[Regression] Intel em0 WoL fix from FreeBSD stable/15

[Phil]

Hi,

Following up on previous reports regarding the Intel em0 (82567LM-3) Wake-on-LAN breaking after a clean software shutdown on FreeBSD 14+ (previously tracked in closed GitHub issue #278: https://github.com/opnsense/src/issues/278), an upstream fix has just been merged in FreeBSD:

    FreeBSD Bug: 295443

    Commit: 46cf612d98e9 (e1000: Rework Wake-on-LAN policy and programming by Kevin Bowling, Aug 31, 2026)

    URL: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=295443

Since this commit fixes the PCH Sx workarounds and keeps the link powered while host wake is armed, is this patch scheduled to be included in an upcoming OPNsense release, and if so, when can we expect it?

Thank you for your amazing work on OPNsense!

Best regards

[franco]

It looks urgent, but not urgent enough to have actually met the 2 week time frame to be added to stable/15.  I'm watching the branch closely in the hope that it lands there before 26.7.5 needs to be pushed out.


Cheers,
Franco

[Phil]

Hi Franco,

Just a quick update on this: I noticed FreeBSD Bug #295443 is now CLOSED / FIXED upstream, and the patch has landed in stable/15.

I tested on 26.7.7 (15.1-RELEASE-p4) and confirmed the regression is still present after a clean shutdown (the card completely powers off despite WOL_MAGIC being present in ifconfig).

Do you know if a sync with stable/15 is planned for a future 26.7.x release to pull this e1000 Sx power fix?

Thanks a lot for your help!

Best regards

[franco]

Hi Phil,

There's a high influx of sizeable commits to stable/15 regarding Intel drivers at the moment (100+ commits) which makes it very difficult for us to navigate as people already report regressions and we need to slow down to sort this out.  I'm doubting a bit that FreeBSD really has a hand on this so we will perhaps see all these regressions in 15.2 either way if we can't pin them down.

If the particular patch is on stable/15 I will see it during review, but I'm not sure which driver batch we're going to tackle for 26.7.8. igc is still pending in all of this and meanwhile e1000/igb receive more commits lately still. We want to separate the driver updates in batches to make it a bit easier to do triage/bisect afterwards.


Cheers,
Franco