On Tue, Sep 20, 2016 at 09:04:56AM +0200, Maxime Coquelin wrote:
> This python script calls 'query-cpus' QMP command to retrieve
> vCPUs thread IDs.
> Thread IDs are then used by taskset to pin vCPUs to physical
> CPUs passed in command line.
>
> In case more vCPUs are present than the number of CPUs assigned
> in command line, multiple vCPUs get pinned to physical CPUs.
>
> Signed-off-by: Maxime Coquelin <maxime.coque...@redhat.com>
> ---
>
> This script is not for production, where libvirt should be used.
> However, I find it useful during development, so I propose to share
> it as it might interest other people.
>
> Note: it seems that tcp socket support is broken with QMP class.
> The bug has already been reported and patch proposed[0].
> So it has only been tested with UNIX socket.
>
> [0]:
> <b9d0e4270773564788dc15a2db8e407119648956.1471108930.git.c...@resonantcircuitlabs.com>
>
> scripts/qmp/qmp-vcpu-pin | 39 +++++++++++++++++++++++++++++++++++++++
> 1 file changed, 39 insertions(+)
> create mode 100755 scripts/qmp/qmp-vcpu-pin
>
> diff --git a/scripts/qmp/qmp-vcpu-pin b/scripts/qmp/qmp-vcpu-pin
> new file mode 100755
> index 0000000..7bf96f7
> --- /dev/null
> +++ b/scripts/qmp/qmp-vcpu-pin
> @@ -0,0 +1,39 @@
> +#!/usr/bin/python
> +# QEMU vCPU pinning tool
> +#
> +# Copyright (C) 2016 Red Hat Inc.
> +#
> +# Authors:
> +# Maxime Coquelin <maxime.coque...@redhat.com>
> +#
> +# This work is licensed under the terms of the GNU GPL, version 2. See
> +# the COPYING file in the top-level directory
> +import argparse
> +import json
> +import os
> +
> +from subprocess import call
> +from qmp import QEMUMonitorProtocol
> +
> +parser = argparse.ArgumentParser(description='Pin QEMU vCPUs to physical
> CPUs')
> +parser.add_argument('-s', '--server', type=str, required=True,
> + help='QMP server path or address:port')
> +parser.add_argument('cpu', type=int, nargs='+',
> + help='Physical CPUs IDs')
> +args = parser.parse_args()
> +
> +devnull = open(os.devnull, 'w')
> +
> +srv = QEMUMonitorProtocol(args.server)
> +srv.connect()
> +
> +for vcpu in srv.command('query-cpus'):
> + vcpuid = vcpu['CPU']
> + tid = vcpu['thread_id']
> + cpuid = args.cpu[vcpuid % len(args.cpu)]
> + print 'Pin vCPU {} (tid {}) to physical CPU {}'.format(vcpuid, tid,
> cpuid)
> + try:
> + call(['taskset', '-pc', str(cpuid), str(tid)], stdout=devnull)
> + except OSError:
> + print 'Failed to pin vCPU{} to CPU{}'.format(vcpuid, cpuid)
> +
> --
> 2.7.4
Reviewed-by: Peter Xu <pet...@redhat.com>
Tested-by: Peter Xu <pet...@redhat.com>
Nit: for TCG, looks like we are using a single thread for all vCPUs.
In that case, I'd suggest we make sure KVM is enabled before running
this script, otherwise raise error (It won't make sense we pin the
same TID for lots of times). I believe that's a prerequisite for this
one.
Thanks,
-- peterx
