From: Igor Mammedov <imammedo@redhat.com>
To: Denis Plotnikov <den-plotnikov@yandex-team.ru>
Cc: pbonzini@redhat.com, qemu-devel@nongnu.org, yc-core@yandex-team.ru
Subject: Re: [PATCH v1] softmmu/vl: make default prealloc-threads work w/o -mem-prealloc
Date: Mon, 17 May 2021 16:33:45 +0200 [thread overview]
Message-ID: <20210517163345.61ce8b3b@redhat.com> (raw)
In-Reply-To: <20210311085918.284903-1-den-plotnikov@yandex-team.ru>
On Thu, 11 Mar 2021 11:59:18 +0300
Denis Plotnikov <den-plotnikov@yandex-team.ru> wrote:
> Preallocation in memory backends can be specified with either global
> QEMU option "-mem-prealloc", or with per-backend property
> "prealloc=true". In the latter case, however, the default for the
> number of preallocation threads is not set to the number of vcpus, but
> remains at 1 instead.
>
> Fix it by setting the "prealloc-threads" sugar property of
> "memory-backend" to the number of vcpus unconditionally.
well, number of prealloc-threads are policy that should be set by user.
it was kept equal to vcpus for compat reasons with "-mem-prealloc",
for if one goes with supplying memory backends explicitly, one is supposed
to set prealloc-threads explicitly as well if default(==1) doesn't
suit user. It could be equal to #vcpus or be a different numbers
depending on host CPU configuration. That was intent when
commit ffac16fab3 was introduced.
Imagine one has a VM with a lot of VCPUs but not so much of of RAM,
setting prealloc-threads to #VCPUs will be wasting CPU resources
at least.
> Fixes: ffac16fab3 ("hostmem: introduce "prealloc-threads" property")
>
> Signed-off-by: Denis Plotnikov <den-plotnikov@yandex-team.ru>
> ---
> softmmu/vl.c | 17 ++++++++++-------
> 1 file changed, 10 insertions(+), 7 deletions(-)
>
> diff --git a/softmmu/vl.c b/softmmu/vl.c
> index ff488ea3e7db..e392e226a2d3 100644
> --- a/softmmu/vl.c
> +++ b/softmmu/vl.c
> @@ -2300,14 +2300,17 @@ static void qemu_validate_options(void)
>
> static void qemu_process_sugar_options(void)
> {
> - if (mem_prealloc) {
> - char *val;
> + char *val;
>
> - val = g_strdup_printf("%d",
> - (uint32_t) qemu_opt_get_number(qemu_find_opts_singleton("smp-opts"), "cpus", 1));
> - object_register_sugar_prop("memory-backend", "prealloc-threads", val,
> - false);
> - g_free(val);
> + val = g_strdup_printf("%d",
> + (uint32_t) qemu_opt_get_number(
> + qemu_find_opts_singleton("smp-opts"), "cpus", 1));
> +
> + object_register_sugar_prop("memory-backend", "prealloc-threads", val,
> + false);
> + g_free(val);
> +
> + if (mem_prealloc) {
> object_register_sugar_prop("memory-backend", "prealloc", "on", false);
> }
>
prev parent reply other threads:[~2021-05-17 14:40 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-11 8:59 [PATCH v1] softmmu/vl: make default prealloc-threads work w/o -mem-prealloc Denis Plotnikov
2021-03-11 10:31 ` Paolo Bonzini
2021-05-05 15:31 ` Paolo Bonzini
2021-05-17 14:33 ` Igor Mammedov [this message]
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=20210517163345.61ce8b3b@redhat.com \
--to=imammedo@redhat.com \
--cc=den-plotnikov@yandex-team.ru \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=yc-core@yandex-team.ru \
/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).