From: Dario Faggioli <raistlin@linux.it>
To: xen-devel@lists.xen.org
Cc: Andre Przywara <andre.przywara@amd.com>,
Ian Campbell <Ian.Campbell@citrix.com>,
Stefano Stabellini <Stefano.Stabellini@eu.citrix.com>,
George Dunlap <george.dunlap@eu.citrix.com>,
Juergen Gross <juergen.gross@ts.fujitsu.com>,
Ian Jackson <Ian.Jackson@eu.citrix.com>
Subject: [PATCH 00 of 11] Automatic NUMA placement for xl
Date: Thu, 31 May 2012 14:11:05 +0200 [thread overview]
Message-ID: <patchbomb.1338466265@Solace> (raw)
Hello Everyone,
This posting follows the RFC subject-ed "Automatically place guest on host's
NUMA nodes with xl" (patchbomb.1334150267@Solace , 4/11/2012),
_however_
the key bits have been pretty much rewritten, aiming at 4.2 inclusion. In fact,
automatic NUMA placement is something xm/xend does, and xl should include
something like that, for the sake of feature parity. For the same reason, many
of the functionalities that were part of the RFC are being kept out (e.g., NUMA
aware scheduling in the hypervisor).
Of course, while at it, all the review comments the RFC got, which involved
patches that are part of this series too, have been considered and applied.
So, here it is, in some more details, what this is all about:
1/11 libxc: abstract xenctl_cpumap to just xenctl_map
2/11 libxl: abstract libxl_cpumap to just libxl_map
3/11 libxc, libxl: introduce xc_nodemap_t and libxl_nodemap
4/11 libxl: expand the libxl_{cpu,node}map API a bit
5/11 libxl: add a new Array type to the IDL
6/11 libxl: introduce libxl_get_numainfo()
Introduce the data structures, calls and infrastructure needed for retrieving
all the information about the NUMA-ness of the system and deal with them at the
toolstack level.
7/11 xen: enhance dump_numa output
8/11 xl: add more NUMA information to `xl info -n'
Are just output and debugging enhancements.
9/11 libxl, xl: enable automatic placement of guests on NUMA nodes
10/11 libxl, xl: heuristics for reordering NUMA placement candidates
Are where the automatic placement is really implemented. Searching for suitable
placement candidates and manipulating these candidates a bit is being done
within libxl, by adding a couple of API calls. This was not like this in the
previous version, but I really think it is something other toolstack wanting to
build on top of libxl could benefit of this, instead of re-implementing
everything from scratch. On the other hand, The key bits of the algorithm
(i.e., which candidate to chose, how, whether or not to mangle it a bit, etc.)
are implemented at the xl level.
11/11 Some automatic NUMA placement documentation
Puts some more details about the implementation and the usage of the new
feature directly in the tree.
