something in the transaction if it goes through - e.g. unlink
and truncate will go through if the beginning of the transaction actually
happens and it's time for InterMezzo to make the record.

The SnapFS (for which the sourceforge code now correctly "cows"
notify_change operations) nesting may be slightly more involved.  But likely
will fit the same shoe.

One thing I don't like about my dealing with Ext3 is that I have to know the
size of the underlying transaction.  I would rather indicate how much I add
to it.  Stephen, is this possible - probably not eh?

- Peter -

-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to [EMAIL PROTECTED]

Reply via email to