Hi,
On Sat, Dec 16, 2000 at 07:08:02PM -0600, Russell Cattelan wrote:
> > There is a very clean way of doing this with address spaces. It's
> > something I would like to see done properly for 2.5: eliminate all
> > knowledge of buffer_heads from the VM layer. It would be pretty
> > simple to remove page->buffers completely and replace it with a
> > page->private pointer, owned by whatever address_space controlled the
> > page. Instead of trying to unmap and flush buffers on the page
> > directly, these operations would become address_space operations.
>
> Yes this is a lot of what page buf would like to do eventually.
> Have the VM system pressure page_buf for pages which would
> then be able to intelligently call the file system to free up cached pages.
> A big part of getting Delay Alloc to not completely consume all the
> system pages, is being told when it's time to start really allocating disk
> space and push pages out.
Delayed allocation is actually much easier, since it's entirely an
operation on logical page addresses, not physical ones --- by
definition you don't have any buffer_heads yet because you haven't
decided on the disk blocks. If you're just dealing with pages, not
blocks, then the address_space is the natural way of dealing with it
already.
Only the full semantics of the flush callback have been missing to
date, and with 2.4.0-test12 even that is mostly solved, since
page_launder will give you the writeback() callbacks you need to flush
things to disk when you start getting memory pressure. You can even
treat the writepage() as an advisory call.
--Stephen
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Dec 23 2000 - 21:00:21 EST