All of lore.kernel.org
 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 v2 net-next 0/2] virtio_net: Expand affinity to arbitrary numbers of cpu and vq
Date: Thu,  9 Aug 2018 18:18:27 -0700	[thread overview]
Message-ID: <20180810011828.199888-1-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.

v2 changes:
Renamed "virtio_net: Make vp_set_vq_affinity() take a mask." to
"virtio: Make vp_set_vq_affinity() take a mask."

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: 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  1:18 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-08-10  1:18 Caleb Raitto [this message]
2018-08-10  1:18 ` [PATCH net-next 1/2] virtio: Make vp_set_vq_affinity() take a mask Caleb Raitto
2018-08-10  1:38   ` Gonglei (Arei)
2018-08-11 19:03 ` [PATCH v2 net-next 0/2] virtio_net: Expand affinity to arbitrary numbers of cpu and vq David Miller
2018-08-12  8:44 ` Michael S. Tsirkin

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=20180810011828.199888-1-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 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.