From: Niklas Cassel via buildroot <buildroot@buildroot.org>
To: Arnout Vandecappelle <arnout@mind.be>
Cc: Niklas Cassel <cassel@kernel.org>,
Damien Le Moal <dlemoal@kernel.org>,
Kilian Zinnecker <kilian.zinnecker@mail.de>,
Thomas Petazzoni <thomas.petazzoni@bootlin.com>,
Niklas Cassel via buildroot <buildroot@buildroot.org>
Subject: Re: [Buildroot] [PATCH 3/5] configs/rock5b: enable eudev to enable automatic module loading
Date: Mon, 9 Sep 2024 11:01:07 +0000 [thread overview]
Message-ID: <Zt7VcrAFvluzqug0@ryzen.lan> (raw)
In-Reply-To: <317c4a83-974e-4a46-9711-f9f82f0fb1b3@mind.be>
On Tue, Sep 03, 2024 at 11:06:13PM +0200, Arnout Vandecappelle wrote:
>
>
> On 03/09/2024 21:56, Niklas Cassel wrote:
> > On Tue, Sep 03, 2024 at 09:26:11PM +0200, Thomas Petazzoni wrote:
> > > On Mon, 2 Sep 2024 12:44:31 +0200
> > > Niklas Cassel via buildroot <buildroot@buildroot.org> wrote:
> > >
> > > > From: Niklas Cassel <cassel@kernel.org>
> > > >
> > > > The arm64 defconfig enables both the R8169 Ethernet driver and the
> > > > PHY_ROCKCHIP_NANENG_COMBO_PHY PHY driver as modules.
> > > >
> > > > Enable udev so that these modules will get automatically loaded by udev
> > > > after the rootfs is mounted from the SD card.
> > > >
> > > > This way, we do not need to build these modules as built-in, and we also
> > > > get the benefit that any device plugged in to the PCIe slot will get
> > > > automatically loaded.
> > > >
> > > > Signed-off-by: Niklas Cassel <cassel@kernel.org>
> > >
> > > I'm not sure what we want to do here, especially when this ends up
> > > needing net.ifnames=0 that you add in PATCH 4/5.
> > >
> > > We do have already a small number of defconfigs that enable eudev. So
> > > far we only use net.ifnames=0 in 4 configurations.
> > >
> > > Basically, I'm trying to make sure we remain consistent and have a
> > > policy that applies to all defconfigs rather than ad-hoc solutions that
> > > are different for each and every defconfig. That's why I'm a bit
> > > reluctant with applying your patches as-is because they only solve the
> > > specific case of this defconfig, in a specific way, without creating a
> > > clear rule that all defconfigs meeting a specific situation should
> > > follow.
> >
> > (Replying with the email that I have subscribed to the list this time.)
> >
> >
> > An alternative to specify net.ifnames=0 on the kernel command line
> > (to disable udev predictable NIC names), is to create an empty file:
> > /etc/udev/rules.d/80-net-name-slot.rules
>
> Well, I think Thomas' comment was primarily about whether we should enable
> eudev and dynamically load modules rather than using builtins.
For boards that support USB, PCIe and HDMI, I think it is a good default
to enable udev (or mdev), as you could see from my other email, enabling
udev will load 45 different kernel modules. If we don't enable udev,
lsmod will show 0 loaded kernel modules.
I don't think we can enable everything as built-in, especially since a
user can plug in basically anything in the PCIe slot or USB slot.
(But I agree that keeping e.g. Ethernet as built-in is good for NFS root.)
E.g. I can plug in a USB to Ethernet adapter, and with udev enabled, the
correct kernel module gets loaded automatically.
>
> Passing net.ifnames=0 is probably the easiest way to disable predictable
> NIC names. Only, it's surprising that we have something like 25 defconfigs
> that enable eudev, but only 4 of them enable that option... We should
> preferably have one "normal" way that this is done.
>
>
> What is BTW not clear to me is _why_ you don't want predictable names.
> What's wrong with BR2_SYSTEM_DHCP="enp0s31f6"?
Well, if you do a:
$ git grep BR2_SYSTEM_DHCP
you see that everyone except:
configs/ls1028ardb_defconfig:BR2_SYSTEM_DHCP="eno0"
is using eth0.
But sure, it would be possible to specify the predictable ifname.
I guess that it is just a bit unintuitive that you would need to
change BR2_SYSTEM_DHCP="eth0" to something else if you happen to enable
eudev. (While if you have mdev or neither mdev or udev, eth0 works fine.)
Kind regards,
Niklas
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot
next prev parent reply other threads:[~2024-09-09 11:01 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-09-02 10:44 [Buildroot] [PATCH 0/5] rock5b additional improvements Niklas Cassel via buildroot
2024-09-02 10:44 ` [Buildroot] [PATCH 1/5] board/radxa/rock5b: use gpt partition table instead of hybrid Niklas Cassel via buildroot
2024-09-02 10:44 ` [Buildroot] [PATCH 2/5] board/radxa/rock5b: use PARTLABEL to specify rootfs Niklas Cassel via buildroot
2024-09-02 10:44 ` [Buildroot] [PATCH 3/5] configs/rock5b: enable eudev to enable automatic module loading Niklas Cassel via buildroot
2024-09-03 19:26 ` Thomas Petazzoni via buildroot
2024-09-03 19:50 ` Arnout Vandecappelle via buildroot
2024-09-03 20:06 ` Niklas Cassel via buildroot
2024-09-03 21:10 ` Arnout Vandecappelle via buildroot
2024-09-09 11:11 ` Niklas Cassel via buildroot
2024-09-03 19:56 ` Niklas Cassel via buildroot
2024-09-03 21:06 ` Arnout Vandecappelle via buildroot
2024-09-09 11:01 ` Niklas Cassel via buildroot [this message]
2024-09-02 10:44 ` [Buildroot] [PATCH 4/5] board/radxa/rock5b: do not use udev predictable NIC names Niklas Cassel via buildroot
2024-09-02 10:44 ` [Buildroot] [PATCH 5/5] configs/rock5b: enable DHCP for eth0 Niklas Cassel via buildroot
2024-09-03 19:24 ` [Buildroot] [PATCH 0/5] rock5b additional improvements Thomas Petazzoni via buildroot
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=Zt7VcrAFvluzqug0@ryzen.lan \
--to=buildroot@buildroot.org \
--cc=Niklas.Cassel@wdc.com \
--cc=arnout@mind.be \
--cc=cassel@kernel.org \
--cc=dlemoal@kernel.org \
--cc=kilian.zinnecker@mail.de \
--cc=thomas.petazzoni@bootlin.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