Thanks a lot and Regards,
Dario
--
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://retis.sssup.it/people/faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)
next reply other threads:[~2012-05-31 12:11 UTC|newest]
Thread overview: 64+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-05-31 12:11 Dario Faggioli [this message]
2012-05-31 12:11 ` [PATCH 01 of 11] libxc: abstract xenctl_cpumap to just xenctl_map Dario Faggioli
2012-05-31 14:10 ` Ian Jackson
2012-05-31 14:47 ` George Dunlap
2012-05-31 14:55 ` George Dunlap
2012-05-31 15:01 ` George Dunlap
2012-05-31 15:08 ` Dario Faggioli
2012-05-31 12:11 ` [PATCH 02 of 11] libxl: abstract libxl_cpumap to just libxl_map Dario Faggioli
2012-05-31 14:11 ` Ian Jackson
2012-05-31 14:54 ` George Dunlap
2012-05-31 12:11 ` [PATCH 03 of 11] libxc, libxl: introduce xc_nodemap_t and libxl_nodemap Dario Faggioli
2012-05-31 14:12 ` Ian Jackson
2012-05-31 14:32 ` Dario Faggioli
2012-05-31 15:41 ` George Dunlap
2012-05-31 16:09 ` Dario Faggioli
2012-05-31 12:11 ` [PATCH 04 of 11] libxl: expand the libxl_{cpu, node}map API a bit Dario Faggioli
2012-05-31 14:13 ` Ian Jackson
2012-05-31 14:30 ` Dario Faggioli
2012-06-08 13:54 ` Ian Jackson
2012-05-31 15:51 ` George Dunlap
2012-05-31 16:01 ` Dario Faggioli
2012-05-31 12:11 ` [PATCH 05 of 11] libxl: add a new Array type to the IDL Dario Faggioli
2012-05-31 15:54 ` George Dunlap
2012-06-08 14:03 ` Ian Jackson
2012-06-08 15:14 ` Dario Faggioli
2012-06-08 15:17 ` Ian Jackson
2012-06-08 15:37 ` Ian Jackson
2012-06-08 15:52 ` Dario Faggioli
2012-06-08 15:57 ` Ian Jackson
2012-06-12 9:02 ` Ian Campbell
2012-06-13 6:59 ` Dario Faggioli
2012-06-18 12:06 ` Dario Faggioli
2012-06-21 14:32 ` Dario Faggioli
2012-06-21 14:35 ` Ian Campbell
2012-06-21 14:35 ` Dario Faggioli
2012-06-26 16:28 ` Ian Jackson
2012-06-26 16:30 ` Ian Campbell
2012-06-26 16:58 ` Dario Faggioli
2012-05-31 12:11 ` [PATCH 06 of 11] libxl: introduce libxl_get_numainfo() Dario Faggioli
2012-05-31 14:22 ` Ian Jackson
2012-05-31 14:57 ` Dario Faggioli
2012-06-01 16:44 ` George Dunlap
2012-06-01 16:58 ` Ian Jackson
2012-05-31 12:11 ` [PATCH 07 of 11] xen: enhance dump_numa output Dario Faggioli
2012-05-31 14:23 ` Ian Jackson
2012-05-31 14:35 ` Dario Faggioli
2012-05-31 12:11 ` [PATCH 08 of 11] xl: add more NUMA information to `xl info -n' Dario Faggioli
2012-05-31 14:24 ` Ian Jackson
2012-05-31 14:40 ` Dario Faggioli
2012-06-01 16:56 ` George Dunlap
2012-05-31 12:11 ` [PATCH 09 of 11] libxl, xl: enable automatic placement of guests on NUMA nodes Dario Faggioli
2012-05-31 15:02 ` Ian Jackson
2012-05-31 16:27 ` Dario Faggioli
2012-05-31 16:42 ` Ian Campbell
2012-05-31 16:56 ` Dario Faggioli
2012-05-31 12:11 ` [PATCH 10 of 11] libxl, xl: heuristics for reordering NUMA placement candidates Dario Faggioli
2012-05-31 12:11 ` [PATCH 11 of 11] Some automatic NUMA placement documentation Dario Faggioli
2012-05-31 15:08 ` Ian Jackson
2012-05-31 15:41 ` Dario Faggioli
2012-06-08 14:01 ` Ian Jackson
2012-06-08 14:03 ` George Dunlap
2012-06-08 14:06 ` Ian Jackson
2012-06-08 14:35 ` Dario Faggioli
2012-06-08 15:19 ` Dario Faggioli
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=patchbomb.1338466265@Solace \
--to=raistlin@linux.it \
--cc=Ian.Campbell@citrix.com \
--cc=Ian.Jackson@eu.citrix.com \
--cc=Stefano.Stabellini@eu.citrix.com \
--cc=andre.przywara@amd.com \
--cc=george.dunlap@eu.citrix.com \
--cc=juergen.gross@ts.fujitsu.com \
--cc=xen-devel@lists.xen.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.