[+Cc Mario Limonciello, the original author on MAC pass-through] On Tue, Jan 11, 2022 at 12:51 AM Jakub Kicinski <kuba@xxxxxxxxxx> wrote: > > On Mon, 10 Jan 2022 11:32:16 +0800 Kai-Heng Feng wrote: > > > > I don't think it's a good idea. On my laptop, > > > > systemd-udev-settle.service can add extra 5~10 seconds boot time > > > > delay. > > > > Furthermore, the external NIC in question is in a USB/Thunderbolt > > > > dock, it can present pre-boot, or it can be hotplugged at any time. > > > > > > IIUC our guess is that this feature used for NAC and IEEE 802.1X. > > > In that case someone is already provisioning certificates to all > > > the machines, and must provide a config for all its interfaces. > > > It should be pretty simple to also put the right MAC address override > > > in the NetworkManager/systemd-networkd/whatever config, no? > > > > If that's really the case, why do major OEMs came up with MAC > > pass-through? Stupid may it be, I don't think it's a solution looking > > for problem. > > I don't know. Maybe due to a limitation in Windows? Maybe it's hard to > do in network manager, too, and we're not seeing something. Or perhaps > simply because they want to convince corporations to buy their > unreasonably expensive docks. > > What I do know is that we need to gain a good understanding of the > motivation before we push any more of such magic into the kernel. Mario, do you know how corporate network and other OS handle MAC pass-through, so we can come up with a more robust design? Kai-Heng > > I may be able to do some testing myself after the Omicron surge is over > in the US.
