<html> <head> <base href="https://bugs.freedesktop.org/" /> </head> <body> <div> <a class="bz_bug_link bz_status_ASSIGNED " title="ASSIGNED --- - systemd-networkd: VLAN on bond fails due to incorrect order of enslavement" href="https://bugs.freedesktop.org/show_bug.cgi?id=76077#c5">Comment # 5</a> on <a class="bz_bug_link bz_status_ASSIGNED " title="ASSIGNED --- - systemd-networkd: VLAN on bond fails due to incorrect order of enslavement" href="https://bugs.freedesktop.org/show_bug.cgi?id=76077">bug 76077</a> from <a class="email" href="mailto:teg@jklm.no" title="Tom Gundersen <teg@jklm.no>"> Tom Gundersen</a> <pre>Finally got around to looking into this properly. It appears Malte's analysis in <a href="show_bug.cgi?id=76077#c4">comment #4</a> is spot on. The situation is that VLAN's cannot be created on devices that has the feature NETIF_F_VLAN_CHALLENGED (some feature...) set. Annoyingly that includes empty bonds (but I don't know why[0]), and I have so far not figured out an elegant way for us to get notified about when this flag changes. I think we should do one of two things (preferably both): 1) hook up the feature flag with rtnetlink so we get notified about changes 2) figure out if empty bonds can be made to work with vlans (then the order will no longer matter). We could attempt to work around this in networkd by trying to add the VLAN every time the number of slaves of a bond changes, but that feels too hacky (I have already applied such hacks only to revert them four times in the past few months), so I'd really prefer if we could fix this in the kernel. [0]: <<a href="http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/drivers/net/bonding/bond_main.c#n1005">http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/drivers/net/bonding/bond_main.c#n1005</a>></pre> </div> <hr> You are receiving this mail because: <ul> <li>You are the QA Contact for the bug.</li> </ul> </body> </html>