]> git.ipfire.org Git - thirdparty/libvirt.git/commit
formatdomain.html.in: fix 'sockets' info in topology element
authorDaniel Henrique Barboza <danielhb413@gmail.com>
Wed, 1 Apr 2020 10:56:37 +0000 (07:56 -0300)
committerMichal Privoznik <mprivozn@redhat.com>
Mon, 6 Apr 2020 13:56:14 +0000 (15:56 +0200)
commit0895a0e75d13874254218e16dc66dcad673671d3
treef9404e09c149f2b57c9dc2b02e2aea971f7e9c0e
parent08071ec0f113bb1fe8dcc263cb6bf87529e8b76b
formatdomain.html.in: fix 'sockets' info in topology element

In the 'topology' element it is mentioned, regarding the sockets
value, "They refer to the total number of CPU sockets".

This is not accurate. What we're doing is calculating the number
of sockets per NUMA node, which can be checked in the current
implementation of virHostCPUGetInfoPopulateLinux(). Calculating
the total number of sockets would break the topology sanity
check nodes*sockets*cores*threads=online_cpus.

This documentation fix is important to avoid user confusion when
seeing the output of 'virsh capabilities' and expecting it to be
equal to the output of 'lscpu'. E.g in a Power 9 host this 'lscpu'
output:

Architecture:        ppc64le
Byte Order:          Little Endian
CPU(s):              160
On-line CPU(s) list: 0-159
Thread(s) per core:  4
Core(s) per socket:  20
Socket(s):           2
NUMA node(s):        2
Model:               2.2 (pvr 004e 1202)
Model name:          POWER9, altivec supported

And this XML output from virsh capabilities:

    <cpu>
      <arch>ppc64le</arch>
      <model>POWER9</model>
      <vendor>IBM</vendor>
      <topology sockets='1' dies='1' cores='20' threads='4'/>
      (...)
    </cpu>

Both are correct, as long as we mention in the Libvirt documentation
that 'sockets' in the topology element represents the number of sockets
per NUMA node.

Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
docs/formatdomain.html.in