From: Koen Kooi <koen.kooi@oss.qualcomm.com>
To: JPEWhacker@gmail.com
Cc: eichest@gmail.com, openembedded-core@lists.openembedded.org,
Stefan Eichenberger <stefan.eichenberger@toradex.com>
Subject: Re: [OE-core] [PATCH] systemd: Allow overriding PNI name policies
Date: Tue, 15 Apr 2025 08:56:26 +0200 [thread overview]
Message-ID: <A1955164-126F-4B34-B796-B2CD3E530AF7@oss.qualcomm.com> (raw)
In-Reply-To: <CAJdd5Ga7-Swcb8Zej8dpuLP5AzuBHiTmFVSatL33Biwjhym+rA@mail.gmail.com>
> Op 14 apr 2025, om 19:03 heeft Joshua Watt via lists.openembedded.org <JPEWhacker=gmail.com@lists.openembedded.org> het volgende geschreven:
>
> On Wed, Apr 9, 2025 at 4:15 AM Stefan Eichenberger via
> lists.openembedded.org <eichest=gmail.com@lists.openembedded.org>
> wrote:
>>
>> From: Stefan Eichenberger <stefan.eichenberger@toradex.com>
>>
>> Add two new variables, PNI_NAME_POLICY and PNI_ALTNAMES_POLICY, to allow
>> overriding the default Predictable Network Interface (PNI) name and
>> alternative name policies respectively.
>>
>> This allows setting different name policies via a .bbappend file if the
>> defaults are not suitable for a specific use case.
>
> It's very difficult to configured systemd in a way that is acceptable
> to everyone, and I'm not sure adding variable "knobs" for every
> possible configuration option is feasible for systemd.
For me, the issue is that OE-core adds the MAC naming feature that upstream has disabled by default, which gives me unpredictable names for various on-SoC network interfaces[1]. Having a way, supported by OE-core, to disable the addition that results in predictable names not being predictable would be great. It would make PNI function as intended :)
> You could do the exact same thing by putting your sed command in a
> do_install:append() function (since you want to set them from a
> bbappend anyway), and in that case it's irrelevant if pni-names is set
> in PACKAGECONFIG or not.
I know and I hate it:
https://github.com/qualcomm-linux/meta-qcom-distro/commit/1820fefb6c67506726277f7f789fd5a4538ea33f
It's fragile and noone on the oe-core side will realize that changing do_install() might break DISTROs that use such bbappend. A proper variable like Stefan proposes would be more obvious and less fragile.
Or we could just drop the bit that adds 'mac' and have DISTROs that want that write a bbappend!
regards,
Koen
[1] The interfaces lacking a MAC in hardware is a separate, but equally frustrating issue.
>
>>
>> Signed-off-by: Stefan Eichenberger <stefan.eichenberger@toradex.com>
>> ---
>> meta/recipes-core/systemd/systemd_257.4.bb | 12 ++++++------
>> 1 file changed, 6 insertions(+), 6 deletions(-)
>>
>> diff --git a/meta/recipes-core/systemd/systemd_257.4.bb b/meta/recipes-core/systemd/systemd_257.4.bb
>> index 24aad11c0a..00edd702fb 100644
>> --- a/meta/recipes-core/systemd/systemd_257.4.bb
>> +++ b/meta/recipes-core/systemd/systemd_257.4.bb
>> @@ -273,6 +273,10 @@ WATCHDOG_TIMEOUT ??= "60"
>> # and the watchdog is enabled. (defaults is no hardware watchdog use)
>> WATCHDOG_RUNTIME_SEC ??= ""
>>
>> +# Predictable network interface name policies
>> +PNI_NAME_POLICY ??= "keep kernel database onboard slot path mac"
>> +PNI_ALTERNATIVE_NAMES_POLICY ??= "database onboard slot path mac"
>> +
>> do_install() {
>> meson_do_install
>>
>> @@ -398,12 +402,8 @@ do_install() {
>> fi
>>
>> if ${@bb.utils.contains('PACKAGECONFIG', 'pni-names', 'true', 'false', d)}; then
>> - if ! grep -q '^NamePolicy=.*mac' ${D}${nonarch_libdir}/systemd/network/99-default.link; then
>> - sed -i '/^NamePolicy=/s/$/ mac/' ${D}${nonarch_libdir}/systemd/network/99-default.link
>> - fi
>> - if ! grep -q 'AlternativeNamesPolicy=.*mac' ${D}${nonarch_libdir}/systemd/network/99-default.link; then
>> - sed -i '/AlternativeNamesPolicy=/s/$/ mac/' ${D}${nonarch_libdir}/systemd/network/99-default.link
>> - fi
>> + sed -i 's/^NamePolicy=.*/NamePolicy=${PNI_NAME_POLICY}/' ${D}${nonarch_libdir}/systemd/network/99-default.link
>> + sed -i 's/^AlternativeNamesPolicy=.*/AlternativeNamesPolicy=${PNI_ALTERNATIVE_NAMES_POLICY}/' ${D}${nonarch_libdir}/systemd/network/99-default.link
>> else
>> # Actively disable Predictable Network Interface Names
>> sed -i 's/^NamePolicy=.*/NamePolicy=/;s/^AlternativeNamesPolicy=.*/AlternativeNamesPolicy=/' ${D}${nonarch_libdir}/systemd/network/99-default.link
>> --
>> 2.45.2
>>
>>
>>
>>
>
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#214806): https://lists.openembedded.org/g/openembedded-core/message/214806
> Mute This Topic: https://lists.openembedded.org/mt/112169516/9418801
> Group Owner: openembedded-core+owner@lists.openembedded.org
> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [koen.kooi@oss.qualcomm.com]
> -=-=-=-=-=-=-=-=-=-=-=-
next prev parent reply other threads:[~2025-04-15 6:56 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-04-09 10:15 [PATCH] systemd: Allow overriding PNI name policies Stefan Eichenberger
2025-04-10 7:34 ` [OE-core] " Koen Kooi
2025-04-14 17:03 ` Joshua Watt
2025-04-15 6:56 ` Koen Kooi [this message]
2025-04-15 7:21 ` Stefan Eichenberger
2025-05-29 20:03 ` Ross Burton
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=A1955164-126F-4B34-B796-B2CD3E530AF7@oss.qualcomm.com \
--to=koen.kooi@oss.qualcomm.com \
--cc=JPEWhacker@gmail.com \
--cc=eichest@gmail.com \
--cc=openembedded-core@lists.openembedded.org \
--cc=stefan.eichenberger@toradex.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