> On Wed, 28 Jun 2000, Marcelo Tosatti wrote:
>
> >The problems were not fix by the protected-balance, but by returning
> >success on any progress made at do_try_to_free_pages().
>
> Returning success on any progress makes a difference only during oom (and
> it also caused my machine to hang and I had to use SYSRQ+E). If you aren't
> oom then it doesn't change anything.
>
> >Alan is going to apply them together anyway, thats why I merged them.
>
> Alan prefers to get them separately AFIK, otherwise I would not keeping
> them separately in first place ;).
Alan,
The balance dirty state patch from Andrea against 2.2.17pre9 is located at
http://bazar.conectiva.com.br/~marcelo/patches/v2.2/2.2.17pre9/balance-dirty-andrea.patch.
As Andrea noted, with 2.2.17pre9 a process may wait for IO completion of a
dirty buffer while there are freeable pages in memory. To fix that, I've
changed try_to_free_buffers to wait on dirty buffers only if:
- do_try_to_free_pages scanning priority is <= 1, meaning that we already
scanned a big amount of pages and still have not found the needed amount
of freeable pages.
or
- the VM is very dirty.
This patch, against 2.2.17pre9+balance-dirty-andrea.patch is located at
http://bazar.conectiva.com.br/~marcelo/patches/v2.2/2.2.17pre9/wait-dirty.patch
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/