Re: [RFC] Generic deferred file writing

[Chris Wedgwood]

On Sun, Dec 31, 2000 at 02:02:34AM +0100, Andrea Arcangeli wrote:

    Yes, we need to add one field to the in-core superblock to do
    this accounting.

How does this work for filesystems like reisefs which do tail merging
and other filesystems which might do sub-clock allocation?

We either need more than one field (or a byte counter) + lock or
perhaps a generic fields + callback to the fs itself.
    
    > estimate than just the data blocks it should not be hard to add
    > an extra callback to the filesystem.
    
    Yes, I was thinking at this callback too. Such a callback is
    nearly the only support we need from the filesystem to provide
    allocate on flush.

I think a callback is the way to go.

  --cw