[Devel] Re: [Lhms-devel] [patch00/05]: Containers(V2)- Introduction

[KAMEZAWA Hiroyuki]

On Wed, 20 Sep 2006 16:31:22 -0700 (PDT)
Christoph Lameter <clameter at sgi.com> wrote:

> On Wed, 20 Sep 2006, Rohit Seth wrote:
> 
> > Absolutely.  Since these containers are not (hard) partitioning the
> > memory in any way so it is easy to change the limits (effectively
> > reducing and increasing the memory limits for tasks belonging to
> > containers).  As you said, memory hot-un-plug is important and it is
> > non-trivial amount of work.
> 
> Maybe the hotplug guys want to contribute to the discussion?
> 
Ah, I'm reading threads with interest.

I think this discussion is about using fake nodes ('struct pgdat')
to divide system's memory into some chunks. Your thought is that
for resizing/adding/removing fake pgdat, memory-hot-plug codes may be useful. 
correct ?

Now, memory-hotplug manages all memory by 'section' and allows adding/(removing)
section to pgdat.
Does this section-size handling meet container people's requirement ?
And do we need freeing page when pgdat is removed ?

I think at least SPARSEMEM is useful for fake nodes because 'struct page'
are not tied to pgdat. (DISCONTIGMEM uses node_start_pfn. SPARSEMEM not.)

-Kame