VLAN problem on some em(4) cards

[Ryan Stone]

tcpdump on em0 will show the vlan-tagged packets if everything is working
correctly.  The fact that they are not showing up is a sign that something
is not working correctly (or that your tcpdump filter isn't account for the
vlan tag)

On Sun, Apr 10, 2016 at 1:44 PM, Christoph Moench-Tegeder <
cmt at burggraben.net> wrote:

> Hi,
> while replacing a switch I ran into a problem with using 802.1q VLANs
> (vlan(4)) on a em(4) card - that is, I could not get the card to work
> with VLANs.
> Details:
> Software is FreeBSD 10.3-RELEASE i386 with a custom kernel (but has
> em(4) and vlan(4) compiled in (not as a module)).
> Hardware is a Soekris net6501-70, the interface in question is (pciconf)
> em2 at pci0:10:0:0:        class=0x020000 card=0x00008086 chip=0x10d38086
> rev=0x00
>     vendor     = 'Intel Corporation'
>     device     = '82574L Gigabit Network Connection'
>     class      = network
>     subclass   = ethernet
>
> The setup I wanted to create was a "router on a stick", that is having
> 3 (for starters) tagged VLANs on that one link between router and
> switch. I was not able to get any packet through to the switch,
> while tcpdump shows the packets on the vlan interface (but not on
> the underlying em interface - I understand that this could be the
> result of hardware vlan processing). I could not ping any host beyond
> the switch nor the switch's management interface itself, I did not
> even receive ARP replies.
> I tried both manual creation of the vlan devices and the rc.conf/
> netif mechanism - with the very same results ("doesn't work").
> Experimenting with the vlanhw* flags on the em interface did not
> result in any change.
>
> To make sure it's not the switch (or my own mistake), I took another
> host with an em interface - this time, it was a
> em0 at pci0:0:25:0:        class=0x020000 card=0x21ce17aa chip=0x15028086
> rev=0x04
>     vendor     = 'Intel Corporation'
>     device     = '82579LM Gigabit Network Connection'
>     class      = network
>     subclass   = ethernet
>
> (again 10.3-RELEASE, but amd64).
> With this interface (connected to the very same switch port) I
> had instantaneous success, so I guess the switch is ok and I'm
> not totally mistaken in my doings.
>
> Could anybody give a hint on how to further debug this or a pointer
> to further reading?
>
> Regards,
> Christoph
>
> --
> Spare Space
> _______________________________________________
> freebsd-stable at freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-stable
> To unsubscribe, send any mail to "freebsd-stable-unsubscribe at freebsd.org"
>