From: Petr Pavlu <petr.pavlu@suse.com>
To: Prarit Bhargava <prarit@redhat.com>, Petr Mladek <pmladek@suse.com>
Cc: David Hildenbrand <david@redhat.com>,
Luis Chamberlain <mcgrof@kernel.org>,
linux-modules@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 2/2] module: Merge same-name module load requests
Date: Wed, 7 Dec 2022 14:23:25 +0100 [thread overview]
Message-ID: <cb149b17-4bbf-71b0-424c-c4f3d1b96952@suse.com> (raw)
In-Reply-To: <2e8d862d-fa3d-0879-9097-215d17cb1fed@redhat.com>
On 12/6/22 13:31, Prarit Bhargava wrote:
>
>>
>>> Could you please boot the machine with 'udev.log_level=debug' and provide me
>>> logs ('journalctl -b -o short-monotonic') from a run with the vanilla kernel
>>> and with the discussed patch?
>>
>
> Petr, I tried to in-line the logs however the email bounced due to its size.
>
> I know this isn't a preferred method of passing information on LKML and
> other lists, but here are links to the logs:
>
> https://people.redhat.com/prarit/4petr/
>
> Both outputs were done with, as requested, 'journalctl -b -o
> short-monotonic'.
>
> vanilla.log is kernel booted with 'udev.log_level=debug'
> with-changeset.log is kernel + patch booted with 'udev.log_level=debug'
Thanks Prarit for re-testing the patch. Both logs in this case actually show
similar startup times.
Vanilla:
[ 68.108176] intel-eaglestream-spr-15.khw3.lab.eng.bos.redhat.com systemd[1]: Startup finished in 55.874s (firmware) + 25.646s (loader) + 35.793s (kernel) + 7.845s (initrd) + 24.469s (userspace) = 2min 29.629s.
With the patch:
[ 68.064826] intel-eaglestream-spr-15.khw3.lab.eng.bos.redhat.com systemd[1]: Startup finished in 54.153s (firmware) + 19.947s (loader) + 35.965s (kernel) + 9.449s (initrd) + 22.650s (userspace) = 2min 22.165s.
The system has 192 CPUs. The vanilla case shows 144x inserts of acpi_cpufreq
and 43x of pcc_cpufreq:
acpi_cpufreq (the first and last recorded insert):
[ 47.485621] intel-eaglestream-spr-15.khw3.lab.eng.bos.redhat.com systemd-udevd[1871]: Inserted module 'acpi_cpufreq'
[...]
[ 53.052401] intel-eaglestream-spr-15.khw3.lab.eng.bos.redhat.com systemd-udevd[1914]: Inserted module 'acpi_cpufreq'
pcc_cpufreq:
[ 47.515221] intel-eaglestream-spr-15.khw3.lab.eng.bos.redhat.com systemd-udevd[1871]: Inserted module 'pcc_cpufreq'
[...]
[ 53.067917] intel-eaglestream-spr-15.khw3.lab.eng.bos.redhat.com systemd-udevd[2040]: Inserted module 'pcc_cpufreq'
Processing inserts of all CPU frequency modules took at least ~5.5 seconds.
It was likely more because not all inserts are recorded in the log, udevd
messages appear to be missing from ~53.1.
With the patch, both modules are attempted to be inserted 192x:
acpi_cpufreq:
[ 50.107403] intel-eaglestream-spr-15.khw3.lab.eng.bos.redhat.com systemd-udevd[1817]: Failed to insert module 'acpi_cpufreq': Device or resource busy
[...]
[ 50.438755] intel-eaglestream-spr-15.khw3.lab.eng.bos.redhat.com systemd-udevd[2016]: Inserted module 'acpi_cpufreq'
pcc_cpufreq:
[ 50.110731] intel-eaglestream-spr-15.khw3.lab.eng.bos.redhat.com systemd-udevd[1849]: Failed to insert module 'pcc_cpufreq': Device or resource busy
[...]
[ 50.579249] intel-eaglestream-spr-15.khw3.lab.eng.bos.redhat.com systemd-udevd[2016]: Inserted module 'pcc_cpufreq'
This shows that the patch reduced the sequence to ~0.5 seconds and its logic
looks to be working as intended.
Thanks,
Petr
next prev parent reply other threads:[~2022-12-07 13:23 UTC|newest]
Thread overview: 43+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-09-19 12:32 [PATCH v2 0/2] module: Merge same-name module load requests Petr Pavlu
2022-09-19 12:32 ` [PATCH v2 1/2] module: Correct wake up of module_wq Petr Pavlu
2022-09-30 20:22 ` Luis Chamberlain
2022-10-14 8:40 ` Petr Mladek
2022-09-19 12:32 ` [PATCH v2 2/2] module: Merge same-name module load requests Petr Pavlu
2022-09-30 20:30 ` Luis Chamberlain
2022-10-15 9:27 ` Petr Pavlu
2022-10-18 18:33 ` Luis Chamberlain
2022-10-18 19:19 ` Prarit Bhargava
2022-10-18 19:53 ` Prarit Bhargava
2022-10-20 7:19 ` Petr Mladek
2022-10-24 13:22 ` Prarit Bhargava
2022-10-24 17:08 ` Luis Chamberlain
2022-10-24 12:37 ` Petr Pavlu
2022-10-24 14:00 ` Prarit Bhargava
2022-11-13 16:44 ` Petr Pavlu
2022-10-19 12:00 ` Petr Pavlu
2022-10-20 7:03 ` Petr Mladek
2022-10-24 17:53 ` Luis Chamberlain
2022-11-12 1:47 ` Luis Chamberlain
2022-11-14 8:57 ` David Hildenbrand
2022-11-14 15:38 ` Luis Chamberlain
2022-11-14 15:45 ` David Hildenbrand
2022-11-15 19:29 ` Luis Chamberlain
2022-11-16 16:03 ` Prarit Bhargava
2022-11-21 16:00 ` Petr Pavlu
2022-11-21 19:03 ` Luis Chamberlain
2022-11-21 19:50 ` David Hildenbrand
2022-11-21 20:27 ` Luis Chamberlain
2022-11-22 13:59 ` Petr Pavlu
2022-11-22 17:58 ` Luis Chamberlain
2022-11-16 16:04 ` David Hildenbrand
2022-11-18 17:32 ` David Hildenbrand
2022-11-28 16:29 ` Prarit Bhargava
2022-11-29 13:13 ` Petr Pavlu
2022-12-02 16:36 ` Petr Mladek
2022-12-06 12:31 ` Prarit Bhargava
2022-12-07 13:23 ` Petr Pavlu [this message]
2022-12-04 19:58 ` Prarit Bhargava
2022-10-14 7:54 ` David Hildenbrand
2022-10-15 9:49 ` Petr Pavlu
2022-10-14 13:52 ` Petr Mladek
2022-10-16 12:25 ` Petr Pavlu
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=cb149b17-4bbf-71b0-424c-c4f3d1b96952@suse.com \
--to=petr.pavlu@suse.com \
--cc=david@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-modules@vger.kernel.org \
--cc=mcgrof@kernel.org \
--cc=pmladek@suse.com \
--cc=prarit@redhat.com \
/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).