linux-crypto.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Caleb Raitto <caleb.raitto@gmail.com>
To: herbert@gondor.apana.org.au, mst@redhat.com, davem@davemloft.net
Cc: arei.gonglei@huawei.com, jasowang@redhat.com,
	netdev@vger.kernel.org, linux-crypto@vger.kernel.org,
	Caleb Raitto <caraitto@google.com>
Subject: [PATCH net-next 0/2] virtio_net: Expand affinity to arbitrary numbers of cpu and vq
Date: Thu,  9 Aug 2018 17:28:38 -0700	[thread overview]
Message-ID: <20180810002840.186635-1-caleb.raitto@gmail.com> (raw)

From: Caleb Raitto <caraitto@google.com>

Virtio-net tries to pin each virtual queue rx and tx interrupt to a cpu if
there are as many queues as cpus.

Expand this heuristic to configure a reasonable affinity setting also
when the number of cpus != the number of virtual queues.

Patch 1 allows vqs to take an affinity mask with more than 1 cpu.
Patch 2 generalizes the algorithm in virtnet_set_affinity beyond
the case where #cpus == #vqs.

Tested:

# 16 vCPU, 16 queue pairs, Debian 9 recent net-next kernel, GCE

# Disable GCE scripts setting affinities during startup.
#
# Add the following to
# /etc/default/instance_configs.cfg.template and reboot:
[InstanceSetup]
set_multiqueue = false

$ cd /proc/irq
$ for i in `seq 24 60` ; do sudo grep ".*" $i/smp_affinity_list;  done
0-15
0
0
1
1
2
2
3
3
4
4
5
5
6
6
7
7
8
8
9
9
10
10
11
11
12
12
13
13
14
14
15
15
0-15
0-15
0-15
0-15

$ cd /sys/class/net/eth0/queues/
$ for i in `seq 0 15` ; do sudo grep ".*" tx-$i/xps_cpus; done
0001
0002
0004
0008
0010
0020
0040
0080
0100
0200
0400
0800
1000
2000
4000
8000

# 16 vCPU, 15 queue pairs
$ sudo ethtool -L eth0 combined 15

$ cd /proc/irq
$ for i in `seq 24 60` ; do sudo grep ".*" $i/smp_affinity_list;  done
0-15
0-1
0-1
2
2
3
3
4
4
5
5
6
6
7
7
8
8
9
9
10
10
11
11
12
12
13
13
14
14
15
15
15
15
0-15
0-15
0-15
0-15

$ cd /sys/class/net/eth0/queues/
$ for i in `seq 0 14` ; do sudo grep ".*" tx-$i/xps_cpus; done
0003
0004
0008
0010
0020
0040
0080
0100
0200
0400
0800
1000
2000
4000
8000

# 16 vCPU, 8 queue pairs
$ sudo ethtool -L eth0 combined 8

$ cd /proc/irq
$ for i in `seq 24 60` ; do sudo grep ".*" $i/smp_affinity_list;  done
0-15
0-1
0-1
2-3
2-3
4-5
4-5
6-7
6-7
8-9
8-9
10-11
10-11
12-13
12-13
14-15
14-15
9
9
10
10
11
11
12
12
13
13
14
14
15
15
15
15
0-15
0-15
0-15
0-15

$ cd /sys/class/net/eth0/queues/
$ for i in `seq 0 7` ; do sudo grep ".*" tx-$i/xps_cpus; done
0003
000c
0030
00c0
0300
0c00
3000
c000

# 15 vCPU, 16 queue pairs
$ sudo ethtool -L eth0 combined 16
$ sudo sh -c "echo 0 > /sys/devices/system/cpu/cpu15/online"

$ cd /proc/irq
$ for i in `seq 24 60` ; do sudo grep ".*" $i/smp_affinity_list;  done
0-15
0
0
1
1
2
2
3
3
4
4
5
5
6
6
7
7
8
8
9
9
10
10
11
11
12
12
13
13
14
14
0
0
0-15
0-15
0-15
0-15

$ cd /sys/class/net/eth0/queues/
$ for i in `seq 0 15` ; do sudo grep ".*" tx-$i/xps_cpus; done
0001
0002
0004
0008
0010
0020
0040
0080
0100
0200
0400
0800
1000
2000
4000
0001

# 8 vCPU, 16 queue pairs
$ for i in `seq 8 15`; \
do sudo sh -c "echo 0 > /sys/devices/system/cpu/cpu$i/online"; done

$ cd /proc/irq
$ for i in `seq 24 60` ; do sudo grep ".*" $i/smp_affinity_list;  done
0-15
0
0
1
1
2
2
3
3
4
4
5
5
6
6
7
7
0
0
1
1
2
2
3
3
4
4
5
5
6
6
7
7
0-15
0-15
0-15
0-15

$ cd /sys/class/net/eth0/queues/
$ for i in `seq 0 15` ; do sudo grep ".*" tx-$i/xps_cpus; done
0001
0002
0004
0008
0010
0020
0040
0080
0001
0002
0004
0008
0010
0020
0040
0080

Caleb Raitto (2):
  virtio_net: Make vp_set_vq_affinity() take a mask.
  virtio_net: Stripe queue affinities across cores.

 drivers/crypto/virtio/virtio_crypto_core.c |  4 +-
 drivers/net/virtio_net.c                   | 46 ++++++++++++++--------
 drivers/virtio/virtio_pci_common.c         |  7 ++--
 drivers/virtio/virtio_pci_common.h         |  2 +-
 include/linux/virtio_config.h              |  7 ++--
 5 files changed, 39 insertions(+), 27 deletions(-)

-- 
2.18.0.597.ga71716f1ad-goog

             reply	other threads:[~2018-08-10  0:28 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-08-10  0:28 Caleb Raitto [this message]
2018-08-10  0:28 ` [PATCH net-next 1/2] virtio_net: Make vp_set_vq_affinity() take a mask Caleb Raitto
2018-08-10  0:48   ` Gonglei (Arei)
2018-08-10  0:28 ` [PATCH net-next 2/2] virtio_net: Stripe queue affinities across cores Caleb Raitto

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=20180810002840.186635-1-caleb.raitto@gmail.com \
    --to=caleb.raitto@gmail.com \
    --cc=arei.gonglei@huawei.com \
    --cc=caraitto@google.com \
    --cc=davem@davemloft.net \
    --cc=herbert@gondor.apana.org.au \
    --cc=jasowang@redhat.com \
    --cc=linux-crypto@vger.kernel.org \
    --cc=mst@redhat.com \
    --cc=netdev@vger.kernel.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).