2.4 todo list update

From: Rik van Riel (riel@conectiva.com.br)
Date: Fri Jan 05 2001 - 09:58:34 EST


Hi Ted,

in the last few weeks quite a few of the bugs listed on your
(excellent) http://linux24.sourceforge.net/ have been fixed.

Here is a list of the VM bugs that are on your list and can
be moved to the "fixed" category:

* truncate->invalidate_inode_pages removes mapping information from
  mapped pages which may be dirty; sync_pte -> crash. (CRITICAL)

        fixed by Linus and Al

* VM: raw I/O data loss (raw IO may arrive in a page which afer it
  is unammped from a process) (CRITICAL)

        fixed by Linus, now page_launder() does the IO
        and try_to_swap_out() only unmaps the pte
 
* VM: Fix the highmem deadlock, where the swapper cannot create low
  memory bounce buffers OR swap out low memory because it has
  consumed all resources {CRITICAL}

        this was never an issue, the pagecache has been
        highmem safe for a long time and the whole bounce
        buffer creation has been removed

* VM: page->mapping->flush() callback in page_lauder() for easier
  integration with journaling filesystem and maybe the network
  filesystems

        page->mapping->writepage(), used from page_launder()
        ... now ext3, reiserfs, xfs and others need to make
        their own ->writepage() function
        ... some semantics are still being discussed, but it's
        mostly ready

* VM: maybe rebalance the swapper a bit... we do page aging now so
  maybe refill_inactive_scan() / shm_swap() and swap_out() need to
  be rebalanced a bit

        moving shm into the page cache permanently and doing
        the page down aging from refill_inactive_scan() seems
        to have fixed most of this
        ... low priority, but may still have some room for
        improvement (consider it fixed)

The following bugs _could_ be fixed ... I'm not 100% certain
but they're probably gone (could somebody confirm/deny?):

* mm->rss is modified in some places without holding the
  page_table_lock

* VFS?VM - mmap/write deadlock (demo code seems to show lock
  is there)

The "probably post 2.4" category VM issues remain ... maybe
we want to add the following 2 items though:

* VM: experiment with different forms of page aging ... maybe
  different aging rates for pages of different ages

* VM: RSS ulimit enforcement (trivial)

regards,

Rik

--
Virtual memory is like a game you can't win;
However, without VM there's truly nothing to loose...

http://www.surriel.com/ http://www.conectiva.com/ http://distro.conectiva.com.br/

- 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 : Sun Jan 07 2001 - 21:00:22 EST