On Wed, 4 Jun 2008 14:03:29 +0900 KAMEZAWA Hiroyuki wrote:

quoted text
> ---
>  Documentation/controllers/memory.txt |   27 +++++-
>  mm/memcontrol.c                      |  156 ++++++++++++++++++++++++++++++++++-
>  2 files changed, 178 insertions(+), 5 deletions(-)
> 

quoted text
> Index: temp-2.6.26-rc2-mm1/Documentation/controllers/memory.txt
> ===================================================================
> --- temp-2.6.26-rc2-mm1.orig/Documentation/controllers/memory.txt
> +++ temp-2.6.26-rc2-mm1/Documentation/controllers/memory.txt
> @@ -237,12 +237,37 @@ cgroup might have some charge associated
>  tasks have migrated away from it. Such charges are automatically dropped at
>  rmdir() if there are no tasks.
>  
> -5. TODO
> +5. Hierarchy Model
> +  the kernel supports following kinds of hierarchy models.

     The kernel supports the following kinds ....

quoted text
> +  (your middle-ware may support others based on this.)

     (Your
quoted text
> +
> +  5-a. Independent Hierarchy
> +  There are no relationship between any cgroups, even among a parent and

           is

quoted text
> +  children. This is the default mode. To use this hierarchy, write 0
> +  to root cgroup's memory.hierarchy_model
> +  echo 0 > .../memory.hierarchy_model.
> +
> +  5-b. Hardwall Hierarchy.
> +  The resource has to be moved from the parent to the child before use it.

                                                                      using it.

quoted text
> +  When a child's limit is set to 'val', val of the resource is moved from
> +  the parent to the child. the parent's usage += val.

                              The parent's usage is incremented by val.
(if that's what you mean)

quoted text
> +  The amount of children's usage is reported by the file
> +
> +  - memory.assigned_to_child
> +
> +  This policy doesn't provide sophisticated automatic resource balancing in
> +  the kernel. But this is very good for strict resource isolation. Users

         kernel, but this ...

quoted text
> +  can get high predictability of behavior of applications if this is used
> +  under proper environments.
> +
> +
> +6. TODO
>  
>  1. Add support for accounting huge pages (as a separate controller)
>  2. Make per-cgroup scanner reclaim not-shared pages first
>  3. Teach controller to account for shared-pages
>  4. Start reclamation when the limit is lowered
> +   (this is already done in Hardwall Hierarchy)
>  5. Start reclamation in the background when the limit is
>     not yet hit but the usage is getting closer
> --

---
~Randy
'"Daemon' is an old piece of jargon from the UNIX operating system,
where it referred to a piece of low-level utility software, a
fundamental part of the operating system."
--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/