From: "Yann E. MORIN" <yann.morin.1998@free.fr>
To: Carlos Alberto Lopez Perez <clopez@igalia.com>
Cc: Martin Bark <martin@barkynet.com>,
"buildroot@buildroot.org" <buildroot@buildroot.org>,
Vincent Fazio <vfazio@xes-inc.com>,
Julien Grossholtz <julien.grossholtz@openest.io>,
Marcus Hoffmann <marcus.hoffmann@othermo.de>
Subject: Re: [Buildroot] [External] - Re: [PATCH 1/1] configs/raspberrypi*: stop setting powersave as the default CPU governor
Date: Tue, 24 Oct 2023 21:07:47 +0200 [thread overview]
Message-ID: <20231024190747.GB2607@scaer> (raw)
In-Reply-To: <8c44f1f6-7ef7-487b-db22-b396497e2a68@igalia.com>
Carlos, All,
Thanks for all the back-n-forth on the subject, I think interested
parties expresed their view, and here's my last take on the topic, see
below...
On 2023-10-24 17:28 +0200, Carlos Alberto Lopez Perez spake thusly:
> On 24/10/2023 17:17, Vincent Fazio wrote:
> >> That is not true. Upstream RPi-OS defaults to the ondemand governor. The
> >> kernel boots with powersave but there is an init scripy that swichts it to
> >> ondemand as soon as the board boots unless the user press the shift key.
> > Sorry, I wasn't very clear here. When I said upstream, I meant
> > "upstream RPi kernel", which default to powersave.
Exactly: the people responsible for maintaing the kernel for the
raspberry pi are the upstream for that kernel. They have decided upon a
default behaviour they want the kernel to have, so this is what we're
following too: it's rpi-upstream's decision to default to powersave.
> > RPi-OS does
> > have a mechanism to switch governors post boot.
RPi-OS is a debian-based distribution using the rpi kernel; they are a
downstream of the rpi kernel, like we are. RPi-OS have done their
integration to switch to using a different behaviour, using a startup
script (or a systemd unit, that's equivalent); they haven't decided to
change the kernel configuration (AFAICS from this thread).
And Buildroot is not RPi-OS.
Besides, some people have started to rely on the fact that a build with
Buildroot will use the powersave governor: Vincent explained a use-case
for which such a default makes sense; on my side I also use (have used)
rpis with powersave because of thermal stability.
> > Again, I think the biggest issue is just informing users and
> > letting them choose the right solution for their deployment.
> I'm all in for informing users and letting them to choose a default.
>
> On the other hand a default selection has to be made for those users
> that don't care or don't have time to read the documentation.
>
> And I repeat: RPi-OS defaults to ondemand governor because they change
> the governor on boot with an init script.
Exactly: they have an init script; they do not change the kernel
configuration.
> The governor powersave is only used for the first seconds of the boot
> process and is opt-in the rest of the time (the user has to press shit
> key meanwhile booting to keep it).
>
> So the current status quo is this one:
>
> 1) RPi default CPU governor of mainline Linux kernel: schedutil or
> ondemand (depends on the kernel version)
> 2) RPi default CPU governor of RPi-OS: ondemand (powersave opt-in if
> user press shit)
> 3) RPi default CPU governor of buildroot: powersave
>
> Why buildroot would want to have a default of powersave that not even
> matches what RPi-OS does?
Because we are *not* RPi-OS. We are not matching against RPi-OS; we are
using rpi-kernel upstream decision.
In another email [0], you wrote;
> The reality nowdays is that all buildroot RPi users are actually running
> with powersave governor by default and I bet 90% of them don't know it.
If they did not notice, then doesn't that mean that the rpi is behaving
good enough for them, and that they in fact do not need the extra speed?
And again, here's a repeat from a previous comment of mine:
The defconfigs are basic configurations that just demonstrate the basic
behaviour of a board, with minimal to no tweaking to deviate from
upstreams. Users are suposed to take those defconfigs, and adapt to
their needs. If they want to change the governor, then they can provide
a startup script that enables the required governor.
[0] https://lore.kernel.org/buildroot/4528d9f7-78ee-1576-37d7-236476d3874f@igalia.com/
Regards,
Yann E. MORIN.
--
.-----------------.--------------------.------------------.--------------------.
| Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ |
| +33 561 099 427 `------------.-------: X AGAINST | \e/ There is no |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. |
'------------------------------^-------^------------------^--------------------'
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot
next prev parent reply other threads:[~2023-10-24 19:08 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-10-20 14:20 [Buildroot] [PATCH 1/1] configs/raspberrypi*: stop setting powersave as the default CPU governor Carlos Alberto Lopez Perez
2023-10-21 21:38 ` Yann E. MORIN
2023-10-23 12:40 ` Marcus Hoffmann via buildroot
2023-10-23 16:25 ` Yann E. MORIN
2023-10-23 20:23 ` Carlos Alberto Lopez Perez
2023-10-24 14:01 ` [Buildroot] [External] - " Vincent Fazio
2023-10-24 14:28 ` Carlos Alberto Lopez Perez
2023-10-24 14:52 ` Vincent Fazio
2023-10-24 15:00 ` Carlos Alberto Lopez Perez
2023-10-24 15:17 ` Vincent Fazio
2023-10-24 15:28 ` Carlos Alberto Lopez Perez
2023-10-24 19:07 ` Yann E. MORIN [this message]
2023-10-24 20:53 ` Carlos Alberto Lopez Perez
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=20231024190747.GB2607@scaer \
--to=yann.morin.1998@free.fr \
--cc=buildroot@buildroot.org \
--cc=clopez@igalia.com \
--cc=julien.grossholtz@openest.io \
--cc=marcus.hoffmann@othermo.de \
--cc=martin@barkynet.com \
--cc=vfazio@xes-inc.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