Re: kern/58124: lagg(4) fails to work with lacp protocol on mcx(4)

[Andrius V <vezhlys%gmail.com@localhost>]

Hi,

Comparing OpenBSD and NetBSD driver, there's one place where
potentially OpenBSD sets full-duplex in comparison to NetBSD:
https://nxr.netbsd.org/xref/src/sys/dev/pci/if_mcx.c#8176
https://github.com/openbsd/src/blob/a938a57b1f753b55309620f7e0ecb50db35645b7/sys/dev/pci/if_mcx.c#L8114

However we don't have this particular "link status". So my guess that
we need to set it in mcx_media_status(), potentially based on media
type, and it is not needed in ifmedia_add() (need to test). Is there a
need for more complex logic, I am not sure.


On Wed, Apr 10, 2024 at 12:06 PM s ymgch <s.ymgch228%gmail.com@localhost> wrote:
>
> Hi,
>
> I suppose you maintain mcx(4). and this PR is related to the driver.
> Could you take a look at the PR?
>
> Best Regards,
> Yamaguchi
>
> On Tue, Apr 9, 2024 at 10:37 AM s ymgch <s.ymgch228%gmail.com@localhost> wrote:
> >
> > Hi,
> >
> > On Tue, Apr 9, 2024 at 4:20 AM Andrius V <vezhlys%gmail.com@localhost> wrote:
> > >
> > >  Hi,
> > >
> > >  You are completely right! Setting the patch helps and lacp protocol
> > >  starts to work. mcx0/mcx1 shows "media: Ethernet autoselect
> > >  (10GBASE-CR1 full-duplex)" now, switch also shows mac address at the
> > >  "LACP Partner System ID". Thank you for identifying the issue. Can the
> > >  patch be applied directly, or more work needs to be done to identify
> > >  when to set full-duplex mode?
> >
> > I don't know whether the patch is correct toward mcx(4) or not, because
> > I'm not familiar with mcx(4). And I don't have any mcx(4) devices.
> > So, I'd like someone to review and commit the patch.
> >
> > -- Yamaguchi