**** BEGIN LOGGING AT Thu Sep 06 02:59:58 2012 Sep 06 11:13:09 so, it's me again, running trunk in kvm Sep 06 11:14:01 dhcp client (udhcpc) can't see dhcp replies Sep 06 11:14:46 i can see it in tcpdump on eth0, but udhcpc just missing it Sep 06 11:15:02 running kvm with option -netdev tap,id=eth0,vhost=on,... -device virtio-net-pci,netdev=eth0,event_idx=off,... Sep 06 11:15:25 this does not happen when kvm is started with -net nic,... -net tap,... Sep 06 11:17:46 if i assign address manually - everything works Sep 06 11:38:25 hm... and i can't login! Sep 06 15:05:32 nbd * r33325 /trunk/package/netifd/Makefile: netifd: update to latest version, fixes premature device release on link loss (fixes #12114, #12112, #11959) Sep 06 15:45:44 guys, i have a problem with udhcpc in kvm. anyone alive? Sep 06 16:07:21 how can i set default HOST_CFLAGS other than "make HOST_CFLAGS=..." ? is there an entry in menuconfig? Sep 06 16:46:44 lol Sep 06 16:48:47 mikrotik is awesome, you suck ;-) Sep 06 16:49:59 "generally causing problems and not working" Sep 06 16:58:24 raorn: read the kvm documentation. its prolly not a openwrt issue but l2 networking stuff Sep 06 17:02:07 https://www.redhat.com/archives/virt-tools-list/2011-October/msg00088.html Sep 06 17:18:36 roh: i don't use kernel bridge on HN, we are running openwrt Sep 06 17:19:03 roh: and i can see dhcp replies inside VM in tcpdump -npi eth0 Sep 06 17:19:20 *openvswitch Sep 06 17:19:35 well.. that you need to debug with these guys then. Sep 06 17:19:46 why? Sep 06 17:19:55 packet is delivered into VM Sep 06 17:20:08 because i do not see why that should be in any way a openwrt specific issue Sep 06 17:20:51 why does everything works if i assign address manually? Sep 06 17:21:12 because thats 'just' l3 and dhcp is evil l2 stuff Sep 06 17:21:33 maybe your mac address on the vm is the same as on the host? Sep 06 17:21:44 or similar evil stuff Sep 06 17:23:30 ipv6 autoconfiguration works fine Sep 06 17:24:26 and no, no evil stuff is happening. VM have it's own unique MAC Sep 06 17:25:50 well.. check your spanning tree settings. i guess its just some timing issue for the virtual switchport. Sep 06 17:25:55 russell--: heh Sep 06 17:26:42 find the " spanning-tree portfast Sep 06 17:26:54 or how its called on your v-eth switch Sep 06 17:27:37 stp disabled Sep 06 17:28:01 the only thing changed is kvm commandline Sep 06 18:15:29 cshore * r33326 /trunk/target/linux/ar71xx/ (8 files in 7 dirs): [ar71xx] dir825b1: Add image for DIR-825-B1 that uses the whole flash (must first be flashed with the 'standard' image) and an image to revert the change to allow reversion to stock firmware. Sep 06 18:59:25 if there is an option in target/linux/x86/config-3.3, i guess it will be applied to x86/kvm_guest target too, if target/linux/x86/kvm_guest/config-3.3 does not have this option (neither set nor unset) Sep 06 18:59:31 am i right? Sep 06 18:59:46 yep Sep 06 19:22:16 so, i've found the problem Sep 06 19:22:55 dhcp replies are rejected by udhcpc because of checksum Sep 06 19:23:15 http://article.gmane.org/gmane.linux.kernel/1003853 Sep 06 19:23:30 complete thread - http://thread.gmane.org/gmane.comp.emulators.kvm.devel/55281 Sep 06 19:25:38 ah Sep 06 19:25:51 reminds me on virtio net issues I had with windows on kvm Sep 06 19:26:59 hm... no CHECKSUM support in iptables... Sep 06 19:28:55 busybox doesn't seem to react on setting CONFIG_UDHCP_DEBUG=9 Sep 06 19:34:36 rx-checksumming is hard-off for virtio-net driver... Sep 06 19:52:44 what is better - fix busybox or package CHECKSUM iptables/kmod target? Sep 06 19:52:55 probably busybox Sep 06 19:58:29 * raorn is looking at "the interface to return the relevant information" Sep 06 19:59:25 do you know what's common for me and busybox? Sep 06 19:59:39 i have no idea how to handle CHECKSUM_PARTIAL too... Sep 06 21:15:06 yessss! Sep 06 21:15:17 i fukken did it! Sep 06 21:15:21 me dances Sep 06 21:17:16 i have a patch, now i need to compose good commit message... Sep 06 21:31:49 phew! patch sent to openwrt-devel@ Sep 06 22:22:35 cool work raorn Sep 07 02:38:43 * russell-- wonders if someone will apply this, pretty straightforward: http://patchwork.openwrt.org/patch/2618/ **** ENDING LOGGING AT Fri Sep 07 02:59:59 2012