From: Pekka Enberg <penberg@kernel.org>
To: Ankita Garg <ankita@in.ibm.com>
Cc: linux-arm-kernel@lists.infradead.org, linux-mm@kvack.org,
linux-kernel@vger.kernel.org,
linux-pm@lists.linux-foundation.org, svaidy@linux.vnet.ibm.com,
thomas.abraham@linaro.org, Dave Hansen <dave@linux.vnet.ibm.com>,
"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>,
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>,
Matthew Garrett <mjg59@srcf.ucam.org>,
Arjan van de Ven <arjan@infradead.org>,
Christoph Lameter <cl@linux.com>
Subject: Re: [PATCH 00/10] mm: Linux VM Infrastructure to support Memory Power Management
Date: Wed, 6 Jul 2011 11:45:41 +0300 [thread overview]
Message-ID: <CAOJsxLHQP=-srK_uYYBsPb7+rUBnPZG7bzwtCd-rRaQa4ikUFg@mail.gmail.com> (raw)
In-Reply-To: <20110629130038.GA7909@in.ibm.com>
Hi Ankita,
[ I don't really know anything about memory power management but
here's my two cents since you asked for it. ]
On Wed, Jun 29, 2011 at 4:00 PM, Ankita Garg <ankita@in.ibm.com> wrote:
> I) Dynamic Power Transition
>
> The goal here is to ensure that as much as possible, on an idle system,
> the memory references do not get spread across the entire RAM, a problem
> similar to memory fragmentation. The proposed approach is as below:
>
> 1) One of the first things is to ensure that the memory allocations do
> not spill over to more number of regions. Thus the allocator needs to
> be aware of the address boundary of the different regions.
Why does the allocator need to know about address boundaries? Why
isn't it enough to make the page allocator and reclaim policies favor using
memory from lower addresses as aggressively as possible? That'd mean
we'd favor the first memory banks and could keep the remaining ones
powered off as much as possible.
IOW, why do we need to support scenarios such as this:
bank 0 bank 1 bank 2 bank3
| online | offline | online | offline |
instead of using memory compaction and possibly something like the
SLUB defragmentation patches to turn the memory map into this:
bank 0 bank 1 bank 2 bank3
| online | online | offline | offline |
> 2) At the time of allocation, before spilling over allocations to the
> next logical region, the allocator needs to make a best attempt to
> reclaim some memory from within the existing region itself first. The
> reclaim here needs to be in LRU order within the region. However, if
> it is ascertained that the reclaim would take a lot of time, like there
> are quite a fe write-backs needed, then we can spill over to the next
> memory region (just like our NUMA node allocation policy now).
I think a much more important question is what happens _after_ we've
allocated and free'd tons of memory few times. AFAICT, memory
regions don't help with that kind of fragmentation that will eventually
happen anyway.
Pekka
next prev parent reply other threads:[~2011-07-06 8:45 UTC|newest]
Thread overview: 70+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-05-27 12:31 [PATCH 00/10] mm: Linux VM Infrastructure to support Memory Power Management Ankita Garg
2011-05-27 12:31 ` [PATCH 01/10] mm: Introduce the memory regions data structure Ankita Garg
2011-05-27 15:30 ` Dave Hansen
2011-05-27 18:20 ` Vaidyanathan Srinivasan
