On 1/8/19 3:00 PM, Ian Kumlien wrote: > On Tue, Jan 8, 2019 at 11:51 PM Stephen Hemminger > <step...@networkplumber.org> wrote: >> On Tue, 8 Jan 2019 23:10:04 +0100 >> Ian Kumlien <ian.kuml...@gmail.com> wrote: >>> On Sun, Jan 6, 2019 at 11:21 PM Ian Kumlien <ian.kuml...@gmail.com> wrote: >>>> >>>> [Sorry for the repost, screwed up the netdev address...] >>>> >>>> Hi, >>>> >>>> Switching from 4.19.x -> 4.20 resulted in DHCP not working for my VM:s. >>>> >>>> My firewall (which also runs the dhcpd) runs VM:s and it does this by >>>> having physical >>>> interfaces attached to bridges - which the VM:s in turn attach to. >>>> >>>> Since 4.20 the VM:s can't use DHCP, it's odd since the requests are >>>> seen - a response is sent but >>>> it never enters the interface attached to the bridge. >>>> >>>> Basically: >>>> VM vnet2: -> br0 -> eno2 -> switch -> eno1 (dhcpd) >>>> dhcpd eno1 -> siwtch and... gone. >>>> >>>> Any clues? >>>> >>>> All the nics are handled by ixgbe >>> >>> So, doing similar tests at work with other drivers works - could it be >>> related to the mac address filter that was added? >>> I don't *really* use VF:s though... (can't really find anything else atm) >>> >>> Will try to test, but the VM:s on this machine is in use. >> >> The default MAC address of the bridge device is the first device assigned >> to the bridge. Remember most VF interfaces will only allow single MAC >> address >> and no promiscious mode. > > Yeah, I'm not running any VF:s and it just seems like the responses > are dropped somewhere > > when looking at "git log v4.19...v4.20 > drivers/net/ethernet/intel/ixgbe/" nothing else really stands out... > The machine is also running NAT for my home network and all of that > works just fine... > > I started with tcpdump, prooving that packets reached all the way > outside but replies never made it, reboorting > with 4.19.13 resulted in replies appearing in the tcpdump. > > I don't quite know where to look - and what can i do to test - i tried > disabling all offloading (due to the UDP > offloading changes) but nothing helped... > > Ideas? Patches? ;)
Running a bisection would certainly help find the offending commit if that is something that you can do? -- Florian