2011-05-27 21:31 ` Dave Hansen
2011-05-29 8:16 ` Ankita Garg
2011-05-31 17:34 ` Dave Hansen
2011-06-02 8:54 ` Ankita Garg
2011-05-27 12:31 ` [PATCH 02/10] mm: Helper routines Ankita Garg
2011-05-27 12:31 ` [PATCH 03/10] mm: Init zones inside memory regions Ankita Garg
2011-05-27 12:31 ` [PATCH 04/10] mm: Refer to zones from " Ankita Garg
2011-05-27 12:31 ` [PATCH 05/10] mm: Create zonelists Ankita Garg
2011-05-27 12:31 ` [PATCH 06/10] mm: Verify zonelists Ankita Garg
2011-05-27 12:31 ` [PATCH 07/10] mm: Modify vmstat Ankita Garg
2011-05-27 12:31 ` [PATCH 08/10] mm: Modify vmscan Ankita Garg
2011-05-27 12:31 ` [PATCH 09/10] mm: Reflect memory region changes in zoneinfo Ankita Garg
2011-05-27 12:31 ` [PATCH 10/10] mm: Create memory regions at boot-up Ankita Garg
2011-05-28 14:39 ` Jean-Christophe PLAGNIOL-VILLARD
2011-05-28 7:56 ` [PATCH 00/10] mm: Linux VM Infrastructure to support Memory Power Management Andrew Morton
2011-05-28 13:16 ` Ankita Garg
2011-06-09 18:52 ` Paul E. McKenney
2011-06-10 0:51 ` Kyungmin Park
2011-06-10 15:11 ` Paul E. McKenney
2011-06-10 15:59 ` Matthew Garrett
2011-06-10 16:55 ` Paul E. McKenney
2011-06-10 17:05 ` Matthew Garrett
2011-06-10 17:19 ` Paul E. McKenney
2011-06-10 17:23 ` Matthew Garrett
2011-06-10 17:52 ` Paul E. McKenney
2011-06-10 18:08 ` Matthew Garrett
2011-06-10 18:47 ` Paul E. McKenney
2011-06-10 19:23 ` Matthew Garrett
2011-06-10 19:37 ` Paul E. McKenney
2011-06-10 20:12 ` Matthew Garrett
2011-06-11 3:02 ` Arjan van de Ven
2011-06-11 17:06 ` Paul E. McKenney
2011-06-11 17:26 ` Arjan van de Ven
2011-06-12 23:07 ` Paul E. McKenney
2011-06-13 14:28 ` Arjan van de Ven
2011-06-13 23:04 ` Paul E. McKenney
2011-06-14 8:51 ` Ankita Garg
2011-06-15 16:53 ` Ankita Garg
2011-06-18 4:08 ` Arjan van de Ven
2011-06-10 17:33 ` Ankita Garg
2011-06-11 17:08 ` Paul E. McKenney
2011-07-12 5:31 ` amit kachhap
2011-06-13 4:47 ` KAMEZAWA Hiroyuki
2011-06-16 4:20 ` Ankita Garg
2011-06-16 9:12 ` KAMEZAWA Hiroyuki
2011-06-17 15:28 ` Ankita Garg
2011-06-19 23:53 ` KAMEZAWA Hiroyuki
2011-06-16 16:04 ` Dave Hansen
2011-06-17 10:03 ` Ankita Garg
2011-06-29 13:00 ` Ankita Garg
2011-06-29 17:06 ` Dave Hansen
2011-06-29 17:42 ` Ankita Garg
2011-06-29 17:59 ` Dave Hansen
2011-06-29 18:17 ` Vaidyanathan Srinivasan
2011-06-30 4:37 ` Ankita Garg
2011-06-29 20:11 ` Andi Kleen
2011-06-30 5:11 ` Ankita Garg
2011-06-29 18:07 ` Vaidyanathan Srinivasan
2011-07-06 8:45 ` Pekka Enberg [this message]
2011-07-06 9:01 ` Pekka Enberg
2011-07-06 16:50 ` Vaidyanathan Srinivasan
2011-07-06 16:41 ` Vaidyanathan Srinivasan
2011-07-06 20:20 ` david
2011-07-07 4:54 ` Ankita Garg
2011-07-07 18:00 ` Pekka Enberg
2011-07-08 1:32 ` david
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAOJsxLHQP=-srK_uYYBsPb7+rUBnPZG7bzwtCd-rRaQa4ikUFg@mail.gmail.com' \
--to=penberg@kernel.org \
--cc=ankita@in.ibm.com \
--cc=arjan@infradead.org \
--cc=cl@linux.com \
--cc=dave@linux.vnet.ibm.com \
--cc=kamezawa.hiroyu@jp.fujitsu.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linux-pm@lists.linux-foundation.org \
--cc=mjg59@srcf.ucam.org \
--cc=paulmck@linux.vnet.ibm.com \
--cc=svaidy@linux.vnet.ibm.com \
--cc=thomas.abraham@linaro.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).